contextbase-plugin-whatsapp-local 0.2.8__py3-none-any.whl

contextbase_plugin_whatsapp_local-0.2.8.dist-info/METADATA

@@ -0,0 +1,13 @@
+Metadata-Version: 2.3
+Name: contextbase-plugin-whatsapp-local
+Version: 0.2.8
+Summary: WhatsApp local plugin for ContextBase
+Author: Alizain Feerasta
+Author-email: Alizain Feerasta <alizain.feerasta@gmail.com>
+Requires-Dist: contextbase-shared-plugins==0.2.8
+Requires-Dist: dagster==1.12.14
+Requires-Dist: dagster-dlt==0.28.14
+Requires-Dist: dlt>=1.26.0
+Requires-Dist: pydantic>=2.12.0
+Requires-Dist: sqlalchemy>=2.0.0
+Requires-Python: >=3.14, <3.15

contextbase_plugin_whatsapp_local-0.2.8.dist-info/RECORD

@@ -0,0 +1,16 @@
+plugin_whatsapp_local/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+plugin_whatsapp_local/binding_config.py,sha256=HKdBxiJBVodGfmKU81wMEgqkVVAe2K8-b-D4OGQL1g8,442
+plugin_whatsapp_local/component.py,sha256=PaRcFXZmLtlAJQqyJgZ82Mx9Zkqqwt_OhBabegFZw2Y,4753
+plugin_whatsapp_local/defs/__init__.py,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
+plugin_whatsapp_local/defs/defs.yaml,sha256=38drNqOp9m6Xxo8pIWwX66iyQQlT-Q8MUqjBoQZ5uc4,66
+plugin_whatsapp_local/models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+plugin_whatsapp_local/models/base.py,sha256=Fp7n4EfxQOxzqm34-D-rYyu60P88eZUIewCBNP_Iuc4,119
+plugin_whatsapp_local/models/ctx.py,sha256=tYoXhusFPtBAbGo1NbrtpKFfWfIez6MTPpoytqfLExA,8511
+plugin_whatsapp_local/models/ingress.py,sha256=kbaUxUyP9VI_UNQMdYhES2xcbJQ_2eOoK7PDTXQFVHk,14763
+plugin_whatsapp_local/models/translators.py,sha256=gIeqiQjnmI7poAniTKyoukeV_pZhiOFZp4DCSIVAnqk,10138
+plugin_whatsapp_local/plugin.json,sha256=xb5rdK4thxvUaS18kYgiua4d7Ac-QnUA5P6beCUiZXY,87
+plugin_whatsapp_local/sources/__init__.py,sha256=U4S_2y3zgLZVfMenHRaJFBW8yqh2mUBuI291LGQVOJ8,35
+plugin_whatsapp_local/sources/snapshot.py,sha256=3iDy_FeIUZSzqUJfdiuZaKWP1yLVkkJ4Uwa6SI3wNzU,8969
+contextbase_plugin_whatsapp_local-0.2.8.dist-info/WHEEL,sha256=i9aSRDivn5iP9LaR1BLQX2GNAuriQWPsFwbbWygTX2k,81
+contextbase_plugin_whatsapp_local-0.2.8.dist-info/METADATA,sha256=otkPzLQCU92lRUrGp_8v-5i_duodSSf69U6HEu8Xzi4,449
+contextbase_plugin_whatsapp_local-0.2.8.dist-info/RECORD,,

contextbase_plugin_whatsapp_local-0.2.8.dist-info/WHEEL

@@ -0,0 +1,4 @@
+Wheel-Version: 1.0
+Generator: uv 0.11.15
+Root-Is-Purelib: true
+Tag: py3-none-any

plugin_whatsapp_local/binding_config.py

@@ -0,0 +1,18 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+from pydantic import Field
+
+from shared_plugins.bindings import BaseBindingConfigModel, ResolvedPath
+
+
+class WhatsAppLocalBindingConfig(BaseBindingConfigModel):
+    data_dir: ResolvedPath = Field(
+        default_factory=lambda: (
+            Path.home()
+            / "Library"
+            / "Group Containers"
+            / "group.net.whatsapp.WhatsApp.shared"
+        ),
+    )

