fix: convert bad img src quotes when parsing weblicp

fix #79
vzhd1701 · Oct 19, 2023 · 080d7e3 · 080d7e3
1 parent f6f31bf
commit 080d7e3
Show file tree

Hide file tree

Showing 3 changed files with 32 additions and 0 deletions.
diff --git a/enex2notion/note_parser/webclip.py b/enex2notion/note_parser/webclip.py
@@ -4,6 +4,7 @@
 
 from enex2notion.note_parser.blocks import parse_note_blocks
 from enex2notion.note_parser.webclip_stages_cleanup import (
+    fix_redundant_img_src_quotes,
     remove_empty_blocks,
     strip_paragraphs,
     wrap_orphans,
@@ -74,6 +75,7 @@ def parse_webclip(note_dom: Tag):
         remove_empty_blocks,
         wrap_orphans,
         strip_paragraphs,
+        fix_redundant_img_src_quotes,
     )
 
     for processor in processors:

diff --git a/enex2notion/note_parser/webclip_stages_cleanup.py b/enex2notion/note_parser/webclip_stages_cleanup.py
@@ -22,6 +22,12 @@ def wrap_orphans(root: Tag):
         _convert_to_paragraph(element)
 
 
+def fix_redundant_img_src_quotes(root: Tag):
+    for e in root.find_all("img"):
+        if e.get("src"):
+            e["src"] = e["src"].strip("\"'")
+
+
 def _convert_to_paragraph(element):
     div = Tag(name="div")
 

diff --git a/tests/test_webclip_parser.py b/tests/test_webclip_parser.py
@@ -1,7 +1,11 @@
+import base64
+
+from enex2notion.enex_types import EvernoteResource
 from enex2notion.note_parser.webclip import parse_webclip
 from enex2notion.notion_blocks.header import NotionSubsubheaderBlock
 from enex2notion.notion_blocks.list import NotionBulletedListBlock
 from enex2notion.notion_blocks.text import NotionTextBlock, TextProp
+from enex2notion.notion_blocks.uploadable import NotionImageBlock
 
 
 def test_empty(parse_html):
@@ -208,3 +212,23 @@ def test_flatten_bad_inline(parse_html):
     assert parse_webclip(test_note) == [
         NotionTextBlock(text_prop=TextProp("test")),
     ]
+
+
+def test_embedded_inline_img_bin_bad_quotes(parse_html, smallest_gif):
+    test_note = parse_html(
+        f"<img src=\"'data:{smallest_gif.mime};"
+        f'base64,{base64.b64encode(smallest_gif.data_bin).decode("utf-8")}\'" />'
+    )
+
+    result_block = parse_webclip(test_note)[0]
+
+    assert result_block == NotionImageBlock(
+        md5_hash=smallest_gif.md5,
+        resource=EvernoteResource(
+            data_bin=smallest_gif.data_bin,
+            size=smallest_gif.size,
+            md5=smallest_gif.md5,
+            mime=smallest_gif.mime,
+            file_name=f"{smallest_gif.md5}.gif",
+        ),
+    )