Markdown reader: allow more attributes in special spans #7307

tarleb · 2021-05-21T14:28:43Z

Spans with "smallcaps" as the first class are converted to SmallCaps
elements. While previously no other classes or attributes were allowed,
additional classes, attributes, and an identifier are not permitted and
kept in a SmallCaps wrapping Span element.

The same change is applied to underline spans, where the first class
must be either "ul" or "underline".

Closes: #4102

tarleb · 2021-05-21T14:29:10Z

I'm having second thoughts about this change, so I'm putting it here as a PR.

jgm · 2021-05-21T16:05:40Z

I think this makes sense, but why require that the special class be the first one?
(I guess it makes pattern-matching faster, but probably only slightly.)

tarleb · 2021-05-21T16:29:20Z

My thought here was that it might give the best balance between new functionality and unintended conversions. I wouldn't be too surprised to see ul be used as a class in other contexts, and the proposed solution still allows to do so.

tarleb · 2021-05-27T13:46:55Z

I've added a second commit (intended to be squashed) that removes the limitation of expecting the special class as the first, allowing it to occur in any position.

jgm · 2021-05-27T15:41:38Z

I really don't know what's best here. Your concern about people wanting to use a span with class ul and not wanting that to be parsed as an Underline element gives me pause. On the other hand, the convention that the special class must occur first seems a bit too "magic" and might surprise people. And even with this, we might get some people who want just a span with class ul.

tarleb · 2021-05-27T17:51:00Z

Maybe we should just keep the current behavior and document the behavior as intentional? Or, but that's a bit of a stretch, we could hide the new behavior behind the attributes extension.

I note that neither the .underline nor .ul class are documented yet in the manual; the changelog only states "Support new Underline element in readers and writers". Removing support for .ul should not support too many problems.

jgm · 2021-05-27T18:42:44Z

I don't think I'd want to remove the current behavior for ul, if that's what you meant. That would cause existing documents to break. It should be documented, though.

jgm · 2021-08-16T04:44:16Z

Note that in a recent commit 649133f .ul and .underline were documented.

Maybe I'll take out the documentation for .ul, just in case we decide to remove it.

See #7307. Motivation: there is talk of removing this.

hftf · 2021-09-18T19:10:28Z

Note that in a recent commit 649133f .ul and .underline were documented.

Maybe I'll take out the documentation for .ul, just in case we decide to remove it.

I believe the Pandoc-flavored Markdown Writer still outputs [a]{.ul} from the input [Underline [Str "a"]], doesn't it?

pandoc/src/Text/Pandoc/Writers/Markdown/Inline.hs

Lines 304 to 317 in 407de98

    
           inlineToMarkdown opts (Underline lst) = do 
        
             variant <- asks envVariant 
        
             contents <- inlineListToMarkdown opts lst 
        
             case variant of 
        
               PlainText -> return contents 
        
               _     | isEnabled Ext_bracketed_spans opts -> 
        
                       return $ "[" <> contents <> "]" <> "{.ul}" 
        
                     | isEnabled Ext_native_spans opts -> 
        
                       return $ tagWithAttrs "span" ("", ["underline"], []) 
        
                         <> contents 
        
                         <> literal "</span>" 
        
                     | isEnabled Ext_raw_html opts -> 
        
                       return $ "<u>" <> contents <> "</u>" 
        
                     | otherwise -> inlineToMarkdown opts (Emph lst)

(Speaking of the inconsistent output of Underline between Pandoc-flavored Markdown and the other Markdown flavors, this could be a good opportunity to decide/discuss whether "raw HTML" like  belongs in Pandoc-flavored Markdown output, even with bracketed_spans or native_spans enabled. See also #5825 (comment); I'll try to get that new issue filed soon.)

jgm · 2021-09-18T22:19:55Z

So, the suggestion is to parse ... as a native Underline element, and to generate it from an Underline element?
That does have a certain attraction. On the other hand, it isn't really consistent with how other raw HTML bits are treated. (.. isn't parsed as Emphasis, for example).

