datamaestro-text 2025.6.30__tar.gz → 2025.9.11__tar.gz

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/.github/workflows/pytest.yml

@@ -25,7 +25,7 @@ jobs:
         python-version: ${{ matrix.python-version }}
     - name: Install dependencies
       run: |
-        python -m pip install --upgrade pip
+        python -m pip install --upgrade pip setuptools
         SETUPTOOLS_SCM_PRETEND_VERSION="0.1-dev" pip install -e '.[dev]'
     - name: Lint with flake8
       run: |

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/.readthedocs.yml

@@ -11,7 +11,7 @@ sphinx:
 build:
   os: ubuntu-20.04
   tools:
-    python: "3.9"
+    python: "3.10"
 
 
 # Install the package

{datamaestro_text-2025.6.30/src/datamaestro_text.egg-info → datamaestro_text-2025.9.11}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datamaestro-text
-Version: 2025.6.30
+Version: 2025.9.11
 Summary: Datamaestro module for text-related datasets
 Author-email: Benjamin Piwowarski <benjamin@piwowarski.fr>
 License: GPL-3
@@ -15,10 +15,10 @@ Classifier: Operating System :: OS Independent
 Classifier: Programming Language :: Python
 Classifier: Programming Language :: Python :: 3
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
-Requires-Python: >=3.8
+Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: datamaestro>=1.4.2
+Requires-Dist: datamaestro>=1.5.0
 Requires-Dist: ir_datasets>=0.5.8
 Requires-Dist: attrs
 Provides-Extra: dev

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/docs/source/api/conversation.rst

@@ -26,6 +26,12 @@ Data classes
 
 .. autoclass:: ConversationTopic
 
+Conversational IR
+-----------------
+
+.. autoxpmconfig:: datamaestro_text.data.conversation.base.ConversationUserTopics
+
+
 Contextual query reformulation
 ------------------------------
 
@@ -34,7 +40,7 @@ Contextual query reformulation
 .. autoclass:: ContextualizedRewrittenQuery
     :members:
 
-CANARD Dataset 
+CANARD Dataset
 
 .. autoxpmconfig:: datamaestro_text.data.conversation.canard.CanardDataset
     :members: iter
@@ -50,7 +56,7 @@ OrConvQA Dataset
 .. autoclass:: datamaestro_text.data.conversation.orconvqa.OrConvQADatasetHistoryEntry
     :members:
 
-QReCC Dataset 
+QReCC Dataset
 
 .. autoclass:: datamaestro_text.data.conversation.qrecc.QReCCDatasetEntry
     :members:
@@ -61,11 +67,11 @@ QReCC Dataset
 
 iKAT Dataset
 
-.. autoclass:: datamaestro_text.data.conversation.ikat.IkatDatasetEntry
+.. autoclass:: datamaestro_text.data.conversation.ikat.IkatConversationTopic
     :members:
 
 .. autoclass:: datamaestro_text.data.conversation.ikat.IkatConversationEntry
     :members:
 
-.. autoxpmconfig:: datamaestro_text.data.conversation.ikat.IkatDataset
+.. autoxpmconfig:: datamaestro_text.data.conversation.ikat.IkatConversations
     :members: iter

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/docs/source/api/ir.rst

@@ -36,11 +36,17 @@ Documents
 
 .. autoxpmconfig:: datamaestro_text.data.ir.Documents
     :members: iter_documents, iter_ids, documentcount
+.. autoxpmconfig:: datamaestro_text.data.ir.csv.Documents
+.. autoxpmconfig:: datamaestro_text.datasets.irds.data.LZ4DocumentStore
+
+
+Dataset-specific documents
+**************************
+
 .. autoxpmconfig:: datamaestro_text.data.ir.cord19.Documents
 .. autoxpmconfig:: datamaestro_text.data.ir.trec.TipsterCollection
-.. autoxpmconfig:: datamaestro_text.data.ir.csv.Documents
 .. autoxpmconfig:: datamaestro_text.data.ir.stores.OrConvQADocumentStore
-.. autoxpmconfig:: datamaestro_text.datasets.irds.data.LZ4DocumentStore
+.. autoxpmconfig:: datamaestro_text.data.ir.stores.IKatClueWeb22DocumentStore
 
 Assessments
 -----------

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "datamaestro-text"
-requires-python = ">=3.8"
+requires-python = ">=3.10"
 keywords = ["dataset manager", "information retrieval", "experiments"]
 description = "Datamaestro module for text-related datasets"
 dynamic = ["version", "readme", "dependencies"]
@@ -44,6 +44,14 @@ fallback_version = "0.0.0-dev"
 [build-system]
 requires = ["setuptools", "setuptools-scm", "wheel"]  # PEP 508 specifications.
 
