discourse-ai/spec/requests/admin/ai_personas_controller_spec.rb

983 lines
34 KiB
Ruby
Raw Normal View History

FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
# frozen_string_literal: true
RSpec.describe DiscourseAi::Admin::AiPersonasController do
fab!(:admin)
fab!(:ai_persona)
fab!(:embedding_definition)
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
fab!(:llm_model)
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
before do
sign_in(admin)
SiteSetting.ai_embeddings_selected_model = embedding_definition.id
SiteSetting.ai_embeddings_enabled = true
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
describe "GET #index" do
it "returns a success response" do
get "/admin/plugins/discourse-ai/ai-personas.json"
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
expect(response).to be_successful
expect(response.parsed_body["ai_personas"].length).to eq(AiPersona.count)
expect(response.parsed_body["meta"]["tools"].length).to eq(
DiscourseAi::Personas::Persona.all_available_tools.length,
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
)
end
it "sideloads llms" do
get "/admin/plugins/discourse-ai/ai-personas.json"
expect(response).to be_successful
expect(response.parsed_body["meta"]["llms"]).to eq(
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
[
{
id: llm_model.id,
name: llm_model.display_name,
vision_enabled: llm_model.vision_enabled,
}.stringify_keys,
],
)
end
it "returns tool options with each tool" do
persona1 = Fabricate(:ai_persona, name: "search1", tools: ["SearchCommand"])
persona2 =
Fabricate(
:ai_persona,
name: "search2",
tools: [["SearchCommand", { base_query: "test" }, true]],
allow_topic_mentions: true,
allow_personal_messages: true,
allow_chat_channel_mentions: true,
allow_chat_direct_messages: true,
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
default_llm_id: llm_model.id,
question_consolidator_llm_id: llm_model.id,
forced_tool_count: 2,
)
persona2.create_user!
get "/admin/plugins/discourse-ai/ai-personas.json"
expect(response).to be_successful
serializer_persona1 = response.parsed_body["ai_personas"].find { |p| p["id"] == persona1.id }
serializer_persona2 = response.parsed_body["ai_personas"].find { |p| p["id"] == persona2.id }
expect(serializer_persona2["allow_topic_mentions"]).to eq(true)
expect(serializer_persona2["allow_personal_messages"]).to eq(true)
expect(serializer_persona2["allow_chat_channel_mentions"]).to eq(true)
expect(serializer_persona2["allow_chat_direct_messages"]).to eq(true)
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
expect(serializer_persona2["default_llm_id"]).to eq(llm_model.id)
expect(serializer_persona2["question_consolidator_llm_id"]).to eq(llm_model.id)
expect(serializer_persona2["user_id"]).to eq(persona2.user_id)
expect(serializer_persona2["user"]["id"]).to eq(persona2.user_id)
expect(serializer_persona2["forced_tool_count"]).to eq(2)
tools = response.parsed_body["meta"]["tools"]
search_tool = tools.find { |c| c["id"] == "Search" }
expect(search_tool["help"]).to eq(I18n.t("discourse_ai.ai_bot.tool_help.search"))
expect(search_tool["options"]).to eq(
{
"base_query" => {
"type" => "string",
"name" => I18n.t("discourse_ai.ai_bot.tool_options.search.base_query.name"),
"description" =>
I18n.t("discourse_ai.ai_bot.tool_options.search.base_query.description"),
},
"max_results" => {
"type" => "integer",
"name" => I18n.t("discourse_ai.ai_bot.tool_options.search.max_results.name"),
"description" =>
I18n.t("discourse_ai.ai_bot.tool_options.search.max_results.description"),
},
"search_private" => {
"type" => "boolean",
"name" => I18n.t("discourse_ai.ai_bot.tool_options.search.search_private.name"),
"description" =>
I18n.t("discourse_ai.ai_bot.tool_options.search.search_private.description"),
},
},
)
expect(serializer_persona1["tools"]).to eq(["SearchCommand"])
expect(serializer_persona2["tools"]).to eq(
[["SearchCommand", { "base_query" => "test" }, true]],
)
end
context "with translations" do
before do
SiteSetting.default_locale = "fr"
TranslationOverride.upsert!(
SiteSetting.default_locale,
"discourse_ai.ai_bot.personas.general.name",
"Général",
)
TranslationOverride.upsert!(
SiteSetting.default_locale,
"discourse_ai.ai_bot.personas.general.description",
"Général Description",
)
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
after do
TranslationOverride.revert!(
SiteSetting.default_locale,
"discourse_ai.ai_bot.personas.general.name",
)
TranslationOverride.revert!(
SiteSetting.default_locale,
"discourse_ai.ai_bot.personas.general.description",
)
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
it "returns localized persona names and descriptions" do
