sed -i 's,<atom>\([^<]\+\)</atom>,`\1`,g'

Change-Id: I530fe998b0c422e83c4324807d42700126bc9883

sed -i 's,<method scope="\([^"]*\)">\([^<]\+\)</method>,\1::\2,g'

Change-Id: I09ee9ecae922e0b28730d1689b802e6041b3ceed

sed -i 's,<method scope="\([^"]*\)">\([^<][^)]\+\)</method>,\1::\2(),g'

Change-Id: I7ceb805d67a79c297fb5f044a8fb226ec2fb366c

sed -i "s,<method>\([a-z][^<]\+\)</method>,\1,g"

Change-Id: I9483d63cce4afd168d47a465b367aed3da45d64b

What is sad about this is that autodoc doesn't even support <method>.

sed -i "s,<method>\([a-z][^<]\+[^)]\)</method>,\1(),g"

Change-Id: I702ef71423ced1d5195f2e0535e73b1bb4d3f6f2

... which does not seem to do anything anyway.

sed -i "s,<arg>\([^<]\+\)</arg>,\1,g"

Change-Id: I90bcf81ea046bc88297ed8eadc560ad31d19830b

This one is apparently often abused to link to a constant group, while
it can only link to constants within a group.

sed -i "s,<const>\([^<]\+\)</const>,\1,g"

Change-Id: Ic3d8099751340e4b046298c861bb659beb351eaf

sed -i 's,<const scope="\([^"]*\)">\([^<]\+\)</const>,\1::\2,g'

Change-Id: I964acc2725e8a6680f5bd8412014e5660510749e

sed -i 's,<member scope="\([^"]*\)">\([^<]\+\)</member>,\1::\2,g'

Change-Id: Idbe0adc5a88e430c99257d4d61e540bf0eeaea8f

sed -i "s,<member>\([^<]\+\)</member>,\1,g"

Change-Id: I74c80f5aa610477fab28faec52e55c3137984aec

These member variables need a # prefix.

sed -i "s,<member>\([^<:]\+[^)]\)</member>,#\1,g"

Change-Id: I721b90f4edae32b5a362479bdbd60423f6427364

sed -i 's,<member scope="\([^"]*\)">\([^<]*::[a-z][^<:]\+[^)]\)</member>,\1::\2(),g'

Change-Id: I265ff28fee73a271f205294e59955b3a1c81d95a

Doxygen would probably recognize these without () too but add them for
consistency.

sed -i "s,<member>\([^<]*::[a-z][^<:]\+[^)]\)</member>,\1(),g"

Change-Id: I2615b99265b75633459e35164e54d9da7fe76b85

Doxygen will only recognize a un-qualified method name as such if it is
followed by "()".

sed -i "s,<member>\([a-z][^<]\+[^)]\)</member>,\1(),g"

Change-Id: I69bc17849e76f3a3d91c6daf0f1be8168a83cfc5

This looks much better now.

sed -i 's,<type scope="\([^"]*\)">\([^<]\+\)</type>,\1::\2,g'

Change-Id: I94cd0f93afa89855b62dadeb229d2b2e1775cd80

Doxygen does not know type element and will recognize strings that
contain capital letter (all API types do) automatically as type.
This patch removes 15k doxygen warnings.

git ls-files | grep \\.idl | xargs sed -i "s,<type>\([^<]\+\)</type>,\1,"

Change-Id: I45c07cf0b115d5fb5353f4aa9719839615ea1150

sed -i "s,<type>\([^<]\+\)</type>,\1,"

Change-Id: Ibdaeb2c396ad29ced05716142b38c6a1c3e49847

sed -i 's,<type scope="\([^"]*\)">\([^<]\+\)</type>,\1::\2,g'

Change-Id: I7eb32acd44132de79cc7ac1e7f78b43bb4814d14

Unsurprisingly doxygen does not support reading a idl_chapter_refs.txt
file like autodoc does.  And it looks like doxygen tag files can not
be used to do this, as they link only in the "other" direction.

But with a bit of hackery we can generate a "dummy" IDL input file, with
emtpy definitions for those types that have DevGuide links.
Doxygen will nicely combine documentation from multiple definitions of a
type, and fortunately does only put in references to the first file that
defines a type, so hopefully adding the generated file as the last input
file is suffficient to make things work nicely.

The generated file is committed to git because there is a bit of a
problem with finding out what sort of entity a name like
"com.sun.star.Foo" refers to, which needs to be in the doxygen input.

Change-Id: I793bc59c8acfecf460e4addea7838c76a3ede77e

Change-Id: I04b8dbecbe62f8d72ca75f6508046294c3409a17

Change-Id: Ibf7ab46bdd08a42ecf475500d35f92b67d4d8036

Change-Id: I9a90bf06a70f6537f9671ef2cd47bf68c3cf5d57

Change-Id: Ic3f2d42121b5423d67c4c79302b02ac495ac9640

- give right path to STRIP_FROM_PATH
- prefix vars
- depend on makefile
- only call cygpath in rule

Change-Id: Iceef191751b82a423be354db1f0a3ca4fc213e04

Change-Id: I273b019409f36d0b4488458e152099493ce81147

Change-Id: I9c301f7d8b5295854e12585fa12df416ed5dbf02

Change-Id: I121f0d3d646ce434ea1451a94a5c911005dc21f9

Project: help  efa1b1723eae3e378fbd299677745be1364e53ce

Change-Id: I0d62ae391b12cec9a057aef26344f3454d2faba4

Change-Id: I9662671d8dbce87993ba67cf8e4fd993ad1eb27a

Change-Id: I746e1fa25137db786f82dda50bfa4b497e94cefd

Change-Id: I2ad35bfbaf40dd1a5b5c9f7f119268b8f30ca093

Also, use a better way to fetch a double array.

Change-Id: I65a5bb9c972620f63e79a84b461245cc7fea1a51

Change-Id: I790db61d2a60cf4074fd2e3291ec42f0a4649b9e

Change-Id: Id9009faa5c9221a04130f87af4fa14e33fe0dc50

Change-Id: Id735c7ef89411581f7bc8d7f10b3edfe8aba0c72

When a fly is anchored at the first paragraph of a ToX, as happens in
the WW8 bugdoc in i#30979, updating the ToX will remove the anchor
node, and bad things happen, including ~SwIndexReg assert and also
eventually a crash when editing the Frame and updating the ToX a few
times.  Probably best to really remove all anchored flys here.

Note that it's not enough to re-anchor to the empty node that is
inserted here because that one will also be removed again.

While we're at it the _DelBookmarks is superfluous because CorrAbs
does exactly that 2 lines above.

Change-Id: Id40ae1d88ad99813b93bea7470f5dc790edcc80c

It have been unused on Mac OS X since since 2008 (commit
da4d070d).

Change-Id: I4b3eabe71a54021903009310a87319c9caecfd4c

Change-Id: I660a895a227da3d28a0b410c56f2395961a5bc4f

This file is Core Text only already.

Change-Id: Ic4c1f44296eaae582091e515d3ca1b4aa227301e