feat: support for google keep (#18) (#85)

* feat: support for google keep

* fix: indentation issues from flake8

* fix: remove updated time from key + rename timestamp

* fix: only use created timestamp for the key

Author: Kiduzk

google_takeout_parser/locales/common.py

@@ -11,6 +11,7 @@
     _parse_location_history,
     _parse_semantic_location_history,
     _parse_chrome_history,
+    _parse_keep
 )
 from ..parse_csv import (  # noqa: F401
     _parse_youtube_comments_csv,

google_takeout_parser/locales/en.py

@@ -10,6 +10,7 @@
     _parse_location_history,
     _parse_semantic_location_history,
     _parse_chrome_history,
+    _parse_keep
 )
 
 
@@ -86,7 +87,8 @@
     r"Groups": None,
     r"Google Play Games Services/Games/.*/(Achievements|Activity|Experience|Scores).html": None,
     r"Hangouts": None,
-    r"Keep": None,
+    r"Keep/.*?.json": _parse_keep,
+    r"Keep/": None,
     r"Maps (your places)": None,
     r"My Maps/.*.kmz": None,  # custom KML maps
     r"Saved/.*.csv": None,  # lists with saved places from Google Maps

google_takeout_parser/models.py

@@ -48,6 +48,19 @@ class LocationInfo(NamedTuple):
     sourceUrl: Optional[Url]
 
 
+class KeepListContent(NamedTuple):
+    textHtml: str
+    text: str
+    isChecked: bool
+
+
+class KeepAnnotation(NamedTuple):
+    description: str
+    source: str
+    title: str
+    url: str
+
+
 # fmt: off
 class BaseEvent(Protocol):
     @property
@@ -284,6 +297,25 @@ def key(self) -> Tuple[str, int]:
         return self.url, int(self.dt.timestamp())
 
 
+@dataclass
+class Keep(BaseEvent):
+    title: str
+    updated_dt: datetime
+    created_dt: datetime
+    listContent: Optional[List[KeepListContent]]
+    textContent: Optional[str]
+    textContentHtml: Optional[str]  # i guess this is good to have, found it in some of the json files
+    color: str
+    annotations: Optional[List[KeepAnnotation]]
+    isTrashed: bool
+    isPinned: bool
+    isArchived: bool
+
+    @property
+    def key(self) -> int:
+        return int(self.created_dt.timestamp())
+
+
 # can't compute this dynamically -- have to write it out
 # if you want to override, override both global variable types with new types
 DEFAULT_MODEL_TYPE = Union[
@@ -296,6 +328,7 @@ def key(self) -> Tuple[str, int]:
     CSVYoutubeComment,
     CSVYoutubeLiveChat,
     PlaceVisit,
+    Keep
 ]
 
 CacheResults = Iterator[Res[DEFAULT_MODEL_TYPE]]

google_takeout_parser/parse_json.py

@@ -21,6 +21,9 @@
     Location,
     PlaceVisit,
     CandidateLocation,
+    Keep,
+    KeepListContent,
+    KeepAnnotation
 )
 from .common import Res
 from .time_utils import parse_json_utc_date
@@ -282,3 +285,45 @@ def _parse_chrome_history(p: Path) -> Iterator[Res[ChromeHistory]]:
             )
         except Exception as e:
             yield e
+
+
+def _parse_keep(p: Path) -> Iterator[Res[Keep]]:
+    json_data = _read_json_data(p)
+    # For google keep, each note is stored as a separate json file,
+    # so technically there is always just one yield value
+    yield Keep(
+        title=json_data["title"],
+        created_dt=datetime.fromtimestamp(
+            json_data["createdTimestampUsec"] / 1_000_000, tz=timezone.utc
+        ),
+        updated_dt=datetime.fromtimestamp(
+            json_data["userEditedTimestampUsec"] / 1_000_000, tz=timezone.utc
+        ),
+        listContent=[
+            KeepListContent(
+                textHtml=content["textHtml"],
+                text=content["text"],
+                isChecked=content["isChecked"]
+            ) for content in json_data.get("listContent", [])
+        ],
+        textContent=(
+            json_data["textContent"]
+            if "textContent" in json_data else None
+        ),
+        textContentHtml=(
+            json_data["textContentHtml"]
+            if "textContentHtml" in json_data else None
+        ),
+        color=json_data["color"],
+        annotations=[
+            KeepAnnotation(
+                description=annotation["description"],
+                source=annotation["source"],
+                title=annotation["title"],
+                url=annotation["url"],
+            ) for annotation in json_data.get("annotations", [])
+        ],
+        isTrashed=json_data["isTrashed"],
+        isPinned=json_data["isPinned"],
+        isArchived=json_data["isArchived"]
+    )

tests/test_json.py

@@ -427,3 +427,78 @@ def test_semantic_location_history_2024(tmp_path_f: Path) -> None:
             ),
         ],
     )
+
+
+def test_keep(tmp_path_f: Path) -> None:
+    data = {
+        "title": "Grocery List",
+        "userEditedTimestampUsec": 1649291142757000,
+        "createdTimestampUsec": 1649291142757000,
+        "listContent": [
+            {
+                "textHtml": "<li>Apples</li>",
+                "text": "Apples",
+                "isChecked": False
+            },
+            {
+                "textHtml": "<li>Milk</li>",
+                "text": "Milk",
+                "isChecked": True
+            }
+        ],
+        "textContent": "Don't forget to buy snacks too.",
+        "textContentHtml": "<p>Don't forget to buy snacks too.</p>",
+        "color": "yellow",
+        "annotations": [
+            {
+                "description": "Weekly grocery run",
+                "source": "Keep",
+                "title": "Shopping Note",
+                "url": "https://keep.google.com/"
+            }
+        ],
+        "isTrashed": False,
+        "isPinned": True,
+        "isArchived": False
+    }
+
+    fp = tmp_path_f / "file"
+    fp.write_text(json.dumps(data))
+    res = list(prj._parse_keep(fp))
+    obj = res[0]
+    assert not isinstance(obj, Exception)
+    assert obj == models.Keep(
+        title="Grocery List",
+        updated_dt=datetime.datetime(
+            2022, 4, 7, 0, 25, 42, 757000, tzinfo=datetime.timezone.utc,
+        ),
+        created_dt=datetime.datetime(
+            2022, 4, 7, 0, 25, 42, 757000, tzinfo=datetime.timezone.utc,
+        ),
+        listContent=[
+            models.KeepListContent(
+                textHtml="<li>Apples</li>",
+                text="Apples",
+                isChecked=False
+            ),
+            models.KeepListContent(
+                textHtml="<li>Milk</li>",
+                text="Milk",
+                isChecked=True
+            )
+        ],
+        textContent="Don't forget to buy snacks too.",
+        textContentHtml="<p>Don't forget to buy snacks too.</p>",
+        color="yellow",
+        annotations=[
+            models.KeepAnnotation(
+                description="Weekly grocery run",
+                source="Keep",
+                title="Shopping Note",
+                url="https://keep.google.com/"
+            )
+        ],
+        isTrashed=False,
+        isPinned=True,
+        isArchived=False
+    )