fix: skip webclip notes before parsing

enex2notion/cli.py

@@ -51,6 +51,10 @@ def upload(self, enex_file: Path):
                 logger.debug(f"Skipping note '{note.title}' (already uploaded)")
                 continue
 
+            if note.is_webclip:
+                logger.warning(f"Skipping note '{note.title}' [WEBCLIPS NOT SUPPORTED]")
+                continue
+
             note_blocks = parse_note(note, self.add_meta)
             if not note_blocks:
                 continue

enex2notion/enex_parser.py

@@ -2,6 +2,7 @@
 import hashlib
 import logging
 import mimetypes
+import re
 import uuid
 from collections import defaultdict
 from dataclasses import dataclass
@@ -33,6 +34,7 @@ class EvernoteNote(object):
     tags: List[str]
     author: str
     url: str
+    is_webclip: bool
     resources: List[EvernoteResource]
     _note_hash: str = None
 
@@ -125,10 +127,24 @@ def _process_note(note_raw: dict):
         tags=note_tags,
         author=note_attrs.get("author", ""),
         url=note_attrs.get("source-url", ""),
+        is_webclip=_is_webclip(note_raw),
         resources=resources,
     )
 
 
+def _is_webclip(note_raw: dict):
+    note_attrs = note_raw.get("note-attributes") or {}
+
+    if "web.clip" in note_attrs.get("source", ""):
+        return True
+    if "webclipper" in note_attrs.get("source-application", ""):
+        return True
+
+    return bool(
+        re.match('<div[^>]+style="[^"]+en-clipped-content[^"]*"', note_raw["content"])
+    )
+
+
 def _convert_resource(resource_raw):
     res_attr = resource_raw.get("resource-attributes", {})
     if not isinstance(res_attr, dict):

tests/test_cli.py

@@ -1,3 +1,5 @@
+import logging
+
 import pytest
 
 from enex2notion.cli import cli
@@ -18,7 +20,7 @@ def mock_api(mocker):
 @pytest.fixture()
 def fake_note_factory(mocker):
     mock_iter = mocker.patch("enex2notion.cli.iter_notes")
-    mock_iter.return_value = [mocker.MagicMock(note_hash="fake_hash")]
+    mock_iter.return_value = [mocker.MagicMock(note_hash="fake_hash", is_webclip=False)]
 
     return mock_iter
 
@@ -101,8 +103,8 @@ def test_done_file(mock_api, fake_note_factory, mocker, fs):
     fs.create_file("done.txt")
 
     fake_note_factory.return_value = [
-        mocker.MagicMock(note_hash="fake_hash1"),
-        mocker.MagicMock(note_hash="fake_hash2"),
+        mocker.MagicMock(note_hash="fake_hash1", is_webclip=False),
+        mocker.MagicMock(note_hash="fake_hash2", is_webclip=False),
     ]
 
     cli(["--token", "fake_token", "--done-file", "done.txt", "fake.enex"])
@@ -143,5 +145,19 @@ def test_bad_file(mock_api, fake_note_factory):
     mock_api["parse_note"].assert_called_once()
 
 
+def test_webclip(mock_api, fake_note_factory, mocker, caplog):
+    fake_note_factory.return_value = [
+        mocker.MagicMock(note_hash="fake_hash1", is_webclip=True),
+    ]
+
+    with caplog.at_level(logging.WARNING):
+        cli(["fake.enex"])
+
+    assert "[WEBCLIPS NOT SUPPORTED]" in caplog.text
+
+    mock_api["upload_note"].assert_not_called()
+    mock_api["parse_note"].assert_not_called()
+
+
 def test_cli_main_import():
     from enex2notion import __main__