+[dependency-groups]
+dev = [
+    "docutils>=0.21.2",
+    "pytest>=8.4.1",
+    "sphinx>=8.1.3",
+    "sphobjinv>=2.3.1.3",
+]
+
 [project.entry-points."datamaestro.repositories"]
 text = "datamaestro_text:Repository"
 irds = "datamaestro_text.datasets.irds:Repository"

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/requirements.txt

@@ -1,3 +1,3 @@
-datamaestro>=1.4.2
+datamaestro>=1.5.0
 ir_datasets>=0.5.8
 attrs

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/config/com/github/aagohary/canard.py

@@ -37,7 +37,7 @@ def main(train, dev, test):
     Each dataset is an instance of :class:`datamaestro_text.data.conversation.CanardDataset`
     """
     return {
-        "train": CanardDataset(path=train),
-        "validation": CanardDataset(path=dev),
-        "test": CanardDataset(path=test),
+        "train": CanardDataset.C(path=train),
+        "validation": CanardDataset.C(path=dev),
+        "test": CanardDataset.C(path=test),
     }

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/config/com/github/apple/ml-qrecc.py

@@ -51,6 +51,7 @@ def main(data: Path) -> Supervised[QReCCDataset, None, QReCCDataset]:
 )
 class Content(LZ4JSONLDocumentStore):
     """QReCC mentionned URLs content"""
+
     @staticmethod
     def __create_dataset__(dataset, options=None):
         ds = reference(reference=main).setup(dataset, options)
@@ -65,7 +66,7 @@ class Content(LZ4JSONLDocumentStore):
             "id",
         ).setup(dataset, options)
 
-        return Content(jsonl_path=store_path)
+        return Content.C(jsonl_path=store_path)
 
     @staticmethod
     def _documents(path: Path):

datamaestro_text-2025.9.11/src/datamaestro_text/config/com/github/ikat.py

@@ -0,0 +1,121 @@
+# 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
+from datamaestro_text.data.ir import Adhoc
+
+from datamaestro.utils import HashCheck
+from datamaestro.context import DatafolderPath
+from datamaestro.download.single import filedownloader
+from datamaestro_text.data.conversation.ikat import IkatConversations
+from datamaestro.download.links import linkfolder
+
+from datamaestro_text.data.ir.stores import IKatClueWeb22DocumentStore
+from datamaestro_text.data.ir.trec import TrecAdhocAssessments
+from datamaestro_text.datasets.irds.helpers import lz4docstore_builder
+
+
+@dataset(as_prepare=True)
+def clueweb22(dataset, options=None) -> IKatClueWeb22DocumentStore:
+    # Number of documents in the dataset
+    count = 116_838_987
+
+    jsonl_folder = linkfolder(
+        "documents", [DatafolderPath("gov.nist.trec.ikat.clueweb22", "jsonl")]
+    ).setup(dataset, options)
+    store_path = lz4docstore_builder(
+        "store",
+        IKatClueWeb22DocumentStore.generator(
+            jsonl_folder,
+            jsonl_folder / "ikat_2023_passages_jsonl.sha256sums",
+            jsonl_folder / "ikat_2023_passages_hashes.tsv.bz2",
+        ),
+        IKatClueWeb22DocumentStore.Document,
+        "id",
+        count_hint=count,
+    ).setup(dataset, options)
+
+    return IKatClueWeb22DocumentStore.C(path=store_path, count=count)
+
+
+@datatags("conversation", "context", "query")
+@datatasks("conversational search", "query rewriting")
+@reference("documents", clueweb22)
+@filedownloader(
+    "topics.json",
+    "https://raw.githubusercontent.com/irlabamsterdam/iKAT/refs/heads/main/2025/data/2025_test_topics.json",
+    checker=HashCheck("16f8444a8d0a8dfe0090f478f185a63c"),
+)
+@dataset(
+    id="2025",
+    url="https://github.com/irlabamsterdam/iKAT/tree/main/2025",
+)
+def test_2025(topics, documents) -> Adhoc.C:
+    """Question-in-context rewriting
+
+    iKAT is a test dataset for question-in-context rewriting that consists of
+    questions each given in a dialog context together with a context-independent
+    rewriting of the question.
+    """
+    return Adhoc.C(
+        topics=ConversationUserTopics.C(conversations=IkatConversations.C(path=topics)),
+        # TODO: add when available
+        assessments=TrecAdhocAssessments.C(path="/to/do"),
+        documents=documents,
+    )
+
+
+@datatags("conversation", "context", "query")
+@datatasks("conversational search", "query rewriting")
+@reference("documents", clueweb22)
+@filedownloader(
+    "qrels",
+    "https://trec.nist.gov/data/ikat/2024-qrels.txt",
+    checker=HashCheck("57f958903ed1c12bbac207f62800814f"),
+)
+@filedownloader(
+    "topics.json",
+    "https://raw.githubusercontent.com/irlabamsterdam/iKAT/refs/heads/main/2024/data/2024_test_topics.json",
+    checker=HashCheck("ad45bc6e7add2081d69ea60a0a4d1203"),
+)
+@dataset(
+    Adhoc,
+    id="2024",
+    url="https://github.com/irlabamsterdam/iKAT/tree/main/2024",
+)
+def test_2024(topics, qrels, documents) -> Adhoc.C:
+    """iKAT 2024 dataset"""
+    return Adhoc.C(
+        topics=ConversationUserTopics.C(conversations=IkatConversations.C(path=topics)),
+        assessments=TrecAdhocAssessments.C(path=qrels),
+        documents=documents,
+    )
+
+
+@datatags("conversation", "context", "query")
+@datatasks("conversational search", "query rewriting")
+@reference("documents", clueweb22)
+@filedownloader(
+    "qrels",
+    "https://trec.nist.gov/data/ikat/2023-qrels.all-turns.txt",
+    checker=HashCheck("79dc121bab25b2245e52a53263e5ad1f"),
+)
+@filedownloader(
+    "topics.json",
+    "https://raw.githubusercontent.com/irlabamsterdam/iKAT/refs/heads/main/2023/data/2023_test_topics.json",
+    checker=HashCheck("684fa0197cdec8c3cfb6a2e586ab83f6"),
+)
+@dataset(
+    Adhoc,
+    id="2023",
+    url="https://github.com/irlabamsterdam/iKAT/tree/main/2023",
+)
+def test_2023(topics, qrels, documents) -> Adhoc.C:
+    """iKAT 2023 dataset"""
+    return Adhoc.C(
+        topics=ConversationUserTopics.C(conversations=IkatConversations.C(path=topics)),
+        assessments=TrecAdhocAssessments.C(path=qrels),
+        documents=documents,
+    )

