Add i18n support for web UI (German + English)

src/document_anonymizer/i18n/__init__.py

@@ -0,0 +1,88 @@
+"""Internationalization support for the web UI."""
+
+from __future__ import annotations
+
+import json
+from functools import lru_cache
+from pathlib import Path
+from typing import TYPE_CHECKING, Any, Literal, TypeGuard, get_args
+
+if TYPE_CHECKING:
+    from collections.abc import Mapping
+
+import structlog
+from jinja2 import pass_context
+
+logger = structlog.get_logger(__name__)
+
+Lang = Literal["de", "en"]
+SUPPORTED_LANGUAGES: set[str] = {"de", "en"}
+DEFAULT_LANGUAGE: Lang = "de"
+
+# Ensure Lang type and SUPPORTED_LANGUAGES stay in sync
+if set(get_args(Lang)) != SUPPORTED_LANGUAGES:  # pragma: no cover
+    msg = "Lang type and SUPPORTED_LANGUAGES are out of sync"
+    raise RuntimeError(msg)
+
+_TRANSLATIONS_DIR = Path(__file__).parent / "translations"
+
+
+def is_supported_lang(value: str) -> TypeGuard[Lang]:
+    """Check if a string is a supported language code (narrows type to Lang)."""
+    return value in SUPPORTED_LANGUAGES
+
+
+def _load_translations(lang: str) -> dict[str, str]:
+    """Normalize unsupported lang codes to default, return cached."""
+    if lang not in SUPPORTED_LANGUAGES:
+        lang = DEFAULT_LANGUAGE
+    return _load_translations_cached(lang)
+
+
+@lru_cache(maxsize=2)  # maxsize matches len(SUPPORTED_LANGUAGES)
+def _load_translations_cached(lang: str) -> dict[str, str]:
+    """Load and cache a translation file."""
+    path = _TRANSLATIONS_DIR / f"{lang}.json"
+    try:
+        with path.open(encoding="utf-8") as f:
+            data: dict[str, str] = json.load(f)
+    except (OSError, json.JSONDecodeError, UnicodeDecodeError):
+        logger.exception("translation_file_load_failed", lang=lang)
+        return {}
+    return data
+
+
+def get_translations(lang: str) -> dict[str, str]:
+    """Get all translations for a language (public API)."""
+    return _load_translations(lang)
+
+
+def translate(
+    key: str, lang: Lang = DEFAULT_LANGUAGE, **kwargs: str | int | float
+) -> str:
+    """Look up a translation key and interpolate any kwargs."""
+    translations = _load_translations(lang)
+    template = translations.get(key)
+    if template is None:
+        logger.warning("translation_key_missing", key=key, lang=lang)
+        return key
+    if kwargs:
+        try:
+            return template.format(**kwargs)
+        except (KeyError, IndexError, ValueError):
+            logger.warning("translation_format_error", key=key, lang=lang)
+            return template
+    return template
+
+
+@pass_context
+def jinja_translate(
+    context: Mapping[str, Any], key: str, **kwargs: str | int | float
+) -> str:
+    """Jinja2 global function: {{ _("key", arg=val) }}.
+
+    The @pass_context decorator injects a jinja2.runtime.Context (Mapping-like).
+    """
+    raw_lang = context.get("lang", DEFAULT_LANGUAGE)
+    lang = raw_lang if is_supported_lang(raw_lang) else DEFAULT_LANGUAGE
+    return translate(key, lang=lang, **kwargs)

src/document_anonymizer/i18n/translations/de.json

@@ -0,0 +1,54 @@
+{
+  "brand.name": "Dokument-Anonymisierer",
+  "brand.subtitle": "Deutsche PII-Erkennung & Schwärzung",
+  "index.heading": "Dokument anonymisieren",
+  "index.subtitle": "Text eingeben oder Datei hochladen. Alle Daten werden nur im Arbeitsspeicher verarbeitet — keine Persistenz.",
+  "index.text_label": "Text",
+  "index.example_btn": "Beispieltext laden",
+  "index.file_label": "Oder Datei hochladen (TXT, PDF)",
+  "index.threshold_label": "Konfidenzschwelle",
+  "index.detect_btn": "PII erkennen",
+  "index.loading": "Analysiere...",
+  "index.placeholder": "Geben Sie hier den zu anonymisierenden Text ein...\n\nBeispiel: Herr Max Mustermann, geboren am 15.03.1985, wohnhaft in 10115 Berlin, Musterstraße 42. IBAN: DE89 3704 0044 0532 0130 00. Steuer-ID: 12345679811. Tel: +49 30 12345678.",
+  "results.heading": "Erkennungsergebnisse",
+  "results.entities_found": "{count} Entität(en) gefunden",
+  "results.preview_heading": "Erkannte PII im Text",
+  "results.review_heading": "Entitäten prüfen und auswählen",
+  "results.tier_high": "Hohe Konfidenz",
+  "results.tier_medium": "Mittlere Konfidenz",
+  "results.tier_low": "Niedrige Konfidenz",
+  "results.strategy_label": "Strategie:",
+  "results.anonymize_btn": "Anonymisieren",
+  "results.redact_btn": "PDF schwärzen",
+  "results.loading": "Anonymisiere...",
+  "results.no_entities": "Keine PII-Entitäten erkannt.",
+  "anonymized.heading": "Ergebnis",
+  "anonymized.entities_anonymized": "{count} Entität(en) anonymisiert",
+  "anonymized.strategy": "Strategie: {strategy}",
+  "anonymized.original_heading": "Original (mit Markierungen)",
+  "anonymized.result_heading": "Anonymisiert",
+  "anonymized.copy_btn": "Text kopieren",
+  "anonymized.download_pdf_btn": "PDF mit Schwärzungen herunterladen",
+  "noscript.text": "JavaScript ist für die interaktive Oberfläche erforderlich. Bitte aktivieren Sie JavaScript oder nutzen Sie die",
+  "noscript.link": "REST API",
+  "footer.text": "Zero-Persistence-Architektur · Keine externen API-Aufrufe · Physische PDF-Schwärzung",
+  "error.no_input": "Bitte Text eingeben oder Datei hochladen.",
+  "error.unknown_strategy": "Unbekannte Strategie: {strategy}",
+  "error.detection_failed": "Fehler bei der PII-Erkennung. (Referenz: {request_id})",
+  "error.anonymization_failed": "Fehler bei der Anonymisierung. (Referenz: {request_id})",
+  "error.entity_parse_failed": "Entitätsauswahl konnte nicht verarbeitet werden.",
+  "error.entity_invalid_format": "Entitätsauswahl hat ein ungültiges Format.",
+  "error.entity_skipped": "{skipped} von {total} ausgewählten Entitäten konnten nicht verarbeitet werden. Bitte erneut versuchen.",
+  "error.pdf_redaction_failed": "PDF-Schwärzung fehlgeschlagen. (Referenz: {request_id})",
+  "error.invalid_pdf": "Ungültige PDF-Daten. Bitte laden Sie die Datei erneut hoch.",
+  "error.incomplete_redaction": "Unvollständige Schwärzung: {unredacted} von {total} erkannten PII-Entitäten konnten im PDF nicht visuell lokalisiert werden. Manuelle Überprüfung empfohlen.",
+  "common.confidence": "Konfidenz: {score}",
+  "review.entities_selected": "{count} von {total} Entitäten ausgewählt",
+  "review.no_selection_warning": "Keine Entitäten ausgewählt. Trotzdem fortfahren?",
+  "review.entity_load_error": "Fehler beim Laden der Entitätsdaten. Bitte führen Sie die Erkennung erneut durch.",
+  "review.download_error": "Fehler beim Herunterladen der PDF.",
+  "app.example_loaded": "Geladen!",
+  "app.example_error": "Fehler!",
+  "app.copied": "Kopiert!",
+  "app.copy_failed": "Kopieren fehlgeschlagen"
+}