tarleb · 2022-07-26T14:29:11Z

The second commit in this PR removes the limitation of the special class (like smallcaps) having to be the first class, making it possible to use markup like [Populus Romanus]{.foo .smallcaps lang="latin"}.

jgm · 2022-07-26T16:36:53Z

src/Text/Pandoc/Readers/Markdown.hs

+    return $ case smallCapsAttr attr of
+      Just ("", [], []) -> B.smallcaps <$> lab
+      Just scAttr       -> B.spanWith scAttr . B.smallcaps <$> lab
+      Nothing           -> case underlineAttr attr of
+        Just ("", [], []) -> B.underline <$> lab
+        Just ulAttr       -> B.spanWith ulAttr . B.underline <$> lab
+        Nothing           -> B.spanWith attr <$> lab
+


Maybe, to be consistent with the recent change to the HTML writer, we should allow [test]{.underline .smallcaps} to turn into Underline [SmallCaps [Str "test"]]? Not sure, what do you think?

This could be done fairly elegantly with a fold over the classes, similarly to what I did in the HTML writer.

Nice! Done.

Spans with "smallcaps" as the first class are converted to *SmallCaps* elements. While previously no other classes or attributes were allowed, additional classes, attributes, and an identifier are not permitted and kept in a *SmallCaps* wrapping *Span* element. The same change is applied to underline spans, where the first class must be either "ul" or "underline". Closes: jgm#4102

tarleb force-pushed the allow-additional-attr-in-special-spans branch from 36872a4 to e0e7eb9 Compare May 21, 2021 15:47

tarleb force-pushed the allow-additional-attr-in-special-spans branch from e0e7eb9 to 455e36c Compare May 27, 2021 13:32

tarleb force-pushed the allow-additional-attr-in-special-spans branch from 455e36c to 78c9bf1 Compare May 27, 2021 15:40

jgm added a commit that referenced this pull request Aug 16, 2021

Remove documentation for .ul shortcut for now.

72447a5

See #7307. Motivation: there is talk of removing this.

tarleb mentioned this pull request Jul 26, 2022

Manual: add mark / highlight documentation #8194

Closed

tarleb force-pushed the allow-additional-attr-in-special-spans branch 3 times, most recently from 9f1621c to 9619f6b Compare July 26, 2022 14:26

tarleb force-pushed the allow-additional-attr-in-special-spans branch from 9619f6b to 32f66cb Compare July 26, 2022 15:21

jgm reviewed Jul 26, 2022

View reviewed changes

tarleb added 2 commits August 1, 2022 09:48

Markdown reader: allow special span classes in any position

40a8418

tarleb force-pushed the allow-additional-attr-in-special-spans branch from 32f66cb to 40a8418 Compare August 1, 2022 09:39

jgm merged commit 7a4afce into jgm:master Aug 1, 2022

tarleb deleted the allow-additional-attr-in-special-spans branch August 1, 2022 18:29

Uh oh!

Markdown reader: allow more attributes in special spans #7307

Markdown reader: allow more attributes in special spans #7307

Uh oh!

Conversation

tarleb commented May 21, 2021

Uh oh!

tarleb commented May 21, 2021

Uh oh!

jgm commented May 21, 2021

Uh oh!

tarleb commented May 21, 2021

Uh oh!

tarleb commented May 27, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jgm commented May 27, 2021

Uh oh!

tarleb commented May 27, 2021

Uh oh!

jgm commented May 27, 2021

Uh oh!

jgm commented Aug 16, 2021

Uh oh!

hftf commented Sep 18, 2021

Uh oh!

jgm commented Sep 18, 2021

Uh oh!

tarleb commented Jul 26, 2022

Uh oh!

jgm Jul 26, 2022

Choose a reason for hiding this comment

Uh oh!

jgm Jul 26, 2022

Choose a reason for hiding this comment

Uh oh!

tarleb Aug 1, 2022

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

tarleb commented May 27, 2021 •

edited

Loading