wip: gotenberg support

xblock_pdf/pdf.py

@@ -1,6 +1,7 @@
 """pdfXBlock main Python class."""
 
 import json
+from logging import getLogger
 
 from django.utils.translation import gettext_noop as _
 from web_fragments.fragment import Fragment
@@ -9,10 +10,12 @@
 from xblock.fields import Boolean, Scope, String
 from xblock.utils.resources import ResourceLoader
 
-from .utils import bool_from_str, is_all_download_disabled
+from .utils import add_asset, convert_to_pdf, error_response, is_all_download_disabled, is_gotenberg_enabled
 
 resource_loader = ResourceLoader(__name__)
 
+logger = getLogger(__name__)
+
 
 @XBlock.needs("i18n")
 class PDFBlock(XBlock):
@@ -114,16 +117,18 @@ def load_pdf(self, *_args, **_kwargs):
         return Response(json.dumps(self.raw_settings), content_type="application/json", charset="utf8")
 
     @XBlock.json_handler
-    def save_pdf(self, data, suffix=""):  # pylint: disable=unused-argument
-        """Save handler."""
-        self.display_name = data["display_name"]
-        self.url = data["url"]
-
-        if not is_all_download_disabled():
-            self.allow_download = bool_from_str(data["allow_download"])
-            self.source_text = data["source_text"]
-            self.source_url = data["source_url"]
-
-        return {
-            "result": "success",
-        }
+    def convert_pdf(self, data, suffix=""):  # pylint: disable=unused-argument
+        """
+        PDF Conversion handling. Basically just a frontend to the Gotenberg service which converts the given URL
+        and then saves it to course assets, returning the URL.
+        """
+        # TODO: Needs permission check.
+        if not is_gotenberg_enabled():
+            return error_response({"error": _("Gotenberg not enabled. PDF Conversion unavailable.")})
+        output_path = f"{self.location}.pdf"
+        url = data["url"]
+        result = convert_to_pdf(url)
+        if result is None:
+            return error_response({"error": _("PDF Conversion failed.")})
+        asset = add_asset(self.location, result, output_path)
+        return {"url": asset.url}

xblock_pdf/tests/test_pdf.py

@@ -4,7 +4,6 @@
 from typing import Any
 from unittest.mock import MagicMock, patch
 
-from django.test import override_settings
 from xblock.field_data import DictFieldData
 from xblock.fields import ScopeIds
 from xblock.test.toy_runtime import ToyRuntime
@@ -52,7 +51,7 @@ def test_download_button():
 
 
 def test_source_url():
-    """Test rendering based on whether or not there's a source URL"""
+    """Test rendering based on whether there's a source URL"""
     block = make_block()
     get_student_content(block)
     content = get_student_content(block)
@@ -62,60 +61,6 @@ def test_source_url():
     assert "Download the source document" in content
 
 
-@override_settings(PDFXBLOCK_DISABLE_ALL_DOWNLOAD=False)
-def test_saves_settings():
-    """Test that PDF settings are saved."""
-    block = make_block()
-    request = mock_handle_request(
-        {
-            "display_name": "Novel application of theory",
-            "url": "https://example.com/nature_article.pdf",
-            "allow_download": "false",
-            "source_text": "Get educated",
-            "source_url": "https://example.com/nature_article.tex",
-        }
-    )
-    block.save_pdf(request)
-    assert block.display_name == "Novel application of theory"
-    assert block.url == "https://example.com/nature_article.pdf"
-    assert not block.allow_download
-    assert block.source_text == "Get educated"
-    assert block.source_url == "https://example.com/nature_article.tex"
-
-
-@override_settings(PDFXBLOCK_DISABLE_ALL_DOWNLOAD=True)
-def test_saves_settings_omits_on_download_disabled_flag():
-    """
-    Test that fields relating to download are ignored when the universal
-    downloads disabled flag is set.
-    """
-    block = make_block()
-    request = mock_handle_request(
-        {
-            "display_name": "Novel application of theory",
-            "url": "https://example.com/nature_article.pdf",
-            # These fields shouldn't be visible on the front end,
-            # but should be dropped if they somehow are.
-            #
-            # Potential future improvement would be saving these
-            # but ignoring them when rendering. This is not currently
-            # the case since the fields are entirely absent from the studio
-            # render, and so would send blank data which would error out.
-            "allow_download": "false",
-            "source_text": "Get educated",
-            "source_url": "https://example.com/nature_article.tex",
-        }
-    )
-    block.save_pdf(request)
-    assert block.display_name == "Novel application of theory"
-    assert block.url == "https://example.com/nature_article.pdf"
-    # Flag will be the default, which is True, even though download will be
-    # disabled in practice.
-    assert block.allow_download
-    assert block.source_text == ""
-    assert block.source_url == ""
-
-
 @patch.object(ToyRuntime, "publish")
 def test_download_event_fires(mock_publish):
     """Test that we fire a download event."""

