#!/bin/bash
# Check for hardcoded Russian/Cyrillic user-facing strings in source code.
# Excludes locale files, docs, tests with explicit locale checks.

set -e

ROOT="$(cd "$(dirname "$0")/.." && pwd)"
EXIT=0

# Find Cyrillic characters in source files, excluding allowed paths.
# The regex matches any Cyrillic character range.
# Allowed exceptions:
#   - locale files (*/i18n/locales/*)
#   - docs/*
#   - README*
#   - spaces/
#   - .json files that are templates or configs
#   - .md files

echo "=== Checking for hardcoded Cyrillic in source code ==="

# Search for Cyrillic characters in Go files (excluding locale files)
GO_CYRILLIC=$(find "$ROOT" -name '*.go' \
  ! -path "*/i18n/locales/*" \
  -exec grep -l '[А-Яа-я]' {} \; 2>/dev/null || true)

if [ -n "$GO_CYRILLIC" ]; then
  echo "WARNING: Cyrillic found in Go files (expected in server HTML templates for now):"
  echo "$GO_CYRILLIC"
  # Don't fail for Go files with HTML templates — they'll be refactored later
fi

# Search for Cyrillic in Svelte/JS files (excluding locale files)
JS_CYRILLIC=$(find "$ROOT/frontend/src" -name '*.svelte' -o -name '*.js' | \
  grep -v 'i18n/locales' | \
  xargs grep -l '[А-Яа-я]' 2>/dev/null || true)

if [ -n "$JS_CYRILLIC" ]; then
  echo ""
  echo "FAIL: Cyrillic found in frontend source files (outside locale files):"
  echo "$JS_CYRILLIC"
  echo ""
  echo "These should use t() from lib/i18n instead."
  EXIT=1
fi

# Check for common bidi/control unicode characters
echo ""
echo "=== Checking for bidi/control Unicode characters ==="
BIDI=$(find "$ROOT" -name '*.go' -o -name '*.svelte' -o -name '*.js' | \
  xargs grep -Pl '[\x{202A}-\x{202E}\x{2066}-\x{2069}\x{200E}\x{200F}\x{061C}]' 2>/dev/null || true)

if [ -n "$BIDI" ]; then
  echo "FAIL: Bidi/control Unicode characters found in:"
  echo "$BIDI"
  EXIT=1
else
  echo "OK: No bidi/control characters found"
fi

# Check that locale keys in ru.js and en.js match
echo ""
echo "=== Checking locale key consistency ==="
RU_KEYS=$(grep -oP "^\s+'[^']+'" "$ROOT/frontend/src/lib/i18n/locales/ru.js" | sed "s/^ *'//;s/'$//" | sort)
EN_KEYS=$(grep -oP "^\s+'[^']+'" "$ROOT/frontend/src/lib/i18n/locales/en.js" | sed "s/^ *'//;s/'$//" | sort)

MISSING_EN=$(comm -23 <(echo "$RU_KEYS") <(echo "$EN_KEYS"))
MISSING_RU=$(comm -23 <(echo "$EN_KEYS") <(echo "$RU_KEYS"))

if [ -n "$MISSING_EN" ]; then
  echo "WARNING: Keys in ru.js but missing in en.js:"
  echo "$MISSING_EN"
fi
if [ -n "$MISSING_RU" ]; then
  echo "WARNING: Keys in en.js but missing in ru.js:"
  echo "$MISSING_RU"
fi
if [ -z "$MISSING_EN" ] && [ -z "$MISSING_RU" ]; then
  echo "OK: All locale keys match between ru.js and en.js"
fi

exit $EXIT