datamaestro-text 2026.1.1__py3-none-any.whl → 2026.2.2__py3-none-any.whl

datamaestro_text/__init__.py

@@ -1,6 +1,6 @@
 import datamaestro
 
-from .version import version, version_tuple
+from .version import version as version, version_tuple as version_tuple
 
 
 class Repository(datamaestro.Repository):

datamaestro_text/config/com/github/ikat.py

@@ -1,6 +1,5 @@
 # See documentation on https://datamaestro.readthedocs.io
 
-import bz2
 from datamaestro.download import reference
 from datamaestro.definitions import datatasks, datatags, dataset
 from datamaestro_text.data.conversation.base import ConversationUserTopics

datamaestro_text/config/com/oscar-corpus.py

@@ -1,4 +1,4 @@
-from datamaestro.definitions import datatasks, datatags, dataset
+from datamaestro.definitions import dataset
 from datamaestro.download.single import filedownloader
 from datamaestro_text.data.text import TextFile
 from datamaestro.utils import HashCheck

datamaestro_text/config/edu/stanford/aclimdb.py

@@ -1,5 +1,5 @@
 from datamaestro.data.ml import FolderBased, Supervised
-from datamaestro.definitions import datatasks, datatags, dataset
+from datamaestro.definitions import dataset
 from datamaestro.download.archive import tardownloader
 
 

datamaestro_text/config/edu/stanford/glove.py

@@ -5,7 +5,6 @@ GloVe is an unsupervised learning algorithm for obtaining vector representations
 """
 
 from datamaestro.definitions import dataset
-from datamaestro.data import Base, Generic
 from datamaestro.download import reference
 from datamaestro.download.archive import zipdownloader
 from datamaestro.download.single import filedownloader

datamaestro_text/config/gov/nist/ir/covid.py

@@ -1,5 +1,4 @@
-"""CORD-19 dataset
-"""
+"""CORD-19 dataset"""
 
 from datamaestro.annotations.agreement import useragreement
 from datamaestro.definitions import datatasks, dataset

datamaestro_text/config/io/metamind/research/wikitext.py

@@ -1,4 +1,4 @@
-from datamaestro.data import Base, File
+from datamaestro.data import File
 from datamaestro.definitions import (
     datatasks,
     datatags,

datamaestro_text/data/conversation/__init__.py

@@ -1,8 +1,8 @@
 from .base import (
-    AnswerEntry,
-    ConversationDataset,
-    ConversationHistory,
-    ConversationHistoryItem,
-    DecontextualizedItem,
-    EntryType,
+    AnswerEntry as AnswerEntry,
+    ConversationDataset as ConversationDataset,
+    ConversationHistory as ConversationHistory,
+    ConversationHistoryItem as ConversationHistoryItem,
+    DecontextualizedItem as DecontextualizedItem,
+    EntryType as EntryType,
 )

datamaestro_text/data/conversation/base.py

@@ -2,13 +2,13 @@ from abc import ABC, abstractmethod
 from enum import Enum
 from datamaestro_text.data.ir.base import IDItem, SimpleTextItem
 from experimaestro import Param
-from typing import Dict, Generic, Iterator, List, Optional, Sequence, Tuple
+from typing import Dict, Iterator, List, Optional, Sequence, Tuple
 from attr import define
 from datamaestro.record import record_type
 from datamaestro.data import Base
 from datamaestro.record import Record, Item
 from datamaestro_text.data.ir import TopicRecord, Topics
-from datamaestro_text.utils.iter import FactoryIterable, LazyList, RangeView
+from datamaestro_text.utils.iter import FactoryIterable, LazyList
 
 # ---- Basic types
 

datamaestro_text/data/conversation/canard.py

@@ -11,7 +11,6 @@ from datamaestro_text.data.conversation.base import (
     EntryType,
 )
 from datamaestro_text.data.ir import IDItem, SimpleTextItem
-import logging
 
 
 @define(kw_only=True)
@@ -82,9 +81,9 @@ class CanardDataset(ConversationDataset, File):
                 )
             else:
                 # The utterance before the last is the last user query
-                assert (
-                    entry.history[-2] == history[-1][SimpleTextItem].text
-                ), f"{entry.dialogue_id} {entry.history} / {history[-4:-1]}"
+                assert entry.history[-2] == history[-1][SimpleTextItem].text, (
+                    f"{entry.dialogue_id} {entry.history} / {history[-4:-1]}"
+                )
 
