unique_toolkit 1.40.0__py3-none-any.whl → 1.40.1__py3-none-any.whl

unique_toolkit/agentic/evaluation/hallucination/utils.py

@@ -1,4 +1,5 @@
 import logging
+from enum import StrEnum
 from string import Template
 
 from unique_toolkit.agentic.evaluation.config import EvaluationMetricConfig
@@ -9,6 +10,7 @@ from unique_toolkit.agentic.evaluation.schemas import (
     EvaluationMetricName,
     EvaluationMetricResult,
 )
+from unique_toolkit.content import ContentReference
 from unique_toolkit.content.schemas import ContentChunk
 from unique_toolkit.language_model.schemas import (
     LanguageModelMessages,
@@ -201,13 +203,58 @@ def _get_user_prompt_default(config: EvaluationMetricConfig):
     )
 
 
+class SourceSelectionMode(StrEnum):
+    FROM_IDS = "FROM_IDS"
+    FROM_ORDER = "FROM_ORDER"
+
+
 def context_text_from_stream_response(
-    response: LanguageModelStreamResponse, selected_chunks: list[ContentChunk]
+    response: LanguageModelStreamResponse,
+    selected_chunks: list[ContentChunk],
+    source_selection_mode: SourceSelectionMode = SourceSelectionMode.FROM_IDS,
 ):
     response_references = response.message.references
-    reference_ids = [reference.source_id for reference in response_references]
-    filtered_contexts: list[str] = []
-    for chunk in selected_chunks:
-        if f"{chunk.id}_{chunk.chunk_id}" in reference_ids:
-            filtered_contexts.append(chunk.text)
-    return filtered_contexts
+    match source_selection_mode:
+        case SourceSelectionMode.FROM_IDS:
+            referenced_chunks = _default_source_selection_mode(
+                response_references, selected_chunks
+            )
+        case SourceSelectionMode.FROM_ORDER:
+            referenced_chunks = _from_order_source_selection_mode(
+                response_references, selected_chunks
+            )
+        case _:
+            raise ValueError(f"Invalid source selection mode: {source_selection_mode}")
+
+    return [chunk.text for chunk in referenced_chunks]
+
+
+def _default_source_selection_mode(
+    references: list[ContentReference], selected_chunks: list[ContentChunk]
+):
+    reference_ids = {reference.source_id for reference in references}
+
+    def build_chunk_id(chunk: ContentChunk) -> str:
+        return f"{chunk.id}_{chunk.chunk_id}"
+
+    referenced_chunks = [
+        chunk for chunk in selected_chunks if build_chunk_id(chunk) in reference_ids
+    ]
+
+    return referenced_chunks
+
+
+def _from_order_source_selection_mode(
+    references: list[ContentReference], selected_chunks: list[ContentChunk]
+):
+    original_chunks_order: list[int] = []
+    for reference in references:
+        for original_index in reference.original_index:
+            if original_index not in original_chunks_order:
+                original_chunks_order.append(original_index)
+
+    referenced_chunks: list[ContentChunk] = []
+    for index in original_chunks_order:
+        referenced_chunks.append(selected_chunks[index])
+
+    return referenced_chunks

{unique_toolkit-1.40.0.dist-info → unique_toolkit-1.40.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: unique_toolkit
-Version: 1.40.0
+Version: 1.40.1
 Summary: 
 License: Proprietary
 Author: Cedric Klinkert
@@ -123,6 +123,9 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), 
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.40.0] - 2025-12-22
+- Add option to use retrieve referenced chunks from their order
+
 ## [1.40.0] - 2025-12-22
 - Add `hide_in_chat` parameter to `upload_to_chat_from_bytes` and `upload_to_chat_from_bytes_async`
 - Hide code interpreter files in chat 

{unique_toolkit-1.40.0.dist-info → unique_toolkit-1.40.1.dist-info}/RECORD

@@ -54,7 +54,7 @@ unique_toolkit/agentic/evaluation/hallucination/constants.py,sha256=SoGmoYti2J33
 unique_toolkit/agentic/evaluation/hallucination/hallucination_evaluation.py,sha256=x5ta2Fum4fE5ySgIXPKlnbTtmV140z0IazSATd0-REg,4092
 unique_toolkit/agentic/evaluation/hallucination/prompts.py,sha256=O3Hi_rOzZlujvnO2wn2jhoPmrYLjzVtRWwxn5Q81m9Y,3405
 unique_toolkit/agentic/evaluation/hallucination/service.py,sha256=WJF1f45uHnYLx1S4TW31bSFobFpV-YlOS3G_zMhuBVU,2512
-unique_toolkit/agentic/evaluation/hallucination/utils.py,sha256=kM5kqUZb5riINr1Oqf3wyrj25o25H2WaJ64haKTJLMo,8213
+unique_toolkit/agentic/evaluation/hallucination/utils.py,sha256=uHKTJw4kJyq0_Gi-EOhbocBAij4_Vzn3dW1wTxAuFg4,9706
 unique_toolkit/agentic/evaluation/output_parser.py,sha256=0FDo8YY_Dc4qlTNeYyQkznzIFj9aX9wMrLOTbhhTl6g,1418
 unique_toolkit/agentic/evaluation/schemas.py,sha256=m9JMCUmeqP8KhsJOVEzsz6dRXUe1uKw-bxRDtn5qwvM,3156
 unique_toolkit/agentic/evaluation/tests/test_context_relevancy_service.py,sha256=4tDxHTApbaTMxN1sNS8WCqj2BweRk6YqZ5_zHP45jto,7977
@@ -212,7 +212,7 @@ unique_toolkit/short_term_memory/service.py,sha256=5PeVBu1ZCAfyDb2HLVvlmqSbyzBBu
 unique_toolkit/smart_rules/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 unique_toolkit/smart_rules/compile.py,sha256=Ozhh70qCn2yOzRWr9d8WmJeTo7AQurwd3tStgBMPFLA,1246
 unique_toolkit/test_utilities/events.py,sha256=_mwV2bs5iLjxS1ynDCjaIq-gjjKhXYCK-iy3dRfvO3g,6410
-unique_toolkit-1.40.0.dist-info/LICENSE,sha256=GlN8wHNdh53xwOPg44URnwag6TEolCjoq3YD_KrWgss,193
-unique_toolkit-1.40.0.dist-info/METADATA,sha256=iEwBy-tYmByLek8I9T4Eck-LCO5eDQKtk8fkAgnhS-g,46331
-unique_toolkit-1.40.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-unique_toolkit-1.40.0.dist-info/RECORD,,
+unique_toolkit-1.40.1.dist-info/LICENSE,sha256=GlN8wHNdh53xwOPg44URnwag6TEolCjoq3YD_KrWgss,193
+unique_toolkit-1.40.1.dist-info/METADATA,sha256=yjUi84bUvdUEGqMC3PLKEDzJXZj-ul3AvIdl3-HouSQ,46421
+unique_toolkit-1.40.1.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+unique_toolkit-1.40.1.dist-info/RECORD,,