v0.1.4

src/chromaIntf.py

@@ -5,6 +5,7 @@ from langchain.chains import RetrievalQA
 from langchain.embeddings import HuggingFaceBgeEmbeddings
 from langchain.retrievers.self_query.base import SelfQueryRetriever
 from langchain.chains.query_constructor.base import AttributeInfo
+from langchain.retrievers.self_query.chroma import ChromaTranslator
 from llm.llmFactory import LLMFactory
 from datetime import datetime
 import baseInfra.dropbox_handler as dbh
@@ -50,17 +51,61 @@ class ChromaIntf():
                                   persist_directory=self.persist_db_directory)
         #self.vectorstore._client.
 
+        # timestamp --> time when added
+        # source --> notes/references/web/youtube/book/conversation, default conversation
+        # title --> of document , will be conversation when source is conversation, default blank
+        # author --> will default to blank
+        #     "Year": 2024,
+        #"Month": 1,
+        #"Day": 3,
+        #"Hour": 11,
+        #"Minute": 29
         self.metadata_field_info = [
             AttributeInfo(
                 name="timestamp",
-                description="Python datetime.timestamp of the document in isoformat, can be used for getting date, year, month, time etc ",
+                description="Python datetime.timestamp of the document in isoformat, should not be used for query",
                 type="str",
             ),
+            AttributeInfo(
+                name="Year",
+                description="Year from the date when the entry was added in YYYY format",
+                type="int",
+            ),
+            AttributeInfo(
+                name="Month",
+                description="Month from the date when the entry was added it is from 1-12",
+                type="int",
+            ),
+            AttributeInfo(
+                name="Day",
+                description="Day of month from the date-time stamp  when the entry was added, it is from 1-31",
+                type="int",
+            ),
+            AttributeInfo(
+                name="Hour",
+                description="Hour from the timestamp when the entry was added",
+                type="int",
+            ),
+            AttributeInfo(
+                name="Minute",
+                description="Minute from the timestamp when the entry was added",
+                type="int",
+            ),
             AttributeInfo(
                 name="source",
                 description="Type of entry",
                 type="string or list[string]",
             ),
+            AttributeInfo(
+                name="title",
+                description="Title or Subject of the entry",
+                type="string",
+            ),
+            AttributeInfo(
+                name="author",
+                description="Author of the entry",
+                type="string",
+            )
             ]
         self.document_content_description = "Information to store for retrival from LLM based chatbot"
         lf=LLMFactory()
@@ -72,6 +117,7 @@ class ChromaIntf():
             self.vectorstore,
             self.document_content_description,
             self.metadata_field_info,
+            structured_query_translator=ChromaTranslator(),
             verbose=True
         )