Add 'keep_cid_imgs' option to HTML to Markdown converter to improve incoming email parsing

lib/html_to_markdown.rb

@@ -136,7 +136,7 @@ class HtmlToMarkdown
   end
 
   def visit_img(node)
-    if is_valid_url?(node["src"]) && is_visible_img?(node)
+    if is_valid_src?(node["src"]) && is_visible_img?(node)
       if @opts[:keep_img_tags]
         @stack[-1].markdown << node.to_html
       else
@@ -147,7 +147,7 @@ class HtmlToMarkdown
   end
 
   def visit_a(node)
-    if is_valid_url?(node["href"])
+    if is_valid_href?(node["href"])
       @stack[-1].markdown << "["
       traverse(node)
       @stack[-1].markdown << "](#{node["href"]})"
@@ -206,14 +206,20 @@ class HtmlToMarkdown
     (lines + [""]).join("\n")
   end
 
-  def is_valid_url?(url)
-    url.present? && (url.start_with?("http") || url.start_with?("www."))
+  def is_valid_href?(href)
+    href.present? && (href.start_with?("http") || href.start_with?("www."))
+  end
+
+  def is_valid_src?(src)
+    return false if src.blank?
+    return true  if @opts[:keep_cid_imgs] && src.start_with?("cid:")
+    src.start_with?("http") || src.start_with?("www.")
   end
 
   def is_visible_img?(img)
-    return false if img["width"].present? && img["width"].to_i == 0
+    return false if img["width"].present?  && img["width"].to_i == 0
     return false if img["height"].present? && img["height"].to_i == 0
-    return false if img["style"].present? && img["style"][/(width|height)\s*:\s*0/]
+    return false if img["style"].present?  && img["style"][/(width|height)\s*:\s*0/]
     true
   end
 

spec/components/html_to_markdown_spec.rb

@@ -3,8 +3,8 @@ require 'html_to_markdown'
 
 describe HtmlToMarkdown do
 
-  def html_to_markdown(html)
-    HtmlToMarkdown.new(html).to_markdown
+  def html_to_markdown(html, opts={})
+    HtmlToMarkdown.new(html, opts).to_markdown
   end
 
   it "remove whitespaces" do
@@ -55,14 +55,15 @@ describe HtmlToMarkdown do
     expect(html_to_markdown(%Q{<a href="foo.bar">Discourse</a>})).to eq("Discourse")
   end
 
-  HTML_WITH_IMG ||= %Q{<img src="https://www.discourse.org/logo.svg" alt="Discourse Logo">}
+  HTML_WITH_IMG     ||= %Q{<img src="https://www.discourse.org/logo.svg" alt="Discourse Logo">}
+  HTML_WITH_CID_IMG ||= %Q{<img src="cid:ii_1525434659ddb4cb" alt="Discourse Logo">}
 
   it "converts <img>" do
     expect(html_to_markdown(HTML_WITH_IMG)).to eq("![Discourse Logo](https://www.discourse.org/logo.svg)")
   end
 
   it "keeps <img> with 'keep_img_tags'" do
-    expect(HtmlToMarkdown.new(HTML_WITH_IMG, keep_img_tags: true).to_markdown).to eq(HTML_WITH_IMG)
+    expect(html_to_markdown(HTML_WITH_IMG, keep_img_tags: true)).to eq(HTML_WITH_IMG)
   end
 
   it "removes empty & invalid <img>" do
@@ -71,6 +72,11 @@ describe HtmlToMarkdown do
     expect(html_to_markdown(%Q{<img src="foo.bar">})).to eq("")
   end
 
+  it "keeps <img> with src='cid:' whith 'keep_cid_imgs'" do
+    expect(html_to_markdown(HTML_WITH_CID_IMG, keep_cid_imgs: true)).to eq("![Discourse Logo](cid:ii_1525434659ddb4cb)")
+    expect(html_to_markdown(HTML_WITH_CID_IMG, keep_img_tags: true, keep_cid_imgs: true)).to eq("<img src=\"cid:ii_1525434659ddb4cb\" alt=\"Discourse Logo\">")
+  end
+
   it "skips hidden <img>" do
     expect(html_to_markdown(%Q{<img src="https://www.discourse.org/logo.svg" width=0>})).to eq("")
     expect(html_to_markdown(%Q{<img src="https://www.discourse.org/logo.svg" height="0">})).to eq("")