ftmq 4.2.4__tar.gz → 4.3.0__tar.gz

{ftmq-4.2.4 → ftmq-4.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: ftmq
-Version: 4.2.4
+Version: 4.3.0
 Summary: followthemoney query dsl and io helpers
 License: AGPLv3+
 Author: Simon Wörpel
@@ -19,17 +19,17 @@ Provides-Extra: postgres
 Provides-Extra: redis
 Provides-Extra: sql
 Requires-Dist: alephclient (>=2.6.0,<3.0.0) ; extra == "aleph"
-Requires-Dist: anystore (>=0.3.11,<0.4.0)
+Requires-Dist: anystore (>=0.4.0,<0.5.0)
 Requires-Dist: click (>=8.2.1,<9.0.0)
 Requires-Dist: click-default-group (>=1.2.4,<2.0.0)
-Requires-Dist: deltalake (>=1.1.4,<2.0.0) ; extra == "lake"
-Requires-Dist: duckdb (>=1.3.2,<2.0.0) ; extra == "lake"
+Requires-Dist: deltalake (>=1.2.0,<2.0.0) ; extra == "lake"
+Requires-Dist: duckdb (>=1.4.1,<2.0.0) ; extra == "lake"
 Requires-Dist: fakeredis (>=2.26.2,<3.0.0) ; extra == "redis"
-Requires-Dist: followthemoney (>=4.2.0,<5.0.0)
+Requires-Dist: followthemoney (>=4.3.0,<5.0.0)
 Requires-Dist: furl (>=2.1.4,<3.0.0) ; extra == "aleph"
-Requires-Dist: nomenklatura (>=4.1.2,<5.0.0)
+Requires-Dist: nomenklatura (>=4.1.9,<5.0.0)
 Requires-Dist: orjson (>=3.10.18,<4.0.0)
-Requires-Dist: pandas (>=2.3.2,<3.0.0) ; extra == "lake"
+Requires-Dist: pandas (>=2.3.3,<3.0.0) ; extra == "lake"
 Requires-Dist: plyvel (>=1.5.1,<2.0.0) ; extra == "level"
 Requires-Dist: psycopg[pool] (>=3.2.9,<4.0.0) ; extra == "postgres"
 Requires-Dist: pyarrow (>=21.0.0,<22.0.0) ; extra == "lake"
@@ -37,7 +37,7 @@ Requires-Dist: pycountry (>=24.6.1,<25.0.0)
 Requires-Dist: pydantic (>=2.11.3,<3.0.0)
 Requires-Dist: pyicu (>=2.15.2,<3.0.0)
 Requires-Dist: redis (>=5.2.1,<6.0.0) ; extra == "redis"
-Requires-Dist: rigour (>=1.3.0,<2.0.0)
+Requires-Dist: rigour (>=1.3.13,<2.0.0)
 Requires-Dist: sqlalchemy (>=2.0.36,<3.0.0) ; extra == "postgres"
 Requires-Dist: sqlalchemy (>=2.0.36,<3.0.0) ; extra == "sql"
 Project-URL: Documentation, https://docs.investigraph.dev/lib/ftmq

{ftmq-4.2.4 → ftmq-4.3.0}/ftmq/__init__.py

@@ -2,7 +2,7 @@ from ftmq.io import smart_read_proxies, smart_write_proxies
 from ftmq.query import Query
 from ftmq.util import make_entity
 
-__version__ = "4.2.4"
+__version__ = "4.3.0"
 __all__ = [
     "smart_read_proxies",
     "smart_write_proxies",

{ftmq-4.2.4 → ftmq-4.3.0}/ftmq/cli.py

@@ -11,6 +11,9 @@ from ftmq.model.dataset import Catalog, Dataset
 from ftmq.model.stats import Collector
 from ftmq.query import Query
 from ftmq.store import get_store
+from ftmq.store.fragments import get_fragments
+from ftmq.store.fragments import get_store as get_fragments_store
+from ftmq.store.fragments.settings import Settings as FragmentsSettings
 from ftmq.util import apply_dataset, parse_unknown_filters
 
 log = get_logger(__name__)
@@ -311,6 +314,61 @@ def store_iterate(
     smart_write_proxies(output_uri, store.iterate())
 
