twitter-archive / twitter-text-rb Goto Github PK

twitter-text-1.5.0 is not allowed to read for other owner. it's following situation.

% gem install twitter-text
% gem unpack twitter-text
% cd twitter-text-1.5.0/lib
% ls -l
drwxr-xr-x 10 hsbt wheel 340  1 12 22:01 twitter-text
-rw-r-----  1 hsbt wheel 564  1 12 22:01 twitter-text.rb
% ls -l twitter-text
-rw-r----- 1 hsbt wheel 22215  1 12 22:01 autolink.rb
-rw-r----- 1 hsbt wheel   447  1 12 22:01 deprecation.rb
-rw-r----- 1 hsbt wheel 12488  1 12 22:01 extractor.rb
-rw-r----- 1 hsbt wheel  2784  1 12 22:01 hit_highlighter.rb
-rw-r----- 1 hsbt wheel 17123  1 12 22:01 regex.rb
-rw-r----- 1 hsbt wheel  1941  1 12 22:01 rewriter.rb
-rw-r----- 1 hsbt wheel   916  1 12 22:01 unicode.rb
-rw-r----- 1 hsbt wheel  4614  1 12 22:01 validation.rb

For example, application user installed twitter-text, but apache2 user used twitter-text. it happend LoadError.

Please adding to other users reading permission and bump version 1.5.1.

Thanks.

The t gem depends on twitter-text to extract mentioned screen names from Tweet text.

Five separate users, all on different operating systems and Ruby versions, reported a LoadError on the line where I require 'twitter-text'.

Originally, I assumed it was an environmental issue, since I was unable to reproduce the problem on my machine, however, when I released a version of the gem (0.9.4), which lazy-loaded twitter-text, users reported the problem had been fixed.

As you can see from this GitHub compare view, the only difference between version 0.9.3, where users were reporting a problem, and 0.9.4, where users reported that it was fixed is the lazy-loading of twitter-text: sferik/t-ruby@v0.9.3...v0.9.4

Again, I am unable to reproduce the error, but I believe that there is a problem with requiring twitter-text as a dependency on some systems, which I thought you'd want to know about.

I am considering reimplementing the extract_mentioned_screen_names functionality that I'm using from twitter-text within the t gem, however I'd prefer to depend on twitter-text than maintain that code myself.

The Developer Display Requirements state (in a roundabout way) that the "@" symbol and the username must link to the Twitter profile (not just the username), but autolink only links the username, not the @ symbol as well.

Background: We're trying to get Site Streams access and this was raised as an issue via email.

A real world URL that fails:

http://www.amazon.com/Jeanne-Córdova/e/B001KC61FK/ref=cm_sw_r_tw_nu_7-WQob0RDA65Y

To better see the unicode character here it is quoted:

http%3A%2F%2Fwww.amazon.com%2FJeanne-C%C3%B3rdova%2Fe%2FB001KC61FK%2Fref%3Dcm_sw_r_tw_nu_7-WQob0RDA65Y

A test that verifies this:

    - description: "Valid unicode url: unicode path"
      text: "http://www.example.com/Bob-Córdova/"
      expected: true

When using <%= auto_link("link @user, please #request") %>

It returns escaped html in the view, I cant find a way to get usable HTML

I am finding it difficult to use different html attributes for different entity types (i.e. url, screen name, entity, cashtag). I have ended up recreating the code here...

https://github.com/twitter/twitter-text-rb/blob/v1.6.1/lib/twitter-text/autolink.rb#L67

...to add type specific attrs. In my specific case, I could work around this by using class names, but was curious if you would accept a patch for type specific attributes? Another solution I was thinking would be passing in a block that gets called in #auto_link_entities to allow for specific attributes to be changed (very basic example below)

  Twitter::Rewriter.rewrite_entities(text, entities) do |entity, chars|
    if entity[:url]
      options = options.merge(block.call(:url) || {})
      link_to_url(entity, chars, options, &block)
    elsif entity[:hashtag]
      options = options.merge(block.call(:hashtag) || {})
      link_to_hashtag(entity, chars, options, &block)
    elsif entity[:screen_name]
      options = options.merge(block.call(:screen_name) || {})
      link_to_screen_name(entity, chars, options, &block)
    elsif entity[:cashtag]
      options = options.merge(block.call(:cashtag) || {})
      link_to_cashtag(entity, chars, options, &block)
    end
  end

The auto_link method raises a RuntimeError when passed the text of a Twitter::Tweet object from the latest version (5.2.0) of the twitter gem.

