pre-processed spellcheck output

Signed-off-by: Frederic BIDON <[email protected]>

Author: fredbi

.github/workflows/doc-update.yml

@@ -25,6 +25,7 @@ name: doc update check
 #     * [ ] should be able to retrieve config files and dictionary from the called ref, not master.
 #     * [ ] should be able to merge config files and dictionary with local definitions on target repo.
 #     * [ ] should be able to work on diff
+#     * [ ] should format the output of spellcheck report
 
 on:
   workflow_call:
@@ -122,6 +123,7 @@ jobs:
       if: ${{ steps.markdownlint.outcome != 'success' && hashFiles(env.lintreport) != '' }}
       id: report-exists
       run: |
+        # summarizes a bit the output from spellcheck
         echo 'report<<EOF' >> $GITHUB_OUTPUT
         cat ${{ env.lintreport }}|sed -e '$a\' >> $GITHUB_OUTPUT
         echo 'EOF' >> $GITHUB_OUTPUT
@@ -423,13 +425,33 @@ jobs:
       reactions: ${{ steps.notify_spelling_report.outputs.reactions }}
     runs-on: ubuntu-latest
     steps:
+    - name: Pre-process spellcheck report
+      id: preprocess
+      env: 
+        MSG: ${{ needs.markdown-spelling.outputs.report }}
+        SED_CMD: .github/workflows/filter.sed # a sed script to parse and reformat the spellcheck report
+        JQ_CMD: .github/workflows/merge.jq # a jq script to dedupe spellcheck reports by file
+        TMPFILE: /tmp/spellcheck-report.txt
+      run: |
+        export MSG
+        printenv MSG > "${TMPFILE}"
+        # produces a JSON object:
+        # {
+        #   "file.md": [ mispelled word [, ...]]
+        # }
+        cat "${TMPFILE}" | sed -n  -f "${SED_CMD}" | jq -s "${JQ_CMD}" > /tmp/preprocessed.txt
+
+        echo 'report<<EOF' >> $GITHUB_OUTPUT
+        cat /tmp/preprocessed.txt|sed -e '$a\' >> $GITHUB_OUTPUT
+        echo 'EOF' >> $GITHUB_OUTPUT
+
     - name: Format PR comment
       id: comment_formatter
       uses: skills/action-text-variables@v3
       with:
         template-vars: >
           {
-            "text": ${{ toJSON(needs.markdown-spelling.outputs.report) }}
+            "text": ${{ steps.preprocess.outputs.report) }}
           }
         template-text: |
           ### ${{ env.spellcheck_comment_title }}
@@ -442,6 +464,7 @@ jobs:
           <br> 
 
           ```
+          ### TODO: range json
           {{ text }}
           ```
 

.github/workflows/filter.sed

@@ -0,0 +1,60 @@
+:scan_input {
+    # main scanner loop, looking for "Misspelled words:" entries
+    /^Misspelled words:/,/^-\{5,\}/ {
+        /^<html\(attribute\|content\)&gt\;/ {
+            # Extract the file name
+            # TODO: be less strict here
+            s/^<html\(attribute\|content\)&gt\;\s\+\(.\+\.md\):.*$/"\2": [/
+            h
+            n  # skip current match
+            n  # skip next line
+    
+            # go decode a section of single words, enclosed between "=====..." lines
+            b parse_section
+        }
+    
+        n
+    
+        b scan_input
+    }
+}
+
+# Back to main loop (cycles)
+d
+
+:parse_section {
+    :word /^[^-]/ {
+        # Append words to the previous line
+        # Trim space around words
+        s/\s\+//g
+        # Quote word
+        s/^/"/
+        s/$/"/
+
+        # Add comma separator
+        s/$/, /
+        H
+        n
+
+        b word
+    }
+
+    /^-\{5,\}$/ {
+        # Print the collected words as a comma-separated array of words
+        x
+
+        s/^/{/g
+        s/\n/ /g
+        s/\s\+/ /g
+        s/\s\+$//g
+        s/\(,\s*\)\?$/ ]/
+        s/$/}/g
+
+        p
+
+        # Start a new cycle
+        d
+    }
+}
+
+d

.github/workflows/merge.jq

@@ -0,0 +1,4 @@
+reduce (.[] | to_entries | .[]) as {$key, $value} (
+  {} ;
+  .[$key] += $value
+) | .[] |= unique