contextbase-plugin-granola 0.2.8__py3-none-any.whl

contextbase_plugin_granola-0.2.8.dist-info/METADATA

@@ -0,0 +1,12 @@
+Metadata-Version: 2.3
+Name: contextbase-plugin-granola
+Version: 0.2.8
+Summary: Granola 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-Python: >=3.14, <3.15

contextbase_plugin_granola-0.2.8.dist-info/RECORD

@@ -0,0 +1,15 @@
+plugin_granola/__init__.py,sha256=G6fbCWFZodK-1_ovN9pgUWDR5--cWs1k7kBruHjcQW4,30
+plugin_granola/binding_config.py,sha256=pswh7ZwSmU879JJE5Me8tS2t2IPvngp-N8e0dQJ6D_0,122
+plugin_granola/component.py,sha256=Pb804dvvmbB17DOXQjl60MWrLonh26LPqvxfjYtW1CY,4018
+plugin_granola/defs/__init__.py,sha256=biedetESbQUyf_8Y3QtxWfvwpfNmfi6LsKWWxc1Tf5E,56
+plugin_granola/defs/defs.yaml,sha256=E0IxqxDTc3JPVjDPxPHOGDU3EQAOcb9igsE6oZ9-YFc,52
+plugin_granola/models/__init__.py,sha256=FEQN98wf1RzFO67kJGVa5LgdGLYz9kdq49ZoQxGG_xM,38
+plugin_granola/models/ctx.py,sha256=alUkc1W_ePlGQWBQtPxB5zXr615J9UqXEmw8kYXW2Tk,2204
+plugin_granola/models/ingress.py,sha256=cATCouL29OYjP5JhhobBZG5t6gVomaSkD2adqHGhoBM,1990
+plugin_granola/models/translators.py,sha256=Sf3Kc9EQ_eb2bYOd27XECWzy6ZQ4MUCCKyU0VVyH0eU,1908
+plugin_granola/plugin.json,sha256=wBtIQq0C5g_N-V1h-l3eSdw1Vv_5D3H9D3xblOco_No,83
+plugin_granola/sources/__init__.py,sha256=OAS7xesKpk54KcG_cVsYD67B0H4wnK0r2hDfVrJbwFY,27
+plugin_granola/sources/snapshot.py,sha256=D8mWU_Xti7kqKLNEM-zfIo6UsMv8EUSAEvVw1HZb4m4,5111
+contextbase_plugin_granola-0.2.8.dist-info/WHEEL,sha256=i9aSRDivn5iP9LaR1BLQX2GNAuriQWPsFwbbWygTX2k,81
+contextbase_plugin_granola-0.2.8.dist-info/METADATA,sha256=-gdpbG-HigOoi9yRCx7_TDMgTXq9UdZC7ufVNMMGNiY,402
+contextbase_plugin_granola-0.2.8.dist-info/RECORD,,

contextbase_plugin_granola-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_granola/__init__.py

@@ -0,0 +1 @@
+"""Granola plugin package."""

plugin_granola/binding_config.py

@@ -0,0 +1,5 @@
+from shared_plugins.bindings import BaseBindingConfigModel
+
+
+class GranolaBindingConfig(BaseBindingConfigModel):
+    pass

plugin_granola/component.py

@@ -0,0 +1,116 @@
+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,
+    require_api_key,
+)
+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 GranolaBindingConfig
+from .sources.snapshot import granola_snapshot_source
+
+PLUGIN_ID = plugin_id_from_module(__file__)
+SNAPSHOT_JOB = "snapshot"
+SNAPSHOT_SOURCE_NAME = dlt_source_name(PLUGIN_ID, SNAPSHOT_JOB)
+SNAPSHOT_NOTE_ASSET_KEY = dagster_dlt_asset_key(SNAPSHOT_SOURCE_NAME, "note")
+SNAPSHOT_TRANSCRIPT_SEGMENT_ASSET_KEY = dagster_dlt_asset_key(
+    SNAPSHOT_SOURCE_NAME,
+    "transcript_segment",
+)
+
+
+def _build_snapshot_specs(
+    partitions_def: dg.PartitionsDefinition,
+    automation_condition: dg.AutomationCondition,
+) -> list[dg.AssetSpec]:
+    return [
+        dg.AssetSpec(
+            key=SNAPSHOT_NOTE_ASSET_KEY,
+            group_name=dagster_asset_group_name(PLUGIN_ID),
+            tags=dagster_asset_tags(PLUGIN_ID),
+            automation_condition=automation_condition,
+            partitions_def=partitions_def,
+        ),
+        dg.AssetSpec(
+            key=SNAPSHOT_TRANSCRIPT_SEGMENT_ASSET_KEY,
+            deps=[SNAPSHOT_NOTE_ASSET_KEY],
+            group_name=dagster_asset_group_name(PLUGIN_ID),
+            tags=dagster_asset_tags(PLUGIN_ID),
+            automation_condition=automation_condition,
+            partitions_def=partitions_def,
+        ),
+    ]
+
+
+class GranolaSyncComponent(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,
+            non_overlapping_automation_condition(
+                dg.AutomationCondition.on_missing()
+                | dg.AutomationCondition.on_cron("*/15 * * * *")
+            ),
+        )
+
+        @dg.multi_asset(
+            specs=snapshot_specs,
+            can_subset=False,
+            name="granola_snapshot",
+            pool=dagster_pool_name(PLUGIN_ID),
+        )
+        def granola_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)
+            parse_binding_config(binding, GranolaBindingConfig)
+
+            source = granola_snapshot_source(
+                binding_id,
+                api_key=require_api_key(binding).api_key,
+            )
+            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="granola_automation_sensor",
+            target=dg.AssetSelection.assets(granola_snapshot_assets),
+            default_status=dg.DefaultSensorStatus.RUNNING,
+            minimum_interval_seconds=30,
+        )
+
+        return dg.Definitions(
+            assets=[granola_snapshot_assets],
+            sensors=[automation_sensor],
+            resources={
+                "dlt_resource": DLT_RESOURCE,
+            },
+        )

