utils.py

"""
Helper functions for the XLIFF serializer
"""

from __future__ import annotations

import datetime
import difflib
import glob
import logging
import os
import uuid
from typing import TYPE_CHECKING

from django.conf import settings
from django.contrib import messages
from django.core import serializers
from django.core.cache import cache
from django.core.files.base import ContentFile
from django.core.files.storage import FileSystemStorage
from django.db import IntegrityError, transaction
from django.forms.models import model_to_dict
from django.utils.html import format_html, format_html_join
from django.utils.translation import gettext_lazy as _
from django.utils.translation import ngettext_lazy
from linkcheck import update_lock

from ..cms.constants import text_directions
from ..cms.forms import PageTranslationForm
from ..cms.models import Page, PageTranslation
from ..cms.utils.file_utils import create_zip_archive
from ..cms.utils.stringify_list import iter_to_string
from ..cms.utils.translation_utils import gettext_many_lazy as __

if TYPE_CHECKING:
    from typing import Any
    from uuid import UUID

    from django.core.serializers.base import DeserializedObject
    from django.http import HttpRequest

    from ..cms.models.languages.language import Language
    from ..cms.models.pages.page import PageQuerySet

upload_storage = FileSystemStorage(location=settings.XLIFF_UPLOAD_DIR)
download_storage = FileSystemStorage(
    location=settings.XLIFF_DOWNLOAD_DIR, base_url=settings.XLIFF_URL
)

logger = logging.getLogger(__name__)


def pages_to_xliff_file(
    request: HttpRequest,
    pages: PageQuerySet,
    target_languages: list[Language],
    only_public: bool = False,
) -> str | None:
    """
    Export a list of page IDs to a ZIP archive containing XLIFF files for a specified target language (or just a single
    XLIFF file if only one page is converted)

    :param request: The current request (used for error messages)
    :param pages: list of pages which should be translated
    :param target_languages: list of target languages (should exclude the region's default language)
    :param only_public: Whether only public versions should be exported
    :return: The path of the generated zip file
    """
    logger.debug(
        "XLIFF export started by %r for pages %r and languages %r",
        request.user,
        pages,
        target_languages,
    )
    # Generate unique directory for this export
    dir_name = uuid.uuid4()
    # Create XLIFF files
    xliff_paths = {}
    for target_language in target_languages:
        for page in pages:
            try:
                xliff_paths[f"{target_language}/{page}"] = page_to_xliff(
                    page, target_language, dir_name, only_public=only_public
                )
            except RuntimeError as e:
                messages.error(request, e)
            except RuntimeWarning as e:
                messages.warning(request, e)
    # Check how many XLIFF files were created
    if not xliff_paths:
        return None
    if len(xliff_paths) == 1:
        # If only one xliff file was created, return it directly instead of creating zip file
        page, xliff_file_path = xliff_paths.popitem()
        xliff_file_url = download_storage.url(
            os.path.relpath(xliff_file_path, settings.XLIFF_DOWNLOAD_DIR)
        )
        logger.info(
            "XLIFF export: %r converted to XLIFF file %r by %r",
            page,
            xliff_file_url,
            request.user,
        )
        return xliff_file_url
    # Generate file path for ZIP archive
    timestamp = datetime.datetime.now().strftime("%Y-%m-%d_%H-%M")
    region = pages[0].region
    if len(target_languages) == 1:
        target_language = target_languages[0]
        zip_name = f"{region.slug}_{timestamp}_{region.get_source_language(target_language.slug).slug}_{target_language.slug}.zip"
    else:
        zip_name = f"{region.slug}_{timestamp}_multiple_languages.zip"
    actual_filename = download_storage.save(f"{dir_name}/{zip_name}", ContentFile(""))
    # Create ZIP archive
    create_zip_archive(
        list(xliff_paths.values()), download_storage.path(actual_filename)
    )
    xliff_file_url = download_storage.url(actual_filename)
    logger.info(
        "XLIFF export: %r converted to XLIFF ZIP archive %r by %r",
        xliff_paths.keys(),
        xliff_file_url,
        request.user,
    )
    return xliff_file_url


def page_to_xliff(
    page: Page, target_language: Language, dir_name: UUID, only_public: bool = False
) -> str:
    """
    Export a page to an XLIFF file for a specified target language

    :param page: Page which should be translated
    :param target_language: The target language (should not be the region's default language)
    :param dir_name: The directory in which the xliff files should be created
    :param only_public: Whether only public versions should be exported
    :raises RuntimeWarning: If the selected page translation does not have a source translation

