reviewdog: upgrade from grep based to fnmatch based, add per-repo set…

Gemfile.lock

@@ -7,6 +7,7 @@ GEM
 
 PLATFORMS
   arm64-darwin-22
+  arm64-darwin-23
   x86_64-linux
 
 DEPENDENCIES

assets/cleaner.rb

@@ -1,20 +1,58 @@
 #!/usr/bin/env ruby
 require 'optparse'
 
-options = {}
+class Matcher
+  def initialize(*blocklist_files)
+    @blocklist = []
+    blocklist_files.each do |blf|
+      next unless File.exist?(blf)
+
+      blocklist = File.read(blf).split("\n").map(&:strip).reject(&:empty?)
+      # remove empty lines and comments
+      blocklist.reject! { |r| r.empty? || r.start_with?('#') }
+
+      # remove all matching lines and report
+      blocklist.reject! do |r|
+        ret = r =~ /^[*@]+$/
+        STDERR.puts "Warning: #{blf} contains a line with only asterisks/at, which will match everything" if ret
+        ret
+      end        
+
+      @blocklist += blocklist
+    end
+  end
+
+  def match?(line)
+    @blocklist.each do |r|
+      return true if File.fnmatch?("*#{r}*", line)
+    end
+    false
+  end
+end
+
+options = {
+  matcher: Matcher.new()
+}
 OptionParser.new do |opts|
   opts.banner = "Usage: reviewdog-adapter.rb [options]"
 
   opts.on("--svgo", "Add SVGO String") do |v|
     options[:svgo] = true
+    options[:matcher] = Matcher.new("#{ENV["SCRIPTPATH"]}/dtd/blocklist.txt")
   end
 
-  opts.on("--assignees", "Add SVGO String") do |v|
+  opts.on("--assignees", "Add Assignees String") do |v|
     options[:assignees] = true
   end
 
-  opts.on("--sveltegrep", "Remove Extracted Script Extension") do |v|
+  opts.on("--sveltegrep", "Remove Extracted Script Extension, and use semgrep blocklist") do |v|
     options[:sveltegrep] = true
+    options[:matcher] = Matcher.new("#{ENV["SCRIPTPATH"]}/semgrep_rules/blocklist.txt")
+  end
+
+  opts.on("--semgrep", "Use semgrep blocklist") do |v|
+    options[:semgrep] = true
+    options[:matcher] = Matcher.new("#{ENV["SCRIPTPATH"]}/semgrep_rules/blocklist.txt")
   end
 end.parse!
 
@@ -39,5 +77,5 @@
     l.gsub!(/$/, "<br>Cc #{ENV['ASSIGNEES']}")
   end
 
-  puts l
+  puts l unless options[:matcher].match?(l)
 end

assets/dtd/blocklist.txt

@@ -1,5 +1,5 @@
-element [^:]+: validity error : ID [^ ]+ already defined
-element [^:]+: validity error : No declaration for attribute data-[^ ]+ of element [^:]+
+element *: validity error : ID * already defined
+element *: validity error : No declaration for attribute data-* of element *
 element style: validity error : Element style does not carry attribute type
-element svg: validity error : Value for attribute version of svg must be [^ ]+
-third_party\/rust\/[^.]+\.svg
+element svg: validity error : Value for attribute version of svg must be *
+third_party/rust/**/*.svg

assets/reviewdog/reviewdog.yml

@@ -14,8 +14,7 @@ runner:
           $([ -n "${GITHUB_BASE_REF+set}" ] && echo "--baseline-commit origin/${GITHUB_BASE_REF:-main}") \
           --json \
       | jq -r '.results[] | "\(.extra.severity[0:1]):\(.path):\(.end.line) \(.extra.message | sub("\n";"<br/>";"g"))<br><br>Source: \(.extra.metadata.source)<br><br>,\(if .extra.metadata.assignees then .extra.metadata.assignees else "null" end | sub("\n";" ";"g"))"' \
-      | grep -f $SCRIPTPATH/semgrep_rules/blocklist.txt -v \
-      | $SCRIPTPATH/cleaner.rb --assignees) 2> reviewdog.semgrep.stderr.log
+      | $SCRIPTPATH/cleaner.rb --semgrep --assignees) 2> reviewdog.semgrep.stderr.log
     errorformat:
       - "%t:%f:%l %m"
   sveltegrep:
@@ -46,7 +45,6 @@ runner:
         '--include=*.extractedscript.html' \
         ./ \
         | jq -r '.results[] | "\(.extra.severity[0:1]):\(.path):\(.end.line) \(.extra.message | sub("\n";"<br/>";"g"))<br><br>Source: \(.extra.metadata.source)<br><br>,\(if .extra.metadata.assignees then .extra.metadata.assignees else "null" end | sub("\n";" ";"g"))"' \
-        | grep -f $SCRIPTPATH/semgrep_rules/blocklist.txt -v \
         | $SCRIPTPATH/cleaner.rb --assignees --sveltegrep && \
       find . -type f -name '*.extractedscript.*' -delete) 2> reviewdog.sveltegrep.stderr.log
     errorformat:
@@ -56,7 +54,6 @@ runner:
     cmd: |
       set -e
       (xargs -0 -n1 -a $SCRIPTPATH/all_changed_files.txt $SCRIPTPATH/xmllint.sh \
-      | egrep -f $SCRIPTPATH/dtd/blocklist.txt -v \
       | $SCRIPTPATH/cleaner.rb --svgo) 2> reviewdog.safesvg.stderr.log
     errorformat:
       - "%f:%l: %m"