                 # The last utterance is the system side
                 history.append(

datamaestro_text/data/conversation/ikat.py

@@ -5,7 +5,6 @@ import logging
 from datamaestro.data import File
 from datamaestro.record import Record
 
-from datamaestro_text.data.ir import Topics
 from datamaestro_text.data.ir.base import (
     IDItem,
     SimpleTextItem,

datamaestro_text/data/conversation/orconvqa.py

@@ -113,9 +113,9 @@ class OrConvQADataset(ConversationDataset, File):
                 if relevance > 0:
                     relevances[rank] = (entry.answer.answer_start, None)
 
-            assert (
-                len(relevances) <= 1
-            ), f"Too many relevance labels ({len(relevances)}) for {entry.query_id}"
+            assert len(relevances) <= 1, (
+                f"Too many relevance labels ({len(relevances)}) for {entry.query_id}"
+            )
 
             history.append(
                 Record(

datamaestro_text/data/embeddings.py

@@ -20,6 +20,7 @@ class WordEmbeddings(Base):
 
 class WordEmbeddingsText(WordEmbeddings, File):
     """Word embeddings as a text word / values"""
+
     encoding: Meta[str] = "utf-8"
 
     def load(self):

datamaestro_text/data/ir/__init__.py

@@ -6,7 +6,7 @@ from functools import cached_property
 import logging
 from pathlib import Path
 from attrs import define
-from typing import Callable, Dict, Iterator, List, Optional, Tuple, Type, TYPE_CHECKING
+from typing import Callable, Dict, Iterator, List, Optional, Tuple, Type
 import random
 from experimaestro import Config
 from datamaestro.definitions import datatasks, Param, Meta

datamaestro_text/data/ir/base.py

@@ -1,7 +1,7 @@
 from abc import ABC, abstractmethod
 from attrs import define
 from typing import List
-from datamaestro.record import Record, Item, record_type
+from datamaestro.record import Record, Item
 
 
 TopicRecord = DocumentRecord = Record

datamaestro_text/data/ir/data.py

@@ -1 +1 @@
-from .base import *
+from .base import *  # noqa: F403

datamaestro_text/data/ir/formats.py

@@ -1,5 +1,5 @@
 from functools import cached_property
-from typing import ClassVar, Tuple, List
+from typing import Tuple, List
 from attrs import define
 from datamaestro.record import record_type
 from ir_datasets.datasets.wapo import WapoDocMedia
@@ -10,6 +10,7 @@ from ir_datasets.datasets.cord19 import Cord19FullTextSection
 @define
 class DocumentWithTitle(TextItem):
     """Web document with title and body"""
+
     title: str
     body: str
 

datamaestro_text/data/ir/stores.py

@@ -82,7 +82,7 @@ class IKatClueWeb22DocumentStore(LZ4DocumentStore):
 
                 file_checksum = hasher.hexdigest()
                 assert file_checksum == checksum, (
-                    f"Expected {checksum}, " f"got {file_checksum} for {filename}"
+                    f"Expected {checksum}, got {file_checksum} for {filename}"
                 )
 
             # Get the MD5 hashes of all the passages

datamaestro_text/data/text.py

@@ -14,6 +14,7 @@ class TrainingText(Supervised):
 
 class TextFolder(Folder):
     "A folder composed of texts"
+
     pass
 
 

datamaestro_text/datasets/__init__.py

@@ -0,0 +1 @@
+# IR datasets integration package

datamaestro_text/datasets/irds/data.py

@@ -1,6 +1,5 @@
 import logging
 from abc import ABC, abstractmethod
-from dataclasses import dataclass
 from functools import cached_property, partial
 from pathlib import Path
 from typing import Dict, Iterator, List, NamedTuple, Tuple, Type
@@ -613,11 +612,7 @@ class Cast2022TopicsHandler(CastTopicsHandler):
             records = []
             nodes: Dict[str, ConversationTreeNode] = {}
 
-            for (
-                query
-            ) in (
-                self.dataset.dataset.queries_iter()
-            ):  # type: _irds.trec_cast.Cast2022Query
+            for query in self.dataset.dataset.queries_iter():  # type: _irds.trec_cast.Cast2022Query
                 parent = nodes[query.parent_id] if query.parent_id else None
 
                 if query.participant == "User":

datamaestro_text/download/tmdb.py

@@ -10,7 +10,6 @@ from collections import namedtuple
 
 from datamaestro.download import Download
 from datamaestro.definitions import AbstractDataset
-from datamaestro.utils import TemporaryDirectory
 