plugin_granola/defs/__init__.py

@@ -0,0 +1 @@
+"""Dagster component definitions for plugin_granola."""

plugin_granola/defs/defs.yaml

@@ -0,0 +1 @@
+type: plugin_granola.component.GranolaSyncComponent

plugin_granola/models/__init__.py

@@ -0,0 +1 @@
+"""Granola ingress and ctx models."""

plugin_granola/models/ctx.py

@@ -0,0 +1,68 @@
+from __future__ import annotations
+
+from typing import Any
+
+from pydantic import AwareDatetime
+from shared_plugins.models import CtxModel, IdStr, NonNegativeInt
+
+NOTE_COLUMNS = {
+    "id": {"description": "Granola note ID. Join to transcript_segment.note_id."},
+    "owner": {
+        "data_type": "json",
+        "description": "Granola owner object with email and optional display name.",
+    },
+    "calendar_event": {
+        "data_type": "json",
+        "description": "Granola calendar metadata when the note is linked to an event; NULL otherwise.",
+    },
+    "attendees": {
+        "data_type": "json",
+        "description": "Granola attendee objects for the note.",
+    },
+    "folder_membership": {
+        "data_type": "json",
+        "description": "Granola folders containing the note, including parent_folder_id for nested folders when present.",
+    },
+    "summary_text": {"description": "Granola-generated summary as plain text."},
+    "summary_markdown": {
+        "description": "Granola-generated summary as Markdown. NULL when the note has no generated summary."
+    },
+}
+
+TRANSCRIPT_SEGMENT_COLUMNS = {
+    "note_id": {"description": "Granola note ID. Join to note.id."},
+    "ordinal": {
+        "description": "0-based segment order within a note. Order by this to reconstruct the transcript."
+    },
+    "speaker": {
+        "data_type": "json",
+        "description": "Raw Granola speaker JSON. diarization_label can be NULL and is not a stable person identity.",
+    },
+    "text": {
+        "description": "Full text for this transcript segment; use note_id plus ordinal for context."
+    },
+}
+
+
+class GranolaNoteRow(CtxModel):
+    id: IdStr
+    object: str
+    title: str | None = None
+    owner: dict[str, Any]
+    created_at: AwareDatetime
+    updated_at: AwareDatetime
+    web_url: str
+    calendar_event: dict[str, Any] | None = None
+    attendees: list[dict[str, Any]]
+    folder_membership: list[dict[str, Any]]
+    summary_text: str
+    summary_markdown: str | None = None
+
+
+class GranolaTranscriptSegmentRow(CtxModel):
+    note_id: IdStr
+    ordinal: NonNegativeInt
+    speaker: dict[str, Any]
+    text: str
+    start_time: AwareDatetime
+    end_time: AwareDatetime