get "/admin/plugins/discourse-ai/ai-personas.json"
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
id = DiscourseAi::Personas::Persona.system_personas[DiscourseAi::Personas::General]
persona = response.parsed_body["ai_personas"].find { |p| p["id"] == id }
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
expect(persona["name"]).to eq("Général")
expect(persona["description"]).to eq("Général Description")
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
end
end
describe "GET #edit" do
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
it "returns a success response" do
get "/admin/plugins/discourse-ai/ai-personas/#{ai_persona.id}/edit.json"
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
expect(response).to be_successful
expect(response.parsed_body["ai_persona"]["name"]).to eq(ai_persona.name)
end
it "includes rag uploads for each persona" do
upload = Fabricate(:upload)
RagDocumentFragment.link_target_and_uploads(ai_persona, [upload.id])
get "/admin/plugins/discourse-ai/ai-personas/#{ai_persona.id}/edit.json"
expect(response).to be_successful
serialized_persona = response.parsed_body["ai_persona"]
expect(serialized_persona.dig("rag_uploads", 0, "id")).to eq(upload.id)
expect(serialized_persona.dig("rag_uploads", 0, "original_filename")).to eq(
upload.original_filename,
)
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
end
describe "POST #create" do
context "with valid params" do
let(:valid_attributes) do
{
name: "superbot",
description: "Assists with tasks",
system_prompt: "you are a helpful bot",
tools: [["search", { "base_query" => "test" }, true]],
top_p: 0.1,
temperature: 0.5,
allow_topic_mentions: true,
allow_personal_messages: true,
allow_chat_channel_mentions: true,
allow_chat_direct_messages: true,
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
default_llm_id: llm_model.id,
question_consolidator_llm_id: llm_model.id,
forced_tool_count: 2,
response_format: [{ key: "summary", type: "string" }],
examples: [%w[user_msg1 assistant_msg1], %w[user_msg2 assistant_msg2]],
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
}
end
it "creates a new AiPersona" do
expect {
post "/admin/plugins/discourse-ai/ai-personas.json",
params: { ai_persona: valid_attributes }.to_json,
headers: {
"CONTENT_TYPE" => "application/json",
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
}
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
expect(response).to be_successful
persona_json = response.parsed_body["ai_persona"]
expect(persona_json["name"]).to eq("superbot")
expect(persona_json["top_p"]).to eq(0.1)
expect(persona_json["temperature"]).to eq(0.5)
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
expect(persona_json["default_llm_id"]).to eq(llm_model.id)
expect(persona_json["forced_tool_count"]).to eq(2)
expect(persona_json["allow_topic_mentions"]).to eq(true)
expect(persona_json["allow_personal_messages"]).to eq(true)
expect(persona_json["allow_chat_channel_mentions"]).to eq(true)
expect(persona_json["allow_chat_direct_messages"]).to eq(true)
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
expect(persona_json["question_consolidator_llm_id"]).to eq(llm_model.id)
expect(persona_json["response_format"].map { |rf| rf["key"] }).to contain_exactly(
"summary",
)
expect(persona_json["examples"]).to eq(valid_attributes[:examples])
persona = AiPersona.find(persona_json["id"])
expect(persona.tools).to eq([["search", { "base_query" => "test" }, true]])
expect(persona.top_p).to eq(0.1)
expect(persona.temperature).to eq(0.5)
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
}.to change(AiPersona, :count).by(1)
end
it "logs staff action when creating a persona" do
# Create the persona
post "/admin/plugins/discourse-ai/ai-personas.json",
params: { ai_persona: valid_attributes }.to_json,
headers: {
"CONTENT_TYPE" => "application/json",
}
expect(response).to be_successful
# Now verify the log was created with the right subject
history =
UserHistory.where(
action: UserHistory.actions[:custom_staff],
custom_type: "create_ai_persona",
).last
expect(history).to be_present
expect(history.subject).to eq("superbot") # Verify subject is set to name
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
end
context "with invalid params" do
it "renders a JSON response with errors for the new ai_persona" do
post "/admin/plugins/discourse-ai/ai-personas.json", params: { ai_persona: { foo: "" } } # invalid attribute
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
expect(response).to have_http_status(:unprocessable_entity)
expect(response.content_type).to include("application/json")
end
end
end