tests/test_enex_parser.py

@@ -35,6 +35,7 @@ def test_iter_notes_single(fs):
             tags=[],
             author="",
             url="",
+            is_webclip=False,
             resources=[],
         ),
     ]
@@ -90,6 +91,7 @@ def test_iter_notes_single_tags(fs):
             tags=["tag1", "tag2"],
             author="",
             url="",
+            is_webclip=False,
             resources=[],
         ),
     ]
@@ -123,6 +125,7 @@ def test_iter_notes_single_one_tag(fs):
             tags=["tag1"],
             author="",
             url="",
+            is_webclip=False,
             resources=[],
         ),
     ]
@@ -137,8 +140,8 @@ def test_iter_notes_single_attrs(fs):
         <created>20211118T085332Z</created>
         <updated>20211118T085920Z</updated>
         <note-attributes>
-        <author>test@user.com</author>
-        <source-url>https://google.com</source-url>
+            <author>test@user.com</author>
+            <source-url>https://google.com</source-url>
         </note-attributes>
         <content>test</content>
       </note>
@@ -157,6 +160,110 @@ def test_iter_notes_single_attrs(fs):
             tags=[],
             author="test@user.com",
             url="https://google.com",
+            is_webclip=False,
+            resources=[],
+        ),
+    ]
+
+
+def test_iter_notes_webclip1(fs):
+    test_enex = """<?xml version="1.0" encoding="UTF-8"?>
+    <!DOCTYPE en-export SYSTEM "http://xml.evernote.com/pub/evernote-export4.dtd">
+    <en-export export-date="20211218T085932Z" application="Evernote" version="10.25.6">
+      <note>
+        <title>test1</title>
+        <created>20211118T085332Z</created>
+        <updated>20211118T085920Z</updated>
+        <note-attributes>
+            <source>web.clip</source>
+        </note-attributes>
+        <content>test</content>
+      </note>
+    </en-export>
+    """
+    fs.create_file("test.enex", contents=test_enex)
+
+    notes = list(iter_notes(Path("test.enex")))
+
+    assert notes == [
+        EvernoteNote(
+            title="test1",
+            created=datetime.datetime(2021, 11, 18, 8, 53, 32, tzinfo=tzutc()),
+            updated=datetime.datetime(2021, 11, 18, 8, 59, 20, tzinfo=tzutc()),
+            content="test",
+            tags=[],
+            author="",
+            url="",
+            is_webclip=True,
+            resources=[],
+        ),
+    ]
+
+
+def test_iter_notes_webclip2(fs):
+    test_enex = """<?xml version="1.0" encoding="UTF-8"?>
+    <!DOCTYPE en-export SYSTEM "http://xml.evernote.com/pub/evernote-export4.dtd">
+    <en-export export-date="20211218T085932Z" application="Evernote" version="10.25.6">
+      <note>
+        <title>test1</title>
+        <created>20211118T085332Z</created>
+        <updated>20211118T085920Z</updated>
+        <note-attributes>
+            <source-application>webclipper.evernote</source-application>
+        </note-attributes>
+        <content>test</content>
+      </note>
+    </en-export>
+    """
+    fs.create_file("test.enex", contents=test_enex)
+
+    notes = list(iter_notes(Path("test.enex")))
+
+    assert notes == [
+        EvernoteNote(
+            title="test1",
+            created=datetime.datetime(2021, 11, 18, 8, 53, 32, tzinfo=tzutc()),
+            updated=datetime.datetime(2021, 11, 18, 8, 59, 20, tzinfo=tzutc()),
+            content="test",
+            tags=[],
+            author="",
+            url="",
+            is_webclip=True,
+            resources=[],
+        ),
+    ]
+
+
+def test_iter_notes_webclip_content(fs):
+    test_enex = """<?xml version="1.0" encoding="UTF-8"?>
+    <!DOCTYPE en-export SYSTEM "http://xml.evernote.com/pub/evernote-export4.dtd">
+    <en-export export-date="20211218T085932Z" application="Evernote" version="10.25.6">
+      <note>
+        <title>test1</title>
+        <created>20211118T085332Z</created>
+        <updated>20211118T085920Z</updated>
+        <note-attributes>
+        </note-attributes>
+        <content><![CDATA[
+        <div style="--en-clipped-content:article;">test</div>
+        ]]></content>
+      </note>
+    </en-export>
+    """
+    fs.create_file("test.enex", contents=test_enex)
+
+    notes = list(iter_notes(Path("test.enex")))
+
+    assert notes == [
+        EvernoteNote(
+            title="test1",
+            created=datetime.datetime(2021, 11, 18, 8, 53, 32, tzinfo=tzutc()),
+            updated=datetime.datetime(2021, 11, 18, 8, 59, 20, tzinfo=tzutc()),
+            content='<div style="--en-clipped-content:article;">test</div>',
+            tags=[],
+            author="",
+            url="",
+            is_webclip=True,
             resources=[],
         ),
     ]
@@ -197,6 +304,7 @@ def test_iter_notes_multi(fs):
             tags=[],
             author="",
             url="",
+            is_webclip=False,
             resources=[],
         ),
         EvernoteNote(
@@ -207,6 +315,7 @@ def test_iter_notes_multi(fs):
             tags=[],
             author="",
             url="",
+            is_webclip=False,
             resources=[],
         ),
     ]
@@ -259,6 +368,7 @@ def test_iter_notes_single_with_resource(fs):
             tags=[],
             author="",
             url="",
+            is_webclip=False,
             resources=[expected_resource],
         ),
     ]
@@ -306,6 +416,7 @@ def test_iter_notes_single_with_noname_resource(fs, mocker):
             tags=[],
             author="",
             url="",
+            is_webclip=False,
             resources=[
                 EvernoteResource(
                     data_bin=(
@@ -367,6 +478,7 @@ def test_iter_notes_single_with_empty_resource(fs, caplog):
             tags=[],
             author="",
             url="",
+            is_webclip=False,
             resources=[expected_resource],
         ),
     ]

tests/test_enex_uploader.py

@@ -111,6 +111,7 @@ def test_upload_note(notion_test_page):
         tags=[],
         author="",
         url="",
+        is_webclip=False,
         resources=[],
     )
 
@@ -135,6 +136,7 @@ def test_upload_note_with_meta(notion_test_page):
         tags=[],
         author="",
         url="",
+        is_webclip=False,
         resources=[],
     )
 
@@ -165,6 +167,7 @@ def test_upload_note_with_file(notion_test_page, tiny_file):
         tags=[],
         author="",
         url="",
+        is_webclip=False,
         resources=[tiny_file],
     )
 
@@ -190,6 +193,7 @@ def test_upload_note_db(notion_test_page):
         tags=[],
         author="",
         url="",
+        is_webclip=False,
         resources=[],
     )
 

tests/test_note_parser.py

@@ -578,6 +578,7 @@ def test_bad_resource(caplog):
         tags=[],
         author="",
         url="",
+        is_webclip=False,
         resources=[],
     )
 
@@ -597,6 +598,7 @@ def test_bad_note(caplog):
         tags=[],
         author="",
         url="",
+        is_webclip=False,
         resources=[],
     )
 
@@ -616,6 +618,7 @@ def test_note_with_meta():
         tags=["tag1", "tag2"],
         author="",
         url="https://google.com",
+        is_webclip=False,
         resources=[],
     )