 APIKEY_KEY = "org.themoviedb.apikey"
 

datamaestro_text/transforms/ir/__init__.py

@@ -2,7 +2,6 @@ import logging
 import gzip
 from abc import ABC, abstractmethod
 from pathlib import Path
-from typing import Type
 from experimaestro import Config, Task, Param, Annotated, pathgenerator, Option, tqdm
 import numpy as np
 from datamaestro.record import RecordType
@@ -131,22 +130,22 @@ class ShuffledTrainingTripletsLines(Task):
 
     def __validate__(self):
         if self.topic_ids:
-            assert self.data.topic_recordtype.has(
-                ir.IDItem
-            ), f"No topic ID in the source data ({self.data.topic_recordtype})"
+            assert self.data.topic_recordtype.has(ir.IDItem), (
+                f"No topic ID in the source data ({self.data.topic_recordtype})"
+            )
         else:
-            assert self.data.topic_recordtype.has(
-                ir.TextItem
-            ), f"No topic text in the source data ({self.data.topic_recordtype})"
+            assert self.data.topic_recordtype.has(ir.TextItem), (
+                f"No topic text in the source data ({self.data.topic_recordtype})"
+            )
 
         if self.doc_ids:
-            assert self.data.document_recordtype.has(
-                ir.IDItem
-            ), "No doc ID in the source data"
+            assert self.data.document_recordtype.has(ir.IDItem), (
+                "No doc ID in the source data"
+            )
         else:
-            assert self.data.document_recordtype.has(
-                ir.TextItem
-            ), "No doc text in the source data"
+            assert self.data.document_recordtype.has(ir.TextItem), (
+                "No doc text in the source data"
+            )
 
     def task_outputs(self, dep):
         return dep(

datamaestro_text/utils/shuffle.py

@@ -50,7 +50,7 @@ def shuffle(
     *,
     memory=MEMORY,
     random=None,
-    tmp_path: Optional[Path] = None
+    tmp_path: Optional[Path] = None,
 ):
     """Shuffle using temporary file"""
     if random is None:

datamaestro_text/version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '2026.1.1'
-__version_tuple__ = version_tuple = (2026, 1, 1)
+__version__ = version = '2026.2.2'
+__version_tuple__ = version_tuple = (2026, 2, 2)
 
 __commit_id__ = commit_id = None

{datamaestro_text-2026.1.1.dist-info → datamaestro_text-2026.2.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datamaestro-text
-Version: 2026.1.1
+Version: 2026.2.2
 Summary: Datamaestro module for text-related datasets
 Project-URL: Homepage, https://github.com/experimaestro/datamaestro_text
 Project-URL: Documentation, https://datamaestro-text.readthedocs.io/en/latest/
@@ -25,12 +25,6 @@ Requires-Dist: attrs
 Requires-Dist: datamaestro>=1.6.2
 Requires-Dist: experimaestro
 Requires-Dist: ir-datasets>=0.5.8
-Provides-Extra: dev
-Requires-Dist: docutils; extra == 'dev'
-Requires-Dist: flake8; extra == 'dev'
-Requires-Dist: pytest; extra == 'dev'
-Requires-Dist: sphinx<8; extra == 'dev'
-Requires-Dist: sphobjinv; extra == 'dev'
 Description-Content-Type: text/markdown
 
  [![pre-commit](https://img.shields.io/badge/pre--commit-enabled-brightgreen?logo=pre-commit&logoColor=white)](https://github.com/pre-commit/pre-commit) [![PyPI version](https://badge.fury.io/py/datamaestro-text.svg)](https://badge.fury.io/py/datamaestro-text)

{datamaestro_text-2026.1.1.dist-info → datamaestro_text-2026.2.2.dist-info}/RECORD