xblock_pdf/utils.py

@@ -1,13 +1,89 @@
 """Utility functions for PDF XBlock."""
 
+import json
+from typing import Any
+from urllib.parse import urlparse
+
+import requests
 from django.conf import settings
+from django.core.files.base import ContentFile
+from opaque_keys.edx.locator import BlockUsageLocator, LibraryUsageLocatorV2
+from webob import Response
+
+
+def is_gotenberg_enabled() -> bool:
+    """
+    Returns if gotenberg is enabled.
+    """
+    return bool(get_gotenberg_host())
+
+
+def get_gotenberg_host() -> str | None:
+    """
+    Returns the hostname of the Gotenberg instance, if configured.
+    Returns None if Gotenberg is not configured.
+    """
+    return getattr(settings, "GOTENBERG_HOST", None)
+
+
+def get_conversion_url() -> str | None:
+    """
+    Get the URL for sending a document for conversion by Gotenberg
+    """
+    return (base_url := get_gotenberg_host()) and f"{base_url}/forms/libreoffice/convert"
+
 
+def add_asset(location: BlockUsageLocator | LibraryUsageLocatorV2, asset: ContentFile, asset_name: str) -> str | None:
+    """
+    Adds an asset for this block. If we aren't in the studio environment, will create ImportErrors.
+    Easily mocked for tests.
+    """
+    from cms.djangoapps.contentstore.asset_storage_handlers import update_course_run_asset
+    from openedx.core.djangoapps.content_libraries.api import add_library_block_static_asset_file
 
-def bool_from_str(str_value):
-    """Convert string from submitted form to boolean."""
-    return str_value.strip().lower() == "true"
+    match location:
+        case BlockUsageLocator():
+            update_course_run_asset(location.course_key, asset, asset_name)
+        case LibraryUsageLocatorV2():
+            add_library_block_static_asset_file(location.lib_key, asset, asset_name)
+
+
+def convert_to_pdf(doc_url: str) -> ContentFile | None:
+    """
+    Uses the Gotenberg service to convert the document at `doc_url` to a PDF file.
+
+    Parameters:
+        doc_url (str): The path or URL to the document to be converted.
+
+    Returns:
+        ContentFile: A django ContentFile prepared for storage.
+    """
+    if not (conversion_url := get_conversion_url()):
+        return None
+    source_url = urlparse(doc_url)
+    filename = source_url.path.split("/")[-1]
+    source_doc_response = requests.get(doc_url, timeout=(10, 120))
+
+    pdf_response = requests.post(
+        conversion_url, files={"file": (filename, source_doc_response.content)}, timeout=(2, 120)
+    )
+    if pdf_response.status_code != 200:
+        return None
+    return ContentFile(pdf_response.content)
 
 
 def is_all_download_disabled():
     """Check if all downloads are disabled or not."""
     return getattr(settings, "PDFXBLOCK_DISABLE_ALL_DOWNLOAD", False)
+
+
+def error_response(data: dict[Any, Any], status: int = 400):
+    """
+    Returns a JSON response object with the appropriate status.
+    """
+    return Response(
+        json.dumps(data),
+        status=status,
+        content_type="application/json",
+        charset="utf8",
+    )