 
+@cli.group()
+def fragments():
+    pass
+
+
+fragments_settings = FragmentsSettings()
+
+
+@fragments.command("list-datasets")
+@click.option(
+    "-i",
+    "--input-uri",
+    default=fragments_settings.database_uri,
+    show_default=True,
+    help="input file or uri",
+)
+@click.option(
+    "-o", "--output-uri", default="-", show_default=True, help="output file or uri"
+)
+def fragments_list_datasets(
+    input_uri: str = fragments_settings.database_uri,
+    output_uri: str = "-",
+):
+    """
+    List datasets within a fragments store
+    """
+    store = get_fragments_store(input_uri)
+    datasets = [ds.name for ds in store.all()]
+    smart_write(output_uri, "\n".join(datasets).encode() + b"\n")
+
+
+@fragments.command("iterate")
+@click.option(
+    "-i",
+    "--input-uri",
+    default=fragments_settings.database_uri,
+    show_default=True,
+    help="fragments store input uri",
+)
+@click.option(
+    "-o", "--output-uri", default="-", show_default=True, help="output file or uri"
+)
+@click.option("-d", "--dataset", required=True, help="Dataset name to iterate")
+def fragments_iterate(
+    input_uri: str = fragments_settings.database_uri,
+    output_uri: str = "-",
+    dataset: str = None,
+):
+    """
+    Iterate all entities from a fragments dataset
+    """
+    fragments = get_fragments(dataset, database_uri=input_uri)
+    smart_write_proxies(output_uri, fragments.iterate())
+
+
 @cli.command("aggregate")
 @click.option(
     "-i", "--input-uri", default="-", show_default=True, help="input file or uri"

{ftmq-4.2.4 → ftmq-4.3.0}/ftmq/store/fragments/dataset.py

@@ -1,4 +1,5 @@
 import logging
+from contextlib import contextmanager
 from datetime import datetime
 from typing import Generator, Iterable, TypeAlias
 
@@ -9,6 +10,7 @@ from normality import slugify
 from sqlalchemy import (
     JSON,
     Column,
+    Connection,
     DateTime,
     String,
     Table,
@@ -43,6 +45,27 @@ except ImportError:
 EntityFragments: TypeAlias = Generator[EntityProxy, None, None]
 
 
+@contextmanager
+def disable_timeout(conn: Connection, store):
+    # for long running iterations (e.g. re-index in OpenAleph), for postgres we
+    # don't want to get cancelled if a idle_in_transaction_timeout is configured
+    # on the server
+    if store.is_postgres:
+        raw_conn = conn.connection.driver_connection
+        with raw_conn.cursor() as cursor:
+            cursor.execute("SET idle_in_transaction_session_timeout = 0")
+    try:
+        yield conn
+    finally:
+        if store.is_postgres:
+            try:
+                raw_conn = conn.connection.driver_connection
+                with raw_conn.cursor() as cursor:
+                    cursor.execute("SET idle_in_transaction_session_timeout = DEFAULT")
+            except Exception:
+                pass  # Connection might be closed
+
+
 class Fragments(object):
     def __init__(self, store, name, origin=NULL_ORIGIN):
         self.store = store
@@ -123,12 +146,13 @@ class Fragments(object):
         # stmt = stmt.order_by(self.table.c.fragment)
         conn = self.store.engine.connect()
         try:
-            conn = conn.execution_options(stream_results=True)
-            for ent in conn.execute(stmt):
-                data = {"id": ent.id, "datasets": [self.name], **ent.entity}
-                if ent.origin != NULL_ORIGIN:
-                    data["origin"] = ent.origin
-                yield data
+            with disable_timeout(conn, self.store) as conn:
+                conn = conn.execution_options(stream_results=True)
+                for ent in conn.execute(stmt):
+                    data = {"id": ent.id, "datasets": [self.name], **ent.entity}
+                    if ent.origin != NULL_ORIGIN:
+                        data["origin"] = ent.origin
+                    yield data
         except Exception:
             self.reset()
             raise
@@ -146,6 +170,10 @@ class Fragments(object):
                 raise
 
     def iterate(self, entity_id=None, skip_errors=False) -> EntityFragments:
+        if entity_id is None:
+            log.info("Using batched iteration for complete dataset.")
+            yield from self.iterate_batched()
+            return
         entity = None
         invalid = None
         fragments = 1
@@ -181,6 +209,39 @@ class Fragments(object):
         if entity is not None:
             yield entity
 