describe "POST #create_user" do
it "creates a user for the persona" do
post "/admin/plugins/discourse-ai/ai-personas/#{ai_persona.id}/create-user.json"
ai_persona.reload
expect(response).to be_successful
expect(response.parsed_body["user"]["id"]).to eq(ai_persona.user_id)
end
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
describe "PUT #update" do
context "with scoped api key" do
it "allows updates with a properly scoped API key" do
api_key = Fabricate(:api_key, user: admin, created_by: admin)
scope =
ApiKeyScope.create!(
resource: "discourse_ai",
action: "update_personas",
api_key_id: api_key.id,
allowed_parameters: {
},
)
put "/admin/plugins/discourse-ai/ai-personas/#{ai_persona.id}.json",
params: {
ai_persona: {
name: "UpdatedByAPI",
description: "Updated via API key",
},
},
headers: {
"Api-Key" => api_key.key,
"Api-Username" => admin.username,
}
expect(response).to have_http_status(:ok)
ai_persona.reload
expect(ai_persona.name).to eq("UpdatedByAPI")
expect(ai_persona.description).to eq("Updated via API key")
scope.update!(action: "fake")
put "/admin/plugins/discourse-ai/ai-personas/#{ai_persona.id}.json",
params: {
ai_persona: {
name: "UpdatedByAPI 2",
description: "Updated via API key",
},
},
headers: {
"Api-Key" => api_key.key,
"Api-Username" => admin.username,
}
expect(response).not_to have_http_status(:ok)
end
end
it "allows us to trivially clear top_p and temperature" do
persona = Fabricate(:ai_persona, name: "test_bot2", top_p: 0.5, temperature: 0.1)
put "/admin/plugins/discourse-ai/ai-personas/#{persona.id}.json",
params: {
ai_persona: {
top_p: "",
temperature: "",
},
}
expect(response).to have_http_status(:ok)
persona.reload
expect(persona.top_p).to eq(nil)
expect(persona.temperature).to eq(nil)
end
it "logs staff action when updating a persona" do
persona = Fabricate(:ai_persona, name: "original_name", description: "original description")
# Update the persona
put "/admin/plugins/discourse-ai/ai-personas/#{persona.id}.json",
params: {
ai_persona: {
name: "updated_name",
description: "updated description",
},
}
expect(response).to have_http_status(:ok)
persona.reload
expect(persona.name).to eq("updated_name")
expect(persona.description).to eq("updated description")
# Now verify the log was created with the right subject
history =
UserHistory.where(
action: UserHistory.actions[:custom_staff],
custom_type: "update_ai_persona",
).last
expect(history).to be_present
expect(history.subject).to eq("updated_name") # Verify subject is set to the new name
end
it "supports updating rag params" do
persona = Fabricate(:ai_persona, name: "test_bot2")
put "/admin/plugins/discourse-ai/ai-personas/#{persona.id}.json",
params: {
ai_persona: {
rag_chunk_tokens: "102",
rag_chunk_overlap_tokens: "12",
rag_conversation_chunks: "13",
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
rag_llm_model_id: llm_model.id,
question_consolidator_llm_id: llm_model.id,
},
}
expect(response).to have_http_status(:ok)
persona.reload
expect(persona.rag_chunk_tokens).to eq(102)
expect(persona.rag_chunk_overlap_tokens).to eq(12)
expect(persona.rag_conversation_chunks).to eq(13)
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
expect(persona.rag_llm_model_id).to eq(llm_model.id)
expect(persona.question_consolidator_llm_id).to eq(llm_model.id)
end
it "supports updating vision params" do
persona = Fabricate(:ai_persona, name: "test_bot2")
put "/admin/plugins/discourse-ai/ai-personas/#{persona.id}.json",
params: {
ai_persona: {
vision_enabled: true,
vision_max_pixels: 512 * 512,
},
}
expect(response).to have_http_status(:ok)
persona.reload
expect(persona.vision_enabled).to eq(true)
expect(persona.vision_max_pixels).to eq(512 * 512)
end
it "does not allow temperature and top p changes on stock personas" do
put "/admin/plugins/discourse-ai/ai-personas/#{DiscourseAi::Personas::Persona.system_personas.values.first}.json",
params: {
ai_persona: {
top_p: 0.5,
temperature: 0.1,
},
}
expect(response).to have_http_status(:unprocessable_entity)
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
context "with valid params" do
it "updates the requested ai_persona" do
put "/admin/plugins/discourse-ai/ai-personas/#{ai_persona.id}.json",
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
params: {
ai_persona: {
name: "SuperBot",
enabled: false,
tools: ["search"],
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
},
}
expect(response).to have_http_status(:ok)
expect(response.content_type).to include("application/json")
ai_persona.reload
expect(ai_persona.name).to eq("SuperBot")
expect(ai_persona.enabled).to eq(false)
expect(ai_persona.tools).to eq([["search", nil, false]])
