deepsights-api 1.2.3__tar.gz → 1.2.4__tar.gz

{deepsights_api-1.2.3 → deepsights_api-1.2.4}/CHANGELOG.md

@@ -2,10 +2,13 @@
 
 <!--next-version-placeholder-->
 
-## v1.2.3 (25-Jan-2025)
+## v1.2.4 (16-Arp-2025)
+
+- Sort order in SCS text search
 
-- Support item text download from content store
+## v1.2.3 (25-Jan-2025)
 
+- Support item text download from cont
 
 ## v1.2.2 (12-Jan-2025)
 

{deepsights_api-1.2.3 → deepsights_api-1.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: deepsights-api
-Version: 1.2.3
+Version: 1.2.4
 Summary: Python library for the DeepSights APIs
 Project-URL: Documentation, https://marketlogicsoftware.github.io/deepsights-api/
 Project-URL: Repository, https://github.com/marketlogicsoftware/deepsights-api

{deepsights_api-1.2.3 → deepsights_api-1.2.4}/deepsights/contentstore/resources/_search.py

@@ -16,9 +16,11 @@
 This module contains the base functions to search the ContentStore.
 """
 
-from typing import List
 from datetime import datetime
+from typing import List
+
 from pydantic import BaseModel
+
 from deepsights.api import API
 from deepsights.utils import (
     promote_exact_matches,
@@ -96,9 +98,9 @@ def contentstore_hybrid_search(
     assert 0 <= vector_fraction <= 1, "Vector fraction must be between 0 and 1"
     assert 0 <= vector_weight <= 1, "Vector weught must be between 0 and 1"
     assert 0 < max_results <= 250, "Maximum results must be between 1 and 250."
-    assert (
-        recency_weight is None or 0 <= recency_weight <= 1
-    ), "Recency weight must be between 0 and 1."
+    assert recency_weight is None or 0 <= recency_weight <= 1, (
+        "Recency weight must be between 0 and 1."
+    )
 
     body = {
         "query": query,
@@ -171,9 +173,9 @@ def contentstore_vector_search(
     assert len(query_embedding) == 1536, "The 'query_embedding' must be of length 1536."
     assert 0 <= min_score <= 1, "Minimum score must be between 0 and 1."
     assert 0 < max_results <= 100, "Maximum results must be between 1 and 100."
-    assert (
-        recency_weight is None or 0 <= recency_weight <= 1
-    ), "Recency weight must be between 0 and 1."
+    assert recency_weight is None or 0 <= recency_weight <= 1, (
+        "Recency weight must be between 0 and 1."
+    )
 
     body = {
         "vector": query_embedding,
@@ -206,7 +208,7 @@ def contentstore_text_search(
     max_results: int = 50,
     offset: int = 0,
     languages: List[str] = None,
-    recency_weight: float = None,
+    sort_descending: bool = True,
     search_from_timestamp: datetime = None,
     search_to_timestamp: datetime = None,
     search_only_ai_allowed_content: bool = True,
@@ -223,7 +225,7 @@ def contentstore_text_search(
         max_results (int, optional): The maximum number of results to return. Defaults to 50.
         offset (int, optional): The offset to start the search from. Defaults to 0.
         languages (List[str], optional): The languages to search for. Defaults to None.
-        recency_weight (float, optional): The weight assigned to recency in result ranking. Defaults to None.
+        most_recent_first (bool, optional): Whether to sort results by most recent first. Defaults to True.
         search_from_timestamp (datetime, optional): The start timestamp for the search. Defaults to None.
         search_to_timestamp (datetime, optional): The end timestamp for the search. Defaults to None.
         search_only_ai_allowed_content (bool, optional): Whether to search only AI-allowed content. Defaults to True.
@@ -233,21 +235,24 @@ def contentstore_text_search(
         List[BaseModel]: The re-ranked search results.
     """
     assert 0 < max_results <= 100, "Maximum results must be between 1 and 100."
-    assert (
-        recency_weight is None or 0 <= recency_weight <= 1
-    ), "Recency weight must be between 0 and 1."
 
     # force proper empty search
     if query is not None and len(query.strip()) == 0:
         query = None
 
+    # determine sort order
+    if query is not None:
+        sort_order = "RELEVANCY_DESC" if sort_descending else "RELEVANCY_ASC"
+    else:
+        sort_order = "PUBLISHED_AT_DESC" if sort_descending else "PUBLISHED_AT_ASC"
+
     body = {
         "query": query,
         "source_items_type": item_type,
         "content_restrictions": "NONE",
         "limit": max_results,
         "offset": offset,
-        "sort": "RELEVANCY_DESC" if query is not None else "PUBLISHED_AT_DESC",
+        "sort": sort_order,
         "languages": languages,
         "published_at": _get_time_filter(search_from_timestamp, search_to_timestamp),
         "content_restrictions": (
@@ -259,5 +264,8 @@ def contentstore_text_search(
     # parse
     results = [search_result(i) for i in response["items"]]
 
-    # re-rank
-    return rerank_by_recency(results, recency_weight=recency_weight)
+    # record rank
+    for rank, result in enumerate(results):
+        result.rank = rank + 1
+
+    return results

