Karel Minarik faa0ee14bd [UTIL] Added a source code generator using the JSON API specs
The generator is based on [Thor](https://github.com/wycats/thor),
a library/framework for command line applications.

The generator will read the JSON API spec file(s), and generate
the Ruby source code (one file per API endpoint) with correct
module namespace, method names, and RDoc documentation.

It will generate a test file for each API endpoint as well.

Currently it only generates Ruby source, but can easily be
extended and adapted to generate source code for other
programming languages.

Usage example:

    $ thor api:code:generate ../../api-spec/*.json --force --verbose
2013-06-16 23:25:08 +02:00

190 lines
6.7 KiB
Ruby

require 'thor'
require 'pathname'
require 'active_support/core_ext/hash/deep_merge'
require 'active_support/inflector/methods'
require 'rest_client'
require 'json'
require 'pry'
module Elasticsearch
module API
module Utils
# controller.registerHandler(RestRequest.Method.GET, "/_cluster/health", this);
PATTERN_REST = /.*controller.registerHandler\(.*(?<method>GET|POST|PUT|DELETE|HEAD|OPTIONS|PATCH)\s*,\s*"(?<url>.*)"\s*,\s*.+\);/
# request.param("index"), request.paramAsBoolean("docs", indicesStatsRequest.docs()), etc
PATTERN_URL_PARAMS = /request.param.*\("(?<param>[a-z_]+)".*/
# controller.registerHandler(GET, "/{index}/_refresh", this)
PATTERN_URL_PARTS = /\{(?<part>[a-zA-Z0-9\_\-]+)\}/
# request.hasContent()
PATTERN_HAS_BODY = /request\.hasContent()/
# Parses the Elasticsearch source code and returns a Hash of REST API information/specs.
#
# Example:
#
# {
# "cluster.health" => [
# { "method" => "GET",
# "path" => "/_cluster/health",
# "parts" => ["index"],
# "params" => ["index", "local", ... ],
# "body" => false
# }
#
def __parse_java_source(path)
path += '/' unless path =~ /\/$/ # Add trailing slash if missing
prefix = "src/main/java/org/elasticsearch/rest/action"
java_rest_files = Dir["#{path}#{prefix}/**/*.java"]
map = {}
java_rest_files.sort.each do |file|
content = File.read(file)
parts = file.gsub(path+prefix, '').split('/')
name = parts[0, parts.size-1].reject { |p| p =~ /^\s*$/ }.join('.')
# Remove the `admin` namespace
name.gsub! /admin\./, ''
# Extract params
url_params = content.scan(PATTERN_URL_PARAMS).map { |n| n.first }.sort
# Extract parts
url_parts = content.scan(PATTERN_URL_PARTS).map { |n| n.first }.sort
# Extract if body allowed
has_body = !!content.match(PATTERN_HAS_BODY)
# Extract HTTP method and path
content.scan(PATTERN_REST) do |method, path|
(map[name] ||= []) << { 'method' => method,
'path' => path,
'parts' => url_parts,
'params' => url_params,
'body' => has_body }
end
end
map
end
extend self
end
# Contains a generator which will parse the Elasticsearch *.java source files,
# extract information about REST API endpoints (URLs, HTTP methods, URL parameters, etc),
# and create a skeleton of the JSON API specification file for each endpoint.
#
# Usage:
#
# $ thor help api:generate:spec
#
# Example:
#
# time thor api:generate:spec \
# --force \
# --verbose \
# --crawl \
# --elasticsearch=/path/to/elasticsearch/source/code
#
# Features:
#
# * Extract the API name from the source filename (eg. `admin/cluster/health/RestClusterHealthAction.java` -> `cluster.health`)
# * Extract the URLs from the `registerHandler` statements
# * Extract the URL parts (eg. `{index}`) from the URLs
# * Extract the URL parameters (eg. `{timeout}`) from the `request.param("ABC")` statements
# * Detect whether HTTP body is allowed for the API from `request.hasContent()` statements
# * Search the <http://elasticsearch.org> website to get proper documentation URLs
# * Assemble the JSON format for the API spec
#
class JsonGenerator < Thor
namespace 'api:spec'
include Thor::Actions
__root = Pathname( File.expand_path('../../..', __FILE__) )
# Usage: thor help api:generate:spec
#
desc "generate", "Generate JSON API spec files from Elasticsearch source code"
method_option :force, type: :boolean, default: false, desc: 'Overwrite the output'
method_option :verbose, type: :boolean, default: false, desc: 'Output more information'
method_option :output, default: __root.join('tmp/out'), desc: 'Path to output directory'
method_option :elasticsearch, default: __root.join('tmp/elasticsearch'), desc: 'Path to directory with Elasticsearch source code'
method_option :crawl, type: :boolean, default: false, desc: 'Extract URLs from Elasticsearch website'
def generate
self.class.source_root File.expand_path('../', __FILE__)
@output = options[:output]
rest_actions = Utils.__parse_java_source(options[:elasticsearch].to_s)
if rest_actions.empty?
say_status 'ERROR', 'Cannot find Elasticsearch source in ' + options[:elasticsearch].to_s, :red
exit(1)
end
rest_actions.each do |name, info|
doc_url = ""
parts = info.reduce([]) { |sum, n| sum |= n['parts']; sum }.reduce({}) { |sum, n| sum[n] = {}; sum }
params = info.reduce([]) { |sum, n| sum |= n['params']; sum }.reduce({}) { |sum, n| sum[n] = {}; sum }
if options[:crawl]
begin
response = RestClient.get "http://search.elasticsearch.org/elastic-search-website/guide/_search?q=#{URI.escape(name.gsub(/\./, ' '))}"
hits = JSON.load(response)['hits']['hits']
if hit = hits.first
if hit['_score'] > 0.2
doc_title = hit['fields']['title']
doc_url = "http://elasticsearch.org" + hit['fields']['url']
end
end
rescue Exception => e
puts "[!] ERROR: #{e.inspect}"
end
end
spec = {
name => {
'documentation' => doc_url,
'methods' => info.map { |n| n['method'] }.uniq,
'url' => {
'path' => info.first['path'],
'paths' => info.map { |n| n['path'] }.uniq,
'parts' => parts,
'params' => params
},
'body' => info.first['body'] ? {} : nil
}
}
json = JSON.pretty_generate(spec, indent: ' ', array_nl: '', object_nl: "\n", space: ' ', space_before: ' ')
# Fix JSON array formatting
json.gsub!(/\[\s+/, '[')
json.gsub!(/, {2,}"/, ', "')
create_file @output.join( "#{name}.json" ), json + "\n"
if options[:verbose]
lines = json.split("\n")
say_status 'JSON',
lines.first + "\n" + lines[1, lines.size].map { |l| ' '*14 + l }.join("\n")
end
end
end
private
end
end
end