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
end
end
context "with system personas" do
it "does not allow editing of system prompts" do
put "/admin/plugins/discourse-ai/ai-personas/#{DiscourseAi::Personas::Persona.system_personas.values.first}.json",
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
params: {
ai_persona: {
system_prompt: "you are not a helpful bot",
},
}
expect(response).to have_http_status(:unprocessable_entity)
expect(response.parsed_body["errors"].join).not_to be_blank
expect(response.parsed_body["errors"].join).not_to include("en.discourse")
end
it "does not allow editing of tools" do
put "/admin/plugins/discourse-ai/ai-personas/#{DiscourseAi::Personas::Persona.system_personas.values.first}.json",
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
params: {
ai_persona: {
tools: %w[SearchCommand ImageCommand],
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
},
}
expect(response).to have_http_status(:unprocessable_entity)
expect(response.parsed_body["errors"].join).not_to be_blank
expect(response.parsed_body["errors"].join).not_to include("en.discourse")
end
it "does not allow editing of name and description cause it is localized" do
put "/admin/plugins/discourse-ai/ai-personas/#{DiscourseAi::Personas::Persona.system_personas.values.first}.json",
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
params: {
ai_persona: {
name: "bob",
2024-01-19 12:51:26 +01:00
description: "the bob",
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
},
}
expect(response).to have_http_status(:unprocessable_entity)
expect(response.parsed_body["errors"].join).not_to be_blank
expect(response.parsed_body["errors"].join).not_to include("en.discourse")
end
it "does allow some actions" do
put "/admin/plugins/discourse-ai/ai-personas/#{DiscourseAi::Personas::Persona.system_personas.values.first}.json",
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
params: {
ai_persona: {
allowed_group_ids: [Group::AUTO_GROUPS[:trust_level_1]],
enabled: false,
priority: 989,
},
}
expect(response).to be_successful
end
end
context "with invalid params" do
it "renders a JSON response with errors for the ai_persona" do
put "/admin/plugins/discourse-ai/ai-personas/#{ai_persona.id}.json",
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
params: {
ai_persona: {
name: "",
},
} # invalid attribute
expect(response).to have_http_status(:unprocessable_entity)
expect(response.content_type).to include("application/json")
end
end
end
describe "GET #export" do
fab!(:ai_tool) do
AiTool.create!(
name: "Test Tool",
tool_name: "test_tool",
description: "A test tool",
script: "function invoke(params) { return 'test'; }",
parameters: [{ name: "query", type: "string", required: true }],
summary: "Test tool summary",
created_by_id: admin.id,
)
end
fab!(:persona_with_tools) do
AiPersona.create!(
name: "ToolMaster",
description: "A persona with custom tools",
system_prompt: "You are a tool master",
tools: [
["SearchCommand", { "base_query" => "test" }, true],
["custom-#{ai_tool.id}", { "max_results" => 10 }, false],
],
temperature: 0.8,
top_p: 0.9,
response_format: [{ type: "string", key: "summary" }],
examples: [["user example", "assistant example"]],
default_llm_id: llm_model.id,
)
end
it "exports a persona as JSON" do
get "/admin/plugins/discourse-ai/ai-personas/#{persona_with_tools.id}/export.json"
expect(response).to be_successful
expect(response.headers["Content-Disposition"]).to include("attachment")
expect(response.headers["Content-Disposition"]).to include("toolmaster.json")
json = response.parsed_body
expect(json["meta"]["version"]).to eq("1.0")
expect(json["meta"]["exported_at"]).to be_present
persona_data = json["persona"]
expect(persona_data["name"]).to eq("ToolMaster")
expect(persona_data["description"]).to eq("A persona with custom tools")
expect(persona_data["system_prompt"]).to eq("You are a tool master")
expect(persona_data["temperature"]).to eq(0.8)
expect(persona_data["top_p"]).to eq(0.9)
expect(persona_data["response_format"]).to eq([{ "type" => "string", "key" => "summary" }])
expect(persona_data["examples"]).to eq([["user example", "assistant example"]])
# Check that custom tool ID is replaced with tool_name
expect(persona_data["tools"]).to include(
["SearchCommand", { "base_query" => "test" }, true],
["custom-test_tool", { "max_results" => 10 }, false],
)
# Check custom tools are exported
expect(json["custom_tools"]).to be_an(Array)
expect(json["custom_tools"].length).to eq(1)
custom_tool = json["custom_tools"].first
expect(custom_tool["identifier"]).to eq("test_tool")
expect(custom_tool["name"]).to eq("Test Tool")
