Definition lists do not work if they are embedded in another block type

[alyjak]

Edited

It looks like the definition lists do not work if they are embedded in another block type for the offset_iter. Both of these examples are failing for me on v0.12.2 with the following panic:

panicked at /home/alyjak/.cargo/registry/src/index.crates.io-6f17d22bba15001f/pulldown-cmark-0.12.2/src/firstpass.rs:1540:38:
called `Option::unwrap()` on a `None` value`.

Offset iter is necessary in this example for it to panic:

use pulldown_cmark::{Parser, Event, Options};
use std::ops::Range;

let mut options = Options::empty();
options.insert(Options::ENABLE_DEFINITION_LIST);

let block_quote = "As defined by the foo:

> The Foo
> :   This defines foo.
>";

format!("{:?}", Parser::new_ext(&block_quote, options).into_offset_iter().collect::<Vec<(Event<'_>, Range<usize>)>>())

A standard parse panics here:

use pulldown_cmark::{Parser, Event, Options};
use std::ops::Range;

let mut options = Options::empty();
options.insert(Options::ENABLE_DEFINITION_LIST);

let triple_def = "A very nested definition
:   Def 1.1
    :   Def 1.1.1
        :   Text

A succeeding paragraph
";

println!("{:?}", Parser::new_ext(&triple_def, options).collect::<Vec<Event<'_>>>())

Thanks for the hard work. I really appreciate this library.

Originally posted by @alyjak in #915 (comment)

[notriddle]

Panicking is definitely never expected behavior.

[alyjak]

Looks like my minimum definition isn't reproducable yet, I'm working on generating a more complete reproduction.

[ytmimi]

I think this is a duplicate of #973, and I believe this is already fixed, just not released yet.

[notriddle]

This problem is already fixed

Here's what I did to figure it out.

1. I wrote these two samples into a repro.txt file:

   Definition list in block quote:
   (<https://github.com/pulldown-cmark/pulldown-cmark/issues/1014>):
   
   ```````````````````````````````` example
   As defined by the foo:
   
   > The Foo
   > :   This defines foo.
   >
   .
   <p>As defined by the foo:</p>
   <blockquote>
   <dl>
   <dt>The Foo</dt>
   <dd>This defines foo.</dd>
   </dl>
   </blockquote>
   ````````````````````````````````
   
   ```````````````````````````````` example
   A very nested definition
   :   Def 1.1
       :   Def 1.1.1
           :   Text
   
   A succeeding paragraph
   .
   <dl>
   <dt>A very nested definition</dt>
   <dd>
   <dl>
   <dt>Def 1.1</dt>
   <dd>
   <dl>
   <dt>Def 1.1.1</dt>
   <dd>Text</dd>
   </dl>
   </dd>
   </dl>
   </dd>
   </dl>
   <p>A succeeding paragraph</p>
   ````````````````````````````````
   
   

2. I wrote a repro.bash script to act as a test harness.

   cat repro.txt >> pulldown-cmark/specs/definition_lists.txt
   ! cargo test --features gen-tests
   result=$?
   git checkout pulldown-cmark/
   exit $result

   IMPORTANT: this script returns 1, a "failure", if the problem fails to reproduce.
   That is, it returns 1 if the parser doesn't crash.

3. I knew, from your panic message, that the problem appeaerd in v0.12.2.
   Luckily, this repository uses tags, so I can run these console commands:

   $ git bisect start master v0.12.2
   Bisecting: 37 revisions left to test after this (roughly 5 steps)
   [a141e065a8ab543e18ad47445e0b95a429886471] Use slice patterns for `scan_eol`
   $ git bisect run bash repro.bash
   >>> git and cargo spew a bunch of stuff
   1f8083a4b9fd2a505c1c3dc5c3d015b97f17680f is the first bad commit
   commit 1f8083a4b9fd2a505c1c3dc5c3d015b97f17680f
   Author: Álvaro Mondéjar Rubio <mondejar1994@gmail.com>
   Date:   Thu Oct 24 18:48:50 2024 +0200
   
       Safer definition lists implementation
   
    pulldown-cmark/specs/definition_lists.txt      | 34 +++++++++++++++++++++++++
    pulldown-cmark/src/tree.rs                     |  2 --
    pulldown-cmark/tests/suite/definition_lists.rs | 35 ++++++++++++++++++++++++++
    3 files changed, 69 insertions(+), 2 deletions(-)
   bisect found first bad commit

   Remember, the "bad" commit is the one that doesn't crash, for the purpose
   of this analysis.

[alyjak]

Awesome! Thank you for looking into this. I'm looking forward to the next release

[Martin1887]

Release 0.13.0 just released 🙂 .