ol-openedx-course-translations 0.1.0__py3-none-any.whl → 0.3.5__py3-none-any.whl

ol_openedx_course_translations/middleware.py

@@ -0,0 +1,143 @@
+"""
+Middleware to set/reset language preference cookie and
+user preference based on course language.
+"""
+
+import re
+
+from django.conf import settings
+from django.http import HttpResponseRedirect
+from django.utils.deprecation import MiddlewareMixin
+from opaque_keys.edx.keys import CourseKey
+from openedx.core.djangoapps.content.course_overviews.models import CourseOverview
+from openedx.core.djangoapps.lang_pref import LANGUAGE_KEY
+from openedx.core.djangoapps.lang_pref import helpers as lang_pref_helpers
+from openedx.core.djangoapps.user_api.preferences.api import set_user_preference
+
+from ol_openedx_course_translations.utils.constants import ENGLISH_LANGUAGE_CODE
+
+
+def should_process_request(request):
+    """
+    Return True if language auto-selection should run for this request.
+    """
+    return (
+        settings.ENABLE_AUTO_LANGUAGE_SELECTION
+        and hasattr(request, "user")
+        and request.user.is_authenticated
+    )
+
+
+def set_language(request, response, language):
+    """
+    Set both cookie and user preference for language.
+    """
+    lang_pref_helpers.set_language_cookie(request, response, language)
+    set_user_preference(request.user, LANGUAGE_KEY, language)
+
+
+def redirect_current_path(request):
+    """
+    Redirect to the same URL to ensure language change takes effect.
+    """
+    return HttpResponseRedirect(request.get_full_path())
+
+
+class CourseLanguageCookieMiddleware(MiddlewareMixin):
+    """
+    LMS middleware that:
+    - Sets language based on course language
+    - Forces English for exempt paths and authoring MFEs
+    """
+
+    COURSE_URL_REGEX = re.compile(
+        rf"^/courses/(?P<course_key>{settings.COURSE_KEY_REGEX})(?:/|$)",
+        re.IGNORECASE,
+    )
+
+    def process_response(self, request, response):
+        """
+        Process the response to set/reset language cookie based on course language.
+        """
+        if not should_process_request(request):
+            return response
+
+        path = getattr(request, "path_info", request.path)
+
+        if self._should_force_english(request, path):
+            return self._force_english_if_needed(request, response)
+
+        course_language = self._get_course_language(path)
+        if not course_language:
+            return response
+
+        return self._apply_course_language(request, response, course_language)
+
+    def _should_force_english(self, request, path):
+        """
+        Determine if English should be forced based on request origin or exempt paths.
+        """
+        return request.META.get(
+            "HTTP_ORIGIN"
+        ) == settings.COURSE_AUTHORING_MICROFRONTEND_URL or any(
+            exempt_path in path
+            for exempt_path in settings.AUTO_LANGUAGE_SELECTION_EXEMPT_PATHS
+        )
+
+    def _force_english_if_needed(self, request, response):
+        """
+        Force language to English if not already set.
+        """
+        cookie_val = lang_pref_helpers.get_language_cookie(request)
+
+        if cookie_val != ENGLISH_LANGUAGE_CODE:
+            set_language(request, response, ENGLISH_LANGUAGE_CODE)
+            return redirect_current_path(request)
+
+        return response
+
+    def _get_course_language(self, path):
+        """
+        Extract course language from the course URL path.
+        """
+        match = self.COURSE_URL_REGEX.match(path)
+        if not match:
+            return None
+
+        try:
+            course_key = CourseKey.from_string(match.group("course_key"))
+            overview = CourseOverview.get_from_id(course_key)
+        except Exception:  # noqa: BLE001
+            return None
+
+        return getattr(overview, "language", None)
+
+    def _apply_course_language(self, request, response, language):
+        """
+        Apply the course language if it differs from the current cookie value.
+        """
+        cookie_val = lang_pref_helpers.get_language_cookie(request)
+        if cookie_val != language:
+            set_language(request, response, language)
+            return redirect_current_path(request)
+
+        return response
+
+
+class CourseLanguageCookieResetMiddleware(MiddlewareMixin):
+    """
+    CMS middleware that always resets language to English.
+    """
+
+    def process_response(self, request, response):
+        """
+        Process the response to reset language cookie to English.
+        """
+        if not should_process_request(request):
+            return response
+
+        cookie_val = lang_pref_helpers.get_language_cookie(request)
+        if cookie_val and cookie_val != ENGLISH_LANGUAGE_CODE:
+            set_language(request, response, ENGLISH_LANGUAGE_CODE)
+
+        return response