expect(custom_tool["description"]).to eq("A test tool")
expect(custom_tool["script"]).to eq("function invoke(params) { return 'test'; }")
expect(custom_tool["parameters"]).to eq(
[{ "name" => "query", "type" => "string", "required" => true }],
)
end
it "handles personas without custom tools" do
persona = Fabricate(:ai_persona, tools: ["SearchCommand"])
get "/admin/plugins/discourse-ai/ai-personas/#{persona.id}/export.json"
expect(response).to be_successful
json = response.parsed_body
expect(json["custom_tools"]).to eq([])
expect(json["persona"]["tools"]).to eq(["SearchCommand"])
end
it "returns 404 for non-existent persona" do
get "/admin/plugins/discourse-ai/ai-personas/999999/export.json"
expect(response).to have_http_status(:not_found)
end
end
describe "POST #import" do
let(:valid_import_data) do
{
meta: {
version: "1.0",
exported_at: Time.zone.now.iso8601,
},
persona: {
name: "ImportedPersona",
description: "An imported persona",
system_prompt: "You are an imported assistant",
temperature: 0.7,
top_p: 0.8,
response_format: [{ type: "string", key: "answer" }],
examples: [["hello", "hi there"]],
tools: ["SearchCommand", ["ReadCommand", { max_length: 1000 }, true]],
},
custom_tools: [],
}
end
it "imports a new persona successfully" do
expect {
post "/admin/plugins/discourse-ai/ai-personas/import.json",
params: valid_import_data,
as: :json
expect(response).to have_http_status(:created)
}.to change(AiPersona, :count).by(1)
persona = AiPersona.find_by(name: "ImportedPersona")
expect(persona).to be_present
expect(persona.description).to eq("An imported persona")
expect(persona.system_prompt).to eq("You are an imported assistant")
expect(persona.temperature).to eq(0.7)
expect(persona.top_p).to eq(0.8)
expect(persona.response_format).to eq([{ "type" => "string", "key" => "answer" }])
expect(persona.examples).to eq([["hello", "hi there"]])
expect(persona.tools).to eq(
["SearchCommand", ["ReadCommand", { "max_length" => 1000 }, true]],
)
end
it "imports a persona with custom tools" do
import_data_with_tools = valid_import_data.deep_dup
import_data_with_tools[:persona][:tools] = [
"SearchCommand",
["custom-my_custom_tool", { param1: "value1" }, false],
]
import_data_with_tools[:custom_tools] = [
{
identifier: "my_custom_tool",
name: "My Custom Tool",
description: "A custom tool for testing",
tool_name: "my_custom_tool",
parameters: [{ name: "param1", type: "string", required: true }],
summary: "Custom tool summary",
script: "function invoke(params) { return params.param1; }",
},
]
expect {
post "/admin/plugins/discourse-ai/ai-personas/import.json",
params: import_data_with_tools,
as: :json
}.to change(AiPersona, :count).by(1).and change(AiTool, :count).by(1)
expect(response).to have_http_status(:created)
persona = AiPersona.find_by(name: "ImportedPersona")
expect(persona).to be_present
tool = AiTool.find_by(tool_name: "my_custom_tool")
expect(tool).to be_present
expect(tool.name).to eq("My Custom Tool")
expect(tool.description).to eq("A custom tool for testing")
expect(tool.script).to eq("function invoke(params) { return params.param1; }")
# Check that the tool reference uses the ID
expect(persona.tools).to include(
"SearchCommand",
["custom-#{tool.id}", { "param1" => "value1" }, false],
)
end
it "prevents importing duplicate personas by default" do
_existing_persona = Fabricate(:ai_persona, name: "ImportedPersona")
post "/admin/plugins/discourse-ai/ai-personas/import.json",
params: valid_import_data,
as: :json
expect(response).to have_http_status(:unprocessable_entity)
expect(response.parsed_body["errors"].join).to include("ImportedPersona")
end
it "allows overwriting existing personas with force=true" do
existing_persona =
Fabricate(:ai_persona, name: "ImportedPersona", description: "Old description")
import_data = valid_import_data.merge(force: true)
expect {
post "/admin/plugins/discourse-ai/ai-personas/import.json", params: import_data, as: :json
}.not_to change(AiPersona, :count)
expect(response).to have_http_status(:ok)
existing_persona.reload
expect(existing_persona.description).to eq("An imported persona")
expect(existing_persona.system_prompt).to eq("You are an imported assistant")
end
it "handles invalid import data gracefully" do
invalid_data = { invalid: "data" }
post "/admin/plugins/discourse-ai/ai-personas/import.json", params: invalid_data, as: :json
expect(response).to have_http_status(:unprocessable_entity)
expect(response.parsed_body["errors"]).to be_present
end
it "handles missing persona data" do
invalid_data = { meta: { version: "1.0" } }
post "/admin/plugins/discourse-ai/ai-personas/import.json", params: invalid_data, as: :json
