IndexError and TypeError while markdown to html conversion

[Arjunsankarlal]

When I try to convert a small markdown content into HTML with the below code,

html = MD.reset().convert(para)

it works without any problem. But when I make swarm multiple concurrent requests with that content I get the following three types of error which occurs randomly. Most of the time a single type of error would occur.

File "/Users/arjun/converter/util.py", line 101, in markdown_to_text
html = MD.reset().convert(para)
File "/Users/arjun/environments/mdtest/lib/python3.6/site-packages/markdown/core.py", line 295, in convert
output = pp.run(output)
File "/Users/arjun/environments/mdtest/lib/python3.6/site-packages/markdown/postprocessors.py", line 75, in run
html = self.md.htmlStash.rawHtmlBlocks[i]
IndexError: list index out of range```

File "/Users/arjun/converter/util.py", line 101, in markdown_to_text
html = MD.reset().convert(para)
File "/Users/arjun/environments/mdtest/lib/python3.6/site-packages/markdown/core.py", line 272, in convert
newRoot = treeprocessor.run(root)
File "/Users/arjun/environments/mdtest/lib/python3.6/site-packages/markdown/treeprocessors.py", line 378, in run
self.ancestors.pop()
IndexError: pop from empty list

File "/Users/arjun/converter/util.py", line 101, in markdown_to_text
html = MD.reset().convert(para)
File "/Users/arjun/environments/mdtest/lib/python3.6/site-packages/markdown/core.py", line 272, in convert
newRoot = treeprocessor.run(root)
File "/Users/arjun/environments/mdtest/lib/python3.6/site-packages/markdown/treeprocessors.py", line 372, in run
self.__handleInline(text), child
File "/Users/arjun/environments/mdtest/lib/python3.6/site-packages/markdown/treeprocessors.py", line 215, in __processPlaceholders
for child in [node] + list(node):
TypeError: 'NoneType' object is not iterable

And the dummy content I passed as input is,

**To edit a record:**

  1. Click the **Red** Button.

  2. In the _main_ page, click the **Home** tab in the lower-left corner.

  3. In the _Home_ page, select an record to view its details.

  4. In the _details_ page, click the **Save& Data** icon ( ![](https://www.abc.com/data?id=1234567&mode=inline&blockId=6wqjmd23e123456789009876543){: style="width: 13px; vertical-align: sub; max-width: 100%" height width } ) from the floating widgets on the left of the screen.

  5. Under _Save & Data Information_ section, click **New Data**.

It was happening randomly for different data. So reproduce the issue with a small content I took this. Generally when it happened the number of requests were low. But to reproduce the issue for this content I had to increase the request count as 200req/sec, and I could reproduce it after 20000, successfully completed requests

My suggestions are (based on the issue order)

1. Iterating based on the length of self.md.htmlStash.rawHtmlBlocks[i]
2. Checking the length before popping
3. Adding null checks

[waylan]

Which version of Markdown are you using?

Which extension, if any are you using?

Are you using a single class instance across threads? If so, that would explain the issue as an instance is not thread safe.

My suggestions are (based on the issue order)

1. Iterating based on the length of self.md.htmlStash.rawHtmlBlocks[i]

We are. That is what i is. I suppose we could actually get the len each time rather than cacheing it. But the fact this error is happening suggests to me one of two things: (1) an extension is misbehaving or (2) the class instance is being used across threads.

2. Checking the length before popping
3. Adding null checks

If I'm not mistaken, issues 2 and 3 should never happen. In other words, my expectation is that the situation should be impossible to get to. Something weird is going on here.

Finally, is there a reason why you can't just call markdown.markdown instead? I ask because, while using a single instance for multiple documents is supported, it is rarely used and not even the favored way in the documentation? I'm always curious why users chose to do this when a simple call to markdown.markdown works so well.

[Arjunsankarlal]

Thanks for the quick response @waylan. As you said it was because of the thread safety problem, using markdown.markdown has solved all the issues. Thanks.