plugin_granola/models/ingress.py

@@ -0,0 +1,64 @@
+from __future__ import annotations
+
+from typing import Literal
+
+from pydantic import AwareDatetime, Field
+from shared_plugins.models import IdStr, IngressModel
+
+
+class GranolaPersonIngress(IngressModel):
+    email: str = Field(strict=True)
+    name: str | None = Field(default=None, strict=True)
+
+
+class GranolaCalendarInviteeIngress(IngressModel):
+    email: str = Field(strict=True)
+
+
+class GranolaCalendarEventIngress(IngressModel):
+    event_title: str = Field(strict=True)
+    invitees: list[GranolaCalendarInviteeIngress] = Field(default_factory=list)
+    organiser: str = Field(strict=True)
+    calendar_event_id: str = Field(strict=True)
+    scheduled_start_time: AwareDatetime
+    scheduled_end_time: AwareDatetime
+
+
+class GranolaFolderMembershipIngress(IngressModel):
+    id: IdStr
+    object: str = Field(strict=True)
+    name: str = Field(strict=True)
+    parent_folder_id: IdStr | None = None
+
+
+class GranolaSpeakerIngress(IngressModel):
+    source: str = Field(strict=True)
+    diarization_label: str | None = Field(default=None, strict=True)
+
+
+class GranolaTranscriptSegmentIngress(IngressModel):
+    speaker: GranolaSpeakerIngress
+    text: str = Field(strict=True)
+    start_time: AwareDatetime
+    end_time: AwareDatetime
+
+
+class GranolaNoteListIngress(IngressModel):
+    id: IdStr
+    object: Literal["note"]
+    title: str | None = Field(default=None, strict=True)
+    owner: GranolaPersonIngress
+    created_at: AwareDatetime
+    updated_at: AwareDatetime
+
+
+class GranolaNoteDetailIngress(GranolaNoteListIngress):
+    web_url: str = Field(strict=True)
+    calendar_event: GranolaCalendarEventIngress | None = None
+    attendees: list[GranolaPersonIngress] = Field(default_factory=list)
+    folder_membership: list[GranolaFolderMembershipIngress] = Field(
+        default_factory=list
+    )
+    summary_text: str = Field(strict=True)
+    summary_markdown: str | None = Field(default=None, strict=True)
+    transcript: list[GranolaTranscriptSegmentIngress] | None = None

plugin_granola/models/translators.py

@@ -0,0 +1,55 @@
+from __future__ import annotations
+
+from collections.abc import Iterable, Iterator
+
+from .ctx import GranolaNoteRow, GranolaTranscriptSegmentRow
+from .ingress import GranolaNoteDetailIngress
+
+
+def granola_notes_to_ctx_models(
+    *,
+    binding_id: str,
+    notes: Iterable[GranolaNoteDetailIngress],
+) -> Iterator[GranolaNoteRow]:
+    for note in notes:
+        yield GranolaNoteRow(
+            ctx_binding_id=binding_id,
+            ctx_source_updated_at=note.updated_at,
+            id=note.id,
+            object=note.object,
+            title=note.title,
+            owner=note.owner.model_dump(mode="json"),
+            created_at=note.created_at,
+            updated_at=note.updated_at,
+            web_url=note.web_url,
+            calendar_event=(
+                note.calendar_event.model_dump(mode="json")
+                if note.calendar_event is not None
+                else None
+            ),
+            attendees=[attendee.model_dump(mode="json") for attendee in note.attendees],
+            folder_membership=[
+                folder.model_dump(mode="json") for folder in note.folder_membership
+            ],
+            summary_text=note.summary_text,
+            summary_markdown=note.summary_markdown,
+        )
+
+
+def granola_transcript_segments_to_ctx_models(
+    *,
+    binding_id: str,
+    notes: Iterable[GranolaNoteDetailIngress],
+) -> Iterator[GranolaTranscriptSegmentRow]:
+    for note in notes:
+        for ordinal, segment in enumerate(note.transcript or []):
+            yield GranolaTranscriptSegmentRow(
+                ctx_binding_id=binding_id,
+                ctx_source_updated_at=note.updated_at,
+                note_id=note.id,
+                ordinal=ordinal,
+                speaker=segment.speaker.model_dump(mode="json"),
+                text=segment.text,
+                start_time=segment.start_time,
+                end_time=segment.end_time,
+            )

plugin_granola/plugin.json

@@ -0,0 +1,7 @@
+{
+	"auth": {
+		"type": "api_key"
+	},
+	"mode": "dagster",
+	"plugin_id": "granola"
+}

plugin_granola/sources/__init__.py

@@ -0,0 +1 @@
+"""Granola dlt sources."""

