This is a major refactor of the links subparser, previously known as anchors subparser.
Closes#355, #534
BREAKING CHANGE: `excludeTrailingPunctuationFromURLs` option was removed. This is now the default behavior
In text, headings only require a single linebreak to be treated as such.
However, in lists, they would require a double linebreak.
Now, the behavior in lists and text is consistent, requiring only a single
linebreak.
Closes#495
This feature was seen as a bit "duh!" since midword asterisks are not really a thing and, for these situations, you can simply escape the asterisk character.
Closes#499
BREAKING CHANGE: literalMidWordAsterisks option was removed and so asterisks will always retain their markdown magic meaning in a source text.
If you're using this feature, and you wish to retain this option, you can find a shim here: <https://gist.github.com/tivie/7f8a88c89ffb00d2afe6c59a25528386>
In some circumstances, on a reference style image, the last capturing
group is ignored, which causes the fucntion argument to return the number
of matches instead of a string (or undefined).
Checking if the title parameter is a string ensures that the title
parameter is actually something that was caught by the regex and not some
metadata.
Closes#585
One or more spaces before the language declaration of a code block is supported by Github.
E.g.
``` html
<div>HTML!</div>
```
``` html
<div>HTML!</div>
```
```html
<div>HTML!</div>
```
Closes#569
With this option enabled, this:
```md
> some text
> some other text
```
witll result in:
```html
<blockquote>
<p>some text</p>
</blockquote>
<blockquote>
<p>some other text</p>
</blockquote>
```
This is the default behavior of GFM.
Closes#477
BREAKING CHANGE: makeHtml subparsers names changed, by prepending 'makehtml.' to them.
Example: 'anchors', subparser is now named 'makehtml.anchors'.
Event names were also changed to reflect this.
Example: 'anchors.before' is now named 'makehtml.anchors.before'.
**To migrate:**
If you have a listener extension, replace the old event name with the new one. Example:
Replace this
```js
showdown.extension('myext', function() {
return [{
type: 'listener',
listeners: {
'anchors.before': function (event, text, converter, options, globals) {
//... some code
return text;
}
}];
});
```
with this
```js
showdown.extension('myext', function() {
return [{
type: 'listener',
listeners: {
'makehtml.anchors.before': function (event, text, converter, options, globals) {
//... some code
return text;
}
}];
});
```
A simple metadata parser can be useful in markdown documents.
This commit introduces the feature, with the following syntax:
--- or ««« at tstart of the document,
(optionally) followed by a alphanumeric format identifier
followed by key value pairs separated by a colon and a space
followed by --- or ÂÂÂ
Also, adds methods for retrieving the parsed metadata, namely:
getMetadata() and getMetadataFormat
Closes#260
Code spans are now hashed after parsing which means extensions
that listen to spanGamut events no longer need to worry about
escaping "custom" magic chars inside code spans.
Closes#464
Syntax is:
```
__double underscores__
or
___triple unserscores___
```
Keep in mind that, with this option enabled, underscore no longer
parses as `<em>` or `<strong>`
Closes#450
Implement support for starting ordered lists at an arbitrary number
Closes#377
BREAKING CHANGE: Since showdown now supports starting ordered lists
at an arbitrary number, list output may differ.
Wrapping base64 strings, which are usually extremely long lines of text, is now supported.
Newlines can be added arbitrarily, as long as they appear after the comma (,) character.
Closes#429
Setting this option to true will prevent showdown from modifying the
prefix. This might result in malformed IDs (if, for instance, the " char is
used in the prefix). Has no effect if prefixHeaderId is set to false.
Closes#409
In code, the option appeared both as 'tableHeaderId' and 'tablesHeaderId',
although only the first form had effect. In documentation was referenced
as 'tablesHeaderId'.
Option is now fixed in code to reflex the documentation and table parser
accepts both forms, with and without an s.
Closes#412
It’s useful for non-Latin texts, where header might be, for example, in Russian, but user wants id to be in English. This feature allows user to set id for header manually, using curly braces:
## Привет, мир {hello-world}
Closes#383
correctly
When a user enters a URL with emphasis or strikethrough, the html output
were incorrect.
Now, URLs inside emphasis or strikethrough are parsed corerctly
Closes#347
If an invalid prefix was passed (a string with spaces, for instance),
the string would be added as is. This ould generate invalid ids.
Also, this makes `prefixHeaderId` option play nicely with `ghCompatibleHeaderId`, since they will
follow the same escaping rules when both options are enabled.
Previously, `<code>` tags were not escaped. This was counter intuitive since ´<pre><code>` tags
were being escaped. Now both pre code and code are escaped.
Closes#339
QML has a bug that changes the behavior or String.search().
This prevents blocks from being correctly unhashified.
This commit works around that bug, using RegExp.test
instead of String.search.
Credits to @qyvlik
Closes#246, Closes#338
Empty lines should not be parsed as paragraphs. This was happening
in determined circumstances.
For instance, when stripping reference style links, `\n\n` was left being,
creating an undesired empty paragraph. This commit fixes the issue.
Closes#334
The way showdown parsed cases suchs as this:
```
**foo **bar
__foo __bar
*foo *bar
_foo _bar
```
was inconsistent. This established that `__` or `**` preceeding a word
would not be parsed as em or strong.
Closes#332
This option enables users to define the generated links in @mentions.
For instance, with ghMentionsOption set to `//mysite.com/{u}/profile`
this text
`@tivie`
will result in this link
`<a href="//mysite.com/tivie/profile">@tivie</a>`
