We don't use source/test. It's kept to give API usage examples, but
it got in the way of a version update (git cl upload keeps timing out).

Also, update update.sh to delete source/test after downloading a new
version from the upstream.

BUG=637001
TEST=None

Review URL: https://codereview.chromium.org/2435373002 .

The file was removed in chromium:
https://codereview.chromium.org/2331893004

This moves it here to keep the logic working for
projects depending on ICU that still have a few gyp bots.

BUG=chromium:474921
R=jochen@chromium.org

Review URL: https://codereview.chromium.org/2335993003 .

Make icu depend on //buildtools/third_party/libc++abi
on mac for sanitizer builds.

BUG=636127
TEST=Libfuzzer Upload Mac ASan builds
R=mmoroz@chromium.org

Review URL: https://codereview.chromium.org/2247953005 .

This is only used in a non-default build configuration.
It's for IANA timezone DB 2016f. The ICU data files for
the default build configuration were updated earlier.

BUG=473288

Review URL: https://codereview.chromium.org/2222353003 .

BUG=626078
TEST=trybot succeeds on non-Windows.
R=rsesek@chromium.org

Review URL: https://codereview.chromium.org/2218673002 .

Private headers are listed in source[] while public headers are listed
in public[] for icuuc and icu18n targets.

This change is to let trybot analyze step recompile icuuc or icu18n even
when the only change in CL is in a header file (public or private).

BUG=632438
TEST=`gn check <out_dir>` does not complain.
TEST=`gn ref <out_dir> third_party/icu/source/common/umutex.h` matches icuuc.
TEST=`gn ref <out_dir> third_party/icu/source/common/unicode/unistr.h`
     matches icuuc.

R=brettw@chromium.org

Review URL: https://codereview.chromium.org/2186343002 .

Add 'data' to copy() to for swarm-testing to work on trybots.
Otherwise, the ICU data bundle would not be copied in isolate_tests step and
all the tests requiring the ICU data would fail on trybots.

BUG=v8:4828
TEST=base_unittests and many other tests pass on trybots (with test isolation)
R=machenbach@chromium.org

Review URL: https://codereview.chromium.org/2197523002 .

The only change affecting the current and future time stamp
is that Asia/Novosibirsk swiches from +06 to +07. The Egyptian
change in 2016f cancelled out the Egyptian change in 2016e.

http://mm.icann.org/pipermail/tz-announce/2016-July/000040.html

All the pre-built data files are updated except for icudt.dll
for Windows (non-default build configuration. The default build
configuration on Windows is covered by this CL).

BUG=473288
TEST=Try the following in d8 or Chrome's JS console:
> (new Date("07/23/2016 12:00Z")).toLocaleString("en", {timeZone:
"Asia/Novosibirsk"})
7/23/2016, 6:00:00 PM
> (new Date("07/24/2016 12:00Z")).toLocaleString("en", {timeZone:
"Asia/Novosibirsk"})
7/24/2016, 7:00:00 PM

Review URL: https://codereview.chromium.org/2183313002 .

Delete three pre-built assembly source files because they're now
generated at build-time.

Update data build scripts and README.chromium accordingly.

Update copy_data.sh and copy_data_android.sh so that the assembly
source files are not copied. Besides, convert the little endian
data bundle to the big endian data bundle for non-Android platforms.

BUG=v8:4828
TEST=Rebuild icu data following the procedure in README.chromium
TEST='gn args <builddir>' with icu_use_data_file set to true or false
TEST=build base_unittests and run with --gtest_filter=ICU*
TEST=build base_unittests and run with --gtest_filter=Message*ormat*
TEST=build 'd8' (v8) and try `(new Date()).toLocaleString("de")`

Review URL: https://codereview.chromium.org/2182883004 .

Add big endian (mips and mips64) support to BUILD.gn

- icudt[lb]_dat.S (assembly source) is generated at build-time from
  icudt[lb].dat (data bundle) when icu_use_data_file is false.
- icudt[lb].dat is copied depending on the endinaness when icu_use_data_file
  is true.