plugin_whatsapp_local/component.py

@@ -0,0 +1,137 @@
+import dagster as dg
+from dagster import AssetExecutionContext
+from dagster_dlt import DagsterDltResource
+from shared_plugins.automation import non_overlapping_automation_condition
+from shared_plugins.bindings import parse_binding_config
+from shared_plugins.control_plane import ControlPlaneClient
+from shared_plugins.dlt import resolve_partition_binding, run_dlt_pipeline
+from shared_plugins.naming import (
+    dagster_asset_group_name,
+    dagster_asset_tags,
+    dagster_dlt_asset_key,
+    dagster_partition_def_name,
+    dagster_pool_name,
+    dlt_source_name,
+    plugin_id_from_module,
+)
+from shared_plugins.resources import DLT_RESOURCE
+
+from .binding_config import WhatsAppLocalBindingConfig
+from .sources.snapshot import whatsapp_local_snapshot_source
+
+PLUGIN_ID = plugin_id_from_module(__file__)
+SNAPSHOT_JOB = "snapshot"
+
+
+def _build_snapshot_specs(
+    partitions_def: dg.PartitionsDefinition,
+    automation_condition: dg.AutomationCondition,
+) -> list[dg.AssetSpec]:
+    snapshot_source_name = dlt_source_name(PLUGIN_ID, SNAPSHOT_JOB)
+    shared = dict(
+        group_name=dagster_asset_group_name(PLUGIN_ID),
+        tags=dagster_asset_tags(PLUGIN_ID),
+        automation_condition=automation_condition,
+        partitions_def=partitions_def,
+    )
+
+    return [
+        dg.AssetSpec(
+            key=dagster_dlt_asset_key(snapshot_source_name, "chat_session"),
+            **shared,
+        ),
+        dg.AssetSpec(
+            key=dagster_dlt_asset_key(snapshot_source_name, "message"),
+            **shared,
+        ),
+        dg.AssetSpec(
+            key=dagster_dlt_asset_key(snapshot_source_name, "contact"),
+            **shared,
+        ),
+        dg.AssetSpec(
+            key=dagster_dlt_asset_key(snapshot_source_name, "group_info"),
+            **shared,
+        ),
+        dg.AssetSpec(
+            key=dagster_dlt_asset_key(snapshot_source_name, "group_member"),
+            **shared,
+        ),
+        dg.AssetSpec(
+            key=dagster_dlt_asset_key(snapshot_source_name, "profile_push_name"),
+            **shared,
+        ),
+        dg.AssetSpec(
+            key=dagster_dlt_asset_key(snapshot_source_name, "message_data_item"),
+            **shared,
+        ),
+        dg.AssetSpec(
+            key=dagster_dlt_asset_key(snapshot_source_name, "aggregate_call_event"),
+            **shared,
+        ),
+        dg.AssetSpec(
+            key=dagster_dlt_asset_key(snapshot_source_name, "call_event"),
+            **shared,
+        ),
+        dg.AssetSpec(
+            key=dagster_dlt_asset_key(snapshot_source_name, "call_event_participant"),
+            **shared,
+        ),
+    ]
+
+
+class WhatsAppLocalSyncComponent(dg.Component):
+    def build_defs(self, context: dg.ComponentLoadContext) -> dg.Definitions:
+        partitions_def = dg.DynamicPartitionsDefinition(
+            name=dagster_partition_def_name(PLUGIN_ID)
+        )
+        snapshot_specs = _build_snapshot_specs(
+            partitions_def=partitions_def,
+            automation_condition=non_overlapping_automation_condition(
+                dg.AutomationCondition.on_missing()
+                | dg.AutomationCondition.on_cron("*/15 * * * *")
+            ),
+        )
+
+        @dg.multi_asset(
+            specs=snapshot_specs,
+            can_subset=True,
+            name="whatsapp_local_snapshot",
+            pool=dagster_pool_name(PLUGIN_ID),
+        )
+        def whatsapp_local_snapshot_assets(
+            context: AssetExecutionContext,
+            dlt_resource: DagsterDltResource,
+            control_plane: dg.ResourceParam[ControlPlaneClient],
+        ):
+            binding = resolve_partition_binding(
+                context=context,
+                control_plane=control_plane,
+                plugin_id=PLUGIN_ID,
+            )
+            binding_id = str(binding.binding_id)
+            cfg = parse_binding_config(binding, WhatsAppLocalBindingConfig)
+
+            source = whatsapp_local_snapshot_source(binding_id, cfg)
+            yield from run_dlt_pipeline(
+                context=context,
+                dlt_resource=dlt_resource,
+                source=source,
+                plugin_id=PLUGIN_ID,
+                binding_id=binding_id,
+                job_name=SNAPSHOT_JOB,
+            )
+
+        automation_sensor = dg.AutomationConditionSensorDefinition(
+            name="whatsapp_local_automation_sensor",
+            target=dg.AssetSelection.assets(whatsapp_local_snapshot_assets),
+            default_status=dg.DefaultSensorStatus.RUNNING,
+            minimum_interval_seconds=30,
+        )
+
+        return dg.Definitions(
+            assets=[whatsapp_local_snapshot_assets],
+            sensors=[automation_sensor],
+            resources={
+                "dlt_resource": DLT_RESOURCE,
+            },
+        )

