Marked: V0.3.6,<h1>,<h2>,<h3>,<h4>,<h5>,<h6>sometimes id are same

[osoc]

test 01 [ok]

//md:

# h1

## h2

//out:
<h1 id="h1">h1</h1>
<h2 id="h2">h2</h1>

---

test 02 [the same id]

//md:
# h1
# h1
//out:
<h1 id="h1">h1</h1>
<h1 id="h1">h1</h1>

test 03 [the same id]

//md:
# 这行只有汉字
# 这行也是
//out:
<h1 id="-">这行只有汉字</h1>
<h1 id="-">这行也是</h1>

---

test 04 [the same id]

//md:
## 这行只有汉字
## 这行也是
//out:
<h2 id="-">这行只有汉字</h1>
<h2 id="-">这行也是</h1>

[destagia]

I have a same problem...

[destagia]

I solved this by overriding marked.Renderer#heading function.

const renderer = new marked.Renderer();
renderer.heading = function(text, level) {
    return `<h${level} id="${text}">${text}</h${level}>`;
};

marked(mdString, { renderer: renderer });

It seems to be natural to assume the above-mentioned behavior is a specification.

[osoc]

I do it：
The full code is here

var ID_TABLE = "0123456789ABCDEF";
var AUTOID = 0x0001;
...
Renderer.prototype.heading = function(text, level, raw) {
  var id, tmp;
  tmp = AUTOID++;
  id = ID_TABLE.charAt((tmp >> 12) & 0x0F);
  id += ID_TABLE.charAt((tmp >> 8) & 0x0F);
  id += ID_TABLE.charAt((tmp >> 4) & 0x0F);
  id += ID_TABLE.charAt(tmp & 0x0F);
  return '<h'
    + level
    + ' id="'
    + id
//    + this.options.headerPrefix
//    + raw.toLowerCase().replace(/[^\w]+/g, '-')
    + '">'
    + text
    + '</h'
    + level
    + '>\n';
};

[joshbruce]

981

[Feder1co5oave]

There's no perfect way to generate IDs. The "naive" way to fix this would be to add a numeric counter to duplicate IDs, but that would require keeping track of all IDs.

[joshbruce]

I still think the easiest way is to not have Marked do it. This is what I used to do back in the day for scrolling table of contents a la Bootstrap - http://getbootstrap.com/2.3.2/getting-started.html

Step one: Convert Markdown to HTML.
Step two (DOM is ready/loaded): Cycle through the headers to add ID attributes and generate the TOC in one shot.

No JS solution = no TOC; therefore, no reason to have the IDs in the first place.

Again, could just be asking Marked to do too much. If a spec has it, we should too but, if a spec has it, it should define the testable rules for implementing the solution, yeah?

Note: I did not use Marked for processing back in the day.

[yzfdjzwl]

@tagia0212
if your header includes (), it will failure. eg:

### do something()

[joshbruce]

1048

[joshbruce]

Closing as most likely going to be deprecated at a later date.

[Grawl]

The problem still exists.

[styfle]

Yes, this problem still exists.

Let's consolidate this issue with a similar one here: #1280

[joshbruce]

Sidenote: GitHub adding detail popovers on hover...fancy. 😎