ol_openedx_course_translations/migrations/0001_add_translation_logs.py

@@ -0,0 +1,84 @@
+# Generated by Django 5.2.9 on 2026-01-15 11:04
+
+import opaque_keys.edx.django.models
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    initial = True
+
+    dependencies = []  # type: ignore[var-annotated]
+
+    operations = [
+        migrations.CreateModel(
+            name="CourseTranslationLog",
+            fields=[
+                (
+                    "id",
+                    models.AutoField(
+                        auto_created=True,
+                        primary_key=True,
+                        serialize=False,
+                        verbose_name="ID",
+                    ),
+                ),
+                (
+                    "source_course_id",
+                    opaque_keys.edx.django.models.CourseKeyField(
+                        db_index=True, max_length=255
+                    ),
+                ),
+                (
+                    "source_course_language",
+                    models.CharField(
+                        help_text="Source language code (e.g., 'EN')", max_length=10
+                    ),
+                ),
+                (
+                    "target_course_language",
+                    models.CharField(
+                        help_text="Target language code for translation (e.g., 'FR')",
+                        max_length=10,
+                    ),
+                ),
+                (
+                    "srt_provider_name",
+                    models.CharField(
+                        help_text="LLM Provider used for SRT translation",
+                        max_length=100,
+                    ),
+                ),
+                (
+                    "srt_provider_model",
+                    models.CharField(
+                        blank=True,
+                        help_text="LLM provider model used for SRT translation",
+                        max_length=100,
+                    ),
+                ),
+                (
+                    "content_provider_name",
+                    models.CharField(
+                        help_text="LLM Provider used for content translation",
+                        max_length=100,
+                    ),
+                ),
+                (
+                    "content_provider_model",
+                    models.CharField(
+                        blank=True,
+                        help_text="LLM provider model used for content translation",
+                        max_length=100,
+                    ),
+                ),
+                (
+                    "command_stats",
+                    models.TextField(
+                        blank=True, help_text="Logs from the translation command"
+                    ),
+                ),
+                ("created_at", models.DateTimeField(auto_now_add=True)),
+                ("updated_at", models.DateTimeField(auto_now=True)),
+            ],
+        ),
+    ]

ol_openedx_course_translations/models.py

@@ -0,0 +1,57 @@
+"""Models for course translations plugin"""
+
+from django.db import models
+from opaque_keys.edx.django.models import CourseKeyField
+
+
+class CourseTranslationLog(models.Model):
+    """Log entry for course translation operations."""
+
+    source_course_id = CourseKeyField(max_length=255, db_index=True)
+    source_course_language = models.CharField(
+        max_length=10,
+        help_text="Source language code (e.g., 'EN')",
+    )
+    target_course_language = models.CharField(
+        max_length=10,
+        help_text="Target language code for translation (e.g., 'FR')",
+    )
+    srt_provider_name = models.CharField(
+        max_length=100,
+        help_text="LLM Provider used for SRT translation",
+    )
+    srt_provider_model = models.CharField(
+        max_length=100,
+        blank=True,
+        help_text="LLM provider model used for SRT translation",
+    )
+    content_provider_name = models.CharField(
+        max_length=100,
+        help_text="LLM Provider used for content translation",
+    )
+    content_provider_model = models.CharField(
+        max_length=100,
+        blank=True,
+        help_text="LLM provider model used for content translation",
+    )
+    command_stats = models.TextField(
+        blank=True, help_text="Logs from the translation command"
+    )
+    created_at = models.DateTimeField(
+        auto_now_add=True,
+    )
+    updated_at = models.DateTimeField(
+        auto_now=True,
+    )
+
+    class Meta:
+        """Meta options for CourseTranslationLog."""
+
+        app_label = "ol_openedx_course_translations"
+
+    def __str__(self):
+        """Return a string representation of the translation log."""
+        return (
+            f"{self.source_course_id} "
+            f"({self.source_course_language} → {self.target_course_language})"
+        )

ol_openedx_course_translations/providers/__init__.py

@@ -0,0 +1 @@
+"""Translation providers for course content."""