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/config/com/microsoft/msmarco/passage.py

@@ -47,7 +47,7 @@ http://www.msmarco.org/dataset.aspx""",
 @dataset(url="https://github.com/microsoft/MSMARCO-Passage-Ranking")
 def collection_etc(data) -> Folder:
     """Documents and some more files"""
-    return Folder(path=data)
+    return Folder.C(path=data)
 
 
 @lua

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/config/com/sentiment140.py

@@ -26,7 +26,7 @@ def english(dir):
 
     If you use this data, please cite Sentiment140 as your source.
     """
-    return {
-        "train": Generic(path=dir / "training.1600000.processed.noemoticon.csv"),
-        "test": Generic(path=dir / "testdata.manual.2009.06.14.csv"),
-    }
+    return Supervised.C(
+        train=Generic.C(path=dir / "training.1600000.processed.noemoticon.csv"),
+        test=Generic.C(path=dir / "testdata.manual.2009.06.14.csv"),
+    )

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/config/edu/stanford/aclimdb.py

@@ -11,6 +11,6 @@ def aclimdb(data):
     Paper http://ai.stanford.edu/~amaas/papers/wvSent_acl2011.pdf
     """
     return {
-        "train": FolderBased(path=data / "train", classes=["neg", "pos"]),
-        "test": FolderBased(path=data / "test", classes=["neg", "pos"]),
+        "train": FolderBased.C(path=data / "train", classes=["neg", "pos"]),
+        "test": FolderBased.C(path=data / "test", classes=["neg", "pos"]),
     }

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/config/gov/nist/trec/tipster.py

@@ -17,7 +17,6 @@ of three CD-ROMs of SGML encoded documents distributed by LDC plus queries and a
 See also https://trec.nist.gov/data/docs_eng.html and https://trec.nist.gov/data/intro_eng.html
 """
 
-from datamaestro.data import Base
 from datamaestro_text.data.ir.trec import TipsterCollection
 from datamaestro.download.links import linkfolder
 from datamaestro.definitions import (

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/config/io/github/thunlp/fewrel.py

@@ -32,4 +32,4 @@ def v1(train, validation):
     Only the train and validation dataset are available. The test set is hidden
     for the leaderboard.
     """
-    return {"train": File(path=train), "validation": File(path=validation)}
+    return {"train": File.C(path=train), "validation": File.C(path=validation)}

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/config/io/metamind/research/wikitext.py