@@ -1,11 +1,11 @@
-datamaestro_text/__init__.py,sha256=hU8jZpkXl3F74qIfqnJl7v4nJ9YxfoR7IpJpUREFNRI,248
-datamaestro_text/version.py,sha256=KRGjQBj37k6x1t02kZiDs0px7bfHmVSimH49hjV1IAU,710
+datamaestro_text/__init__.py,sha256=MP7ShYx32k5irdgml1PjnmSofzioYQh9rzUEcHs5eys,276
+datamaestro_text/version.py,sha256=PcJXzZYuv0SaBM1rOymP9IhKDJxqcLKUPHINlOD-hL0,710
 datamaestro_text/config/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datamaestro_text/config/ai/quac.yaml,sha256=h1D7UJo2z1nZ_9MXpDHuQNJG9Ma2oayUdJV6lyumAIg,1103
-datamaestro_text/config/com/oscar-corpus.py,sha256=6F2RYOyE9_5uq_t8VrTggWxcFzefFPrmcxQXvXhfia8,723
+datamaestro_text/config/com/oscar-corpus.py,sha256=gEWz8Nxpv7VXU8X-vfRZLwPfq0KXtkGSNtsfoqfcUI0,702
 datamaestro_text/config/com/sentiment140.py,sha256=itfBEgcOniECXKOw8I2dhzyS9LOMsltMLfKK6NGRpVY,1293
 datamaestro_text/config/com/fastml/goodbooks-10k.yaml,sha256=5ZABxUnBFs2ZnCXtBH8YoBiPb3SocRRdH1DLSfVWF-Y,1172
-datamaestro_text/config/com/github/ikat.py,sha256=DCayX-t2OBeW5bOJvRxoQgIH3vy-__mYzdmVcnayAkk,4230
+datamaestro_text/config/com/github/ikat.py,sha256=nAmBre9zNlnGhx-C50EvLGvHqtoB7Ce-mZUZqM_ymO8,4219
 datamaestro_text/config/com/github/aagohary/canard.py,sha256=5fLwCLNBGM_7--naTCDayAMYLvK3yTD8auaEf-dqrb4,1768
 datamaestro_text/config/com/github/apple/ml-qrecc.py,sha256=zP3w7A9KSvJVCo44OaB1az1pDKWxE6qXS4qFm3hqg3Y,3064
 datamaestro_text/config/com/github/prdwb/orconvqa.py,sha256=MFJYaxJoqJQ6kMDHa2PIE0zoxYTA8Kyl26-vzFoMML0,3032
@@ -15,8 +15,8 @@ datamaestro_text/config/com/microsoft/msmarco/passage.py,sha256=hN2KOdi6ToHlodoz
 datamaestro_text/config/com/smashwords/bookcorpus.py,sha256=LgUcnR-z99kTrZj6QaCLuLrj1bG-wHMM5GlVNmbrY2k,851
 datamaestro_text/config/edu/cornell/nlvr.yaml,sha256=9Yk5VZMncSmrP7JNuGXqExksgX5nQ_Zfnlps8hWze3Q,921
 datamaestro_text/config/edu/stanford/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datamaestro_text/config/edu/stanford/aclimdb.py,sha256=QtriReAVsbJlxkgfJWQCZdCeJ9LswYnOR9mFrgghL9c,647
-datamaestro_text/config/edu/stanford/glove.py,sha256=bXxwiJqT8alPs5MwwxVuY0xBO8g1QWhm2L3AKVskTlA,2391
+datamaestro_text/config/edu/stanford/aclimdb.py,sha256=gv_4IauUCURbMzMWpSMyx3qgOAXVQuBwKR-mMaKExpc,626
+datamaestro_text/config/edu/stanford/glove.py,sha256=FiVYbzQMD11CiKfklrggtm7YXBCevyTXXwhehRd65H8,2348
 datamaestro_text/config/edu/stanford/im2p.yaml,sha256=JoToNyEPpmwdyLFedCBot5ypyw7p9rzi12mGXJuZin0,2909
 datamaestro_text/config/edu/upenn/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datamaestro_text/config/edu/upenn/ldc/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -24,7 +24,7 @@ datamaestro_text/config/edu/upenn/ldc/aquaint.py,sha256=YEU3kIXzv71Vjg9bUoXyQ-vS
 datamaestro_text/config/fr/granddebat.py,sha256=JRLC3q6o-XhJECjAh40w2p40pCSRw9K3-YMDUpdNwMM,7016
