All doc conversion to HTML (#2171)

.github/scripts/convert_documents.py

@@ -0,0 +1,76 @@
+#! /usr/bin/env python3
+
+import io
+import json
+import os
+import markdown as md
+from weasyprint import HTML
+from pathlib import Path
+from functions import findAllFiles
+
+
+def convertMdToHTML(file_path):
+    output = io.BytesIO()
+    try:
+        # Parse the Markdown file
+        md.markdownFromFile(
+            input=str(file_path),
+            output=output,
+            extensions=[
+                "markdown.extensions.attr_list",
+                "markdown.extensions.toc",
+                "tables",
+            ],
+        )
+        content = output.getvalue().decode("utf-8")
+    except OSError:
+        content = None
+    finally:
+        output.close()
+
+    return content
+
+
+def main():
+    script_path = os.path.dirname(__file__)
+    root_path = os.path.abspath(os.path.join(script_path, os.pardir, os.pardir))
+
+    # Get list of documents to convert
+    with open(os.path.join(script_path, "convert_sources.json")) as f:
+        json_data = json.load(f)
+        pdf_files = json_data["pdf"]
+        html_files = json_data["html"]
+        files_to_convert = list(set(pdf_files + html_files))
+
+    files_list = findAllFiles(root_path)
+
+    for locale, filenames in files_list.items():
+        for f in filenames:
+            if f in files_to_convert:
+                locale_folder = os.path.join(root_path, locale)
+                source_file = os.path.join(locale_folder, f)
+
+                # Convert Markdown to HTML
+                html_content = convertMdToHTML(source_file)
+
+                # Save HTML file if requested, creating `html` folder if missing
+                if f in html_files:
+                    Path(os.path.join(locale_folder, "html")).mkdir(exist_ok=True)
+                    html_dest_file = os.path.join(
+                        locale_folder, "html", f"{f.rstrip('.md')}.html"
+                    )
+                    with open(html_dest_file, "w", encoding="utf-8") as f:
+                        f.write(html_content)
+
+                # Save PDF file if requested, creating `pdf` folder if missing
+                if f in pdf_files:
+                    Path(os.path.join(locale_folder, "pdf")).mkdir(exist_ok=True)
+                    pdf_dest_file = os.path.join(
+                        locale_folder, "pdf", f"{f.rstrip('.md')}.pdf"
+                    )
+                    # Convert HTML to PDF
+                    HTML(string=html_content).write_pdf(pdf_dest_file)
+
+
+if __name__ == "__main__":
+    main()

.github/scripts/pdf_sources.json → .github/scripts/convert_sources.json

@@ -1,5 +1,6 @@
 {
-    "files": [
+    "html": [],
+    "pdf": [
         "firefox_relay_privacy_notice.md",
         "firefox_relay_tos.md",
         "hubs_privacy_notice.md",

.github/workflows/pdf_generation.yml → .github/workflows/doc_generation.yml

@@ -1,4 +1,4 @@
-name: PDF generation
+name: Convert documents to PDF/HTML
 
 on:
   push:
@@ -7,7 +7,7 @@ on:
     paths:
       - '**.md'
       - '**.json'
-      - '.github/workflows/pdf_generation.yml'
+      - '.github/workflows/doc_generation.yml'
       - '.github/requirements.txt'
       # Ignore Markdown files in the root
       - '!*.md'
@@ -16,7 +16,7 @@ on:
 
 jobs:
   fetch:
-    name: Generate PDF
+    name: Generate documents
     runs-on: ubuntu-latest
     steps:
       - name: Install Linux packages
@@ -41,36 +41,36 @@ jobs:
           # Set arbitrary EPOCH date to get reproducible PDF generation.
           # https://github.com/Kozea/WeasyPrint/issues/1553
           export SOURCE_DATE_EPOCH=1658210455
-          python .github/scripts/convert_to_pdf.py
+          python .github/scripts/convert_documents.py
 
       - name: Set up git credentials
         run : |
           git config --global user.email '[email protected]'
           git config --global user.name 'MozMEAO Bot'
 
-      - name: Commit changes to pdf_updates branch
+      - name: Commit changes to doc_updates branch
         continue-on-error: true
         run: |
           # Commit changes. Failure is allowed if there is nothing to commit.
-          git checkout -B pdf_updates
+          git checkout -B doc_updates
           git add .
-          git commit -m "Convert documents to PDF"
-          git push -f origin pdf_updates
+          git commit -m "Convert documents"
+          git push -f origin doc_updates
 
       - name: Open pull request
         continue-on-error: true
         run: |
           # continue-on-error is needed in case there is already an open PR
-          changes=$(git diff --name-only main pdf_updates -- | wc -l | awk '{print $1}')
+          changes=$(git diff --name-only main doc_updates -- | wc -l | awk '{print $1}')
           if [[ "$changes" = "0" ]];
           then
             echo "No changes."
           else
             # Create pull request
             gh pr create \
-              --title "[pdf] Update documents in PDF format" \
-              --body "Convert documents to PDF" \
-              --head pdf_updates \
+              --title "[pdf] Update converted documents" \
+              --body "Convert documents to PDF/HTML" \
+              --head doc_updates \
               --base main \
               --label l10n \
               --reviewer @mozilla/legal-l10n