expect(response).to have_http_status(:unprocessable_entity)
end
it "logs staff action when importing a new persona" do
post "/admin/plugins/discourse-ai/ai-personas/import.json",
params: valid_import_data,
as: :json
expect(response).to have_http_status(:created)
history =
UserHistory.where(
action: UserHistory.actions[:custom_staff],
custom_type: "create_ai_persona",
).last
expect(history).to be_present
expect(history.subject).to eq("ImportedPersona")
end
it "logs staff action when updating an existing persona" do
_existing_persona = Fabricate(:ai_persona, name: "ImportedPersona")
import_data = valid_import_data.merge(force: true)
post "/admin/plugins/discourse-ai/ai-personas/import.json", params: import_data, as: :json
expect(response).to have_http_status(:ok)
history =
UserHistory.where(
action: UserHistory.actions[:custom_staff],
custom_type: "update_ai_persona",
).last
expect(history).to be_present
expect(history.subject).to eq("ImportedPersona")
end
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
describe "DELETE #destroy" do
it "destroys the requested ai_persona" do
expect {
delete "/admin/plugins/discourse-ai/ai-personas/#{ai_persona.id}.json"
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
expect(response).to have_http_status(:no_content)
}.to change(AiPersona, :count).by(-1)
end
it "logs staff action when deleting a persona" do
# Capture persona details before deletion
_persona_id = ai_persona.id
persona_name = ai_persona.name
# Delete the persona
delete "/admin/plugins/discourse-ai/ai-personas/#{ai_persona.id}.json"
expect(response).to have_http_status(:no_content)
# Now verify the log was created with the right subject
history =
UserHistory.where(
action: UserHistory.actions[:custom_staff],
custom_type: "delete_ai_persona",
).last
expect(history).to be_present
expect(history.subject).to eq(persona_name) # Verify subject is set to name
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
it "is not allowed to delete system personas" do
expect {
delete "/admin/plugins/discourse-ai/ai-personas/#{DiscourseAi::Personas::Persona.system_personas.values.first}.json"
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
expect(response).to have_http_status(:unprocessable_entity)
expect(response.parsed_body["errors"].join).not_to be_blank
# let's make sure this is translated
expect(response.parsed_body["errors"].join).not_to include("en.discourse")
}.not_to change(AiPersona, :count)
end
end
describe "#stream_reply" do
fab!(:llm) { Fabricate(:llm_model, name: "fake_llm", provider: "fake") }
let(:fake_endpoint) { DiscourseAi::Completions::Endpoints::Fake }
before { fake_endpoint.delays = [] }
after { fake_endpoint.reset! }
it "ensures persona exists" do
post "/admin/plugins/discourse-ai/ai-personas/stream-reply.json"
expect(response).to have_http_status(:unprocessable_entity)
# this ensures localization key is actually in the yaml
expect(response.body).to include("persona_name")
end
it "ensures question exists" do
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
ai_persona.update!(default_llm_id: llm.id)
post "/admin/plugins/discourse-ai/ai-personas/stream-reply.json",
params: {
persona_id: ai_persona.id,
user_unique_id: "site:test.com:user_id:1",
}
expect(response).to have_http_status(:unprocessable_entity)
expect(response.body).to include("query")
end
it "ensure persona has a user specified" do
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
ai_persona.update!(default_llm_id: llm.id)
post "/admin/plugins/discourse-ai/ai-personas/stream-reply.json",
params: {
persona_id: ai_persona.id,
query: "how are you today?",
user_unique_id: "site:test.com:user_id:1",
}
expect(response).to have_http_status(:unprocessable_entity)
expect(response.body).to include("associated")
end
def validate_streamed_response(raw_http, expected)
lines = raw_http.split("\r\n")
header_lines, _, payload_lines = lines.chunk { |l| l == "" }.map(&:last)
preamble = (<<~PREAMBLE).strip
HTTP/1.1 200 OK
Content-Type: text/plain; charset=utf-8
Transfer-Encoding: chunked
Cache-Control: no-cache, no-store, must-revalidate
Connection: close
X-Accel-Buffering: no
X-Content-Type-Options: nosniff
PREAMBLE
expect(header_lines.join("\n")).to eq(preamble)
parsed = +""
context_info = nil
payload_lines.each_slice(2) do |size, data|
size = size.to_i(16)
data = data.to_s
expect(data.bytesize).to eq(size)
if size > 0
json = JSON.parse(data)
parsed << json["partial"].to_s
context_info = json if json["topic_id"]
end
end
expect(parsed).to eq(expected)
context_info
end
it "is able to create a new conversation" do
Jobs.run_immediately!