 datamaestro_text/config/gov/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datamaestro_text/config/gov/nist/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datamaestro_text/config/gov/nist/ir/covid.py,sha256=wn2E7sQ8M6pAucVD4sKJYImyzKUKphyiDFJD0oYRCbg,4004
+datamaestro_text/config/gov/nist/ir/covid.py,sha256=i9xxZcrKeX1gezK_TE68oropMF9PKHX2ofyREEUWYPY,4003
 datamaestro_text/config/gov/nist/trec/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datamaestro_text/config/gov/nist/trec/adhoc.py,sha256=vcFaLlZ-chwDt013MYI8bYZ2ug39jPaeimsiok_sqfU,11035
 datamaestro_text/config/gov/nist/trec/clueweb.yaml,sha256=sm4UbdtMzWoDVPsewtVDS7Vj2jBOdgp18Xqo1X4ysQc,792
@@ -35,7 +35,7 @@ datamaestro_text/config/gov/nist/trec/web.yaml,sha256=iNIJ-PCeLwpUaJByJesyCqfFJo
 datamaestro_text/config/io/github/rajpurkar/squad.yaml,sha256=JchkTruEhVd0uOTGgek2tOgBL4UT3G3KVASvZcNigLg,1143
 datamaestro_text/config/io/github/thunlp/fewrel.py,sha256=rg_qAnMrXYUZhQYxA12r_Npl0ggyfTLJQjdSCjU0QxM,1228
 datamaestro_text/config/io/metamind/research/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datamaestro_text/config/io/metamind/research/wikitext.py,sha256=DjyBmG74JvuMt9RpMwuLAnxzOdByIWsk4VnXgkJp1NM,2307
+datamaestro_text/config/io/metamind/research/wikitext.py,sha256=jw_CbBbradIUp_mrhG-z3rfa4_0ybvIBSkDqJvGLCCI,2301
 datamaestro_text/config/net/mattmahoney/enwiki.yaml,sha256=HCUn3s0AydXX3BjJ6yUXY0vGLGWSBkOCaDhQ4PA2Adg,2452
 datamaestro_text/config/org/acm/recsys/cb2014.yaml,sha256=5SAK3Am1k0HFugSSCIQN5mLPBfr1zZZAkhLrSH5pHQc,1274
 datamaestro_text/config/org/cocodataset/index.yaml,sha256=KISJChMeKwlZbSnHmRcGMsm6jbcFGVe1aA4GhP2fzqw,474
@@ -43,48 +43,49 @@ datamaestro_text/config/org/grouplens/movielens.py,sha256=tV6OSTDdtjll1dQBCsYIls
 datamaestro_text/config/org/universaldependencies/french.py,sha256=etedb3_SC-fV5Oa2rM4_smZk6t4CPiNvU4C4keUFZHY,2214
 datamaestro_text/config/uk/ac/ucl/cs/qangaroo.yaml,sha256=IBy82CDNNLjJPNPzues1EgDXu0A5WDvUFeVNSOyrIpI,1137
 datamaestro_text/data/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datamaestro_text/data/embeddings.py,sha256=qdeZ4m6drFNkqW_4pKxJKdJHtOnDRs0P7QVJ7AO6xFE,1127
+datamaestro_text/data/embeddings.py,sha256=YMoNLyVvaOt86bq_8X71_Fgu7EYYI71vr67xSQsi57I,1128
 datamaestro_text/data/recommendation.py,sha256=wHV_9SXSclouuXaBmvwg3ncZLOFfdKRZG3IHkkPJX9Y,279
 datamaestro_text/data/tagging.py,sha256=yWm7bNLks77cAySa1aZNCmLFxTTqhqXZ0PaoaEYU6hI,697
-datamaestro_text/data/text.py,sha256=_9J7-j06BOIZ1HsOcBsWy_asanAKkDoZIsWWMMj4tm4,496
-datamaestro_text/data/conversation/__init__.py,sha256=esOWnSImMlQs5QtfxUt559ABLd6a5wwoNpj9XtIq71Y,159
-datamaestro_text/data/conversation/base.py,sha256=rLOP2dqvMr2E9ONKgPF6CKUCAEHcXt5WqabJyo76AEI,7508
