Extend -default to include "it" and "pl", so we can retire -us-euro.

I've also modified icu_dat_generator.py to list the languages included in
each subset. (But not the locales, because that list is too long to be
convenient, plus we now warn if you don't have all the locales for a given
language, so it's just the languages that are interesting anyway.)

We need to keep -us as long as voles is supported, so -default still isn't
the default (and -us still exists), sadly.

Bug: 3139942
Change-Id: If92879fa95b6a10288612b5b4350f22985472919

6 files changed, 40 insertions, 305 deletions

diff --git a/stubdata/Android.mk b/stubdata/Android.mk
index a1b2fd6c..f9809b32 100644
--- a/stubdata/Android.mk
+++ b/stubdata/Android.mk
@@ -37,11 +37,17 @@ include $(CLEAR_VARS)
 
 # Build configuration:
 #
-# "Large" includes all the supported locales.
-# Japanese includes US and Japan.
-# US-Euro is needed for IT or PL builds
-# Default is suitable for CS, DE, EN, ES, FR, NL
-# US has only EN and ES
+# all includes all ICU's locale data.
+#
+# icudt44l-large.txt includes all the most-used
+# locales (http://googleblog.blogspot.com/2008/07/hitting-40-languages.html):
+#   ar, bg, ca, cs, da, de, el, en, es, fa, fi, fr, he, hi, hr, hu, id, it, ja,
+#   ko, lt, lv, nb, nl, pl, ps, pt, rm, ro, ru, sk, sl, sr, sv, th, tr, uk, vi,
+#   zh.
+#
+# icudt44l-default.txt includes cs, de, en, es, fr, it, nl, pl.
+# icudt44l-us-japan.txt includes en, ja.
+# icudt44l-us.txt includes en, es (for voles-eng).
 
 config := $(word 1, \
             $(if $(findstring ar,$(PRODUCT_LOCALES)),large) \
@@ -66,12 +72,8 @@ config := $(word 1, \
             $(if $(findstring uk,$(PRODUCT_LOCALES)),large) \
             $(if $(findstring zh,$(PRODUCT_LOCALES)),large) \
             $(if $(findstring ja,$(PRODUCT_LOCALES)),us-japan) \
-            $(if $(findstring it,$(PRODUCT_LOCALES)),us-euro) \
-            $(if $(findstring pl,$(PRODUCT_LOCALES)),us-euro) \
-            $(if $(findstring cs,$(PRODUCT_LOCALES)),default) \
-            $(if $(findstring de,$(PRODUCT_LOCALES)),default) \
-            $(if $(findstring fr,$(PRODUCT_LOCALES)),default) \
-            $(if $(findstring nl,$(PRODUCT_LOCALES)),default) \
+            $(if $(findstring it,$(PRODUCT_LOCALES)),default) \
+            $(if $(findstring pl,$(PRODUCT_LOCALES)),default) \
             us)
 
 include $(LOCAL_PATH)/root.mk
diff --git a/stubdata/icu_dat_generator.py b/stubdata/icu_dat_generator.py
index 4656d834..851d9af2 100755
--- a/stubdata/icu_dat_generator.py
+++ b/stubdata/icu_dat_generator.py
@@ -163,7 +163,15 @@ def GenResIndex(input_file):
   if VERBOSE:
     for kind, locales in kind_to_locales.items():
       print "%s=%s" % (kind, sorted(locales))
-    print "every_locale=" % sorted(every_locale)
+
+  # Print a human-readable list of the languages supported.
+  every_language = set()
+  for locale in every_locale:
+    language = re.sub(r"(_.*)", "", locale)
+    if len(language) == 2:
+      every_language.add(language)
+  input_basename = os.path.basename(input_file)
+  print "%s includes %s." % (input_basename, ", ".join(sorted(every_language)))
 
