Change management of attachments

weasyprint/document.py

@@ -143,8 +143,7 @@ def __init__(self, title=None, authors=None, description=None,
         #: Extracted from the ``<meta name=dcterms.modified>`` element in HTML
         #: and written to the ``/ModDate`` info field in PDF.
         self.modified = modified
-        #: File attachments, as a list of tuples of URL and a description or
-        #: :obj:`None`. (Defaults to the empty list.)
+        #: A list of :class:`attachments <Attachment>`, empty by default.
         #: Extracted from the ``<link rel=attachment>`` elements in HTML
         #: and written to the ``/EmbeddedFiles`` dictionary in PDF.
         self.attachments = attachments or []

weasyprint/html.py

@@ -19,7 +19,7 @@ def read_text(package, resource):
 
 import re
 
-from . import CSS, css
+from . import CSS, Attachment, css
 from .css import get_child_text
 from .css.counters import CounterStyle
 from .formatting_structure import boxes
@@ -281,7 +281,8 @@ def get_html_metadata(html):
             if url is None:
                 LOGGER.error('Missing href in <link rel="attachment">')
             else:
-                attachments.append((url, attachment_title))
+                attachment = Attachment(url=url, description=attachment_title)
+                attachments.append(attachment)
     return {
         'title': title,
         'description': description,

weasyprint/pdf/__init__.py

@@ -2,7 +2,7 @@
 
 import pydyf
 
-from .. import VERSION
+from .. import VERSION, Attachment
 from ..html import W3C_DATE_RE
 from ..logger import LOGGER, PROGRESS_LOGGER
 from ..matrix import Matrix
@@ -245,11 +245,15 @@ def generate_pdf(document, target, zoom, **options):
                 pdf.info[key] = pydyf.String(value)
 
     # Embedded files
-    attachments = metadata.attachments + (options['attachments'] or [])
+    attachments = metadata.attachments.copy()
+    if options['attachments']:
+        for attachment in options['attachments']:
+            if not isinstance(attachment, Attachment):
+                attachment = Attachment(attachment)
+            attachments.append(attachment)
     pdf_attachments = []
     for attachment in attachments:
-        pdf_attachment = write_pdf_attachment(
-            pdf, attachment, document.url_fetcher, compress)
+        pdf_attachment = write_pdf_attachment(pdf, attachment, compress)
         if pdf_attachment is not None:
             pdf_attachments.append(pdf_attachment)
     if pdf_attachments:

weasyprint/pdf/anchors.py

@@ -253,7 +253,7 @@ def add_annotations(links, matrix, document, pdf, page, annot_files, compress):
             # above about multiple regions won't always be correct, because
             # two links might have the same href, but different titles.
             annot_files[annot_target] = write_pdf_attachment(
-                pdf, (annot_target, None), document.url_fetcher, compress)
+                pdf, Attachment(annot_target), compress)
         annot_file = annot_files[annot_target]
         if annot_file is None:
             continue
@@ -281,18 +281,11 @@ def add_annotations(links, matrix, document, pdf, page, annot_files, compress):
         page['Annots'].append(annot.reference)
 
 
-def write_pdf_attachment(pdf, attachment, url_fetcher, compress):
+def write_pdf_attachment(pdf, attachment, compress):
     """Write an attachment to the PDF stream."""
     # Attachments from document links like <link> or <a> can only be URLs.
     # They're passed in as tuples
-    url = ''
-    if isinstance(attachment, tuple):
-        url, description = attachment
-        attachment = Attachment(
-            url=url, url_fetcher=url_fetcher, description=description)
-    elif not isinstance(attachment, Attachment):
-        attachment = Attachment(guess=attachment, url_fetcher=url_fetcher)
-
+    url = None
     uncompressed_length = 0
     stream = b''
     try: