Adjust calendar locale data trimming on Andorid

On Android, add Islamic calendar to Arabic and Farsi locales and Indian calendar to Hindi locale. Also add both variants of Ethiopic calendars to am/root. There was a 'typo' in the locale trimming script for Android ('arabic' instead of 'islamic' was specified.). As a result, Date formatting in ar-SA (where the default calendar is Islamic) failed for the lack of locale data. Add back calendar entries to data/locales/root.txt. They're necessary to support locale-specific calendars on Android (e.g. islamic on ar/fa, japanese on ja, hebrew on he, etc). Adjust the drop list for the data file for cast. More locale variants have been added since it's made. Drop locale bundles for newly added variants. Drop zone/unit data for 8 locales Android Chrome does not yet support. This saves over 100 kB. For all platforms, drop a few more items not yet used in Chromium/v8. With this change, the net data size decreased on all platforms, paving the way to add more locale variants in a subsequent CL. Previous This CL Platform 6375056 6235856 android 4916608 4714896 cast 10268240 10258960 common 880512 879104 flutter 6361376 6195600 ios TBR=ftang@chromium.org,gsathya@chromium.org Bug: v8:8432 Test: v8: intl/regress-8432-* Change-Id: Iafd94c08ea4e88b33aed4724309636c3e224bdc2 Reviewed-on: https://chromium-review.googlesource.com/c/1325943 Reviewed-by: Jungshik Shin <jshin@chromium.org>
denoland · Nov 14, 2018 · ecae5c0 · ecae5c0
1 parent 45f655f
commit ecae5c0
Show file tree

Hide file tree

Showing 9 changed files with 156 additions and 64 deletions.
diff --git a/android/icudtl.dat b/android/icudtl.dat
diff --git a/cast/cast-removed-resources.txt b/cast/cast-removed-resources.txt
diff --git a/cast/icudtl.dat b/cast/icudtl.dat
diff --git a/cast/patch_locale.sh b/cast/patch_locale.sh
@@ -82,11 +82,32 @@ do
      /^\}$/p' ${target}
 done
 
+echo Creating minimum locale data in ${langdatapath}
+for lang in ${EXTRA_LANGUAGES}
+do
+  target=lang/${lang}.txt
+  [  -e ${target} ] || { echo "missing ${lang}"; continue; }
+  echo Overwriting ${target} ...
+
+  # Do not include '%%Parent' line on purpose.
+  sed -n -r -i \
+    '1, /^'${lang}'\{$/p
+     /^    "%%ALIAS"\{/p
+     /^    Languages\{$/, /^    \}$/ {
+       /^    Languages\{$/p
+       /^        '${lang}'\{.*\}$/p
+       /^    \}$/p
+     }
+     /^\}$/p' ${target}
+done
+
 echo Overwriting curr/reslocal.mk to drop the currency names
 echo for ${EXTRA_LANGUAGES}
 for lang in ${EXTRA_LANGUAGES}
 do
   sed -i -e '/'$lang'.txt/ d' curr/reslocal.mk
+  sed -i -e '/'$lang'.txt/ d' zone/reslocal.mk
+  sed -i -e '/'$lang'.txt/ d' unit/reslocal.mk
 done
 
 # Remove exemplar cities in timezone data.
@@ -117,20 +138,31 @@ for i in locales/*.txt; do
       EXTRA_CAL='dangi'
       ;;
     am)
-      EXTRA_CAL='ethiopic'
+      EXTRA_CAL='ethiopic|ethiopic-amete-alem'
       ;;
     he)
       EXTRA_CAL='hebrew'
       ;;
     ar)
-      EXTRA_CAL='arabic'
+      # Other Islamic calendar formats are not in locales other than root.
+      # ar-SA's default is islamic-umalqura, but its format entries are
+      # specified in root via aliases.
+      EXTRA_CAL='islamic'
       ;;
     fa)
-      EXTRA_CAL='persian'
+      EXTRA_CAL='persian|islamic'
       ;;
     ja)
       EXTRA_CAL='japanese'
       ;;
+    # When adding other Indian locales for Android,
+    # add 'indian' calendar to them as well.
+    hi)
+      EXTRA_CAL='indian'
+      ;;
+    root)
+      EXTRA_CAL='buddhist|chinese|roc|dangi|ethiopic|ethiopic-amete-alem|japanese|hebrew|islamic|islamic-(umalqura|civil|tbla|rgsa)|persian|indian'
+      ;;
     *)
       EXTRA_CAL=''
       ;;
@@ -152,4 +184,16 @@ for i in locales/*.txt; do
           }' -i $i
 done
 
+# Delete Japanese era display names in root. 'ja' has Japanese era names
+# so that root does not need them.
+# The same is true of eras and monthNames for Islamic calendar.
+sed -r -i \
+  '/^        japanese\{$/,/^        \}$/ {
+     /^            eras\{/,/^            \}$/d
+   }
+   /^        islamic\{$/,/^        \}$/ {
+     /^            eras\{/,/^            \}$/d
+     /^            monthNames\{/,/^            \}$/d
+   }'  locales/root.txt
+
 echo DONE.
diff --git a/common/icudtb.dat b/common/icudtb.dat
diff --git a/common/icudtl.dat b/common/icudtl.dat
diff --git a/flutter/icudtl.dat b/flutter/icudtl.dat
diff --git a/ios/icudtl.dat b/ios/icudtl.dat
diff --git a/scripts/trim_data.sh b/scripts/trim_data.sh
@@ -17,7 +17,12 @@ function filter_locale_data {
        /^    AuxExemplarCharacters\{$/, /^    \}$/d
        /^    ExemplarCharacters\{.*\}$/d
        /^    ExemplarCharacters\{$/, /^    \}$/d
-       /^        (mon|tue|wed|thu|fri|sat|sun)(|-short|-narrow)\{$/, /^        \}$/d' ${langpath}
+       /^    ExemplarCharactersNumbers\{.*\}$/d
+       /^    ExemplarCharactersPunctuation\{.*\}$/d
+       /^    ExemplarCharactersPunctuation\{$/, /^    \}$/d
+       /^        (mon|tue|wed|thu|fri|sat|sun)(|-short|-narrow)\{$/, /^        \}$/d
+       /^        (mon|tue|wed|thu|fri|sat|sun)(|-short|-narrow)\{.*\}$/d
+       /^        (mon|tue|wed|thu|fri|sat|sun)-(short|narrow):alias\{.*\}$/d' ${langpath}
     # Delete empty blocks. Otherwise, locale fallback fails.
     # See crbug.com/v8/8414 .
     sed -r -i \