fix: add proper parsing for missing note attributes

enex2notion/enex_parser.py

@@ -5,64 +5,15 @@
 import re
 import uuid
 from collections import defaultdict
-from dataclasses import dataclass
 from datetime import datetime
 from pathlib import Path
-from typing import List
 from xml.etree import ElementTree
 
 from dateutil.parser import isoparse
 
-logger = logging.getLogger(__name__)
-
+from enex2notion.enex_types import EvernoteNote, EvernoteResource
 
-@dataclass(frozen=True)
-class EvernoteResource(object):
-    data_bin: bytes
-    size: int
-    md5: str
-    mime: str
-    file_name: str
-
-
-@dataclass
-class EvernoteNote(object):
-    title: str
-    created: datetime
-    updated: datetime
-    content: str  # noqa: WPS110
-    tags: List[str]
-    author: str
-    url: str
-    is_webclip: bool
-    resources: List[EvernoteResource]
-    _note_hash: str = None
-
-    def resource_by_md5(self, md5):
-        for resource in self.resources:
-            if resource.md5 == md5:
-                return resource
-        return None
-
-    @property
-    def note_hash(self):
-        if self._note_hash is None:
-            hashable = [
-                self.title,
-                self.created.isoformat(),
-                self.updated.isoformat(),
-                self.content,
-                "".join(self.tags),
-                self.author,
-                self.url,
-            ]
-
-            s1_hash = hashlib.sha1()
-            for h in hashable:
-                s1_hash.update(h.encode("utf-8"))
-            self._note_hash = s1_hash.hexdigest()  # noqa: WPS601
-
-        return self._note_hash
+logger = logging.getLogger(__name__)
 
 
 def iter_notes(enex_file: Path):
@@ -108,30 +59,39 @@ def _etree_to_dict(t):  # noqa: WPS210, WPS231, C901
 
 
 def _process_note(note_raw: dict):
+    if not note_raw:
+        note_raw = {}
+
     note_attrs = note_raw.get("note-attributes") or {}
 
     note_tags = note_raw.get("tag", [])
     if isinstance(note_tags, str):
         note_tags = [note_tags]
 
-    note_resources = note_raw.get("resource", [])
-    if isinstance(note_resources, dict):
-        note_resources = [note_resources]
-    resources = [_convert_resource(r) for r in note_resources]
+    now = datetime.now()
 
     return EvernoteNote(
-        title=note_raw["title"],
-        created=isoparse(note_raw["created"]),
-        updated=isoparse(note_raw["updated"]),
-        content=note_raw["content"],
+        title=note_raw.get("title", "Untitled"),
+        created=isoparse(note_raw.get("created", now.isoformat())),
+        updated=isoparse(note_raw.get("updated", now.isoformat())),
+        content=note_raw.get("content", ""),
         tags=note_tags,
         author=note_attrs.get("author", ""),
         url=note_attrs.get("source-url", ""),
         is_webclip=_is_webclip(note_raw),
-        resources=resources,
+        resources=_parse_resources(note_raw),
     )
 
 
+def _parse_resources(note_raw):
+    note_resources = note_raw.get("resource", [])
+
+    if isinstance(note_resources, dict):
+        note_resources = [note_resources]
+
+    return [_convert_resource(r) for r in note_resources]
+
+
 def _is_webclip(note_raw: dict):
     note_attrs = note_raw.get("note-attributes") or {}
 
@@ -141,7 +101,9 @@ def _is_webclip(note_raw: dict):
         return True
 
     return bool(
-        re.match('<div[^>]+style="[^"]+en-clipped-content[^"]*"', note_raw["content"])
+        re.match(
+            '<div[^>]+style="[^"]+en-clipped-content[^"]*"', note_raw.get("content", "")
+        )
     )
 
 

enex2notion/enex_types.py

@@ -0,0 +1,53 @@
+import hashlib
+from dataclasses import dataclass
+from datetime import datetime
+from typing import List
+
+
+@dataclass(frozen=True)
+class EvernoteResource(object):
+    data_bin: bytes
+    size: int
+    md5: str
+    mime: str
+    file_name: str
+
+
+@dataclass
+class EvernoteNote(object):
+    title: str
+    created: datetime
+    updated: datetime
+    content: str  # noqa: WPS110
+    tags: List[str]
+    author: str
+    url: str
+    is_webclip: bool
+    resources: List[EvernoteResource]
+    _note_hash: str = None
+
+    def resource_by_md5(self, md5):
+        for resource in self.resources:
+            if resource.md5 == md5:
+                return resource
+        return None
+
+    @property
+    def note_hash(self):
+        if self._note_hash is None:
+            hashable = [
+                self.title,
+                self.created.isoformat(),
+                self.updated.isoformat(),
+                self.content,
+                "".join(self.tags),
+                self.author,
+                self.url,
+            ]
+
+            s1_hash = hashlib.sha1()
+            for h in hashable:
+                s1_hash.update(h.encode("utf-8"))
+            self._note_hash = s1_hash.hexdigest()  # noqa: WPS601
+
+        return self._note_hash

enex2notion/enex_uploader.py

@@ -1,7 +1,7 @@
 from notion.block import CollectionViewPageBlock, PageBlock
 from progress.bar import Bar
 