plugin_whatsapp_local/defs/__init__.py

@@ -0,0 +1 @@
+

plugin_whatsapp_local/defs/defs.yaml

@@ -0,0 +1,2 @@
+type: plugin_whatsapp_local.component.WhatsAppLocalSyncComponent
+

plugin_whatsapp_local/models/base.py

@@ -0,0 +1,7 @@
+from __future__ import annotations
+
+from sqlalchemy.orm import DeclarativeBase
+
+
+class Base(DeclarativeBase):
+    pass

plugin_whatsapp_local/models/ctx.py

@@ -0,0 +1,260 @@
+from __future__ import annotations
+
+from pydantic import AwareDatetime
+from shared_plugins.models import CtxModel
+
+
+class ChatSessionRow(CtxModel):
+    pk: int
+    ent: int | None = None
+    opt: int | None = None
+    archived: int | None = None
+    contact_abid: int | None = None
+    flags: int | None = None
+    hidden: int | None = None
+    identity_verification_epoch: int | None = None
+    identity_verification_state: int | None = None
+    message_counter: int | None = None
+    removed: int | None = None
+    session_type: int | None = None
+    spotlight_status: int | None = None
+    unread_count: int | None = None
+    group_info_pk: int | None = None
+    last_message_pk: int | None = None
+    properties_pk: int | None = None
+    last_message_date: AwareDatetime | None = None
+    location_sharing_end_date: AwareDatetime | None = None
+    contact_identifier: str | None = None
+    contact_jid: str | None = None
+    etag: str | None = None
+    last_message_text: str | None = None
+    partner_name: str | None = None
+    saved_input: str | None = None
+
+
+class MessageRow(CtxModel):
+    pk: int
+    ent: int | None = None
+    opt: int | None = None
+    child_messages_delivered_count: int | None = None
+    child_messages_played_count: int | None = None
+    child_messages_read_count: int | None = None
+    data_item_version: int | None = None
+    doc_id: int | None = None
+    enc_retry_count: int | None = None
+    filtered_recipient_count: int | None = None
+    flags: int | None = None
+    group_event_type: int | None = None
+    is_from_me: int | None = None
+    message_error_status: int | None = None
+    message_status: int | None = None
+    message_type: int | None = None
+    sort: int | None = None
+    spotlight_status: int | None = None
+    starred: int | None = None
+    chat_session_pk: int | None = None
+    group_member_pk: int | None = None
+    last_session_pk: int | None = None
+    media_item_pk: int | None = None
+    message_info_pk: int | None = None
+    parent_message_pk: int | None = None
+    message_date: AwareDatetime | None = None
+    sent_date: AwareDatetime | None = None
+    from_jid: str | None = None
+    media_section_id: str | None = None
+    phash: str | None = None
+    stanza_id: str | None = None
+    text: str | None = None
+    to_jid: str | None = None
+
+
+class ContactRow(CtxModel):
+    pk: int
+    ent: int | None = None
+    opt: int | None = None
+    disappearing_mode_duration: int | None = None
+    phone_status: int | None = None
+    section_sort: int | None = None
+    sort: int | None = None
+    source: int | None = None
+    spotlight_status: int | None = None
+    sync_policy: int | None = None
+    sync_with_server_state: int | None = None
+    about_expiration_timestamp: AwareDatetime | None = None
+    about_timestamp: AwareDatetime | None = None
+    disappearing_mode_timestamp: AwareDatetime | None = None
+    last_updated: AwareDatetime | None = None
+    about_emoji: str | None = None
+    about_text: str | None = None
+    business_name: str | None = None
+    full_name: str | None = None
+    given_name: str | None = None
+    highlighted_name: str | None = None
+    identifier: str | None = None
+    last_name: str | None = None
+    lid: str | None = None
+    lid_hash: str | None = None
+    localized_phone_number: str | None = None
+    notes: str | None = None
+    phone_number: str | None = None
+    phone_number_label: str | None = None
+    pn_hash: str | None = None
+    search_token_list: str | None = None
+    section_title: str | None = None
+    unique_id: str | None = None
+    username: str | None = None
+    whatsapp_id: str | None = None
+    device_list: bytes | None = None
+
+
+class GroupInfoRow(CtxModel):
+    pk: int
+    ent: int | None = None
+    opt: int | None = None
+    state: int | None = None
+    chat_session_pk: int | None = None
+    last_message_owner_pk: int | None = None
+    creation_date: AwareDatetime | None = None
+    subject_timestamp: AwareDatetime | None = None
+    creator_jid: str | None = None
+    owner_jid: str | None = None
+    picture_id: str | None = None