    :raises RuntimeError: When an unexpected error occurs during serialization

    :return: The path of the generated XLIFF file
    """
    target_page_translation = (
        page.get_public_translation(target_language.slug)
        if only_public
        else page.get_public_or_draft_translation(target_language.slug)
    )
    if not target_page_translation:
        # Create temporary target translation
        target_page_translation = PageTranslation(
            page=page,
            language=target_language,
        )
    source_translation = (
        target_page_translation.public_source_translation
        if only_public
        else target_page_translation.public_or_draft_source_translation
    )
    if not source_translation:
        source_language = page.region.get_source_language(target_language.slug)
        logger.warning(
            "Page translation %r does not have a %ssource translation in %r and therefore cannot be exported to XLIFF.",
            target_page_translation,
            "published " if only_public else "",
            source_language,
        )
        raise RuntimeWarning(
            _(
                "Page {} does not have a {}source translation in {} and "
                "therefore cannot be exported as XLIFF for translation to {}."
            ).format(
                target_page_translation.readable_title,
                _("published") + " " if only_public else "",
                source_language,
                target_language,
            )
        )

    try:
        xliff_content = serializers.serialize(
            settings.XLIFF_EXPORT_VERSION,
            [target_page_translation],
            only_public=only_public,
        )
    except Exception as e:
        # All these error should already have been prevented
        logger.exception(
            "An unexpected error has occurred while exporting %r to XLIFF: %s",
            target_page_translation,
            e,
        )
        raise RuntimeError(
            __(
                _("An unexpected error has occurred while exporting page {}.").format(
                    target_page_translation.readable_title
                ),
                _("Please try again later or contact an administrator."),
            )
        ) from e

    filename = (
        f"{page.region.slug}_{source_translation.language.slug}_{target_language.slug}_"
        f"{page.id}_{source_translation.version}_{source_translation.slug}.xliff"
    )

    actual_filename = download_storage.save(
        f"{dir_name}/{filename}", ContentFile(xliff_content)
    )
    logger.debug("Created XLIFF file %r", actual_filename)

    # Set "currently in translation" status for existing target translations
    if settings.REDIS_CACHE:
        target_page_translation.all_versions.invalidated_update(
            currently_in_translation=True
        )
    else:
        target_page_translation.all_versions.update(currently_in_translation=True)

    return download_storage.path(actual_filename)


def xliffs_to_pages(
    request: HttpRequest, xliff_dir: str
) -> dict[str, list[DeserializedObject]]:
    """
    Export a page to an XLIFF file for a specified target language

    :param request: The current request (used for error messages)
    :param xliff_dir: The directory containing the xliff files
    :return: A dict of all page translations as ``DeserializedObject``
    """
    # Check if result is cached
    if cached_result := cache.get(f"xliff-{xliff_dir}"):
        logger.debug(
            "Returning cached result for deserializing all XLIFF files of %r",
            xliff_dir,
        )
        return cached_result
    # Get all xliff files in the given directory (sort for deterministic order)
    xliff_file_paths = sorted(glob.glob(f"{xliff_dir}/**/*.xliff", recursive=True))
    page_translations = {}
    for xliff_file_path in xliff_file_paths:
        with open(xliff_file_path, "r", encoding="utf-8") as xliff_file:
            logger.debug(
                "Deserializing XLIFF file %r",
                xliff_file_path,
            )
            xliff_file_path_rel = os.path.relpath(xliff_file_path, xliff_dir)
            try:
                # Try to deserialize the file
                page_translations[xliff_file_path_rel] = list(
                    serializers.deserialize("xliff", xliff_file.read())
                )
            except Page.DoesNotExist:
                logger.error(
                    "The page of XLIFF file %r does not exist.",
                    xliff_file_path,
                )
                messages.error(
                    request,
                    __(
                        _(
                            'The page referenced in XLIFF file "{}" could not be found.'
                        ).format(xliff_file_path_rel),
                        _("Please contact the administrator."),
                    ),
                )
            # In this case, we want to catch all exceptions because the import of the other files should work even if
            # some xliff files are broken or other unexpected errors occur
            except Exception as e:  # pylint: disable=broad-except
                # All these error should already have been prevented, so probably the XLIFF file is broken.
                logger.exception(
                    "An unexpected error has occurred while importing XLIFF file %r: %s",
                    xliff_file_path,
                    e,
                )
                messages.error(
                    request,
                    __(
                        _(
                            'An unexpected error has occurred while importing XLIFF file "{}".'
                        ).format(xliff_file_path_rel),
                        _("Please try again later or contact an administrator."),
                    ),
                )
    # Store deserialized objects in cache
    cache.set(f"xliff-{xliff_dir}", page_translations)
    return page_translations


def get_xliff_import_diff(request: HttpRequest, xliff_dir: str) -> list[dict[str, Any]]:
    """
    Show the XLIFF import diff