src/document_anonymizer/i18n/translations/en.json

@@ -0,0 +1,54 @@
+{
+  "brand.name": "Document Anonymizer",
+  "brand.subtitle": "German PII Detection & Redaction",
+  "index.heading": "Anonymize document",
+  "index.subtitle": "Enter text or upload a file. All data is processed in memory only — no persistence.",
+  "index.text_label": "Text",
+  "index.example_btn": "Load example text",
+  "index.file_label": "Or upload file (TXT, PDF)",
+  "index.threshold_label": "Confidence threshold",
+  "index.detect_btn": "Detect PII",
+  "index.loading": "Analyzing...",
+  "index.placeholder": "Enter text to anonymize here...\n\nExample: Herr Max Mustermann, born on 15.03.1985, residing at 10115 Berlin, Musterstraße 42. IBAN: DE89 3704 0044 0532 0130 00. Tax ID: 12345679811. Tel: +49 30 12345678.",
+  "results.heading": "Detection results",
+  "results.entities_found": "{count} entity(ies) found",
+  "results.preview_heading": "Detected PII in text",
+  "results.review_heading": "Review and select entities",
+  "results.tier_high": "High confidence",
+  "results.tier_medium": "Medium confidence",
+  "results.tier_low": "Low confidence",
+  "results.strategy_label": "Strategy:",
+  "results.anonymize_btn": "Anonymize",
+  "results.redact_btn": "Redact PDF",
+  "results.loading": "Anonymizing...",
+  "results.no_entities": "No PII entities detected.",
+  "anonymized.heading": "Result",
+  "anonymized.entities_anonymized": "{count} entity(ies) anonymized",
+  "anonymized.strategy": "Strategy: {strategy}",
+  "anonymized.original_heading": "Original (with highlights)",
+  "anonymized.result_heading": "Anonymized",
+  "anonymized.copy_btn": "Copy text",
+  "anonymized.download_pdf_btn": "Download redacted PDF",
+  "noscript.text": "JavaScript is required for the interactive interface. Please enable JavaScript or use the",
+  "noscript.link": "REST API",
+  "footer.text": "Zero-persistence architecture · No external API calls · Physical PDF redaction",
+  "error.no_input": "Please enter text or upload a file.",
+  "error.unknown_strategy": "Unknown strategy: {strategy}",
+  "error.detection_failed": "PII detection failed. (Reference: {request_id})",
+  "error.anonymization_failed": "Anonymization failed. (Reference: {request_id})",
+  "error.entity_parse_failed": "Entity selection could not be processed.",
+  "error.entity_invalid_format": "Entity selection has an invalid format.",
+  "error.entity_skipped": "{skipped} of {total} selected entities could not be processed. Please try again.",
+  "error.pdf_redaction_failed": "PDF redaction failed. (Reference: {request_id})",
+  "error.invalid_pdf": "Invalid PDF data. Please upload the file again.",
+  "error.incomplete_redaction": "Incomplete redaction: {unredacted} of {total} detected PII entities could not be visually located in the PDF. Manual review recommended.",
+  "common.confidence": "Confidence: {score}",
+  "review.entities_selected": "{count} of {total} entities selected",
+  "review.no_selection_warning": "No entities selected. Continue anyway?",
+  "review.entity_load_error": "Error loading entity data. Please run detection again.",
+  "review.download_error": "Error downloading PDF.",
+  "app.example_loaded": "Loaded!",
+  "app.example_error": "Error!",
+  "app.copied": "Copied!",
+  "app.copy_failed": "Copy failed"
+}