Get file names from InMemoryDocumentStore

[anirbansaha96]

I'm using InMemoryDocumentStore as my document_store. At a point, I'm using PDToTextConverter followed by writing them into the document store.

Is there any way to get a list of files in the InMemoryDocumentStore().

Also in my answer output, with the command print_answers(prediction, details="all") I'm getting a document doument_id, is there any way to leverage this information to perhaps get the filename.

[tanaysoni]

Hi @anirbansaha96,

Is there any way to get a list of files in the InMemoryDocumentStore().

You can use InMemoryDocumentStore.get_all_documents() method.

doument_id, is there any way to leverage this information to perhaps get the filename.

You can use InMemoryDocumentStore.get_document_by_id(). The returned Document object should have a name if it was supplied during indexing.

[anirbansaha96]

@tanaysoni The output for my command InMemoryDocumentStore.get_all_documents(document_store) is
[Document(id='6257a1cdb8e5f13804b65b3b8125d509', text='Security best practices for Azure solutions', external_source_id=None, question=None, query_score=None, meta={}, tags=None)]

But when I'm searching InMemoryDocumentStore.get_document_by_id(document_store,'6257a1cdb8e5f13804b65b3b8125d509') or InMemoryDocumentStore.get_document_by_id(document_store,id='6257a1cdb8e5f13804b65b3b8125d509'), it should ideally give me an output as the filename, like you mentioned. But it is showing the following error:

KeyError                                  Traceback (most recent call last)
<ipython-input-15-ce74eccb6d51> in <module>()
----> 1 InMemoryDocumentStore.get_document_by_id(document_store,'62b8c57494d145acdcecf3950754fa61')

1 frames
/content/haystack/haystack/database/memory.py in _convert_memory_hit_to_document(self, hit, doc_id)
     69         document = Document(
     70             id=doc_id,
---> 71             text=hit[0].get('text', None),
     72             meta=hit[0].get('meta', {}),
     73             query_score=hit[1],

KeyError: 0

[anirbansaha96]

@tanaysoni, wanted to add some information in case it helps to pinpoint the issue.
With the command document_store.get_all_documents() , I get the output id='8096d30a73bca646151018cbd42fa977'.
However with print_answers(prediction, details="all"), the same document has 'document_id': '1843'

[anirbansaha96]

@tanaysoni Has there been changes made to print_answers( #something , details="all") because it is no longer showing document_id like before.

[tanaysoni]

Hi @anirbansaha96,

With #217, you can now get file names using the ID for all document stores. Here's an example:

from haystack.database.memory import InMemoryDocumentStore

document_store = InMemoryDocumentStore()

test_docs = [
    {"name": "testing the finder 1", "text": "testing the finder with pyhton unit test 1", 'meta': {'url': 'url'}},
    {"name": "testing the finder 2", "text": "testing the finder with pyhton unit test 2", 'meta': {'url': 'url'}},
    {"name": "testing the finder 3", "text": "testing the finder with pyhton unit test 3", 'meta': {'url': 'url'}}
]
document_store.write_documents(test_docs)

print(document_store.get_document_by_id("e97e6fbebbc591fe7214e0bf26ec5dbf").meta["name"])

[tanaysoni]

Has there been changes made to print_answers( #something , details="all") because it is no longer showing document_id like before.

I am not able to reproduce this issue with tutorials 1 & 3 on the latest master. If possible, could you share a code snippet to help reproduce?

The issue is print_answers() is deleting keys from the passed results dicts. It should get resolved with #230.

[anirbansaha96]

@tanaysoni There is still the same error, print_answers() prints a document id as 'document_id': '742', however print(document_store.get_document_by_id("742").meta["name"]) gives KeyError:

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-21-b74f2b2ca29a> in <module>()
----> 1 print(document_store_Customer_Content.get_document_by_id("742").meta["name"])

/content/haystack/haystack/database/memory.py in get_document_by_id(self, id)
     65 
     66     def get_document_by_id(self, id: str) -> Document:
---> 67         document = self._convert_memory_hit_to_document(self.docs[id], doc_id=id)
     68         return document
     69 

KeyError: '742'

[anirbansaha96]

@tanaysoni one update to pinpoint this issue:

1. document_store.get_all_documents() gives id='30af7155e4a229c32317768ec16954eb' and using print(document_store.get_document_by_id("30af7155e4a229c32317768ec16954eb").meta["name"]) gives me the correct name.

2. However print_answers() provides a document id 742 which doesn't work with print(document_store.get_document_by_id("742").meta["name"])

[anirbansaha96]

Thank You #232 will solve this issue hopefully.

[tanaysoni]

Hi @anirbansaha96, thank you for raising the issue. It is now resolved with #232.

[anirbansaha96]

Thank You, I've checked it. It works fine now. Thank You!

[sophgit]

Hi!

I'm trying to use the elasticsearch retriever like this:
retriever.retrieve("When is Britta on vacation?")
and would like to return the document name and/or text. I know that I can get the name and text by doing this
document_store.get_document_by_id("5c4fc733-a69b-479f-bce6-2fc517623cd9").text
document_store.get_document_by_id("5c4fc733-a69b-479f-bce6-2fc517623cd9").meta["name"]

However, the retriever.retrieve returns something like [<haystack.schema.Document at 0x7efcd596b510>].
How can I get the document id of the retrieved document, so I can access the documents text or name?

Thank you!!

[anirbansaha96]

@tanaysoni

[tanaysoni]

Hi @sophgit, you can use document.id to get the document ids of the retrieved documents.

It seems you're using an earlier version of Haystack. With the current master branch, the representation of a document(in debugger/console, etc) is changed to be human-readable rather than the cryptic object notation.

To upgrade to the laster master branch, you can follow the installation guide.

[sophgit]

aah, thank you @tanaysoni, it worked with the current master branch.