#!/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

# Explicitly allowed Go files that may contain Cyrillic (test data, SQL migrations, i18n catalog).
# These are NOT user-facing strings — they are test fixtures, embedded SQL, or locale definitions.
ALLOWED_GO_CYRILLIC=(
  "internal/core/worklog/worklog_test.go"
  "internal/core/worklog/worklog.go"
  "internal/core/smoke_test.go"
  "internal/core/nodes/types.go"
  "internal/core/nodes/repository_test.go"
  "internal/core/plugins/manager_test.go"
  "internal/core/actions/action_test.go"
  "internal/core/actions/action.go"
  "internal/core/files/file.go"
  "internal/core/storage/migrations_008.sql.go"
  "internal/gui/index.html.go"
  "internal/i18n/catalog.go"
  "cmd/verstak-gui/main.go"
  "internal/core/templates/safename.go"
  "internal/core/templates/safename_test.go"
  "cmd/verstak-gui/vault_layout_test.go"
)

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)

# Filter out allowed files
FILTERED=""
for f in $GO_CYRILLIC; do
  rel="${f#$ROOT/}"
  skip=0
  for allowed in "${ALLOWED_GO_CYRILLIC[@]}"; do
    if [ "$rel" = "$allowed" ]; then
      skip=1
      break
    fi
  done
  if [ "$skip" -eq 0 ]; then
    FILTERED="$FILTERED $f"
  fi
done

if [ -n "$FILTERED" ]; then
  echo ""
  echo "FAIL: Cyrillic found in Go source files (outside allowed exceptions):"
  echo "$FILTERED"
  echo ""
  echo "These should use i18n.T() instead of hardcoded Russian strings."
  EXIT=1
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 match between ru and en for frontend AND Go
echo ""
echo "=== Checking locale key consistency ==="

# Frontend JS locales
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 "$MISSING_EN" ]; then
  echo "FAIL: Keys in frontend ru.js but missing in en.js:"
  echo "$MISSING_EN"
  EXIT=1
fi
if [ -n "$MISSING_RU" ]; then
  echo "FAIL: Keys in frontend en.js but missing in ru.js:"
  echo "$MISSING_RU"
  EXIT=1
fi
if [ -z "$MISSING_EN" ] && [ -z "$MISSING_RU" ]; then
  echo "OK: All frontend locale keys match between ru.js and en.js"
fi

# Go locales
GO_RU_KEYS=$(grep -oP '"[^"]+":\s*"' "$ROOT/internal/i18n/locales/ru.json" | sed 's/":.*//' | sed 's/"//g' | sort)
GO_EN_KEYS=$(grep -oP '"[^"]+":\s*"' "$ROOT/internal/i18n/locales/en.json" | sed 's/":.*//' | sed 's/"//g' | sort)

GO_MISSING_EN=$(comm -23 <(echo "$GO_RU_KEYS") <(echo "$GO_EN_KEYS"))
GO_MISSING_RU=$(comm -23 <(echo "$GO_EN_KEYS") <(echo "$GO_RU_KEYS"))

if [ -n "$GO_MISSING_EN" ]; then
  echo "FAIL: Keys in Go ru.json but missing in en.json:"
  echo "$GO_MISSING_EN"
  EXIT=1
fi
if [ -n "$GO_MISSING_RU" ]; then
  echo "FAIL: Keys in Go en.json but missing in ru.json:"
  echo "$GO_MISSING_RU"
  EXIT=1
fi
if [ -z "$GO_MISSING_EN" ] && [ -z "$GO_MISSING_RU" ]; then
  echo "OK: All Go locale keys match between ru.json and en.json"
fi

exit $EXIT