fix(docworker): Fix WeasyPrint and FontTools config

Author: MarekSuchanek

.cspell/dictionary.txt

@@ -88,6 +88,7 @@ thechapter
 setul
 setulcolor
 bbding
+fontconfig
 
 # Project specific terms
 docworker
@@ -118,5 +119,3 @@ odend
 heighta
 xurl
 nosniff
-
-

packages/dsw-document-worker/dsw/document_worker/handlers.py

@@ -1,5 +1,6 @@
 import os
 import pathlib
+import tempfile
 
 from . import consts
 from .cli import load_config_str
@@ -10,10 +11,16 @@ def lambda_handler(event, context):
     config_path = pathlib.Path(os.getenv(consts.VAR_APP_CONFIG_PATH, '/var/task/application.yml'))
     workdir_path = pathlib.Path(os.getenv(consts.VAR_WORKDIR_PATH, '/var/task/templates'))
 
-    config = load_config_str(config_path.read_text(encoding=consts.DEFAULT_ENCODING))
-    try:
-        doc_worker = DocumentWorker(config, workdir_path)
-        doc_worker.run_once()
-    except Exception as e:
-        SentryReporter.capture_exception(e)
-        raise e
+    with tempfile.TemporaryDirectory() as tmpdir:
+        cache_dir = pathlib.Path(tmpdir) / '.cache'
+        fontconfig_tmp = cache_dir / 'fontconfig'
+        os.environ['XDG_CACHE_HOME'] = cache_dir.as_posix()
+        fontconfig_tmp.mkdir(parents=True, exist_ok=True)
+
+        config = load_config_str(config_path.read_text(encoding=consts.DEFAULT_ENCODING))
+        try:
+            doc_worker = DocumentWorker(config, workdir_path)
+            doc_worker.run_once()
+        except Exception as e:
+            SentryReporter.capture_exception(e)
+            raise

packages/dsw-document-worker/dsw/document_worker/templates/steps/base.py

@@ -21,6 +21,10 @@ def __init__(self, template, options: dict[str, str]):
         extras_str: str = self.options.get(self.OPTION_EXTRAS, '')
         self.extras: set[str] = set(extras_str.split(','))
 
+    @staticmethod
+    def initialize_step():
+        pass
+
     def requires_via_extras(self, requirement: str) -> bool:
         return requirement in self.extras
 
@@ -38,6 +42,7 @@ def raise_exc(self, message: str):
 
 
 def register_step(name: str, step_class: type[Step]):
+    step_class.initialize_step()
     STEPS[name.lower()] = step_class
 
 

packages/dsw-document-worker/dsw/document_worker/templates/steps/conversion.py

@@ -1,3 +1,7 @@
+import logging
+
+import weasyprint
+
 from ... import consts
 from ...context import Context
 from ...conversions import Pandoc, RdfLibConvert
@@ -15,14 +19,21 @@ class WeasyPrintStep(Step):
     OUTPUT_FORMAT = FileFormats.PDF
 
     def __init__(self, template, options: dict):
-        import weasyprint
-
         super().__init__(template, options)
         # PDF options
-        self.wp_options = weasyprint.DEFAULT_OPTIONS
+        self.wp_options = weasyprint.DEFAULT_OPTIONS.copy()
         self.wp_update_options(options)
         self.wp_zoom = float(options.get('pdf.zoom', '1'))
 
+    @staticmethod
+    def initialize_step():
+        # mute weasyprint logger
+        logging.getLogger('weasyprint').setLevel(logging.WARNING)
+        # mute fontTools loggers
+        for logger_name in logging.root.manager.loggerDict:
+            if logger_name.startswith('fontTools'):
+                logging.getLogger(logger_name).setLevel(logging.WARNING)
+
     def wp_update_options(self, options: dict):
         optimize_size = tuple(options.get('render.optimize_size', 'fonts').split(','))
         self.wp_options.update({
@@ -57,7 +68,7 @@ def execute_follow(self, document: DocumentFile, context: dict) -> DocumentFile:
             zoom=self.wp_zoom,
             font_config=None,  # not used now (should be in CSS)
             counter_style=None,  # not used now (should be in CSS)
-            options=self.options,
+            **self.wp_options,
         )
         return DocumentFile(
             file_format=self.OUTPUT_FORMAT,