ol_openedx_course_translations/providers/base.py

@@ -0,0 +1,278 @@
+"""Base classes for translation providers."""
+
+import logging
+from abc import ABC, abstractmethod
+from pathlib import Path
+
+import srt
+
+logger = logging.getLogger(__name__)
+
+MAX_SUBTITLE_TRANSLATION_RETRIES = 1
+
+
+def load_glossary(target_language: str, glossary_directory: str | None = None) -> str:
+    """
+    Load a glossary for the given language from the glossary directory.
+
+    Args:
+        target_language: Target language code
+        glossary_directory: Path to glossary directory
+
+    Returns:
+        Glossary content as string, empty if not found or directory not provided
+    """
+    if not glossary_directory:
+        return ""
+
+    glossary_dir_path = Path(glossary_directory)
+    if not glossary_dir_path.exists() or not glossary_dir_path.is_dir():
+        logger.warning("Glossary directory not found: %s", glossary_dir_path)
+        return ""
+
+    glossary_file_path = glossary_dir_path / f"{target_language.lower()}.txt"
+    if not glossary_file_path.exists():
+        logger.warning(
+            "Glossary file not found for language %s: %s",
+            target_language,
+            glossary_file_path,
+        )
+        return ""
+
+    return glossary_file_path.read_text(encoding="utf-8-sig").strip()
+
+
+class TranslationProvider(ABC):
+    """Abstract base class for translation providers."""
+
+    def __init__(self, primary_api_key: str, repair_api_key: str | None = None):
+        """
+        Initialize translation provider with API keys.
+
+        Args:
+            primary_api_key: API key for primary translation service
+            repair_api_key: API key for repair service (DeepL API key)
+        """
+        self.primary_api_key = primary_api_key
+        self.repair_api_key = repair_api_key
+
+    def translate_srt_with_validation(
+        self,
+        subtitle_list: list[srt.Subtitle],
+        target_language: str,
+        glossary_file: str | None = None,
+    ) -> list[srt.Subtitle]:
+        """
+        Translate SRT subtitles with timestamp validation and repair.
+
+        Performs translation, validates timestamps, and attempts repair
+        if validation fails using DeepL.
+
+        Args:
+            subtitle_list: List of subtitle objects to translate
+            target_language: Target language code
+            glossary_file: Path to glossary directory (optional)
+
+        Returns:
+            List of translated subtitle objects with validated timestamps
+
+        Raises:
+            ValueError: If timestamp repair fails after validation
+        """
+        log = logger.getChild("TranslationProvider")
+        log.info("  🌐 Translating subtitles to %s...", target_language)
+
+        # Try translation with retries
+        translated_subtitles = []
+        for attempt in range(MAX_SUBTITLE_TRANSLATION_RETRIES + 1):
+            if attempt > 0:
+                log.info("  🔧 Retrying subtitle translations...")
+
+            translated_subtitles = self.translate_subtitles(
+                subtitle_list, target_language, glossary_file
+            )
+
+            log.info(
+                "  🔍 %sValidating translated subtitles...",
+                "Re-" if attempt > 0 else "",
+            )
+
+            if self._validate_timestamps(subtitle_list, translated_subtitles):
+                log.info(
+                    "  ✅ Timestamps validated successfully%s.",
+                    " on retry" if attempt > 0 else "",
+                )
+                return translated_subtitles
+
+            log.warning(
+                "  ❌ Timestamp %svalidation failed.", "re-" if attempt > 0 else ""
+            )
+
+        repaired_subtitles = self._repair_timestamps_with_deepl(
+            subtitle_list, target_language
+        )
+
+        log.info("  🔍 Re-validating repaired subtitles...")
+        if self._validate_timestamps(subtitle_list, repaired_subtitles):
+            log.info("  ✅ Timestamps repaired and validated successfully.")
+            return repaired_subtitles
+
+        log.error("  ❌ Timestamp repair failed. Translation cannot proceed.")
+        raise ValueError(  # noqa: TRY003
+            "Subtitle timestamp repair failed - timestamps could not be validated"  # noqa: EM101
+        )
+
+    def _validate_timestamps(
+        self, original: list[srt.Subtitle], translated: list[srt.Subtitle]
+    ) -> bool:
+        """
+        Validate that timestamps and cue numbers are preserved.
+
+        Checks for cue count mismatches, index mismatches, timestamp mismatches,
+        and blank translations.
+
+        Args:
+            original: Original subtitle list
+            translated: Translated subtitle list
+
+        Returns:
