#3414 webpage: add mechanism to deal with corrupted translation files

Signed-off-by: Patrizio Bekerle <[email protected]>

Author: pbek

webpage/scripts/find-corrupted-files.sh

@@ -0,0 +1,55 @@
+#!/bin/sh
+
+# Script to identify markdown files with severe HTML corruption
+# These files will be excluded from the build
+
+echo "🔍 Scanning for corrupted translation files..." >&2
+
+# Array to store corrupted files (sh compatible)
+corrupted_files=""
+
+# Find all markdown files in translation directories (any 2-letter language code)
+find src -name "*.md" -type f 2>/dev/null | grep -E 'src/[a-z]{2}(-[A-Z]{2})?/' | while IFS= read -r file; do
+  # Check for common corruption patterns
+  has_error=false
+  error_details=""
+
+  # Pattern 1: Tags with numbers like </0>, </1> or <0>, <1>, including with spaces like </0 >
+  if grep -qE '</?[0-9]+\s*>' "$file" 2>/dev/null || grep -qE '</\s*[0-9]+\s*>' "$file" 2>/dev/null; then
+    has_error=true
+    error_details="${error_details}numeric tag (<0>, </0>, </0 >, etc); "
+  fi
+
+  # Pattern 2: Unclosed code tags followed by table syntax
+  if grep -q '<code>[^<]*`.*|' "$file" 2>/dev/null; then
+    has_error=true
+    error_details="${error_details}unclosed <code> with table syntax; "
+  fi
+
+  # Pattern 3: Mixing HTML table with markdown table syntax
+  if grep -q '<table' "$file" 2>/dev/null && grep -q '<code>.*|.*|' "$file" 2>/dev/null; then
+    has_error=true
+    error_details="${error_details}mixed HTML/markdown table; "
+  fi
+
+  # Pattern 4: Orphaned closing paragraph tags or other malformed closing tags
+  if grep -qE '^[^<]*</p>|</[0-9 ]+>' "$file" 2>/dev/null; then
+    has_error=true
+    error_details="${error_details}orphaned/malformed closing tag; "
+  fi
+
+  # Pattern 5: Opening tags with numbers
+  if grep -qE '<[0-9]+[^>]*>' "$file" 2>/dev/null; then
+    has_error=true
+    error_details="${error_details}numeric opening tag; "
+  fi
+
+  if [ "$has_error" = true ]; then
+    corrupted_files="$corrupted_files$file
+"
+    echo "  ⚠️  $file" >&2
+    echo "      └─ Errors: $error_details" >&2
+    # Output to stdout for the build script to capture
+    echo "$file"
+  fi
+done

webpage/scripts/run-build.sh

@@ -2,9 +2,41 @@
 
 set -e
 
-echo "⚠️  Removing severely corrupted translation files..."
-# Remove files that are too corrupted to fix automatically
-rm -f src/ko/getting-started/markdown.md && echo "  ✓ Removed src/ko/getting-started/markdown.md" || true
+echo "⚠️  Replacing corrupted translation files with English versions..."
+# Replace corrupted files with English versions instead of removing them
+# This allows the build to complete successfully while corrupted translations are being fixed
+
+replace_with_english() {
+  corrupted_file="$1"
+  # Extract the path after the language code (e.g., src/ko/editor/time-formats.md -> editor/time-formats.md)
+  relative_path=$(echo "$corrupted_file" | sed -E 's|^src/[a-z]{2}(-[A-Z]{2})?/||')
+  english_file="src/$relative_path"
+
+  if [ -f "$english_file" ]; then
+    cp "$english_file" "$corrupted_file" && echo "  ✓ Replaced $corrupted_file with English version"
+  else
+    rm -f "$corrupted_file" && echo "  ✓ Removed $corrupted_file (no English version found)"
+  fi
+}
+
+# Run detection script and replace found files
+if [ -x "./scripts/find-corrupted-files.sh" ]; then
+  corrupted_files=$(./scripts/find-corrupted-files.sh | grep "^src/")
+  if [ -n "$corrupted_files" ]; then
+    echo "$corrupted_files" | while read -r file; do
+      if [ -f "$file" ]; then
+        replace_with_english "$file"
+      fi
+    done
+  fi
+else
+  # Fallback: manually specified files
+  [ -f "src/ko/getting-started/markdown.md" ] && replace_with_english "src/ko/getting-started/markdown.md"
+  [ -f "src/ko/editor/time-formats.md" ] && replace_with_english "src/ko/editor/time-formats.md"
+  [ -f "src/fa/editor/time-formats.md" ] && replace_with_english "src/fa/editor/time-formats.md"
+  [ -f "src/fa/contributing/translation.md" ] && replace_with_english "src/fa/contributing/translation.md"
+  [ -f "src/it/getting-started/overview.md" ] && replace_with_english "src/it/getting-started/overview.md"
+fi
 
 #echo ""
 #echo "🔧 Fixing HTML issues in markdown files..."