Part 1:  https://codereview.chromium.org/2162393003
Part 2:  https://codereview.chromium.org/2165403003

It works on Linux (both Chrome and v8) and Mac (v8).
Android was also tested by building base_unittests target with
icu_use_data_file=true/false.

v8 does not yet support mips/mips64 in GN so that this CL cannot be
tested with target_cpu={mips,mips64}.

With Chrome on Mac, TEST below has to be done with
https://codereview.chromium.org/2181043003 for http://crbug.com/630929.

Windows should not be affected at all.

BUG=v8:4828
TEST='gn args <builddir>' with icu_use_data_file set to true or false
TEST=build base_unittests and run with --gtest_filter=ICU*
TEST=build base_unittests and run with --gtest_filter=Message*ormat*
TEST=build 'd8' (v8) and try `(new Date()).toLocaleString("de")`
R=machenbach@chromium.org

Review URL: https://codereview.chromium.org/2174993002 .

BUG=
R=tandrii@chromium.org

Review URL: https://codereview.chromium.org/2182943002 .

Follow-up CL to https://codereview.chromium.org/2162393003

1. make_data_assembly now accepts '--mac' to generate assembly source
for Mac

2.  Fix icu.gyp to support all platforms

BUG=v8:4828
TEST='d8' is built correct with icu_use_data_file set to either 0 or 1 on Mac/Linux
TEST=run `GYP_DEFINES="target_arch=mips" ./gypfiles/gyp_v8` and make sure
that ninja files use 'b' data/assembly file for Big Endian on Mac/Linux

R=machenbach@chromium.org

Review URL: https://codereview.chromium.org/2165403003 .

Add a script that generates an assembly file from a .dat file. This is
needed for generating big endian assembly file after using icupkg to
convert little endian icudtl.dat to big endian icudtb.dat. Also the
icu.gyp file is modified so big endian architectures use appropriate
files.

Patch by miran.karic@  ( https://codereview.chromium.org/1967523002/)
with a couple of fixes:

1. Two errors mentioned against PS#9 in the above CL.
2. Support copying icu data file for Big Endian targets.

Besides, icudtb.dat was added to common. icudtb.dat was created by
running 'icupkg -tb icudt56l.dat icudt56b.dat' and renaming icudt56b.dat
to icudtb.dat.

BUG=v8:4828
TEST='d8' is built correct with icu_use_data_file set to either 0 or 1.
TEST=run `GYP_DEFINES="target_arch=mips" ./gypfiles/gyp_v8` and make sure
that ninja files use 'b' data/assembly file for Big Endian.

Review URL: https://codereview.chromium.org/2162393003 .

Patch from Miran Karic <miran.karic@imgtec.com>.

I will compare utf8/utf32 fuzzer performance and coverage on
ClusterFuzz.

BUG=pdfium:38
R=jshin@chromium.org

Review URL: https://codereview.chromium.org/2119023002 .

What's new in 2016d is found at

  http://mm.icann.org/pipermail/tz-announce/2016-April/000038.html

Rebuilt ICU data/assembly files are checked in (not shown in the
codereview due to their sizes).

While I'm at it, add scripts/LICENSE file that is identical to
LICENSE at the top of the Chromium tree. Because LICENSE in
third_party/icu is for ICU and is not applicable to files in scripts/.

BUG=473288
TBR=mark
TEST=In JavaScript console, run the following.
      apr30_2016_1200 = new Date("04/30/2016 12:00Z")
      may01_2016_1200 = new Date("05/01/2016 12:00Z")
      apr30_2016_1200.toLocaleString("en", {timeZone: "America/Caracas"})
      may01_2016_1200.toLocaleString("en", {timeZone: "America/Caracas"})

  On April 30, 2016, Caracas is 4:30 behind UTC. On May 1, it's 4:00 behind.

Review URL: https://codereview.chromium.org/1985243002 .

Adding fuzzers for:
  - NumberFormat::parse
  - BreakIterator
  - ucasemap

BUG=chromium:474921
R=jshin@chromium.org, vogelheim@chromium.org

Review URL: https://codereview.chromium.org/1994903002 .