+    def iterate_batched(self, skip_errors=False, batch_size=10_000) -> EntityFragments:
+        """
+        For large datasets an overall sort is not feasible, so we iterate in
+        sorted batched IDs.
+        """
+        for entity_ids in self.get_sorted_id_batches(batch_size):
+            yield from self.iterate(entity_id=entity_ids, skip_errors=skip_errors)
+
+    def get_sorted_id_batches(
+        self, batch_size=10_000
+    ) -> Generator[list[str], None, None]:
+        """
+        Get sorted ID batches to speed up iteration and useful to parallelize
+        processing of iterator Entities
+        """
+        last_id = None
+        with self.store.engine.connect() as conn:
+            while True:
+                stmt = select(self.table.c.id).distinct()
+                if last_id is not None:
+                    stmt = stmt.where(self.table.c.id > last_id)
+                stmt = stmt.order_by(self.table.c.id).limit(batch_size)
+                try:
+                    res = conn.execute(stmt)
+                    entity_ids = [r.id for r in res.fetchall()]
+                    if not entity_ids:
+                        return
+                    yield entity_ids
+                    last_id = entity_ids[-1]
+                except Exception:
+                    self.reset()
+                    raise
+
     def statements(
         self,
         entity_ids: Iterable[str] | None = None,

{ftmq-4.2.4 → ftmq-4.3.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ftmq"
-version = "4.2.4"
+version = "4.3.0"
 description = "followthemoney query dsl and io helpers"
 authors = [{ name = "Simon Wörpel", email = "simon.woerpel@pm.me" }]
 license = "AGPLv3+"
@@ -15,10 +15,10 @@ classifiers = [
 ]
 requires-python = ">=3.11,<4"
 dependencies = [
-    "anystore (>=0.3.11,<0.4.0)",
-    "followthemoney (>=4.2.0,<5.0.0)",
-    "nomenklatura (>=4.1.2,<5.0.0)",
-    "rigour (>=1.3.0,<2.0.0)",
+    "anystore (>=0.4.0,<0.5.0)",
+    "followthemoney (>=4.3.0,<5.0.0)",
+    "nomenklatura (>=4.1.9,<5.0.0)",
+    "rigour (>=1.3.13,<2.0.0)",
     "click (>=8.2.1,<9.0.0)",
     "click-default-group (>=1.2.4,<2.0.0)",
     "orjson (>=3.10.18,<4.0.0)",
@@ -33,9 +33,9 @@ sql = ["sqlalchemy (>=2.0.36,<3.0.0)"]
 postgres = ["sqlalchemy (>=2.0.36,<3.0.0)", "psycopg[pool] (>=3.2.9,<4.0.0)"]
 redis = ["redis (>=5.2.1,<6.0.0)", "fakeredis (>=2.26.2,<3.0.0)"]
 lake = [
-    "duckdb (>=1.3.2,<2.0.0)",
-    "pandas (>=2.3.2,<3.0.0)",
-    "deltalake (>=1.1.4,<2.0.0)",
+    "duckdb (>=1.4.1,<2.0.0)",
+    "pandas (>=2.3.3,<3.0.0)",
+    "deltalake (>=1.2.0,<2.0.0)",
     "pyarrow (>=21.0.0,<22.0.0)",
 ]
 aleph = ["furl (>=2.1.4,<3.0.0)", "alephclient (>=2.6.0,<3.0.0)"]
@@ -51,10 +51,10 @@ Issues = "https://github.com/dataresearchcenter/ftmq/issues"
 
 [tool.poetry.group.dev.dependencies]
 pytest = ">=7.4.3,<9.0.0"
-pytest-cov = ">=4.1,<7.0"
+pytest-cov = ">=4.1,<8.0"
 pytest-env = "^1.1.1"
 black = ">=23.11,<26.0"
-isort = "^6.0.1"
+isort = "^6.1.0"
 mypy = "^1.17.1"
 pre-commit = "^4.0.1"
 flake8 = ">=6.1,<8.0"
@@ -62,7 +62,7 @@ ipdb = "^0.13.13"
 bump2version = "^1.0.1"
 mkdocs = "^1.6.1"
 mkdocs-autorefs = "^1.4.3"
-mkdocstrings-python = "^1.18.0"
+mkdocstrings-python = "^1.18.2"
 mkdocs-material = "^9.6.18"
 mkdocs-click = "^0.9.0"