Questions about the API flow

[watdafox]

Hello,

I'm developping a node.js module for the XML API, to make easier downloading and uploading subtitles.

Code: Select all

var OS = require('mymodule'); var OpenSubtitles = new OS('UserAgent', 'Username', 'Password'); OpenSubtitles.search({ imdbid: 'tt000000', path: '/home/user/breakdance.avi' }).then(function (subtitles) { console.log('%s subtitles found', subtitles.length); }).catch(function (error) { console.error(error); });

The goal is to have a fully-featured wrapper, entirely asynchronous and easy-to-use.

But I have some question:

1) if a user is registered on OpenSubtitles.org as VIP, or some other account without ads, and that the API queries "SearchSubtitles" using a token linked to that premium account, will the response contain links to subtitles without ads? Or is it only by passing by "DownloadSubtitles" ? (the token is sent with SearchSubtitles, so it would make sense that the returned links are ad-free)

2) Why does login in as user: 'wzqkejqkzj' & password: 'iiiiiiii' with a fake UserAgent (say 'QJZHEJHJZ') work?

EDIT: I'm sorry to edit this so much, but when I find the answer by myself, I prefer to take the related questions out rather than making someone take time to answer something non-needed.

[oss]

1. For now SearchSubtitles() returns "clean" links without ads for VIP, but further checks are done, when downloading subtitles, so even you pass clean link, you might get ad-supported subtitles. You can use standard HTTP download and not use XML-RPC download

2. for me it doesnt, I get: "[status] => 414 Unknown User Agent"

[watdafox]

I see, then my auth flow must be broken somehow. Because I get a token back when I use:

Code: Select all

var OS = require('opensubtitles-api'); var Os = new OS('HJJJJEZQK', 'eqizjekj', 'qjzkje'); Os.login() .then(function(token) { console.log(token) });

(that moment when you have to debug why you can bypass authentication without even trying to, lol)

[watdafox]

Regarding the auth, I figured it out. I was only checking for "response.token" and OpenSubtitles sends back a token even if the status is 414. I added a check for the status as well.

Thanks

[watdafox]

Ok, I block on uploading subtitles. Somehow it always responds 403+402 (hash & content don't match + invalid srt format).

It must be the gzip+b64 encode. For unknown reason though, it seems to work perfectly on my end.

Could someone double-check these please:

- subtitle content (plain text): http://pastebin.com/raw.php?i=h8G6fmKa
- subtitle hash (md5): 0f12fd8aadb9496885b25a505cd5d9be
- gzip+b64 subcontent: http://pastebin.com/raw.php?i=ci59ZuGA

This is the code I'm using:

Code: Select all

fs.readFile(subpath, function (err, data) { if (err) reject(err); zlib.deflateRaw(data, function (err, buffer) { if (err) reject(err); resolve(buffer.toString('base64')); }); });

EDIT: finally I managed to upload something... it's zlib.deflate, not deflateRaw. For some reason, if I trust the docs on both node.js and OpenSubtitles, and compare with the results, I can manage to send gzipped+base64 encoded data only if gzipped with header.

[oss]

good you solved it.

Well, we are using

http://php.net/manual/en/function.gzuncompress.php

for decompress...

it is opposite of

http://php.net/manual/en/function.gzcompress.php

from docs

Note:
This is not the same as gzip compression, which includes some header data. See gzencode() for gzip compression.

[watdafox]

Module has been pushed to npm:

https://www.npmjs.com/package/opensubtitles-api

[oss]

looks really great. I would add to docs that one have to register UserAgent with OpenSubtitles.org to use that module and link here:
http://trac.opensubtitles.org/projects/ ... vReadFirst

Also it would be great to have somewhere link directly to our website www.opensubtitles.org

thanks for great work.

[watdafox]

added to the todo list, thanks.

I have another question regarding UA though, doc says to avoid wierd versionning because it's matched by regex. So, is one of these UA better than the other:

- MyApp v0.3.8 => more easy to match, lacks hotfix and prerelease state
- MyApp v0.3.8-2 => more accurate atm
- MyApp v0.3.8-beta2 => our application will use that versionning in the future

[oss]

currently we are using this:

Code: Select all

if (preg_match('/(.*?)[\s_]+[vV]?(\d+(?:\.\d+)*)/', $useragent, $matches)) {

- user agent can be any string
- i can change regexp to match your versions, I think this can do:

Code: Select all

/(.*?)[\s_]+[v]?(\d+(?:[\.\d\-]|alpha|beta|pre|rc|ga|rtm)*)/i

[watdafox]

I based myself on http://semver.org/

------

I don't know the first think about php (it's what you're using, right?), but on nodejs you have "semver" module.

Example:

Code: Select all

semver = require('semver'); semver('0.3.8-rc1+20150813');

will output:

Code: Select all

Object { build: { 0: '20150813' } major: 0 minor: 3 patch: 8 prerelease: { 0: 'rc1' } raw: '0.3.8-rc1+20150813' version: '0.3.8-rc1' }

------

The regex you posted seem to work fine too, except for 1 part: it leaves a trailing '-' with 'UserAgent v0.3.8-LOL' for example.

[oss]

ok my regexp match all of that, so I will update code.

"UserAgent v0.3.8-LOL" - is not according semver, right ?