Here's an example:

require 'twitter' # 5.2.0
require 'twitter-text' # 1.7.0

client = Twitter::REST::Client.new do |config|
  config.consumer_key        = "YOUR_CONSUMER_KEY"
  config.consumer_secret     = "YOUR_CONSUMER_SECRET"
  config.access_token        = "YOUR_ACCESS_TOKEN"
  config.access_token_secret = "YOUR_ACCESS_SECRET"
end

tweet = client.user_timeline('gem').first

Twitter::Autolink.auto_link(tweet.text)
# => RuntimeError: can't modify frozen String

This was working as expected in version 4.8.1, but I haven't tested any of the versions in between.

lib/autolink.rb includes a custom version of link_to. This clobbers Rails' link_to definition and fails on link_to calls that utilize a block. link_to should be enhanced, renamed, or the following should be removed from the documentation:

# For Ruby on Rails you want to add this to app/helpers/application_helper.rb
module ApplicationHelper
  include Twitter::Autolink
end

Would a patch for any of these solutions be accepted?

would allow us to kill a git dependencies :)
@caniszczyk @jakl

A conformance spec is currently broken on master.

  1) Failure:
test_urls Autolink URLs with unicode chars in them(ConformanceTest) [test/conformance_test.rb:126]:
<"See: <a href=\"http://example.com/tsa-pre✓™\">http://example.com/tsa-pre✓™</a> is a link"> expected but was
<"See: <a href=\"http://example.com/tsa-pre\">http://example.com/tsa-pre</a>✓™ is a link">

In this case, the unicode characters are not being included in the matched URL when we expect them to be.

I believe @psychs addressed the rationale behind what should and shouldn't match in #91, but I don't believe the spec is clear enough which unicode codepoint ranges should be considered part of the URL, and which shouldn't.

A solution to this issue would be to fix the spec; alternatively, I can take on the task of fixing it given documentation on what codepoint ranges should be considered part of the URL.

My mention string has hyphen in between when I am trying to create link it break because of '-'.

mention_str = "sunny-b"
puts Twitter::Autolink.auto_link(mention_str, :username_include_symbol => true)

O/P - sunny-b with invalid URL

The tweet_length calculation code update (#61) will be really helpful for us. Can we get a new release that includes it? Thanks!

In this file

I got "text.scan(Twitter::Regex[:auto_link_hashtags]) do |before, hash, hash_text, hash|"

Why there are two "hash" ?

The auto_link_usernames_or_lists can't deal with html well.

for example:

auto_link_usernames_or_lists('<p>@blankyao</p>')

it return the text without any link.

tweet_length should normalize line breaks of the string. eg. twee length of hello\n of hello\r\n should be equal. I think same thing happen on twitter.com side.

I'm trying to extract the hashtags from the text of this Tweet using the Twitter::Extractor.extract_hashtags method.

It's hard to see on the Twitter website, but the text string contains various newline characters:

"#bieber\n#justinbieber \n#bangersgarlingistrending \nRT if you are mega excited about biebs' 'secret' show coming in January to The Metro"

I expect the result to be:

["bieber", "justinbieber", "bangersgarlingistrending"]

instead of just:

["bangersgarlingistrending"]

I am not sure if this can be done now but I would like to open up the url's in a seperate browser window. Is this possible or does this functionality need to be added?

[feature request]
Would it be possible to convert t.co domains to the full / original domain when autolinking them?

I was playing with it in my rails console (jruby-1.6.7, rails 3.2) and found things were not quite correct if I extract mentions/hashtags with japanese or chinese text with indices, the screen name was correctly returned, but the indices is not correct.