+    picture_path: str | None = None
+    source_jid: str | None = None
+    subject_owner_jid: str | None = None
+
+
+class GroupMemberRow(CtxModel):
+    pk: int
+    ent: int | None = None
+    opt: int | None = None
+    contact_abid: int | None = None
+    is_active: int | None = None
+    is_admin: int | None = None
+    sender_key_sent: int | None = None
+    chat_session_pk: int | None = None
+    recent_group_chat_pk: int | None = None
+    contact_identifier: str | None = None
+    member_jid: str | None = None
+
+
+class ProfilePushNameRow(CtxModel):
+    pk: int
+    ent: int | None = None
+    opt: int | None = None
+    jid: str | None = None
+    push_name: str | None = None
+
+
+class MessageDataItemRow(CtxModel):
+    pk: int
+    ent: int | None = None
+    opt: int | None = None
+    index: int | None = None
+    owns_thumbnail: int | None = None
+    type: int | None = None
+    message_pk: int | None = None
+    date: AwareDatetime | None = None
+    chat_jid: str | None = None
+    content1: str | None = None
+    content2: str | None = None
+    matched_text: str | None = None
+    section_id: str | None = None
+    sender_jid: str | None = None
+    summary: str | None = None
+    thumbnail_path: str | None = None
+    title: str | None = None
+
+
+# ---------------------------------------------------------------------------
+# CallHistory.sqlite
+# ---------------------------------------------------------------------------
+
+
+class AggregateCallEventRow(CtxModel):
+    pk: int
+    ent: int | None = None
+    opt: int | None = None
+    incoming: int | None = None
+    missed: int | None = None
+    missed_reason: int | None = None
+    video: int | None = None
+    first_date: AwareDatetime | None = None
+    link_token: str | None = None
+
+
+class CallEventRow(CtxModel):
+    pk: int
+    ent: int | None = None
+    opt: int | None = None
+    bytes_received: int | None = None
+    bytes_sent: int | None = None
+    is_lightweight: int | None = None
+    outcome: int | None = None
+    aggregate_call_event_pk: int | None = None
+    aggregate_call_event_fok: int | None = None
+    date: AwareDatetime | None = None
+    duration: float | None = None
+    call_id_string: str | None = None
+    group_call_creator_user_jid_string: str | None = None
+    group_jid_string: str | None = None
+    reminder_id: str | None = None
+    scheduled_id: str | None = None
+
+
+class CallEventParticipantRow(CtxModel):
+    pk: int
+    ent: int | None = None
+    opt: int | None = None
+    outcome: int | None = None
+    call_event_pk: int | None = None
+    call_event_fok: int | None = None
+    jid_string: str | None = None
+
+
+CHAT_SESSION_COLUMNS: dict[str, dict[str, str]] = {
+    "session_type": {
+        "description": "1 = group chat (1:1 with group_info), 0 = 1:1 DM. Other values exist for broadcasts/statuses."
+    },
+}
+
+
+MESSAGE_COLUMNS: dict[str, dict[str, str]] = {
+    "is_from_me": {
+        "description": "When 1, the phone owner sent the message and group_member_pk is NULL. When 0, resolve sender via group_member_pk."
+    },
+    "from_jid": {
+        "description": "In group messages this is the GROUP's jid, not the sender's. Use group_member_pk for sender identity."
+    },
+}
+
+
+GROUP_MEMBER_COLUMNS: dict[str, dict[str, str]] = {
+    "member_jid": {
+        "description": "Join to contact via lid for @lid jids, via whatsapp_id for @s.whatsapp.net jids — neither column alone covers both."
+    },
+}
+
+
+AGGREGATE_CALL_EVENT_COLUMNS: dict[str, dict[str, str]] = {
+    "incoming": {
+        "description": "1 if the call group was received, 0 if placed by self. Combine with call_event.group_call_creator_user_jid_string to identify the peer."
+    },
+}
+
+
+CALL_EVENT_COLUMNS: dict[str, dict[str, str]] = {
+    "group_call_creator_user_jid_string": {
+        "description": "Call initiator's jid — set for ALL calls despite the name, not just group calls. For aggregate_call_event.incoming=0 (outgoing) it equals self; for incoming=1 it equals the peer."
+    },
+    "outcome": {
+        "description": "Undocumented enum at the source (values 0, 1, 4, 5 observed). Prefer `duration > 0` to detect an answered call."
+    },
+}
+
+
+CALL_EVENT_PARTICIPANT_COLUMNS: dict[str, dict[str, str]] = {
+    "jid_string": {
+        "description": "Other-party jid for this call leg. For outgoing 1:1 this is the peer; if it differs from call_event.group_call_creator_user_jid_string on an incoming call, the call is almost always multi-party (not 1:1)."
+    },
+}