-datamaestro_text/data/conversation/canard.py,sha256=FI02rPNuzSsA_uXOcgcuQn31d3QahmHndzYCV1Zm8rk,3305
-datamaestro_text/data/conversation/ikat.py,sha256=djxTv0vHLXIUPzfmwpB25fyUWJNVPqthTtDbQ6CzmQo,4363
-datamaestro_text/data/conversation/orconvqa.py,sha256=RL9cpK7QK2dsk_j2e4Th8wzGFq-G3ww_EjdM1eKh-8o,3764
+datamaestro_text/data/text.py,sha256=Lln4eoegU9B27oS-2mv3eEQC6MyRBgVhoewQ2-YNxEQ,497
+datamaestro_text/data/conversation/__init__.py,sha256=Kk7FxPz_0oGO2PtIa8zH7UBqbCUsywTHfA-yKd_KO6c,284
+datamaestro_text/data/conversation/base.py,sha256=gF_-izQ1ijX7w49pKQvjfjUVzrX3VSHXxcqVIPWmAfY,7488
+datamaestro_text/data/conversation/canard.py,sha256=aYpkHzuJWGT3-myFNUjCYAtvG3gVh_d3Zc5lyiasQ04,3290
+datamaestro_text/data/conversation/ikat.py,sha256=hoGqHUWyT8BhC_ouUmnwoh93B2jGLHn8uc6npKP4Sl8,4319
+datamaestro_text/data/conversation/orconvqa.py,sha256=zNp02jyYgny0qtIFOMjmrUy7hG8VKWcELHWrg3FBCc0,3764
 datamaestro_text/data/conversation/qrecc.py,sha256=es4GmqPtE63A7O_GARe8Zy3rQvuLEhAvUA7CfN_nMeA,2562
 datamaestro_text/data/debate/__init__.py,sha256=PzCV3Bd9fmonE-OQp4VtK1NglH42-iv34WAWUIU-eYk,187
 datamaestro_text/data/debate/granddebat.py,sha256=4-HMfgvF2bPru56D3hkA1E2bN3dgIUmcvX9eOIXroLA,2176
-datamaestro_text/data/ir/__init__.py,sha256=ZRJrUeeUyD1ncMN5JINVvFJ2lDr3KsbgiiEBJkczSi0,9814
-datamaestro_text/data/ir/base.py,sha256=uwIiKn0ryK5DWUQsEegeTs6bye0uAOGl0XVm_cvV3ZU,1506
+datamaestro_text/data/ir/__init__.py,sha256=oYI7eIScg-olxPh95XBgTK-E2PunieXvqQPlrRlHU8M,9799
+datamaestro_text/data/ir/base.py,sha256=ksluGOOzOwbdZ2SPnwiDMMUhBa6P1Ti2sr6Ch5xXUgg,1493
 datamaestro_text/data/ir/cord19.py,sha256=yu1Khgy6AZjH2BPQKdnxDid0vQgQ8zvb8-FQlHH-GOU,1465
 datamaestro_text/data/ir/csv.py,sha256=0jnaV-wKLgslH7izR-xP_RX7l90vykQTn3bPhaCFR-c,1027
-datamaestro_text/data/ir/data.py,sha256=ERmmOxz_9mUkIuccNbzUa5Y6gVLLVDdyc4cCxbCCUbY,20
-datamaestro_text/data/ir/formats.py,sha256=eyP7PJ6A4Pd1uv3nbeU1N_Q3Bee7XSYTaYsiHP1MFns,3639
+datamaestro_text/data/ir/data.py,sha256=6ASVsyVVfiSd1m8C8QTrxVLnFVmtoW3d9c9nQ07zlbY,34
+datamaestro_text/data/ir/formats.py,sha256=rKflCuY8UBpXC3nltBqzC4waWYoxuyP91xJvG7p690Y,3630
 datamaestro_text/data/ir/huggingface.py,sha256=G71VFDN-SllZy4LFxumEbCumEJvb5-orAbLemHjWhiA,956
-datamaestro_text/data/ir/stores.py,sha256=A4Ew0L4P6iLLmDKhxqjjVkHcz797BHf4d76YguGkB1A,4412
+datamaestro_text/data/ir/stores.py,sha256=rdOwYCG_NzHSsUQpJ1aneiA2SDWrcfdi16aY-df852U,4408
 datamaestro_text/data/ir/trec.py,sha256=IOtQRMUz8zx-dYEMR2NIIM6qXEUjsV0eVOhGvKIRJK4,1974
 datamaestro_text/data/ir/utils.py,sha256=6-GhXVtgkBZGhIs2-ODZua_3DmKjSSVydStpHDqbAwE,833
+datamaestro_text/datasets/__init__.py,sha256=ORn-Q1gGibg-N5grVc7MqOYfExels3FRI51oQ4xI1QA,34
 datamaestro_text/datasets/irds/__init__.py,sha256=Tq0HN1qojnZYLBumM59BuTkz7r0gcu-5OXmDDLgPpAc,707
