Cloze detection broken for multi-line cloze inside html tag

[leezu]

Reproduction Steps

1. Create a cloze containing the following html on Anki Desktop:

<p>Cloze in html tag with {{c1::multi-line
string}}</p>

Note the difference to

<p>Cloze in html tag with {{c1::string}}</p>

or

Cloze with {{c1::multi-line
string}}

2. Sync to AnkiDroid
3. Review notes

Expected Result

All 3 notes should show up as Cloze with [...].

Actual Result

The <p>Cloze in html tag with {{c1::multi-line string}}</p> note does not show up correctly.
Instead Edit this note and add some cloze deletions. (help) is displayed.

After clicking "Edit note" on AnkiDroid without making any changes to the note, the note is displayed correctly.
Checking on Anki Desktop, it turns out that clicking Edit note on AnkiDroid changed the html of the note to:

<p>Cloze in html tag with {{c1::multi-line<br>string}}</p>

Debug info

AnkiDroid Version = 2.9beta7

Android Version = 10

ACRA UUID = b8969252-15d5-4075-92a1-81871187ab4b

Research

Enter an [ x ] character to confirm the points below:

[X] I have read the support page and am reporting a bug or enhancement request specific to AnkiDroid

[X] I have checked the manual and the FAQ and could not find a solution to my issue

[X] I have searched for similar existing issues here and on the user forum

[mikehardy]

interesting! great report
The perturbation of the note (saving it back as something different without editing) is tracked elsewhere #3304 which can be summed as "if you edit but make no changes, no changes should be made" (a reasonable stance)

This issue should be focused on "the close identification regexp seems to be different on AnkiDroid than Anki Desktop"

The python code they're using in Anki Desktop should be examined and whatever part makes it work multiline should be brought in to here

This should be a great first issue since Anki Desktop (code available here https://github.com/dae/anki/) is so easy to inspect

[mikehardy]

thanks by the way for confirming this in the beta so we know it is a current issue - that would have been one of my first requests :-)

[adamwolf]

Leonard, if you are interested in digging into this, it's a good first issue. If you aren't, let me know--I was just in this code and I can probably hammer this out in a few minutes since you have a great test case.

…

On Sat, Aug 31, 2019, 9:24 AM Mike Hardy ***@***.***> wrote: thanks by the way for confirming this in the beta so we know it is a current issue - that would have been one of my first requests :-) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#5421?email_source=notifications&email_token=AAAIWYKSKGC5UCFCHU6CLQLQHJ5JFA5CNFSM4ISTHQGKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5TNW2I#issuecomment-526834537>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAAIWYKKJ5E7KA2KPPU26B3QHJ5JFANCNFSM4ISTHQGA> .

[leezu]

@mikehardy Both the AnkiDroid and Anki regexes seem the same and explicitly enable matching new-lines via the (?s) inline flag.

https://github.com/dae/anki/blob/8a2a4d827f1afcab746585ede2ac08833d679757/anki/template/template.py#L7

Anki-Android/AnkiDroid/src/main/java/com/ichi2/libanki/template/Template.java

Line 48 in 95087f5

public static final String clozeReg = "(?si)\\{\\{(c)%s::(.*?)(::(.*?))?\\}\\}";

The buggy behaviour only occurs when wrapping the cloze in <p></p> tags.

@adamwolf I don't have an Android dev setup ready to verify any changes, so it would be appreciated if you could go ahead ;) Thanks for taking care of it!

[adamwolf]

I'm out this weekend for a US holiday but I will be back at it early next week.

…

On Sat, Aug 31, 2019, 1:41 PM Leonard Lausen ***@***.***> wrote: @mikehardy <https://github.com/mikehardy> Both the AnkiDroid and Anki regexes seem the same and explicitly enable matching new-lines via the (?s) inline flag. https://github.com/dae/anki/blob/8a2a4d827f1afcab746585ede2ac08833d679757/anki/template/template.py#L7 https://github.com/ankidroid/Anki-Android/blob/95087f50410ff58fe5a8e914ca6462b53b74fe53/AnkiDroid/src/main/java/com/ichi2/libanki/template/Template.java#L48 The buggy behaviour only occurs when wrapping the cloze in <p></p> tags. @adamwolf <https://github.com/adamwolf> I don't have an Android dev setup ready to verify any changes, so it would be appreciated if you could go ahead ;) Thanks for taking care of it! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5421?email_source=notifications&email_token=AAAIWYKFCT72YK2XMRZZPVLQHK3OJA5CNFSM4ISTHQGKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5TS5TI#issuecomment-526855885>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAAIWYM4LIGN3NVYRI4BQL3QHK3OJANCNFSM4ISTHQGA> .

[adamwolf]

(I'm a few minutes into this, adding some tests, and... we have at least a few bugs here, folks!)

Thanks @leezu. Good catch!

[mikehardy]

Nothing like good user reports (yes, thank you @leezu ) + automated testing + a developer ready to go (thank you @adamwolf ) :-)

[adamwolf]

Can someone confirm this is fixed in the latest betas?

[leezu]

It's fixed, thank you!