# trust level 0
SiteSetting.ai_bot_allowed_groups = "10"
fake_endpoint.fake_content = ["This is a test! Testing!", "An amazing title"]
ai_persona.create_user!
ai_persona.update!(
allowed_group_ids: [Group::AUTO_GROUPS[:trust_level_0]],
FEATURE: PDF support for rag pipeline (#1118) This PR introduces several enhancements and refactorings to the AI Persona and RAG (Retrieval-Augmented Generation) functionalities within the discourse-ai plugin. Here's a breakdown of the changes: **1. LLM Model Association for RAG and Personas:** - **New Database Columns:** Adds `rag_llm_model_id` to both `ai_personas` and `ai_tools` tables. This allows specifying a dedicated LLM for RAG indexing, separate from the persona's primary LLM. Adds `default_llm_id` and `question_consolidator_llm_id` to `ai_personas`. - **Migration:** Includes a migration (`20250210032345_migrate_persona_to_llm_model_id.rb`) to populate the new `default_llm_id` and `question_consolidator_llm_id` columns in `ai_personas` based on the existing `default_llm` and `question_consolidator_llm` string columns, and a post migration to remove the latter. - **Model Changes:** The `AiPersona` and `AiTool` models now `belong_to` an `LlmModel` via `rag_llm_model_id`. The `LlmModel.proxy` method now accepts an `LlmModel` instance instead of just an identifier. `AiPersona` now has `default_llm_id` and `question_consolidator_llm_id` attributes. - **UI Updates:** The AI Persona and AI Tool editors in the admin panel now allow selecting an LLM for RAG indexing (if PDF/image support is enabled). The RAG options component displays an LLM selector. - **Serialization:** The serializers (`AiCustomToolSerializer`, `AiCustomToolListSerializer`, `LocalizedAiPersonaSerializer`) have been updated to include the new `rag_llm_model_id`, `default_llm_id` and `question_consolidator_llm_id` attributes. **2. PDF and Image Support for RAG:** - **Site Setting:** Introduces a new hidden site setting, `ai_rag_pdf_images_enabled`, to control whether PDF and image files can be indexed for RAG. This defaults to `false`. - **File Upload Validation:** The `RagDocumentFragmentsController` now checks the `ai_rag_pdf_images_enabled` setting and allows PDF, PNG, JPG, and JPEG files if enabled. Error handling is included for cases where PDF/image indexing is attempted with the setting disabled. - **PDF Processing:** Adds a new utility class, `DiscourseAi::Utils::PdfToImages`, which uses ImageMagick (`magick`) to convert PDF pages into individual PNG images. A maximum PDF size and conversion timeout are enforced. - **Image Processing:** A new utility class, `DiscourseAi::Utils::ImageToText`, is included to handle OCR for the images and PDFs. - **RAG Digestion Job:** The `DigestRagUpload` job now handles PDF and image uploads. It uses `PdfToImages` and `ImageToText` to extract text and create document fragments. - **UI Updates:** The RAG uploader component now accepts PDF and image file types if `ai_rag_pdf_images_enabled` is true. The UI text is adjusted to indicate supported file types. **3. Refactoring and Improvements:** - **LLM Enumeration:** The `DiscourseAi::Configuration::LlmEnumerator` now provides a `values_for_serialization` method, which returns a simplified array of LLM data (id, name, vision_enabled) suitable for use in serializers. This avoids exposing unnecessary details to the frontend. - **AI Helper:** The `AiHelper::Assistant` now takes optional `helper_llm` and `image_caption_llm` parameters in its constructor, allowing for greater flexibility. - **Bot and Persona Updates:** Several updates were made across the codebase, changing the string based association to a LLM to the new model based. - **Audit Logs:** The `DiscourseAi::Completions::Endpoints::Base` now formats raw request payloads as pretty JSON for easier auditing. - **Eval Script:** An evaluation script is included. **4. Testing:** - The PR introduces a new eval system for LLMs, this allows us to test how functionality works across various LLM providers. This lives in `/evals`
2025-02-14 12:15:07 +11:00
default_llm_id: llm.id,
allow_personal_messages: true,
system_prompt: "you are a helpful bot",
)
io_out, io_in = IO.pipe
post "/admin/plugins/discourse-ai/ai-personas/stream-reply.json",
params: {
persona_name: ai_persona.name,
query: "how are you today?",
user_unique_id: "site:test.com:user_id:1",
preferred_username: "test_user",
custom_instructions: "To be appended to system prompt",
},
env: {
"rack.hijack" => lambda { io_in },
}
# this is a fake response but it catches errors
expect(response).to have_http_status(:no_content)
raw = io_out.read
context_info = validate_streamed_response(raw, "This is a test! Testing!")
