Criteria search on a list of points

[drew-wks]

I'm stumped. Point does not show up in search despite containing the value.

Qdrant Cloud web interface show the following metadata exists for a payload says:

Pasted here:
{ "page": 1, "source": "References/ALAUXs/2022/038_22_AUXSCOUT_PROGRAM_SOP_REVISION.pdf" }

Using python in a jupyter notebook client.scroll confirms the 'page' and 'source' keys are accessible as expected:

import qdrant_client.http.models as models

client.scroll(
    collection_name=qdrant_collection_name, 
    scroll_filter=models.Filter(
        must=[
            models.FieldCondition(
                key="page",
                match=models.MatchValue(value=1),
            ),
        ]
    ),
)

([Record(id='000a7b22-2c06-431b-b8de-c614c996566c', payload={'metadata': {'page': 1, 'source': 'References/ALAUXs/2022/038_22_AUXSCOUT_PROGRAM_SOP_REVISION.pdf'}, 'page_content': "f. Several changes to reflect SSDT, SAT, YPT, and SSALBT requirements for flotilla, \ndivision, and district -level AS staff officers. \n \n3. The revised AUXSCOUT SOP (AUX -SOP-002(D)) will be posted on the Sea Scouts page of \nthe RBS Outreach Directorate wiki site: \nhttps://auxbdeptwiki.cgaux.org/index.php?title=Sea_Scouts ; on the Auxiliary national Youth \nPrograms website: http://wow.uscgaux.info/content.php?unit=S -DEPT&category=auxiliary -\nyouth -programs ; and on the Chief Director of Auxiliary web site (SOP section): http://agroup/ -\nbx.wow.uscgaux.info/content.php?unit=BX -GROUP&category=sop . \n \n4. Internet release is authorized. \n \n***For many reasons including the value of keeping communication li nes clear and open as well \nas facilitating access to training and educational tools, all Auxiliarists are urged to have their own \nemail address and to keep it updated in AUXDATA.*** \n \n*All ALAUX's are posted on the Chief Director of Auxiliary web site located at: CHDIRAUX \nALAUX \n \nIf you have a question regarding this ALAUX, please seek resolution within your Chain of \nLeadership and Management (COLM) including up to your servicing District Director of \nAuxiliary (DIRAUX). If your question still cannot be resolved after that, then please email \nCGAUX@uscg.mil ."}, vector=None)], '00214575-205a-4b5f-9086-e6e33edf6e0e')

Yet the three code snippets below return 'none'. I tried treating 1 as a string and about everything else I could think of.

import qdrant_client.http.models as models

client.scroll(
    collection_name=qdrant_collection_name, 
    scroll_filter=models.Filter(
        must=[
            models.FieldCondition(
                key="page",
                match=models.MatchValue(value=1),
            ),
        ]
    ),
)

([], None)

This attempt was unsuccessful as well.

import qdrant_client.http.models as models

client.scroll(
    collection_name=qdrant_collection_name, 
    scroll_filter=models.Filter(
        must=[
            models.FieldCondition(
                key="source", 
                match=models.MatchValue(value="References/ALAUXs/2022/038_22_AUXSCOUT_PROGRAM_SOP_REVISION.pdf")
            ),
        ]
    ),
    limit=1,
    with_payload=True,
    with_vectors=False,
)

([], None)

Likewise this approach returned nothing

import qdrant_client.http.models as models

client.scroll(
    collection_name=qdrant_collection_name,
    scroll_filter=models.Filter(
        must=[
            models.FieldCondition(
                key="source",
                match=models.MatchText(text="AUXSCOUT"),
            )
        ]
    )
)

([], None)

Any ideas???

[drew-wks]

SOLUTION
Since I used Langchain's .from_documents to place the docs into qdrant. I just found a clue from another post about Langchain. when using Langchain, the Qdrant client.scroll key has to be prefixed with "metadata." so in my case it should be key="metadata.page" and key="metadata.source". I just tried it on the code aboce and they all works.Shout out to the Qdrant discord server.