salyh
commented
10 years ago If i read and understood well your search for "ambiguous" in the imapriverdata index succeeded (which means attachment data is searched correctly) but the hits.hits result document contains base64 encoded data? Or does your search for "ambiguous" does not work at all?
Can you try to omit "type" : "nested" for the attachments mapping to see if this is the cause? Pls also have a look here: https://github.com/salyh/elasticsearch-river-imap/issues/10#issuecomment-50125929
I neither of the above work pls. post (maybe at http://pastebin.com) the output of
http://elasticsearchserver:9200/test/_mapping?pretty=true
and
http://elasticsearchserver:9200/imapriverdata/_mapping?pretty=true
because the only problem i can think of (currently) is a incorrect mapping.
vanderstaaij
dan-lind
Hi,
I'm having trouble accessing the Tika-extracted text content of attachments when downloaded with river-imap.
I have installed mapper-attachments and also the langdetect and icu_tokenizer plugins. I'm still on ElasticSearch 1.1.0 now and thus using river-imap 0.7b20 and mapper-attachments 2.0.0.
To test if the mapper-attachments plugin is working and I can extract text, I have created an index called test, created a mapping for a type called mapper. I have based this test on what is described here: https://gist.github.com/dadoonet/5310075
POST http://localhost:9200/test:
POST http://localhost:9200/test/mapper/_mapping:
I can index a document:
PUT http://localhost:9200/test/mapper/66d2006d-d269-5de7-5aa8-53f5b7b730e8:
The indexed document looks as such:
GET http://localhost:9200/test/mapper/66d2006d-d269-5de7-5aa8-53f5b7b730e8:
I can perform the following search: POST http://localhost:9200/test/mapper/_search with the following payload (searching for the term "ambiguous"):
This is the result which includes the file field and the extracted content:
To me this confirms the mapper-attachment plugin is working and I can extract my text.
So, my wish is to extract the text from attachments stored by river-imap as well. Therefore I made the following configuration of river-imap, containing type mappings and index settings:
As you can see I have configured content to be of type attachment, but with extended configuration for the mapper-attachments plugin.
I also made some changes to the mapping in order to do nested query searches on attachments, headers and to, which seem to be accepted.
When started, river-imap nicely syncs all messages from the IMAP server to the imapriverdata index. It creates the index nicely with the settings and mappings as specified in the above configuration.
When I perform a search I want to see the extracted text, just like when using the mapper-attachment in the test I described above. I perform this search:
POST http://localhost:9200/imapriverdata/mail/_search with the following payload:
this results in the following:
As you can see, attachments.content shows the base64 encoded data, as opposed to my test above with just the mapper plugin. I'd say the attachments.content field is configured exactly the same as in the test.
Any suggestion why this is happening?