@@ -30,9 +30,9 @@ def WikiText(data, type):
     https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/
     """
     return {
-        "train": File(path=data / ("wiki.train.%s" % type)),
-        "validation": File(path=data / ("wiki.valid.%s" % type)),
-        "test": File(path=data / ("wiki.test.%s" % type)),
+        "train": File.C(path=data / ("wiki.train.%s" % type)),
+        "validation": File.C(path=data / ("wiki.valid.%s" % type)),
+        "test": File.C(path=data / ("wiki.test.%s" % type)),
     }
 
 

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/config/org/grouplens/movielens.py

@@ -31,10 +31,10 @@ def small(ds) -> Movielens:
     100,000 ratings and 3,600 tag applications applied to 9,000 movies by 600 users (as of 9/2018)
     """
     return {
-        "ratings": csv.Generic(path=ds / "ratings.csv", names_row=0),
-        "links": csv.Generic(path=ds / "links.csv", names_row=0),
-        "movies": csv.Generic(path=ds / "movies.csv", names_row=0),
-        "tags": csv.Generic(path=ds / "tags.csv", names_row=0),
+        "ratings": csv.Generic.C(path=ds / "ratings.csv", names_row=0),
+        "links": csv.Generic.C(path=ds / "links.csv", names_row=0),
+        "movies": csv.Generic.C(path=ds / "movies.csv", names_row=0),
+        "tags": csv.Generic.C(path=ds / "tags.csv", names_row=0),
     }
 
 
@@ -46,8 +46,8 @@ def full(ds) -> Movielens:
     27,000,000 ratings and 1,100,000 tag applications applied to 58,000 movies by 280,000 users (as of 9/2018)
     """
     return {
-        "ratings": csv.Generic(path=ds / "ratings.csv", names_row=0),
-        "links": csv.Generic(path=ds / "links.csv", names_row=0),
-        "movies": csv.Generic(path=ds / "movies.csv", names_row=0),
-        "tags": csv.Generic(path=ds / "tags.csv", names_row=0),
+        "ratings": csv.Generic.C(path=ds / "ratings.csv", names_row=0),
+        "links": csv.Generic.C(path=ds / "links.csv", names_row=0),
+        "movies": csv.Generic.C(path=ds / "movies.csv", names_row=0),
+        "tags": csv.Generic.C(path=ds / "tags.csv", names_row=0),
     }

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/config/org/universaldependencies/french.py

@@ -34,9 +34,9 @@ def gsd(ds) -> Supervised:
     is updated since 2015 independently from the previous source.
     """
     return {
-        "train": CoNLL_U(path=ds / "fr_gsd-ud-train.conllu"),
-        "test": CoNLL_U(path=ds / "fr_gsd-ud-dev.conllu"),
-        "validation": CoNLL_U(path=ds / "fr_gsd-ud-test.conllu"),
+        "train": CoNLL_U.C(path=ds / "fr_gsd-ud-train.conllu"),
+        "test": CoNLL_U.C(path=ds / "fr_gsd-ud-dev.conllu"),
+        "validation": CoNLL_U.C(path=ds / "fr_gsd-ud-test.conllu"),
     }
 
 

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/data/conversation/base.py

@@ -1,10 +1,13 @@
 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 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
+from datamaestro_text.data.ir import TopicRecord, Topics
 from datamaestro_text.utils.iter import FactoryIterable, LazyList, RangeView
 
 # ---- Basic types
@@ -120,20 +123,17 @@ class ConversationNode:
         ...
 
     @abstractmethod
-    def parent(self) -> Optional["ConversationNode"]:
-        ...
+    def parent(self) -> Optional["ConversationNode"]: ...
 
     @abstractmethod
-    def children(self) -> List["ConversationNode"]:
-        ...
+    def children(self) -> List["ConversationNode"]: ...
 
 
 class ConversationTree(ABC):
     """Represents a conversation tree"""
 
     @abstractmethod
-    def root(self) -> ConversationNode:
-        ...
+    def root(self) -> ConversationNode: ...
 