{deepsights_api-1.2.3 → deepsights_api-1.2.4}/deepsights/contentstore/resources/news.py

@@ -99,7 +99,7 @@ class NewsResource(APIResource):
         max_results: int = 30,
         offset: int = 0,
         languages: List[str] = None,
-        recency_weight: float = None,
+        sort_descending: bool = True,
         search_from_timestamp: datetime = None,
         search_to_timestamp: datetime = None,
     ):
@@ -112,7 +112,7 @@ class NewsResource(APIResource):
             max_results (int, optional): The maximum number of search results to return. Defaults to 30.
             offset (int, optional): The offset to start the search from. Defaults to 0.
             languages (List[str], optional): The list of languages to search for. Defaults to None.
-            recency_weight (float, optional): The weight to assign to recency in the search ranking. Defaults to None.
+            sort_descending (bool, optional): Whether to sort the results in descending order. Defaults to True.
             search_from_timestamp (datetime, optional): The start timestamp for the search. Defaults to None.
             search_to_timestamp (datetime, optional): The end timestamp for the search. Defaults to None.
 
@@ -128,7 +128,7 @@ class NewsResource(APIResource):
             ),
             query=query,
             max_results=max_results,
-            recency_weight=recency_weight,
+            sort_descending=sort_descending,
             languages=languages,
             offset=offset,
             search_from_timestamp=search_from_timestamp,

{deepsights_api-1.2.3 → deepsights_api-1.2.4}/deepsights/contentstore/resources/secondary.py

@@ -99,7 +99,7 @@ class SecondaryResource(APIResource):
         max_results: int = 50,
         offset: int = 0,
         languages: List[str] = None,
-        recency_weight: float = None,
+        sort_descending: bool = True,
         search_from_timestamp: datetime = None,
         search_to_timestamp: datetime = None,
     ):
@@ -112,7 +112,7 @@ class SecondaryResource(APIResource):
             max_results (int, optional): The maximum number of search results to return. Defaults to 50.
             offset (int, optional): The offset to start the search from. Defaults to 0.
             languages (List[str], optional): The languages to search for. Defaults to None.
-            recency_weight (float, optional): The weight to assign to recency in the search ranking. Defaults to None.
+            sort_descending (bool, optional): Whether to sort the results in descending order. Defaults to True.
             search_from_timestamp (datetime, optional): The start timestamp for the search. Defaults to None.
             search_to_timestamp (datetime, optional): The end timestamp for the search. Defaults to None.
 
@@ -130,7 +130,7 @@ class SecondaryResource(APIResource):
             max_results=max_results,
             offset=offset,
             languages=languages,
-            recency_weight=recency_weight,
+            sort_descending=sort_descending,
             search_from_timestamp=search_from_timestamp,
             search_to_timestamp=search_to_timestamp,
         )

{deepsights_api-1.2.3 → deepsights_api-1.2.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "deepsights-api"
-version = "1.2.3"
+version = "1.2.4"
 description = "Python library for the DeepSights APIs"
 readme = "README.md"
 requires-python = ">=3.10"

{deepsights_api-1.2.3 → deepsights_api-1.2.4}/tests/contentstore/test_news_search.py

@@ -163,51 +163,43 @@ def test_news_text_search_with_date():
         assert result.publication_date <= end and result.publication_date >= start
 
 
-def test_news_text_search_with_recency_low():
+def test_news_text_search_descending():
     """
-    Test the news text search function with a low recency weight.
-
-    This test verifies that the news text search function returns results
-    when the recency weight is set to a low value. It checks that the
-    returned results have valid IDs, ranks, score ranks, and age ranks,
-    and that the final rank is identical to the score rank.
-
-    Note: This test assumes the existence of a `test_query` variable.
+    Test news text search with descending publication date sort order.
+    Verifies results have valid IDs, sequential ranks, and descending dates.
     """
-
     results = ds.contentstore.news.text_search(
-        query=test_query,
+        query=None,
         max_results=10,
-        recency_weight=0.00001,
+        sort_descending=True,
     )
 
     assert len(results) > 0
     for ix, result in enumerate(results):
         assert result.id is not None
         assert result.rank == ix + 1
+        if ix < len(results) - 1:
+            assert result.publication_date >= results[ix + 1].publication_date
 
 
-def test_news_text_search_with_recency_high():
+def test_news_text_search_ascending():
     """
-    Test the news text search function with a high recency weight.
-
-    This test verifies that the news text search function returns results
-    when the recency weight is set to a high value. It checks that the
-    returned results have valid IDs, ranks, score ranks, and age ranks,
-    and that the final rank is identical to the age rank.
-
-    Note: This test assumes the existence of a `test_query` variable.
+    Test news text search with ascending publication date sort order.
+    Verifies results have valid IDs, sequential ranks, and ascending dates.
     """
+
     results = ds.contentstore.news.text_search(
-        query=test_query,
+        query=None,
         max_results=10,
-        recency_weight=0.99999,
+        sort_descending=False,
     )
 
