commit 2823bdd7ed9022fafa0c3443ffde90e522d92b52
author Frank Tang <ftang@chromium.org> Tue Dec 11 15:39:45 2018 -0800
committer Jungshik Shin <jshin@chromium.org> Wed Dec 12 00:03:08 2018 +0000
tree 4b9089f7557d0fadbf4cce800cccdab6a1b83ebb
parent 407b39301e71006b68bd38e770f35d32398a7b14

i18n: Fix new Intl.DateTimeFormat("ja-u-tz-uslax")

Fix the locale parsing with tz which resolved to a tz name which
have a _ in it. "uslax" is resolved to "Los_Angles" and cause that.

TEST=In v8 or Chrome console, new Intl.DateTimeFormat("ja-u-tz-uslax") works fine.
resolvedOptions().timeZone is 'America/Los_Angenles'.
Bug: v8:8469
Change-Id: I996790b1ce59fbb5f6b0a460ff8acdb674894df1
Reviewed-on: https://chromium-review.googlesource.com/c/1372810
Reviewed-by: Jungshik Shin <jshin@chromium.org>

README.chromium

diff --git a/README.chromium b/README.chromium
index 5fbf9b3..e266952 100644
--- a/README.chromium
+++ b/README.chromium
@@ -266,3 +266,11 @@
   - Fix:
     https://github.com/unicode-org/icu/pull/265
     https://github.com/unicode-org/icu/pull/278
+
+11. Locale parsing with tz unicode extension.
+
+  - patches/locale_id_tz_parsing.patch
+  - upstream bug:
+    https://unicode-org.atlassian.net/browse/ICU-11053
+  - Fix:
+    https://github.com/unicode-org/icu/pull/300

patches/locale_id_tz_parsing.patch

diff --git a/patches/locale_id_tz_parsing.patch b/patches/locale_id_tz_parsing.patch
new file mode 100644
index 0000000..910eb55
--- /dev/null
+++ b/patches/locale_id_tz_parsing.patch
@@ -0,0 +1,19 @@
+diff --git a/source/common/locid.cpp b/source/common/locid.cpp
+index e0dcc8a8..5d584bea 100644
+--- a/source/common/locid.cpp
++++ b/source/common/locid.cpp
+@@ -569,9 +569,13 @@ Locale& Locale::init(const char* localeID, UBool canonicalize)
+         variantBegin = length;
+ 
+         /* after uloc_getName/canonicalize() we know that only '_' are separators */
++        /* But _ could also appeared in timezone such as "en@timezone=America/Los_Angeles" */
+         separator = field[0] = fullName;
+         fieldIdx = 1;
+-        while ((separator = uprv_strchr(field[fieldIdx-1], SEP_CHAR)) != 0 && fieldIdx < UPRV_LENGTHOF(field)-1) {
++        char* at = uprv_strchr(fullName, '@');
++        while ((separator = uprv_strchr(field[fieldIdx-1], SEP_CHAR)) != 0 &&
++               fieldIdx < UPRV_LENGTHOF(field)-1 &&
++               (at == nullptr || separator < at)) {
+             field[fieldIdx] = separator + 1;
+             fieldLen[fieldIdx-1] = (int32_t)(separator - field[fieldIdx-1]);
+             fieldIdx++;

source/common/locid.cpp

diff --git a/source/common/locid.cpp b/source/common/locid.cpp
index e0dcc8a..5d584be 100644
--- a/source/common/locid.cpp
+++ b/source/common/locid.cpp
@@ -569,9 +569,13 @@
         variantBegin = length;
 
         /* after uloc_getName/canonicalize() we know that only '_' are separators */
+        /* But _ could also appeared in timezone such as "en@timezone=America/Los_Angeles" */
         separator = field[0] = fullName;
         fieldIdx = 1;
-        while ((separator = uprv_strchr(field[fieldIdx-1], SEP_CHAR)) != 0 && fieldIdx < UPRV_LENGTHOF(field)-1) {
+        char* at = uprv_strchr(fullName, '@');
+        while ((separator = uprv_strchr(field[fieldIdx-1], SEP_CHAR)) != 0 &&
+               fieldIdx < UPRV_LENGTHOF(field)-1 &&
+               (at == nullptr || separator < at)) {
             field[fieldIdx] = separator + 1;
             fieldLen[fieldIdx-1] = (int32_t)(separator - field[fieldIdx-1]);
             fieldIdx++;