system_prompt = fake_endpoint.previous_calls[-2][:dialect].prompt.messages.first[:content]
expect(system_prompt).to eq("you are a helpful bot\nTo be appended to system prompt")
expect(context_info["topic_id"]).to be_present
topic = Topic.find(context_info["topic_id"])
last_post = topic.posts.order(:created_at).last
expect(last_post.raw).to eq("This is a test! Testing!")
user_post = topic.posts.find_by(post_number: 1)
expect(user_post.raw).to eq("how are you today?")
# need ai persona and user
expect(topic.topic_allowed_users.count).to eq(2)
expect(topic.archetype).to eq(Archetype.private_message)
expect(topic.title).to eq("An amazing title")
expect(topic.posts.count).to eq(2)
tool_call =
DiscourseAi::Completions::ToolCall.new(name: "categories", parameters: {}, id: "tool_1")
fake_endpoint.fake_content = [tool_call, "this is the response after the tool"]
# this simplifies function calls
fake_endpoint.chunk_count = 1
ai_persona.update!(tools: ["Categories"])
# lets also unstage the user and add the user to tl0
# this will ensure there are no feedback loops
new_user = user_post.user
new_user.update!(staged: false)
Group.user_trust_level_change!(new_user.id, new_user.trust_level)
# double check this happened and user is in group
personas = AiPersona.allowed_modalities(user: new_user.reload, allow_personal_messages: true)
expect(personas.count).to eq(1)
io_out, io_in = IO.pipe
post "/admin/plugins/discourse-ai/ai-personas/stream-reply.json",
params: {
persona_id: ai_persona.id,
query: "how are you now?",
user_unique_id: "site:test.com:user_id:1",
preferred_username: "test_user",
topic_id: context_info["topic_id"],
},
env: {
"rack.hijack" => lambda { io_in },
}
# this is a fake response but it catches errors
expect(response).to have_http_status(:no_content)
raw = io_out.read
context_info = validate_streamed_response(raw, "this is the response after the tool")
topic = topic.reload
last_post = topic.posts.order(:created_at).last
expect(last_post.raw).to end_with("this is the response after the tool")
# function call is visible in the post
expect(last_post.raw[0..8]).to eq("<details>")
user_post = topic.posts.find_by(post_number: 3)
expect(user_post.raw).to eq("how are you now?")
expect(user_post.user.username).to eq("test_user")
expect(user_post.user.custom_fields).to eq(
{ "ai-stream-conversation-unique-id" => "site:test.com:user_id:1" },
)
expect(topic.posts.count).to eq(4)
end
end
FEATURE: UI to update ai personas on admin page (#290) Introduces a UI to manage customizable personas (admin only feature) Part of the change was some extensive internal refactoring: - AIBot now has a persona set in the constructor, once set it never changes - Command now takes in bot as a constructor param, so it has the correct persona and is not generating AIBot objects on the fly - Added a .prettierignore file, due to the way ALE is configured in nvim it is a pre-req for prettier to work - Adds a bunch of validations on the AIPersona model, system personas (artist/creative etc...) are all seeded. We now ensure - name uniqueness, and only allow certain properties to be touched for system personas. - (JS note) the client side design takes advantage of nested routes, the parent route for personas gets all the personas via this.store.findAll("ai-persona") then child routes simply reach into this model to find a particular persona. - (JS note) data is sideloaded into the ai-persona model the meta property supplied from the controller, resultSetMeta - This removes ai_bot_enabled_personas and ai_bot_enabled_chat_commands, both should be controlled from the UI on a per persona basis - Fixes a long standing bug in token accounting ... we were doing to_json.length instead of to_json.to_s.length - Amended it so {commands} are always inserted at the end unconditionally, no need to add it to the template of the system message as it just confuses things - Adds a concept of required_commands to stock personas, these are commands that must be configured for this stock persona to show up. - Refactored tests so we stop requiring inference_stubs, it was very confusing to need it, added to plugin.rb for now which at least is clearer - Migrates the persona selector to gjs --------- Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com> Co-authored-by: Martin Brennan <martin@discourse.org>
2023-11-21 16:56:43 +11:00
end