Are you planning to support CommonMark?

Like Stackoverflow and Github (the stackoverflow are prettier)

There is a space in markdown process result, I fix it by change Utils.java, patch here :)

<a href="http://test.com">test</a>

to

<p><a href="http: //test.com">test</a></p>

diff --git a/src/main/java/com/github/rjeschke/txtmark/Utils.java b/src/main/java/com/github/rjeschke/txtmark/Utils.java
index 2e19c08..5339893 100644
--- a/src/main/java/com/github/rjeschke/txtmark/Utils.java
+++ b/src/main/java/com/github/rjeschke/txtmark/Utils.java
@@ -530,7 +530,7 @@
            pos = readRawUntil(out, in, pos, '/', '>');
            if(in.charAt(pos) == '/')
            {
-               out.append(" /");
+               out.append("/"); // a 'space' prefix will cause 'http://test.com' become 'http: //test.com'
                pos = readRawUntil(out, in, pos + 1, '>');
                if(pos == -1)
                    return -1;

Is there any way to convert html to markdown. Basically reverse of Processor.process(str) which converts markdown to html.

Hi,
First, congrats for your simple and efficient tool !
Is it possible to generate all links with the target="_blank". Is there any configuration trick to achieve this ?

Thanks !
Mathieu

It currently allows users to "escape" the place they are supposed to be at with for example the following HTML template:

<p>
 This contains the text: 
 <div id="user-content">${mdContent}</div>
</p>

One can then create Markdown content such as:

In the box
</div></p>
<p>
 Outside the box, looks like part of the site now 
 <a href="malicious-link">go there to reset your password please</a>
</p>

I need a way to escape every unescaped < character from the input.

Compare the result of the following markdown (use copy+paste – there’s a difference between 2. and 3.):

foo
`` `class Object` `` is a *meta literal*.
bar

foo
 `` `class Object` `` is a *meta literal*.
bar

foo
 `` `class Object` `` is a *meta literal*.
bar

(that’s a No-Break Space at the beginning of the second line)
4.

foo
other `` `class Object` `` is a *meta literal*.
bar

Here’s how the markdown is rendered by Github:

foo
class Object is a meta literal.
bar

foo
class Object is a meta literal.
bar

foo
class Object is a meta literal.
bar

foo
other class Object is a meta literal.
bar

(The line breaks are incorrect because GFM always inserts line breaks for newlines.)

And here’s how the markdown is rendered by txtmark (as used by ceylon doc):

foo
bar

foo
bar

foo class Object is a meta literal. bar

foo other class Object is a meta literal. bar

That is, if a line starts with nothing other than spaces (meaning the identity of U+0020, not the Unicode category “Separator, Space [Zs]”, which also contains the No-Break Space U+00A0 as seen in 3.) and then a multi-backtick code block containing backticks `code` with backticks`, then strange behavior occurs.

^{(originally reported as ceylon/ceylon-compiler#1553)}

Hi,
Any plans for having a way to get the markdown from a given HTML code snippet?
Thanks

Can you explain more please how to use [$PROFILE$]: extended? I have added this at the beggining of text processed by txtmark but it doesnt work. I probably do not undestand the concept ...

Thank you
David

The output of this code:

Configuration CONFIGURATION = Configuration.builder().enablePanicMode().build();
String result = Processor.process("`<p>My code</p>`", CONFIGURATION);

is:

<p><code>&amp;lt;p&gt;My code&amp;lt;/p&gt;</code></p>

One method is to convert it to HTML and then parse it to extract plain text . I was wondering if there is direct way to do it

First, thanks for your hard work. I was surprised you haven't pushed it to Maven Central however... Putting your artifact in Maven Central will extend the reach of your project and make it easier for people to download your work.

