This code:

func main() {
	fmt.Printf("%#v\n",
		mention.GetTags('@', strings.NewReader("[email protected]")))
	fmt.Printf("%#v\n",
		mention.GetTags('@', strings.NewReader("[email protected]"), '.'))
}

Produces the following output:

[]string{"example.com"}
[]string{"example"}

Which is unexpected – at least it is in our use case. If there is a user with the handle @example and someone writes send an email to [email protected]! then this user will be matched if . is in the terminator list. I think most people have . in the terminator list, since otherwise a @handle at the end of a sentence produces the wrong result.

I wanted to write a patch to fix this, but I'm not sure what the best way to handle this would be, as ,@handle or /@handle should be a match. Maybe ignoring [any-unicode-letter]@handle would be a good solution?

Maybe related to #16
Sometimes a single character can count as three characters (ie the prime symbol which looks like a single quote). this causes the Index value of a tag to be incremented by 2. While this may work fine for golang, when you send this data to another stack (say a webUI), they may not have that increment for symbols.
So the problem is that in golang the index value of a tag may be 15 (for example), but from javascript's perspective, the correct answer is 13. The languages figure/count indexes differently.

https://play.golang.org/p/_YhJTv3s6vb

"how does it feel to be rejected? #loner tt ggg sjdsj dj #link"
[loner linker]

"how does it feel to be rejected? # it is #loner tt ggg sjdsj dj #linker"
Expect: [loner linker]
Got: [ it is #loner tt ggg sjdsj dj #linker]

Add to documentation and examples how to use mention with custom terminators

@gernest i've refactored the code base again to give us more information about the tags other than just what they are. We now return (in v2 branch) a struct rather than strings.

Heres what i'm thinking... we create a new branch called v1 from master, and have that branch be the default branch for the repo, but also have v2 (which has breaking changes) that people can import via go get github.com/gernest/mention.v2 or get original version via go get github.com/gernest/mention.v1 or not include a version at all. Thoughts on versioning this repo?

Thoughts on the changes in v2??

Twitter has an official list of characters that are allowed/disallowed in hashtags and mentions. For example, hashtags can't start with a number, and cannot contain special characters (some are allowed like _ (underscore)).

Create a function that gets mentions/hashtags according to these standard set by Twitter. May make sense to do this in a subdirectory creating an organizational location to support other social media sites in the future like facebook, instagram, etc.

Since termIndexes is a []int and some "raw" byte indexes are accessed (e.g. str[t+1] == byte(prefix)), multibyte prefixes won't work. This:

mention.GetTags('🔥', "I'm on 🔥fire!")

Should match fire, but doesn't match anything.

Failing test case:

diff --git i/mention_test.go w/mention_test.go
index 423ed9b..904af66 100644
--- i/mention_test.go
+++ w/mention_test.go
@@ -139,6 +139,9 @@ func (s *MentionSuite) TestGetTags(c *C) {

        // use default terminators
        c.Assert(GetTags('@', "hello @test"), DeepEquals, []Tag{{'@', "test", 6}})
+
+       // Test multibyte start characters.
+       c.Assert(GetTags('🔥', "I'm on 🔥fire!"), DeepEquals, []Tag{{'🔥', "fire", 6}})
 }

 func BenchmarkGetTags(b *testing.B) {

Right side after apostrophe should't be a part of hashtag.
In case of @bob's cookies only @bob is tag.

If i GetTags on "### #. # the #foo #bar #baz #baz." and supply all special characters as runes (ie var specials = []rune(".=+-~,/ <>?[]{})(*&^%$#@!;:\"'|\\")), the resulting answer i get back is

achievements_test.go:27:
    c.Check(tags, DeepEquals, []string{"bar", "baz", "foo"})
... obtained []string = []string{"#", ".", "bar", "baz", "foo"}
... expected []string = []string{"bar", "baz", "foo"}

I only want my tags to be alphanumeric, and I can't seem to get that to work out of the box.