TS3 searching. Encoding::CompatibilityError & sphinxql: syntax error

[mcmegavolt]

In production, when I make request with Latin characters I get this error:

ActionView::Template::Error (sphinxql: syntax error, unexpected IDENT, expecting ')' 
near 's Magazine 2012', 'event_core', '*toefl* *provider* *tutor* *posting* *event* *tag*',
'<span class="match">' AS before_match, '</span>' AS after_match, ' &#8230; '
AS chunk_separator)'):

And when make request with Cyrillic characters I get this:

Encoding::CompatibilityError
(incompatible encoding regexp match (US-ASCII regexp with UTF-8 string)): 
app/controllers/search_controller.rb:7:in `index'

Production: Sphinx 2.0.4-release (r3135) thinking-sphinx-3.0.0

But in development I have problem only with Cyrillic with the same error. Searching with Latin queries in development is working fine.

[mcmegavolt]

I updated my bundle (I guess riddle gem was updated) Now, In production, when I make request with Latin characters (query is "TOEFL" in this case) I get results like this:

Unreadable symbols is Cyrillic that's showing incorrectly.

When I make request with Cyrillic characters I get the same Encoding::CompatibilityError as in the previous message.

I still can't find the answer ho to fix it :(

[pat]

Hi Aleksandr

Sorry I've not responded yet to this issue... I'm not yet sure what the cause is. I'll try to look into it soon, but if you're able to create a test application that reproduces the issue, that'd be fantastic.

Cheers

Pat

[pat]

Hi Aleksandr

Can you provide the full stacktrace of the error you're seeing when you search for Cycillic characters?

[che-shr-cat]

Hi Pat,

First of all thanks for the module.

I encountered somehow similar problem, when on my development with utf-8 searches the app crashed with the error "incompatible encoding regexp match (US-ASCII regexp with UTF-8 string)" leading to the line 28 of lib/thinking_sphinx/search/query.rb. I tried many things, nothing helped me. Then finally I've added at the head of this file the encoding line:

# encoding: utf-8

and the problem disappear.

I didn't find any mentions of the same problem in other places, and I'm not sure this was the best fix.

[pat]

@che-shr-cat thanks for that - I just pushed a commit with that line, so future releases will have it sorted :)