BUG=610673
R=alexclarke@chromium.org, dpranke@chromium.org

Review URL: https://codereview.chromium.org/1991833002 .

Cherry pick  http://bugs.icu-project.org/trac/changeset/38183
to add the support for Emoji properties.

This will be followed by a Blink change to use the API.
( https://codereview.chromium.org/1856353002 )

There's no need to rebuild ICU data because Unicode character
properties are hard-coded in the C++ source files that are
patched by the cherry-pick above.

source/data/in/{pnames,uprops}.icu are updated just for the
record, but they're not actually used. The same is true
of files in source/data/unidata.

BUG=583378
TEST=None
TBR=drott

Review URL: https://codereview.chromium.org/1864463002 .

BUG=575007
R=jshin@chromium.org

Review URL: https://codereview.chromium.org/1838103003 .

1. Update the IANA tz data to 2016c

What's new in 2016b and 2016c are found at

  http://mm.icann.org/pipermail/tz-announce/2016-March/000036.html (2016b)
  http://mm.icann.org/pipermail/tz-announce/2016-March/000037.html (2016c)

2. Locale data fixes
  - en-AU date format fix from the upstream
  - ar and fa: Prepend 'percent sign' with RTL mark (U+200F). From Android.
  - tr: Use ₺ (U+20BA; Turkish Lira Sign) instead of 'TL'. This is to revert a locale patch picked up from Google's internal build of ICU. (Android also uses U+20BA).

In addition, icudtl.dat (the prebuilt ICU data file for platforms other than
Android is moved out of source/data/in to common/. This way, the data build
steps for non-Android and Android can be unified and a bit more streamlined.
icu.gyp and BUILD.gn are updated accordingly as well as README.chromium.

BUG=598000
TEST=See bug comment 0 and comment 1
R=mark@chromium.org

Review URL: https://codereview.chromium.org/1823293002 .

region.cpp has an unnecessary static initializer for a type
UnicodeString and triggers a 'si check' failure.

Upstream bug: http://bugs.icu-project.org/trac/ticket/12408

BUG=575007
TEST=`tools/linux/dump-static-initializers.py chrome` (static build) does not have an entry from region.cpp
TEST=`tools/linux/dump-static-initializers.py libicui18n.so` (shared build) has no entry.
TBR=thakis@chromium.org

Review URL: https://codereview.chromium.org/1816843002 .

ICU in Chromium is compiled with exception disabled [1], but
U_NOEXCEPT is set to 'noexcept'  regardless of whether _HAS_EXCEPTION is set to 0 or (in source/common/unicode/platform.h ). [2].

Set U_NOEXCEPT to empty in our build files to avoid this issue.

[1] http://bugs.icu-project.org/trac/ticket/8343
[2] http://bugs.icu-project.org/trac/ticket/12406

BUG=575007
TEST=Build on Windows with MSVS 2015 succeeds.
R=thakis@chromium.org

Review URL: https://codereview.chromium.org/1816673002 .

BUG=297668
R=dpranke@chromium.org

Review URL: https://codereview.chromium.org/1755473004 .

This is, among other things, to make sure that a change is not made
against ToT while ICU is in the middle of update to a new version.
For such a change, a branch has to be used.

BUG=None

Review URL: https://codereview.chromium.org/1713163002 .

Update the copyright years in LICENSE (taken from license.html) and
IANA timezone DB version in README.chromium.

Incorporate the format fixes made by Mostyn Bramley-Moore <mostynb@opera.com>
in https://codereview.chromium.org/1530343003/

BUG=575007
TEST=None
TBR=mostynb@opera.com

Review URL: https://codereview.chromium.org/1678243002 .

1. ast,an,wa locale data (minimal)
   - make up the minimal locale data for 3 languages.
   - update source/data/{locale,lang}/reslocal.mk to have 3 languages
   - update scripts/data_file_to_preserve.txt to have files for 3 locales
     listed.
   - In chromium, these languages have to be added to the A-L list to show
     up in the A-L pull-down in settings.

2. IANA timezone db updte to 2016a
  See http://mm.icann.org/pipermail/tz-announce/2016-January/000035.html
  for the change.