-from enex2notion.enex_parser import EvernoteNote
+from enex2notion.enex_types import EvernoteNote
 from enex2notion.note_parser_dom import logger
 from enex2notion.note_uploader import upload_block
 

enex2notion/note_parser.py

@@ -1,6 +1,6 @@
 from bs4 import BeautifulSoup
 
-from enex2notion.enex_parser import EvernoteNote
+from enex2notion.enex_types import EvernoteNote
 from enex2notion.note_parser_dom import logger, parse_note_dom
 from enex2notion.notion_blocks import TextProp
 from enex2notion.notion_blocks_container import NotionCalloutBlock

enex2notion/note_parser_e_media.py

@@ -6,7 +6,7 @@
 from bs4 import Tag
 from w3lib.url import parse_data_uri
 
-from enex2notion.enex_parser import EvernoteResource
+from enex2notion.enex_types import EvernoteResource
 from enex2notion.notion_blocks_embeddable import NotionImageEmbedBlock
 from enex2notion.notion_blocks_uploadable import (
     NotionAudioBlock,

enex2notion/note_uploader.py

@@ -2,7 +2,7 @@
 from notion.block import FileBlock
 from notion.settings import S3_URL_PREFIX
 
-from enex2notion.enex_parser import EvernoteResource
+from enex2notion.enex_types import EvernoteResource
 from enex2notion.notion_blocks_uploadable import NotionUploadableBlock
 
 

tests/conftest.py

@@ -8,7 +8,7 @@
 from notion.block import PageBlock
 from notion.client import NotionClient
 
-from enex2notion.enex_parser import EvernoteResource
+from enex2notion.enex_types import EvernoteResource
 
 
 @pytest.fixture()

tests/test_enex_parser.py

@@ -5,7 +5,8 @@
 import pytest
 from dateutil.tz import tzutc
 
-from enex2notion.enex_parser import EvernoteNote, EvernoteResource, iter_notes
+from enex2notion.enex_parser import iter_notes
+from enex2notion.enex_types import EvernoteNote, EvernoteResource
 
 
 def test_iter_notes_single(fs):
@@ -486,3 +487,34 @@ def test_iter_notes_single_with_empty_resource(fs, caplog):
         notes[0].resource_by_md5("d41d8cd98f00b204e9800998ecf8427e")
         == expected_resource
     )
+
+
+def test_iter_notes_single_empty(fs, mocker):
+    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>
+      </note>
+    </en-export>
+    """
+    fs.create_file("test.enex", contents=test_enex)
+
+    fake_now = datetime.datetime(2021, 11, 18, 8, 0, 0, tzinfo=tzutc())
+    mock_iter = mocker.patch("enex2notion.enex_parser.datetime")
+    mock_iter.now.return_value = fake_now
+
+    test_note = list(iter_notes(Path("test.enex")))[0]
+
+    assert test_note.note_hash == "479b590db4f4d8817f93d01af51f21c894815920"
+    assert test_note == EvernoteNote(
+        title="Untitled",
+        created=fake_now,
+        updated=fake_now,
+        content="",
+        tags=[],
+        author="",
+        url="",
+        is_webclip=False,
+        resources=[],
+        _note_hash="479b590db4f4d8817f93d01af51f21c894815920",
+    )

tests/test_enex_uploader.py

@@ -6,7 +6,7 @@
 from dateutil.tz import tzutc
 from notion.block import CalloutBlock, CollectionViewPageBlock, FileBlock, PageBlock
 
-from enex2notion.enex_parser import EvernoteNote
+from enex2notion.enex_types import EvernoteNote
 from enex2notion.enex_uploader import get_import_root, upload_note
 from enex2notion.enex_uploader_modes import get_notebook_database, get_notebook_page
 from enex2notion.note_parser import parse_note

tests/test_note_parser.py

@@ -7,7 +7,7 @@
 from bs4 import BeautifulSoup
 from dateutil.tz import tzutc
 
-from enex2notion.enex_parser import EvernoteNote, EvernoteResource
+from enex2notion.enex_types import EvernoteNote, EvernoteResource
 from enex2notion.note_parser import parse_note
 from enex2notion.note_parser_dom import parse_note_dom
 from enex2notion.notion_blocks import (