mapnik/scripts/markdown-hyperlinks.pl
Mickey Rose 70a511165b scripts/markdown-hyperlinks: prevent unwanted replacements
- don't replace @var in code spans with link to github user
- don't replace 1234567... in links with link to commit
- replace hexadecimal numbers looking like commit hashes only
  after replacing raw URIs - this together with the previous point
  prevents inadvertent replacement of long numbers in links

[skip ci]
2018-07-09 14:32:00 +02:00

56 lines
2.1 KiB
Perl
Executable file

#! /usr/bin/env perl
#
# Re-generate hyperlinks in place:
#
# perl -i scripts/markdown-hyperlinks.pl CHANGELOG.md
#
# Generate to another file:
#
# scripts/markdown-hyperlinks.pl CHANGELOG.md > CHANGELOG-autolink.md
use strict;
use warnings;
my $user = qr/ [a-zA-Z] [a-zA-Z0-9]* /x;
my $repo = qr/ [a-zA-Z] [a-zA-Z0-9.-]* /x;
while (<>) {
# make links from @username references
# (except when escaped like \@foobar or in code like `where !@barman! = 'Moe'`)
s"(?: (`++) .*? \g{-1} (*SKIP) (*FAIL) )? # skip over code spans
(?<! \[ | \\ ) # no match after [ or \
\@ ($user) \b
(?! \] ) # no match before ]
"[\@$2](https://github.com/$2)"xg;
# make links from #1234 references (except when escaped like \#5)
# we can't tell whether the number refers to an issue or a pull request,
# luckily link to issues/1234 works in either case
s;(?<! \[ | \\ ) \# ([0-9]+) \b (?! \] )
;[\#$1](https://github.com/mapnik/mapnik/issues/$1);xg;
# make shortcut links from raw URIs (which GFM turns into proper links,
# but doesn't contract even though it could)
# - issues
s;(?<! \] \( ) (https://github\.com/mapnik/mapnik/(?:issues|pull)/([0-9]+))
;[\#$2]($1);xg;
s;(?<! \] \( ) (https://github\.com/($user/$repo)/(?:issues|pull)/([0-9]+))
;[$2\#$3]($1);xg;
# - commit hashes
s;(?<! \] \( ) (https://github\.com/mapnik/mapnik/commit/([0-9a-f]{7})[0-9a-f]{0,33}) \b
;[$2]($1);xg;
s;(?<! \] \( ) (https://github\.com/($user/$repo)/commit/([0-9a-f]{7})[0-9a-f]{0,33}) \b
;[$2\@$3]($1);xg;
# make links from commit hashes
# (accept 7 or 9-40 hex digits, but not 8 which could be a date)
s"(?: (`++) .*? \g{-1} (*SKIP) (*FAIL) )? # skip over code spans
(?: \[.*?\] \(.*?\) (*SKIP) (*FAIL) )? # skip over links
(?<! / ) # no match after /
\b (([0-9a-f]{7}) # 7 digits for link text
([0-9a-f]{2,33})?) \b # maybe 2-33 more digits
"[$3](https://github.com/mapnik/mapnik/commit/$2)"xg;
print;
}