It's not incredibly difficult to do either, and Sonatype will help you through the process if you get stuck.
http://maven.apache.org/guides/mini/guide-central-repository-upload.html#Other_Projects
https://docs.sonatype.org/display/Repository/Sonatype+OSS+Maven+Repository+Usage+Guide

Not sure if this is intentional but

• item
• item

1. item
2. item

returns a single unnumbered list containing all items

It needs to check that start + 1 < in.length().

This code block does not work properly:

Hello !

Is there a way where i can strip all the markdowns to get just pure text (not html) ?
I'd like this feature to do a simple substringed-preview, the one with the "read more" link.

Thank you,
Albert

com.github.rjeschke.txtmark.cmd.HlUtils.highlight is designed to highlight code blocks. However, passing an unchecked argument to this API can lead to the execution of arbitrary commands. For instance, first, we create an instance of CodeBlockEmitter and specify the parameter program of CodeBlockEmitter to ”calc.exe”:

Configuration.Builder builder = Configuration.builder();
Class clazz = Class.forName("com.github.rjeschke.txtmark.cmd.CodeBlockEmitter");
Constructor constructor = clazz.getDeclaredConstructors()[0];
constructor.setAccessible(true);
Object cb = constructor.newInstance("UTF-8", "calc.exe");

Second, we set CodeBlockEmitter to the instance that we just created.

builder.setCodeBlockEmitter((BlockEmitter) cb);
builder.forceExtentedProfile();
Configuration config = builder.build();
System.out.println(Processor.process("```java\n123\n```\n", config));

Finally, malicious programs “calc.exe” would be executed.

Configuration configuration = Configuration.builder()
    .enableSafeMode()
    .build();
String markdown = "Hello **world**  \nThis should come after newline";
String result = Processor.process(markdown, configuration);
System.out.println(result);

The output is this :

<p>Hello <strong>world</strong>  />
This should come after newline</p>

The expected result is this :

<p>Hello <strong>world</strong>  <br/>
This should come after newline</p>

I'm using this great library (thanks!) in conjunction with another library... http://code.google.com/p/pagedown/

One difference in the implementation is that pagedown will automatically make a regular url into a link. Heck, that is even similar to the way that the Github flavored markdown works (as you can see, the link above is made clickable).

Would it be possible to get an extension to txtmark to allow that as well? It just seems natural to allow people to write urls like that.

when a line contains only a '<' character.

I'm not able to submit a pull request, but here's a patch:

$ git diff 108cee6b209a362843729c032c2d982625d12d98
diff --git a/src/main/java/com/github/rjeschke/txtmark/Line.java b/src/main/java/com/github/rjeschke/txtmark/Line.java
index 4ef1c43..2afcbc0 100644
--- a/src/main/java/com/github/rjeschke/txtmark/Line.java
+++ b/src/main/java/com/github/rjeschke/txtmark/Line.java
@@ -517,7 +517,7 @@ class Line
         final LinkedList<String> tags = new LinkedList<String>();
         final StringBuilder temp = new StringBuilder();
         int pos = this.leading;
-        if (this.value.charAt(this.leading + 1) == '!')
+        if (this.leading + 1 < this.value.length() && this.value.charAt(this.leading + 1) == '!')
         {
             if (this.readXMLComment(this, this.leading) > 0)
             {

private int readXMLComment(final Line firstLine, final int start)
    {
        Line line = firstLine;
        if (start + 3 < line.value.length())
        {
            if (line.value.charAt(2) == '-' && line.value.charAt(3) == '-')
            {
...

I think that should be

            if (line.value.charAt(start + 2) == '-' && line.value.charAt(start + 3) == '-')

Hi!

I tested your markdown processor but I couldn't find a way to escape html.
I know that the markdown documentation specify that HTML is possible, but it could be great to have a "security" parameter in case the markdown is used by other users (such as comments).

Moreover, I think you would be the first markdown processor in java that include a safe mode parameter :p

• Need to find some time to review the pull request. (!)
• Switch from LGPL to Apache v2. (!)

