* fix: Debounce the scroll event, to make Chrome happy
* Address review comments
* fix: Prevent overscrolling from scrolling the main content
Co-authored-by: Robert Sparks <rjsparks@nostrum.com>
* fix: Avoid numeric IDs on NomCom pages
* fix: Correctly group last names with lowercase/i18n parts
* Remove some unneeded CSS
* fix: Remove extra div closing tag
* fix: import moment.js library with moment-timezone support in wrapper
Packaging changes broke the mechanism used to make moment.tz available
to javascript code. This eliminates the separate moment.js and
moment-timezone-with-data-10-year-range.js wrappers and instead
imports the timezone-augmented version directly. This provides both
the timezone support and the regular moment features.
* refactor: put moment on window instead of global for consistency
* fix: use DST-friendly tz calculation for TimeSlot end times
Around DST changes, localize(time + duration).astimezone() is not
the same as localize(time).astimezone() + duration. The latter allows
tz-aware times to do their magic and give correct clock times.
* fix: refactor/correct calculation of week-view event start/end times
There were bugs in the timestamp calculations around DST switchover.
This corrects those and tries to clarify and improve documentation in
the relevant code.
* Unicode messages are triggered by both db content and tests
* Make ids unique
* Avoid "No value found" message on page
* Strip HTML from history entries, it's often broken
* Check HTML sources for occurrences of "** No value found for" and fix them
* Fix another occurrence of "** No value found for"
* Fix more occurrences of "** No value found for"
* Fix document revision stripping
* Force breaks of long (garbage) words
* Check URL validity before urlizing them
* Handle some additional corner cases
* Linkify action items
* Don't create profile/email links for System
* Handle headings with HTML elements in them better
* Fix comment
* Fix another occurrence of "** No value found for"
* Better I-D URLization that handles more edge cases. Also, test for them.
* Remove print
* Handle charters better
* Cache for one day
* Update vnu.jar
* Fix py2 -> py3 issue
* Run pyupgrade
* test: Add default-jdk to images
* test: Add option to also validate HTML with vnu.jar
Since it's already installed in bin. Don't do this by default, since it
increases the time needed for tests by ~50%.
* fix: Stop the urlizer from urlizing in linkified mailto: text
* More HTML fixes
* More HTML validation fixes
* And more HTML fixes
* Fix floating badge
* Ignore unicode errors
* Only URLize docs that are existing
* Final fixes
* Don't URLize everything during test-crawl
* Feed HTML into vnu using python rather than Java to speed things up
* Allow test-crawl to start vnu on a different port
* Increase retry count to vnu. Restore batch size to 30.
* More HTML validation fixes
* Use urllib3 to make requests to vnu, since overriding requests_mock is tricky
* Undo commit of unmodified file
* Also urlize ftp links
* Fix matching of file name
* More HTML fixes
* Add `is_valid_url` filter
* weekday -> data-weekday
* urlencode URLs
* Add and use vnu_fmt_message. Bump vnu max buffer.
* Simplify doc_exists
* Don't add tab link to mail archive if the URL is invalid
* Run urlize_ietf_docs before linkify
Reduces the possibility of generating incorrect HTML
* Undo superfluous change
* Runner fixes
* Consolidate vnu message filtering into vnu_filter_message
* Correctly handle multiple persons with same name
* Minimze diff
* Fix HTML nits
* Print source snippet in vnu_fmt_message
* Only escape if there is something to escape
* Fix snippet
* Skip crufty old IPR declarations
* Only include modal when needed. Add handles.
* Fix wordwrap+linkification
* Update ietf/doc/templatetags/ietf_filters.py
* Update ietf/doc/templatetags/tests_ietf_filters.py
* Don't right-align second column
* feat: Make the "report a bug" links more visible
Also rework the page footer somewhat, while we're here.
* Revert empty change
Co-authored-by: Nicolas Giard <github@ngpixel.com>
* fix: Correctly linkify all current TLDs
* Pass a list to the build_*_re functions, not a string
* Need to sort TLDs by length to force longer ones to match first
* chore: silence incorrect mypy complaint.
Co-authored-by: Robert Sparks <rjsparks@nostrum.com>
Co-authored-by: Nicolas Giard <github@ngpixel.com>
* test: add tests of the person_link tag
* fix: do not prematurely dereference status change RelatedDocuments
The urlize_related_source_list template tag expects the RelatedDocument
instances, not the source Document.
* test: add test cases for status changes on document_main view
They seem to be one of the only bs5 elements that do not by default wrap, which
causes display issues on narrow screens when there are lots of page elements,
e.g., on http://127.0.0.1:8000/doc/draft-ietf-ipwave-vehicular-networking/
Adding `flex-wrap` (per https://github.com/twbs/bootstrap/issues/23504) makes
this better, but unfortunately doesn't fix it completely, because the layout
still jumps when the window is narrowed by the used. It does fix the issue on
first load and after a reload though.
Also rewrites scroll_to_element() using javascript. This seems to be less flaky
than the ActionChains implementation. Slight change in behavior - now scrolls
the requested element to the middle of the window instead of barely into view.
The old code would first substitute the "rfx<nnnn>" part, and then again
on the full draft name. This now only ever applies one substitute,
and does longer ones first.