-datamaestro_text/datasets/irds/data.py,sha256=5ZtJTEV9qtbl_Do4VR6EvYoxPTlsRkkjoBunXDLfmHI,23012
+datamaestro_text/datasets/irds/data.py,sha256=sIU7_rt4I1E9rjkIGcpNfbD5mtO97vxFsUDmouRMDV4,22914
 datamaestro_text/datasets/irds/datasets.py,sha256=CJ8MA44XCwIQGZTzYIJnR-qFm890rUZZB7C3lKIwNyY,5627
 datamaestro_text/datasets/irds/helpers.py,sha256=fGE-fbuJbXdTzl1qo55247jzn9cvApY-d82GJBgfY1E,3982
 datamaestro_text/datasets/irds/utils.py,sha256=m30JLIrV_HgilN11TvY9dGTyumES6LLzWZDUAMT915M,1425
-datamaestro_text/download/tmdb.py,sha256=kU_Vz9jhznlyeKMHziVu58IHoWv8zPu6CZTHVNQvmu4,4009
+datamaestro_text/download/tmdb.py,sha256=sfnSUJwGSjBsLNVVhT30db2m0R8mrRkDZpbpBUt7GMg,3960
 datamaestro_text/interfaces/plaintext.py,sha256=cWfS_xjqZxQ0EV4Ax5BEarZ4lnhQ1I7mc_vgfBgE76w,885
 datamaestro_text/interfaces/trec.py,sha256=GrP0N_Hcj5f73KS6CSzkyN4aaI-XoBJ19oVMpHVR3QM,3579
 datamaestro_text/test/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datamaestro_text/test/test_datasets.py,sha256=hD1pe-CjEnOj0JNqDLasz_97ltFzMbYfyYBy1QyYdf8,202
 datamaestro_text/test/test_documented.py,sha256=Kl90fz_r_dsV0oXE1Mad34GiQyZ9dc6bfZKNtxP0O2s,453
 datamaestro_text/transforms/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datamaestro_text/transforms/ir/__init__.py,sha256=Pb8C-jwjtCur6gU-Lv4AosSFFKw2o6jMQcJE1A_5PD8,6555
+datamaestro_text/transforms/ir/__init__.py,sha256=7D6wurKVQf-f2mu1I3tT-baQbKo7yRCxW8pOHh-MSjM,6539
 datamaestro_text/utils/__init__.py,sha256=2449YLTAtKJzkmt84Mu8sBRCCveNs5fiaqTCK_p5ha0,3340
 datamaestro_text/utils/files.py,sha256=jhcirufgTztEkx1hs2-qgScEcHnIcGF_BjCeuCSsNv0,2838
 datamaestro_text/utils/iter.py,sha256=QBajeSPLHvkeh6BCTZDSqWlOYNjwUDvgTTZ_YxJntXw,2701
 datamaestro_text/utils/randomstream.py,sha256=_-boH4IIqN8qcl3IktjpNp9vmF4TWRzHUSNVwg7WAr8,973
-datamaestro_text/utils/shuffle.py,sha256=o8JTz3mr0lYWyv0zEh91jEK12ci1etMiUnzh5GkOHCM,3490
-datamaestro_text-2026.1.1.dist-info/METADATA,sha256=PnX4hQbQmrRUUBily4GEDJ53HsE-u1QqhK7gcv5nWeU,2103
-datamaestro_text-2026.1.1.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-datamaestro_text-2026.1.1.dist-info/entry_points.txt,sha256=lO1P5hE183L5qEEVHlG8d_ik0HNXnX7Eo87cQLdcl-Y,111
-datamaestro_text-2026.1.1.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
-datamaestro_text-2026.1.1.dist-info/RECORD,,
+datamaestro_text/utils/shuffle.py,sha256=xXzgBQ8An7tKboxI0z123Tl6ywXI4S0tWf8MnfOon0c,3491
+datamaestro_text-2026.2.2.dist-info/METADATA,sha256=cHXRhpnNO6sliuE09Jg-eHJtr2kl1Z4Dy3mE1RCGELA,1886
+datamaestro_text-2026.2.2.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+datamaestro_text-2026.2.2.dist-info/entry_points.txt,sha256=lO1P5hE183L5qEEVHlG8d_ik0HNXnX7Eo87cQLdcl-Y,111
+datamaestro_text-2026.2.2.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
+datamaestro_text-2026.2.2.dist-info/RECORD,,