     assert len(results) > 0
     for ix, result in enumerate(results):
         assert result.id is not None
         assert result.rank == ix + 1
+        if ix < len(results) - 1:
+            assert result.publication_date <= results[ix + 1].publication_date
 
 
 def test_news_vector_search():
@@ -360,9 +352,7 @@ def test_news_hybrid_search_only_text():
     )
 
     text_results = ds.contentstore.news.text_search(
-        query=test_query,
-        max_results=results,
-        recency_weight=0.0,
+        query=test_query, max_results=results
     )
 
     assert len(hybrid_results) == results
@@ -450,19 +440,19 @@ def test_news_hybrid_search_with_vector_high():
     """
     hybrid_results = ds.contentstore.news.search(
         query=test_query,
-        max_results=10,
-        vector_weight=0.99999,
+        max_results=5,
+        vector_weight=0.9999999999,
         vector_fraction=1.0,
         recency_weight=0.0,
     )
 
     vector_results = ds.contentstore.news.vector_search(
         query_embedding=test_embedding,
-        max_results=10,
+        max_results=5,
         recency_weight=0.0,
     )
 
-    assert len(hybrid_results) == 10
+    assert len(hybrid_results) == 5
     for ix, result in enumerate(hybrid_results):
         assert equal_results(result, vector_results[ix])
 
@@ -487,7 +477,6 @@ def test_news_hybrid_search_with_vector_low():
     text_results = ds.contentstore.news.text_search(
         query=test_query,
         max_results=results,
-        recency_weight=0.0,
     )
 
     assert len(hybrid_results) == results

{deepsights_api-1.2.3 → deepsights_api-1.2.4}/tests/contentstore/test_secondary_search.py

@@ -164,51 +164,43 @@ def test_secondary_text_search_with_date():
         assert result.publication_date <= end and result.publication_date >= start
 
 
-def test_secondary_text_search_with_recency_low():
+def test_secondary_text_search_descending():
     """
-    Test the secondary text search function with a low recency weight.
-
-    This test verifies that the secondary text search function returns results
-    when the recency weight is set to a low value. It checks that the
-    returned results have valid IDs, ranks, score ranks, and age ranks,
-    and that the final rank is identical to the score rank.
-
-    Note: This test assumes the existence of a `test_query` variable.
+    Test secondary text search with descending publication date sort order.
+    Verifies results have valid IDs, sequential ranks, and descending dates.
     """
-
     results = ds.contentstore.secondary.text_search(
-        query=test_query,
+        query=None,
         max_results=10,
-        recency_weight=0.00001,
+        sort_descending=True,
     )
 
     assert len(results) > 0
     for ix, result in enumerate(results):
         assert result.id is not None
         assert result.rank == ix + 1
+        if ix < len(results) - 1:
+            assert result.publication_date >= results[ix + 1].publication_date
 
 
-def test_secondary_text_search_with_recency_high():
+def test_secondary_text_search_ascending():
     """
-    Test the secondary text search function with a high recency weight.
-
-    This test verifies that the secondary text search function returns results
-    when the recency weight is set to a high value. It checks that the
-    returned results have valid IDs, ranks, score ranks, and age ranks,
-    and that the final rank is identical to the age rank.
-
-    Note: This test assumes the existence of a `test_query` variable.
+    Test secondary text search with ascending publication date sort order.
+    Verifies results have valid IDs, sequential ranks, and ascending dates.
     """
+
     results = ds.contentstore.secondary.text_search(
-        query=test_query,
+        query=None,
         max_results=10,
-        recency_weight=0.99999,
+        sort_descending=False,
     )
 
     assert len(results) > 0
     for ix, result in enumerate(results):
         assert result.id is not None
         assert result.rank == ix + 1
+        if ix < len(results) - 1:
+            assert result.publication_date <= results[ix + 1].publication_date
 
 
 def test_secondary_vector_search():
@@ -359,7 +351,6 @@ def test_secondary_hybrid_search_only_text():
     text_results = ds.contentstore.secondary.text_search(
         query=test_query,
         max_results=5,
-        recency_weight=0.0,
     )
 
     assert len(hybrid_results) == 5

{deepsights_api-1.2.3 → deepsights_api-1.2.4}/tests/data/test_data.json

@@ -1,9 +1,9 @@
 {
     "comment": "text-ada-002 embedding for 'What are emerging food moments with Gen Z consumers?'",
     "question": "What are emerging food moments with Gen Z consumers?",
-    "document_id": "900aa7c7-60f4-4b2e-bc33-9dba20f4508f",
-    "document_page_id": "3d861e4e-37cd-4a7b-b4cf-6986556f9544",
-    "report_id": "7e9943f3-d9b8-4b36-86b0-c5ef939fcc71",
+    "document_id": "b3d728a6-6ef8-4c31-9949-dcc49439f4f1",
+    "document_page_id": "319eb5ab-e701-402e-881f-be3ccede65b3",
+    "report_id": "44126fc2-2aff-47c0-ac3f-d86bd6afcca2",
     "answerV2_id": "7511f5dd-bb5d-4082-8799-0d1733d61afc",
     "embedding": [
         -0.0012380050029605627,