Then I created a spec in my rails project and copied from twitter-text-rb/spec/extrator_spec into it, and I got failed tests if I run it in my rails project. I created a gist for the results. https://gist.github.com/2204246

Is there any settings to be done for it could be used in a rails project?

I have the Encoding.default_external and Encoding.default_internal already set to Encoding::UTF_8

We should provide semantic classes for styling purposes: e.g., twitter-hashsymbol, twitter-atsymbol, twitter-emoji, twitter-text, etc.

We can probably drop / deprecate customizable tags because the correct tag is span.

Equivalent js issue: twitter-archive/twitter-text-js#110

are you generally supporting Rails' auto_link or is twitter_text's autolink to be used only for Tweets? the impression i got from the twitter-text docu was that part of the gem's functionality is to extend the default auto_link's capabilities to link to tweet authors etc.

when i include Twitter::Autolink in my application helper emails are no longer auto_linked in my views. when i remove the include they work again. in the views i have a normal call like:
= auto_link @text

i'm using:

• the latest twitter/twitter-text-rb (github path in gemfile, pulled just now)
• ruby-1.9.3-p0
• Rails 3.1.1
• rails_autolink (1.0.4) (via Rubygems)

simple demo in my rails console (as far as i understand it helper.auto_link uses 'pure' default auto_link while plain auto_link, after the Twitter::Autolink include, uses a method extende by twitter-text):

ruby-1.9.3-p0 :002 > helper.auto_link '[email protected] hahaha'
=> "<a href="mailto:[email protected]">[email protected] hahaha"

ruby-1.9.3-p0 :003 > include Twitter::Autolink
=> Object
ruby-1.9.3-p0 :004 > helper.auto_link '[email protected] hahaha'
=> "<a href="mailto:[email protected]">[email protected] hahaha"

ruby-1.9.3-p0 :006 > auto_link '[email protected] hahaha'
=> "[email protected] hahaha"
uby-1.9.3-p0 :007 > auto_link "This is a #tweet with a @username"
=> "This is a <a href="http://twitter.com/search?q=%23tweet\" title="#tweet" class="tweet-url hashtag" rel="nofollow">#tweet with a @<a class="tweet-url username" href="http://twitter.com/username\" rel="nofollow">username"

ruby-1.9.3-p0 :008 > helper.auto_link "This is a #tweet with a @username"
=> "This is a #tweet with a @username"

Is there a way to not autolink URLs?

From what I understand, twitter-text-rb isn't really meant for URLs, so I prefer letting https://github.com/vmg/rinku deal with that.

Installing this gem via sudo give root user read access to the files.

Install gem:

sudo gem install 'twitter-text' --version 1.4.13

Start irb

require 'rubygems'
require 'twitter-text'
=> LoadError: no such file to load -- twitter-text

All files in /usr/lib/ruby/gems/1.8/gems/twitter-text-1.4.13/lib permissions of 660 (rw-rw----). I needed to manually add read permissions for non-root users to all files in the lib directory.

I just has this test:

with jruby-1.6.7.2 1.9 mode:

$ jruby -S test.rb
=== original ===
@baozi hi you are a dog #dog
@baozi 嗨你是一条狗诶 #狗狗
@baozi ねえ、あなたは犬だ #犬
=== autolinked ===
<a class="tweet-url username" href="https://twitter.com/baozi" rel="nofollow">@baozi</a> hi you are a dog <a class="tweet-url hashtag" href="https://twitter.com/#!/search?q=%23dog" rel="nofollow" title="#dog">#dog</a>
@baozi 嗨你是一条狗诶 #狗<a class="tweet-url username" href="https://twitter.com/baozi" rel="nofollow">狗baozi</a><a class="tweet-url hashtag" href="https://twitter.com/#!/search?q=%23狗狗" rel="nofollow" title="#狗狗">@狗狗</a>@baozi 嗨你是一条狗诶 #狗狗
@baozi ねえ、あなたは犬だ #<a class="tweet-url username" href="https://twitter.com/baozi" rel="nofollow">犬baozi</a><a class="tweet-url hashtag" href="https://twitter.com/#!/search?q=%23犬" rel="nofollow" title="#犬">@犬</a>@baozi ねえ、あなたは犬だ #犬

The autolink breaks when test contains Chinese/Japanese. But this only happens with jruby, if I use MRI ruby, everything is fine.