plugin_granola/sources/snapshot.py

@@ -0,0 +1,150 @@
+from __future__ import annotations
+
+from collections.abc import Iterable, Iterator, Mapping
+from typing import Any
+
+import dlt
+from dlt.sources.rest_api import rest_api_resources
+from shared_plugins.naming import (
+    dlt_resource_name,
+    dlt_source_name,
+    plugin_id_from_module,
+)
+from shared_plugins.resources import ctx_dlt_transformer
+
+from ..models.ctx import (
+    NOTE_COLUMNS,
+    TRANSCRIPT_SEGMENT_COLUMNS,
+    GranolaNoteRow,
+    GranolaTranscriptSegmentRow,
+)
+from ..models.ingress import GranolaNoteDetailIngress
+from ..models.translators import (
+    granola_notes_to_ctx_models,
+    granola_transcript_segments_to_ctx_models,
+)
+
+PLUGIN_ID = plugin_id_from_module(__file__)
+JOB = "snapshot"
+GRANOLA_BASE_URL = "https://public-api.granola.ai/v1"
+GRANOLA_INITIAL_UPDATED_AFTER = "1970-01-01T00:00:00Z"
+GRANOLA_PAGE_SIZE = 30
+GRANOLA_NOTES_LIST_RESOURCE_NAME = "granola_notes_list"
+GRANOLA_NOTE_DETAILS_RESOURCE_NAME = "granola_note_details"
+
+
+def build_granola_note_details_resource(*, api_key: str) -> Any:
+    normalized_api_key = api_key.strip()
+    if not normalized_api_key:
+        raise RuntimeError("Granola API key cannot be empty.")
+
+    resources = rest_api_resources(
+        {
+            "client": {
+                "base_url": GRANOLA_BASE_URL,
+                "auth": {
+                    "type": "bearer",
+                    "token": normalized_api_key,
+                },
+                "headers": {
+                    "Accept": "application/json",
+                },
+            },
+            "resources": [
+                {
+                    "name": GRANOLA_NOTES_LIST_RESOURCE_NAME,
+                    "selected": False,
+                    "endpoint": {
+                        "path": "notes",
+                        "method": "GET",
+                        "paginator": {
+                            "type": "cursor",
+                            "cursor_path": "cursor",
+                            "cursor_param": "cursor",
+                        },
+                        "data_selector": "notes",
+                        "params": {
+                            "page_size": GRANOLA_PAGE_SIZE,
+                            "updated_after": "{incremental.start_value}",
+                        },
+                        "incremental": {
+                            "cursor_path": "updated_at",
+                            "initial_value": GRANOLA_INITIAL_UPDATED_AFTER,
+                        },
+                    },
+                },
+                {
+                    "name": GRANOLA_NOTE_DETAILS_RESOURCE_NAME,
+                    "selected": False,
+                    "endpoint": {
+                        "path": f"notes/{{resources.{GRANOLA_NOTES_LIST_RESOURCE_NAME}.id}}",
+                        "method": "GET",
+                        "paginator": "single_page",
+                        "data_selector": "$",
+                        "params": {
+                            "include": "transcript",
+                        },
+                    },
+                },
+            ],
+        }
+    )
+    resources_by_name = {resource.name: resource for resource in resources}
+    return resources_by_name[GRANOLA_NOTE_DETAILS_RESOURCE_NAME]
+
+
+def _ingress_note_details_from_raw(
+    raw_notes: Mapping[str, Any] | Iterable[Mapping[str, Any]],
+) -> Iterator[GranolaNoteDetailIngress]:
+    if isinstance(raw_notes, Mapping):
+        raw_notes = (raw_notes,)
+    for raw_note in raw_notes:
+        yield GranolaNoteDetailIngress.model_validate(raw_note)
+
+
+@dlt.source(name=dlt_source_name(PLUGIN_ID, JOB))
+def granola_snapshot_source(
+    binding_id: str,
+    *,
+    api_key: str,
+) -> tuple[Any, ...]:
+    note_details_resource = build_granola_note_details_resource(api_key=api_key)
+
+    @dlt.transformer(
+        data_from=note_details_resource,
+        name="granola_ingress_note_details",
+        selected=False,
+    )
+    def ingress_note_details_resource(
+        raw_note: Mapping[str, Any] | Iterable[Mapping[str, Any]],
+    ) -> Iterator[GranolaNoteDetailIngress]:
+        yield from _ingress_note_details_from_raw(raw_note)
+
+    @ctx_dlt_transformer(
+        data_from=ingress_note_details_resource,
+        name=dlt_resource_name("note"),
+        write_disposition="merge",
+        primary_key=("_ctx_binding_id", "id"),
+        columns=NOTE_COLUMNS,
+    )
+    def note_resource(
+        note: GranolaNoteDetailIngress,
+    ) -> Iterator[GranolaNoteRow]:
+        yield from granola_notes_to_ctx_models(binding_id=binding_id, notes=[note])
+
+    @ctx_dlt_transformer(
+        data_from=ingress_note_details_resource,
+        name=dlt_resource_name("transcript_segment"),
+        write_disposition="merge",
+        primary_key=("_ctx_binding_id", "note_id", "ordinal"),
+        columns=TRANSCRIPT_SEGMENT_COLUMNS,
+    )
+    def transcript_segment_resource(
+        note: GranolaNoteDetailIngress,
+    ) -> Iterator[GranolaTranscriptSegmentRow]:
+        yield from granola_transcript_segments_to_ctx_models(
+            binding_id=binding_id,
+            notes=[note],
+        )
+
+    return (note_resource, transcript_segment_resource)