    :param request: The current request (used for error messages)
    :param xliff_dir: The directory containing the xliff files
    :return: A dict containing data about the imported xliff files
    """
    diff: dict[str, dict] = {}
    for xliff_file, deserialized_objects in xliffs_to_pages(request, xliff_dir).items():
        for deserialized in deserialized_objects:
            page_translation = deserialized.object
            # The prefetched translations now also contain the new deserialized object with id None, so we have to delete
            # the cached property and query it from the database again.
            try:
                del page_translation.page.prefetched_translations_by_language_slug
            except AttributeError:
                pass
            existing_translation = page_translation.latest_version or PageTranslation(
                page=page_translation.page,
                language=page_translation.language,
            )
            if (translation_key := get_translation_key(existing_translation)) in diff:
                # Show global error to indicate duplicate translation
                messages.error(
                    request,
                    format_html(
                        __(
                            _(
                                "Page <b>{}</b> from file <b>{}</b> was also translated in file <b>{}</b>."
                            ),
                            _(
                                "Please check which of the files contains the most recent version and upload only this file."
                            ),
                            _(
                                "If you confirm this import, only the first file will be imported and the latter will be ignored."
                            ),
                        ),
                        existing_translation.readable_title,
                        diff[translation_key]["file"],
                        xliff_file,
                    ),
                )
                # Show warning in the section of the first occurrence of this duplicate
                diff[translation_key]["errors"].append(
                    {
                        "level_tag": "warning",
                        "message": format_html(
                            _(
                                "This page was also translated in file <b>{}</b>, which will be ignored."
                            ),
                            xliff_file,
                        ),
                    }
                )
                # Skip this duplicated translation
                continue
            diff[translation_key] = {
                "file": xliff_file,
                "existing": existing_translation,
                "import": page_translation,
                "source_diff": {
                    "title": "\n".join(
                        list(
                            difflib.unified_diff(
                                [existing_translation.title],
                                [page_translation.title],
                                lineterm="",
                            )
                        )[2:]
                    ),
                    "content": "\n".join(
                        list(
                            difflib.unified_diff(
                                existing_translation.content.splitlines(),
                                page_translation.content.splitlines(),
                                lineterm="",
                            )
                        )[2:]
                    ),
                },
                "right_to_left": page_translation.language.text_direction
                == text_directions.RIGHT_TO_LEFT,
                "errors": get_xliff_import_errors_and_clean_translation(
                    request, page_translation, add_message_if_unchanged=True
                )[0],
            }
    # Throw away the translation keys, only take the value dictionaries
    return list(diff.values())


def xliff_import_confirm(
    request: HttpRequest, xliff_dir: str, machine_translated: bool
) -> bool:
    """
    Confirm the XLIFF import and write changes to database

    :param request: The current request (used for error messages)
    :param xliff_dir: The directory containing the xliff files
    :param machine_translated: A flag indicating the import was marked as machine translated
    :return: Whether the import was successful
    """
    success = True

    successful_imports = []
    imports_without_changes = []

    # Acquire linkcheck lock to avoid race conditions between post_save signal and links.delete()
    with update_lock:
        # Iterate over all xliff files
        for xliff_file, deserialized_objects in xliffs_to_pages(
            request, xliff_dir
        ).items():
            # Iterate over all objects of one xliff file
            # (typically, one xliff file contains exactly one page translation)
            for deserialized in deserialized_objects:
                page_translation = deserialized.object
                errors, has_changed = get_xliff_import_errors_and_clean_translation(
                    request, page_translation
                )
                if errors:
                    logger.warning(
                        "XLIFF import of %r not possible because validation of %r failed with the errors: %r",
                        xliff_file,
                        page_translation,
                        errors,
                    )

                    messages.error(
                        request,
                        format_html(
                            "{} <ul>{}</ul>",
                            _(
                                "Page {} could not be imported successfully because of the errors:"
                            ).format(page_translation.readable_title),
                            format_html_join(
                                "",
                                "<li><i icon-name='alert-triangle' class='pb-1'></i>{}</li>",
                                [[error["message"]] for error in errors],
                            ),
                        ),
                    )

