Acording to the spec, multi paragraph (or block) list item requires subblocks
to be indented 4 spaces (or 1 tab). Although, this is mentioned in the documentation,
Showdown didn't enforce this rule in sublists because other implementations,
such as GFM also didn't. However, in some edge cases, this led to inconsistent behavior,
as shown in issue #299. This commit makes 4 space indentation in sublists
mandatory.
BREAKING CHANGE: syntax for sublists is more restrictive. Before, sublists SHOULD be
indented by 4 spaces, but indenting 2 spaces would work. Now, sublists MUST be
indented 4 spaces or they won't work.
With this input:
```md
* one
* two
* three
```
Before (ouput):
```html
<ul>
<li>one
<ul>
<li>two
<ul><li>three</li></ul>
<li>
</ul>
</li>
<ul>
```
After (output):
```html
<ul>
<li>one</li>
<li>two
<ul><li>three</li></ul>
</li>
</ul>
```
To migrate either fix source md files or activate the option `disableForced4SpacesIndentedSublists` (coming in v1.5.0):
```md
showdown.setOption('disableForced4SpacesIndentedSublists', true);
```
Nested ul and ol lists behave inconsistently in the requirement
of having 3 spaces to be considered a nested list.
This fix changes the requirement to only one space in
both cases.
Closes#299
Using the simplifiedAutoLink option does not return the expected GFM behaviour when parsing links without a http prefix.
Previously, `www.google.com` would be parsed into `<a href="www.google.com">www.google.com</a>`.
With this fix, showdown behaves like GFM, and the result is `<a href="http://www.google.com">www.google.com</a>`
Closes#284, closes#285
When an html comment was followed by a a long line of dashes, it would
freeze the parser as the lookahead in the html comment parser regex was
very slow. The regex was modified and simplified, so no lookahead is
needed anymore.
Closes#276
When using html pre/code tags to wrap github's fenced code block syntax,
showdown would parsed them instead of treating them like plain code.
Closes#229
Enable parsing markdown inside HTML blocks if those blocks have an attribute called markdown="1".
This feature is EXPERIMENTAL! As such, the behavior might change on future releases.
Closes#178
This feature enables hashing span elements that should not be touched by
showdown. For instance, `<code>` tags in markdown source should not be
parsed by showdown, so the text inside them remains unchanged.
This is made possible by a new exciting internal feature,
matchRecursiveRegExp.
Closes#196, Closes#175, Partially reverts 5f043ca
The spec states that you can be lazy and only put the `>` before the first line of a hard-wrapped paragraph.
It also states that blocquotes can contain any other md element inside.
This means headings and horizontal rules should be included in the blockquote but, right now, are treated as
independent entities
Closes#191
Fix ghCodeBlocks not being correctly parsed inside lists. Also, as a side
effect, fixes issues with consecutive lists and extra paragraphs being
added into lists.
Closes#142, Closes#183, Closes#184
When literalMidWordUnderscoresis set to true, em and strong tags that start or end a paragraph don't get parsed as such.
This fixes this issue.
Closes#174
Cannot read property 'trim' of undefined happens when the parser is fed a malformed table.
This happens in live previews (for instance, when using Angularjs).
This feature enables users to select a preset/flavor.
A flavor is just a preset of options, a shortcut so users don't have to set each option one by one.
Closes#164
Github Flavored Markdown supports tasklist by `[x]` or `[ ]` after list item marker.
This commit adds this feature to showdown through an option called "tasklists".
Related to #164
GFM support fenced codeblocks. Showdown, since very early, adopted this too.
It is now possible to disable GFM codeblocks with the option "ghCodeBlocks" set to false.
It is enabled by default
Github Flavored Markdown supports a specific table syntax. Table support was already available as an extension.
With this commit, the feature was moved to core, adding this feature to showdown through an option called "tables".
Related to #164
Github Flavored Markdown supports strikethrough (`<del>`) syntax using double tilde `~~` delimiters.
This commit adds this feature to showdown through an option called "strikethrough".
Related to #164
Github Flavored Markdown does not parse underscores in the middle of a word as emphasis/bold.
This commit adds this feature to showdown through an option called "literalMidWordUnderscores".
Related to #164
Github Flavored Markdown detects urls and mails embeded in the text without any extra markup or delimiter.
This commit adds this feature to showdown through an option called "simplifiedAutoLink".
Related to #164
This feature allows users to define the image dimensions using markdown syntax:
```
```
To enable this feature, use the option `parseImgDimensions`.
Closes#143
Consecutive lists we're previously being condensed into one unique list, with odd paragraph output.
This fix correctly splits lists, but does not change the weird paragraph output
closes#142
According to spec, the title attribute in link definitions can be wrapped in single quotes. Previously, showdown didn't support this.
Now the title attribute can be wrapped in single quotes.
Old extensions that register themselves in `showdown.extensions` can be loaded and validated using the new extension loading mechanism.
However, a warn is issued, alerting users and developers that the extension should be updated to use the new mechanism
BREAKING CHANGE: Deprecates `showdown.extensions` property. To migrate, you should use the new method `showdown.extension(<ext name>, <extension>)` to register the extension.
Passing an option to a specific converter affects other instances of the converter since options are merged into showdown's global options.
This commit fixes that.
Closes#153
Some HTML5 block elements were not being properly ignored. This caused problems in rendering markdown as showdown would add additional <br /> to some block elements.
This commit should fix this issue.
Closes#90, closes#140, closes#147
If two headers have similar texts, the generated id could be equal. In order to prevent id clash:
- A unique suffix is added if a header id already exists
- Option to add a prefix to header id
- Update of correspondent tests
- (credits to nicovalencia)
Closes#81, closes#82
When the text is pulled from indented HTML elements, ex:
```
<body>
<div>
## Content to be converted
</div>
</body>
```
it no longer becomes wrapped in code/pre tags. A new option is also available (allowBlockIndents) that when set to false, reverts to the previous behavior