Configuration.Builder#setCodeBlockEmitter(BlockEmitter emitter)

Example code:

Configuration.Builder c = Configuration.builder();
c.setCodeBlockEmitter(new CodeBlockEmitter());
c.setAllowSpacesInFencedCodeBlockDelimiters(true);
String res = Processor.process(markdown, c.build());

The configuration does seem to be adding, nor executing the CodeBlockEmitter() class. The code for it is similar to the example code.

So I have this statement

~ ~ ~ ~ ~
if (a > 3) {
  moveShip(5 * gravity, DOWN);
}

What I get is for Processor.process(..)
is

a > 3) {
veShip(5 * gravity, DOWN);

Which is not the desired result.

Could txtmark support table parse?

Link references are being processed in the wrong place. The following Markdown:

## Example

<pre>
[1]: blah
</pre>

Is converted into this:

<h2>Example</h2>
<pre>
</pre>

When it should be:

<h2>Example</h2>
<pre>
[1]: blah
</pre>

This may explain why your Markdown parser is faster than others!

I have the following html:

• one <br />
• two <br />
• three <br />

... but this is not interpreted as a list . Is it because of br tags?

David

Would it be possible to enable/disable and set the value of the rel= attribute on generated hrefs?

Triple backticks are rendered as <pre> when I set forceExtended in the config. However there are a couple of issues with the text.

1. Everything from the block appears on the same line. (This could be intended, but I'm not certain)
2. The first few characters on every line disappears

Example

```
go test -cover // shows you what percentage of the code is covered by tests
go tool cover -html=coverage.out //opens a web browser which shows you which lines aren't covered
```

random text

```
func main() {
    return nil
}
```

renders as

forceExtendedProfile = True

<p>From <a href="http://blog.golang.org/cover">a post</a> on the Go Blog</p>
    <pre><code>est -cover // shows you what percentage of the code is covered by tests
    ool cover -html=coverage.out //opens a web browser which shows you which lines aren't covered
    </code></pre>
    <p>random text</p>
    <pre><code> main() {
    return nil
    </code></pre>

forceExtendedProfile = False

<p>From <a href="http://blog.golang.org/cover">a post</a> on the Go Blog</p>
    <p><code>`
    go test -cover // shows you what percentage of the code is covered by tests
    go tool cover -html=coverage.out //opens a web browser which shows you which lines aren't covered
    </code>`</p>
    <p>random text</p>
    <p><code>`
    func main() {
    return nil
    }
    </code>`</p>

Is this project still maintained?

Added a custom block emitter to process blocks like

``` dot

digraph M1{ 
	node[shape=box width=1.1]
	dot[label="Graphviz\nDOT"]
	zestCode[label="Zest\ngraph"]
	zestVis[label="SWT\napp"]
	image[label="Image\nfile"]
	
	dot->image//[label=" Graphviz"]
	dot->zestCode[constraint=false color=black style=dashed label="            " dir=both]
	zestCode->zestVis//[label=" Zest"]	
}

~~~

Problem is that when the trailing delimiter of the code block is three back-ticks: ```, the list of lines delivered to the custom BlockEmitt#emitBlock includes every line from the first line of fenced code to the end of the document.

If the trailing delimiter is changed to ~~~, as show above, TxtMark behaves correctly.

If the custom block emitter is not used, the problem remains when using the trailing back-tick delimiter.

Suggests that something in this Dot content is messing with TxtMark. Just don't see what it is.

Switch to Apache v2 license.

Hi,

How does [$PROFILE$]: extended work. I put it on top of a string and ran it through markdown.geHtmlContent(myString) and it prints [$PROFILE$]: extended without executing it.

Is there a plan to support Girhub style triple back ticks for code blocks? It seems that it does not work at the moment.

I'd be happy to add it if you could give me some guidance on how to start.
Thanks

The maven repository dates back two years ago.

Are you not publishing there anymore or are you not publishing snapshots? Both are highly appreciated

txtmark.Processor.process(post.content)

where is the reference to txtmark?