                    success = False
                else:
                    # Check if previous version already exists
                    if existing_translation := page_translation.latest_version:
                        # Delete link objects of existing translation
                        existing_translation.links.all().delete()
                    page_translation.machine_translated = machine_translated
                    # Confirm import and write changes to the database
                    try:
                        # Use encapsulated transaction to allow rollback in case of IntegrityError
                        with transaction.atomic():
                            page_translation.save()
                    except IntegrityError as e:
                        logger.error(
                            "%s when importing new version for %r from %r by %r: %s",
                            type(e).__name__,
                            existing_translation,
                            xliff_file,
                            request.user,
                            e,
                        )
                        messages.error(
                            request,
                            format_html(
                                __(
                                    _(
                                        "Page {} from the file <b>{}</b> could not be imported."
                                    ),
                                    _(
                                        "Check if you have uploaded any other conflicting files for this page."
                                    ),
                                    _(
                                        "If the problem persists, contact an administrator."
                                    ),
                                ),
                                page_translation.readable_title,
                                xliff_file,
                            ),
                        )
                    else:
                        if has_changed:
                            logger.info(
                                "%r of XLIFF file %r was imported successfully by %r",
                                page_translation,
                                xliff_file,
                                request.user,
                            )
                            successful_imports.append(page_translation.readable_title)
                        else:
                            logger.info(
                                "%r of XLIFF file %r was imported without changes by %r",
                                existing_translation,
                                xliff_file,
                                request.user,
                            )
                            imports_without_changes.append(
                                page_translation.readable_title
                            )

    if successful_imports:
        messages.success(
            request,
            ngettext_lazy(
                "Page {} was imported successfully.",
                "Pages {} were imported successfully.",
                len(successful_imports),
            ).format(iter_to_string(successful_imports, quotation_char="")),
        )

    if imports_without_changes:
        messages.info(
            request,
            ngettext_lazy(
                "Page {} was imported without changes.",
                "Pages {} were imported without changes.",
                len(imports_without_changes),
            ).format(iter_to_string(imports_without_changes, quotation_char="")),
        )

    return success


def get_xliff_import_errors_and_clean_translation(
    request: HttpRequest,
    page_translation: PageTranslation,
    add_message_if_unchanged: bool = False,
) -> tuple[list, bool]:
    """
    Validate an imported page translation and return all errors
    As a side effect, a unique slug is generated for the translation if it does not yet have one.
    Additionally, the content of the translation will be cleaned by a PageTranslationForm.

    :param request: The current request (used for error messages)
    :param page_translation: The page translation which is being imported
    :param add_message_if_unchanged: Whether a message should be added if no changes are detected
    :return: All errors of this XLIFF import
    """
    error_messages = []
    # Get current region
    region = request.region
    # Check whether user can import the page translation
    if not request.user.has_perm("cms.change_page_object", page_translation.page):
        error_messages.append(
            {
                "level_tag": "error",
                "message": _(
                    "You don't have the permission to import this page."
                ).format(page_translation.readable_title),
            }
        )
    # Check whether the page translation belongs to the current region
    if not page_translation.page.region == region:
        error_messages.append(
            {
                "level_tag": "error",
                "message": _("This page belongs to another region ({}).").format(
                    page_translation.page.region.full_name,
                ),
            }
        )
    # Retrieve existing page translation in the target language
    # The prefetched translations now also contain the new deserialized object with id None, so we have to delete
    # the cached property and query it from the database again.
    try:
        del page_translation.latest_version
    except AttributeError:
        pass
    try:
        del page_translation.page.prefetched_translations_by_language_slug
    except AttributeError:
        pass
    existing_translation = page_translation.latest_version
    # Validate page translation
    page_translation_form = PageTranslationForm(
        data=model_to_dict(page_translation),
        instance=existing_translation,
        additional_instance_attributes={
            "creator": request.user,
            "language": page_translation.language,
            "page": page_translation.page,
        },
    )
    if not page_translation_form.is_valid():
        error_messages.extend(
            [
                {"level_tag": message["type"], "message": message["text"]}
                for message in page_translation_form.get_error_messages()
            ]
        )
    else:
        # Use the unique slug generated by the form's clean-method for imported page translations
        page_translation.slug = page_translation_form.instance.slug
        # Use the cleaned content from the form
        page_translation.content = page_translation_form.instance.content
    if add_message_if_unchanged and not page_translation_form.has_changed():
        error_messages.append(
            {
                "level_tag": "info",
                "message": _("No changes detected."),
            }
        )
    return error_messages, page_translation_form.has_changed()


def get_translation_key(translation: PageTranslation) -> str:
    """
    Return a unique key for each translation to be able to find duplicates.
    We cannot use the translation id here because the objects have not been written to the database.

    :param translation: The translation
    :return: A unique key for this ``Page``/``Language``/``version`` combination
    """
    return f"{translation.page.id}-{translation.language.id}-{translation.version}"