     @abstractmethod
     def __iter__(self) -> Iterator[ConversationNode]:
@@ -253,5 +253,30 @@ class ConversationDataset(Base, ABC):
     @abstractmethod
     def __iter__(self) -> Iterator[ConversationTree]:
         """Return an iterator over conversations"""
-        for i in range(len(self)):
-            yield self.get(i)
+        ...
+
+
+class ConversationUserTopics(Topics):
+    """Extract user topics from conversations"""
+
+    conversations: Param[ConversationDataset]
+
+    topic_recordtype = record_type(IDItem, SimpleTextItem)
+
+    def iter(self) -> Iterator[TopicRecord]:
+        """Returns an iterator over topics"""
+        # Extracts topics from conversations, Each user query is a topic (can perform retrieval on it)
+        # TODO: merge with xpmir.learning.DatasetConversationBase -> same logic
+        
+        records: List[TopicRecord] = []
+        for conversation in self.conversations.__iter__():
+            nodes = [
+                node
+                for node in conversation
+                if node.entry[EntryType] == EntryType.USER_QUERY
+            ]
+            for node in nodes:
+                records.append(
+                    node.entry.update(ConversationHistoryItem(node.history()))
+                )
+        return iter(records)

{datamaestro_text-2025.6.30 → datamaestro_text-2025.9.11}/src/datamaestro_text/data/conversation/ikat.py

@@ -1,10 +1,11 @@
-from typing import Iterator, List, Optional
+from typing import Iterator, List
 from attr import define, field
 import json
 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,
@@ -12,7 +13,6 @@ from datamaestro_text.data.ir.base import (
 
 
 from .base import (
-    AnswerDocumentURL,
     AnswerEntry,
     ConversationTree,
     EntryType,
@@ -21,6 +21,25 @@ from .base import (
 )
 from . import ConversationDataset
 
+# Keys to change in the dataset entries for compatibility across different years
+
+KEY_MAPPINGS = {
+    # Keys to replace: Target Key
+    "turns": "responses",
+    "utterance": "user_utterance",
+    "ptkb_provenance": "relevant_ptkbs",
+    "response_provenance": "citations",
+}
+
+
+def norm_dict(entry: dict) -> dict:
+    """Convert keys in the entry to match the expected format."""
+    normalized = {}
+    for k, v in entry.items():
+        # Check for direct mapping, then try lowercase mapping
+        new_key = KEY_MAPPINGS.get(k) or KEY_MAPPINGS.get(k.lower()) or k.lower()
+        normalized[new_key] = v
+    return normalized
 
 
 @define(kw_only=True)
@@ -47,7 +66,7 @@ class IkatConversationEntry:
 
 
 @define(kw_only=True)
-class IkatDatasetEntry:
+class IkatConversationTopic:
     """A query with past history"""
 
     number: str
@@ -60,14 +79,20 @@ class IkatDatasetEntry:
     """The personal knowledge base associated with the user"""
 
     responses: List[IkatConversationEntry] = field(
-        converter=lambda items: [IkatConversationEntry(**item) if isinstance(item, dict) else item for item in items]
+        converter=lambda items: [
+            IkatConversationEntry(**item) if isinstance(item, dict) else item
+            for item in map(norm_dict, items)
+        ]
     )
     """The list of responses to the query"""
 
 
-class IkatDataset(ConversationDataset, File):
+class IkatConversations(ConversationDataset, File):
+    """A dataset containing conversations from the IKAT project"""
 
-    def entries(self) -> Iterator[IkatDatasetEntry]:
+    """Keys to change in the dataset entries for compatibility across different years"""
+
+    def entries(self) -> Iterator[IkatConversationTopic]:
         """Reads all conversation entries from the dataset file."""
         with self.path.open("rt") as fp:
             raw_data = json.load(fp)
@@ -75,12 +100,13 @@ class IkatDataset(ConversationDataset, File):
         logging.debug("Loaded %d entries from %s", len(raw_data), self.path)
         logging.debug(f"raw data has keys {raw_data[0].keys()}")
 
-        processed_data = []
         for entry in raw_data:
-            processed_data.append(IkatDatasetEntry(**{key.lower(): value for key, value in entry.items()}))
-
-        logging.debug(f"First parsed data sample: {processed_data[0]}")
-        return iter(processed_data)
+            try:
+                normalized_entry = norm_dict(entry)
+                yield IkatConversationTopic(**normalized_entry)
+            except Exception as e:
+                logging.warning(f"Failed to parse entry: {e}")
+                raise e
 
     def __iter__(self) -> Iterator[ConversationTree]:
         for entry in self.entries():
@@ -88,7 +114,7 @@ class IkatDataset(ConversationDataset, File):
 
             for turn in entry.responses:
                 turn: IkatConversationEntry = turn  # Ensure type is correct
-                query_id = f"{entry.number}#{turn.turn_id}"
+                query_id = f"{entry.number}_{turn.turn_id}"
 
                 # USER QUERY record
                 history.append(
@@ -117,4 +143,3 @@ class IkatDataset(ConversationDataset, File):
             # Ensure reverse if needed for compatibility (optional)
             history.reverse()
             yield SingleConversationTree(entry.number, history)
-