+            True if validation passes, False otherwise
+        """
+        issues = []
+        if len(original) != len(translated):
+            issues.append(
+                f"Cue count mismatch: original {len(original)}, "
+                f"translated {len(translated)}"
+            )
+
+        for i, (orig, trans) in enumerate(zip(original, translated)):
+            if orig.index != trans.index:
+                issues.append(
+                    f"Cue {i + 1}: index mismatch ({orig.index} vs {trans.index})"
+                )
+            if orig.start != trans.start or orig.end != trans.end:
+                issues.append(f"Cue {i + 1}: timestamp mismatch")
+            if orig.content.strip() and not trans.content.strip():
+                issues.append(f"Cue {i + 1}: translation is BLANK")
+
+        if issues:
+            logger.warning("Translation validation found issues:")
+            for issue in issues[:10]:
+                logger.warning("  - %s", issue)
+            if len(issues) > 10:  # noqa: PLR2004
+                logger.warning("  ... and %s more issues", len(issues) - 10)
+            return False
+        return True
+
+    def _repair_timestamps_with_deepl(
+        self,
+        original: list[srt.Subtitle],
+        target_lang: str,
+    ) -> list[srt.Subtitle]:
+        """
+        Repair misaligned timestamps using DeepL translation.
+
+        Uses DeepL to retranslate subtitles with proper timestamp preservation.
+
+        Args:
+            original: Original subtitle list with correct timestamps
+            target_lang: Target language code
+
+        Returns:
+            List of repaired subtitles with corrected timestamps
+        """
+        if not self.repair_api_key:
+            logger.warning("   No repair API key available, skipping repair.")
+            return original
+
+        logger.info("  🔧 Repairing timestamps using DeepL...")
+
+        try:
+            # Import DeepL provider for repair
+            from ol_openedx_course_translations.providers.deepl_provider import (  # noqa: PLC0415
+                DeepLProvider,
+            )
+
+            # Create DeepL provider instance for repair
+            deepl_provider = DeepLProvider(self.repair_api_key, None)
+
+            # Use DeepL to translate with proper timestamp preservation
+            repaired_subtitles = deepl_provider.translate_subtitles(
+                original, target_lang, None
+            )
+
+            logger.info("  ✅ DeepL repair completed.")
+            return repaired_subtitles  # noqa: TRY300
+
+        except Exception as e:  # noqa: BLE001
+            logger.error("  ❌ DeepL repair failed: %s", e)  # noqa: TRY400
+            # Fallback: return original with empty content to preserve structure
+            return [
+                srt.Subtitle(
+                    index=sub.index,
+                    start=sub.start,
+                    end=sub.end,
+                    content="",
+                )
+                for sub in original
+            ]
+
+    @abstractmethod
+    def translate_subtitles(
+        self,
+        subtitle_list: list[srt.Subtitle],
+        target_language: str,
+        glossary_file: str | None = None,
+    ) -> list[srt.Subtitle]:
+        """
+        Translate SRT subtitles.
+
+        Args:
+            subtitle_list: List of subtitle objects to translate
+            target_language: Target language code
+            glossary_file: Path to glossary directory (optional)
+
+        Returns:
+            List of translated subtitle objects
+        """
+
+    @abstractmethod
+    def translate_text(
+        self,
+        source_text: str,
+        target_language: str,
+        tag_handling: str | None = None,
+        glossary_file: str | None = None,
+    ) -> str:
+        """
+        Translate plain text or HTML/XML.
+
+        Args:
+            source_text: Text to translate
+            target_language: Target language code
+            tag_handling: How to handle XML/HTML tags (optional)
+            glossary_file: Path to glossary directory (optional)
+
+        Returns:
+            Translated text
+        """
+
+    @abstractmethod
+    def translate_document(
+        self,
+        input_file_path: Path,
+        output_file_path: Path,
+        source_language: str,
+        target_language: str,
+        glossary_file: str | None = None,
+    ) -> None:
+        """
+        Translate document file.
+
+        Args:
+            input_file_path: Path to input file
+            output_file_path: Path to output file
+            source_language: Source language code
+            target_language: Target language code
+            glossary_file: Path to glossary directory (optional)
+        """