The code I use to test is here: https://gist.github.com/3045668

http://www.w3.org/Addressing/URL/url-spec.txt

Using 1.6.1:

>  Twitter::Autolink.auto_link("@Bob hello")
=> "@<a class=\"tweet-url username\" href=\"https://twitter.com/bob\" rel=\"nofollow\">bob</a> hello"

Is there a reason why Autolink would drop capitalization?

It seems that the auto_link method is wrongly detecting email addresses as mentions and linking them.

Twitter::Autolink.auto_link_usernames_or_lists 'My email is [email protected]'
 => "My email is robert@<a class=\"tweet-url username\" href=\"https://twitter.com/weboffins\" rel=\"nofollow\">weboffins</a>.com" 

This feels counter intuitive to me as it creates invalid links, and also present conflicts when used with the rinku gem for linking emails with mailto://

Source string (e.g.):

extract_urls("http://www.flickr.com/photos/7511731@N06/7684402828/in/pool-89594630@N00")

Expected:

["http://www.flickr.com/photos/7511731@N06/7684402828/in/pool-89594630@N00"]

Actual:

["http://www.flickr.com/photos/7511731@N06/7684402828/in/pool-89594630"]

In practice we lose some context for the image. (Twitter redirects to just http://www.flickr.com/photos/7511731@N06/7684402828/).

Hi,

I've just started using this gem, and am encountering a problem when auto_linking URLs. I've added 'include Twitter::AutoLink' into my ApplicationHelper, however when I do the following, I'm getting unexpected results:

text = "@gavinlaking"
"@gavinlaking"

auto_link(text)
"@<a class=\"tweet-url username\" href=\"http://twitter.com/gavinlaking\" rel=\"nofollow\">gavinlaking</a>"

Okay so far... so let's try a URL:

text = "http://www.example.com"
"http://www.example.com"

auto_link(text)
"<a href=\"http://www.example.com\"{:xmlns=>\"http://www.w3.org/1999/xhtml\", \"xml:lang\"=<{:rel=>\"nofollow\", :class=<nil}, :lang=>{:rel=<\"nofollow\", :class=>nil}}<http://www.example.com>/a<"

It looks like the brackets are the wrong way around, and also we have a hash in the middle of our HTML.

Any ideas?

Latin: $$$
Full width: ＄＄＄
Small: ﹩﹩﹩ (We could consider the small size hashtag add too)

Should be a simple add here:
https://github.com/twitter/twitter-text-rb/blob/master/lib/twitter-text/regex.rb#L284

We already do this for hashtags elsewhere:
https://github.com/twitter/twitter-text-rb/blob/master/lib/twitter-text/regex.rb#L165

I'm using the Java library and am wondering how I can make sure the protocol is included the extracted href. Not including the protocol in the link can cause issues when a non https external url is linked to from a site with SSL. Is there a way to ensure the protocol is included instead of producing href="//..?

I just spent ages trying to get this to work on ruby 1.8.7 and rubygems 1.5.3. Now I have it working. At the top of my script I needed to do:

require 'rubygems'
$KCODE='u'
require 'twitter'
require 'twitter-text'
include Twitter::Autolink 

KCODE='u' is a bit of magic I found at the bottom of a discussion on a google cached copy of the old issue tracker here: http://webcache.googleusercontent.com/search?q=cache:iAS36VNIoLYJ:https://github.com/mzsanford/twitter-text-rb/issues/8+twitter-text+requires+the+%24KCODE+variable+be+set+to+%27UTF8%27&cd=5&hl=en&ct=clnk&client=ubuntu

(took me a while to find!) As I say, I got it working, so this is maybe just a comment on the documentation, or not even that.