Topic on Extension talk:CirrusSearch

Question: Can this extention search inside documents which have been uploaded to the wiki like PDF's?

Yes, this is possible since gerrit:101252 that resolved bugzilla:6421. See also the first item in m:Tech/News/2014/01.

I've just backported most of the features in Cirrus' master branch to the REL1_22 branch, including this. If you want to try it make sure to get the new version of the Elastica plugin on its REL1_22 branch as well and rebuild your index.

Do we need to force index the pdf files? I'm seeing no results from pdfs.

First of all try a null edit and wait some time (at most few hours) for the job queue; report back if that wasn't enough.

I've been working on a method that parses document files (PDFs, Word, PPT, etc.) using Tika to extract the document text, and then re-insert the extracted text into the file_text field of the WIKI_general_first index inside Elasticsearch. On this point, I have a couple of questions: 1) Does this sound like the proper method to provide searchable text from documents in CirrusSearch? 2) Has anyone else done anything similar?

On point 2, the reason I ask is that for some documents I'm extracting text from, the resulting text can be huge (100s of MBs) and can grind the search to a hault for some queries (mostly for terms which there aren't many of inside the index).

Any pointers would be greatly appreciated.

There is a working extension for this that I have been using for years: https://www.mediawiki.org/wiki/Extension_talk:FileIndexer#Resubmit --SmartK (talk) 15:32, 26 March 2015 (UTC)

Hello Everyone,

I have just added the extension CirrusSearch with all the dependencies. I am not able to search through PDF, txt and, docx. Please help!

Regards,

I am also not able to index documents. I have Image Authorization configured.

I got PDF search working but not for *.doc files (PDF search works on MW 1.26.2 and MW 1.28.2). You need at least to

• set up elasticsearch with „Mapper Attachments Plugin“ (for ElasticSearch 1.x (below version REL1_28) + elasticsearch-mapper-attachments version 2.7.1, ElasticSearch 2.x since REL1_28 + plugin manager will autodetect the right version of mapper-attachments)
• activate Extension:PdfHandler
• build the index exactly as the README of Extension:CirrusSearch documents it

1. Does anybody have a solution for searching inside *.doc files yet?
2. Did I miss some configuration to set up?
3. Would it need an FileHandler for doc files to get it working?

Does anyone know how to do this with elasticsearch 5.5 and Mediawiki 1.29.0 (MW1.29 does now require elasticseach 5 and higher). The mapper-attachments have been depreciated. The new thing is: "ingest attachment plugin". See here: https://www.elastic.co/guide/en/elasticsearch/plugins/5.5/ingest-attachment.html

The Extension:ExtendedSearch of BlueSpice 3.0 offers full-text search in articles and files (i.e. Microsoft Office documents & PDF files) via Elasticsearch 6+ and the Ingest Attachment Processor Plugin

A new option: Extension:TikaAllTheFiles. This extension uses Tika to do content extraction (text and/or metadata), and provides the content to CirrusSearch for indexing.