3. Pre-built-data files are updated.

Data (icudtl.dat) size changes between 54.1 and 56.1
  non-Android platforms: 10,124,096 bytes (net change: -83,840)
  Android: 6,560,080 bytes (net change: +291,840 / 4.66%)

BUG=575007,474333
R=jsbell@chromium.org

Review URL: https://codereview.chromium.org/1665113004 .

See http://crbug.com/417850 for UTF-32
See http://bugs.icu-project.org/trac/ticket/12147 for C4275 warning
from MSVC.

BUG=575007
TEST=Layout tests fast/encoding/*
TEST=MSVC does not emit C4275 warning any more
R=jsbell@chromium.org

Review URL: https://codereview.chromium.org/1664293002 .

patches/wordbrk.patch was not applied by mistake while updating ICU
to 56.1.

This affects how we break domain names (separated by full stop). We want
them to be split at full stop at the cost of 'breaking' cases like
"e.g." and "i.e.".

All the ICU data files were rebuilt for the change.

BUG=575007
TEST=gfx_unittests --gtest_filter=RenderText*Word*
TEST=layout tests: css1/text_properties/text_transform.html fast/text/window-find.html editing/text-iterator/findString.html css2.1/t1605-c545-txttrans-00-b-ag.html

Review URL: https://codereview.chromium.org/1659833002 .

Set fLastSetLiteral to U_SENTINEL in the ctor and check if it's still
U_SENTINEL before adding the range.

BUG=582470
TEST=See comments 13 and 15 in the bug
TBR=mmoroz

Review URL: https://codereview.chromium.org/1656003004 .

This is to cherry-pick a fix for
  http://bugs.icu-project.org/trac/ticket/12130

It's done by merging commit '2b12f877' (for m49 branch) to
master. (ICU's m49 branch was fixed first because ICU master was
in the middle of upgrading to ICU 56.1).

BUG=575206
TEST=See comment 51 in the bug
TBR=mmoroz@chromium.org

Review URL: https://codereview.chromium.org/1662453002 .

MSVC complains about '*/* foobar */'. Add a space between '*' and '/'.

ICU uses a deprecated API internally and '-Wno-deprecated...' flag
has to be turned on on Android as well as on Linux.

Do not use ''-Wno-reorder' any more because it's not necessary.

BUG=575007
TEST=Once rolled, Android and Windows build go through.
R=thakis@chromium.org

Review URL: https://codereview.chromium.org/1646623006 .

* Update the pre-built ICU data files for all platforms

  source/data/in/icudtl.dat for non-Android platforms
  {linux,mac}/icudt*.S for linux/mac
  android/icudtl.dat and android/icudt*.S for Android
  windows/icudt.dll for Windows

* Update Android data trimming script

  1. Make sure that 'default' calendar is kept in locales where it's relevant
    : root, th, fa, ar_SA, etc.
  2. Add a minimal region data to work around a bug in ICU with pool.res
     handling

* Update gn and gyp files
* And add a TODO comment to update.sh to automate the build file update.
* Add it_CH to the locale list.
* Add sr_Latn to unit/reslocal.mk (required by sh) and
  line_normal_fi to brkitr/brklocal.mk (referred to in brkitr/fi.txt) in
  place of line_fi.

* Update and add scripts for data building
* Completely rewrite README.chromium
* Check-in the prebuilt ICU data files/assembly sources for
  Linux,Mac,Windows,Chrome OS and Android.

BUG=575007
TEST=Blink layout tests, webkit unittests
TEST=All bots can build successfully
TEST=net_unittests --gtest_filter="*ilenameUtil*"
TEST=net_unittests --gtest_filter="*IDN*" (pending bug 336973)
TEST=base_unittests --gtest_filter="*Conv*"
TEST=browser_tests --gtest_filter="*ncoding*"
TEST=base_unittests --gtest_filter="*essage*"
TEST=ui_base_unittests --gtest_filter="*ormat*"
TEST=ui_base_unittests --gtest_filter="L10n*"
R=mark@chromium.org

Review URL: https://codereview.chromium.org/1639543006 .