   # Find cases where we've included only part of a locale's data.
   missing_files = []
diff --git a/stubdata/icudt44l-default.dat b/stubdata/icudt44l-default.dat
index ffb6305d..97a5fe23 100644
--- a/stubdata/icudt44l-default.dat
+++ b/stubdata/icudt44l-default.dat
diff --git a/stubdata/icudt44l-default.txt b/stubdata/icudt44l-default.txt
index c4e1a9e1..a3a68ec0 100644
--- a/stubdata/icudt44l-default.txt
+++ b/stubdata/icudt44l-default.txt
@@ -58,9 +58,14 @@ coll/fr.res
 coll/fr_CA.res
 coll/fr_FR.res
 coll/fr_LU.res
+coll/it.res
+coll/it_CH.res
+coll/it_IT.res
 coll/nl.res
 coll/nl_BE.res
 coll/nl_NL.res
+coll/pl.res
+coll/pl_PL.res
 
 curr/res_index.res
 curr/root.res
@@ -101,7 +106,9 @@ curr/es_UY.res
 curr/fr.res
 curr/fr_CA.res
 curr/fr_LU.res
+curr/it.res
 curr/nl.res
+curr/pl.res
 
 lang/res_index.res
 lang/root.res
@@ -116,8 +123,10 @@ lang/es.res
 lang/es_AR.res
 lang/es_CL.res
 lang/fr.res
+lang/it.res
 lang/nl.res
 lang/nl_BE.res
+lang/pl.res
 
 region/res_index.res
 region/root.res
@@ -131,8 +140,10 @@ region/en_ZW.res
 region/es.res
 region/es_CL.res
 region/fr.res
+region/it.res
 region/nl.res
 region/nl_BE.res
+region/pl.res
 
 zone/res_index.res
 zone/pool.res
@@ -150,7 +161,9 @@ zone/en_ZW.res
 zone/es.res
 zone/fr.res
 zone/fr_CA.res
+zone/it.res
 zone/nl.res
+zone/pl.res
 
 res_index.res
 root.res
@@ -204,9 +217,14 @@ fr.res
 fr_CA.res
 fr_FR.res
 fr_LU.res
+it.res
+it_CH.res
+it_IT.res
 nl.res
 nl_BE.res
 nl_NL.res
+pl.res
+pl_PL.res
 
 gsm-03.38-2000.cnv
 ibm-1089_P100-1995.cnv
diff --git a/stubdata/icudt44l-us-euro.dat b/stubdata/icudt44l-us-euro.dat
deleted file mode 100644
index 4fdbeb0d..00000000
--- a/stubdata/icudt44l-us-euro.dat
+++ /dev/null
diff --git a/stubdata/icudt44l-us-euro.txt b/stubdata/icudt44l-us-euro.txt
deleted file mode 100644
index 91d68778..00000000
--- a/stubdata/icudt44l-us-euro.txt
+++ /dev/null
@@ -1,293 +0,0 @@
-brkitr/res_index.res
-brkitr/root.res
-brkitr/char.brk
-brkitr/en.res
-brkitr/en_US.res
-brkitr/en_US_POSIX.res
-brkitr/line.brk
-brkitr/sent.brk
-brkitr/title.brk
-brkitr/word.brk
-brkitr/word_POSIX.brk
-brkitr/thaidict.ctd
-
-coll/res_index.res
-coll/invuca.icu
-coll/ucadata.icu
-coll/root.res
-coll/cs.res
-coll/cs_CZ.res
-coll/de.res
-coll/de_.res
-coll/de_AT.res
-coll/de_BE.res
-coll/de_CH.res
-coll/de_DE.res
-coll/de_LU.res
-coll/de__PHONEBOOK.res
-coll/en.res
-coll/en_AU.res
-coll/en_BE.res
-coll/en_BW.res
-coll/en_CA.res
-coll/en_GB.res
-coll/en_HK.res
-coll/en_IE.res
-coll/en_IN.res
-coll/en_MT.res
-coll/en_NZ.res
-coll/en_PH.res
-coll/en_SG.res
-coll/en_US.res
-coll/en_US_POSIX.res
-coll/en_VI.res
-coll/en_ZA.res
-coll/en_ZW.res
-coll/es.res
-coll/es_AR.res
-coll/es_CL.res
-coll/es_CO.res
-coll/es_EC.res
-coll/es_ES.res
-coll/es_GT.res
-coll/es_HN.res
-coll/es_MX.res
-coll/es_PR.res
-coll/es_US.res
-coll/es_UY.res
-coll/fr.res
-coll/fr_BE.res
-coll/fr_CA.res
-coll/fr_CH.res
-coll/fr_FR.res
-coll/fr_LU.res
-coll/it.res
-coll/it_CH.res
-coll/it_IT.res
-coll/nl.res
-coll/nl_BE.res
-coll/nl_NL.res
-coll/pl.res
-coll/pl_PL.res
-
-curr/res_index.res
-curr/root.res
-curr/pool.res
-curr/supplementalData.res
-curr/cs.res
-curr/de.res
-curr/de_BE.res
-curr/de_LU.res
-curr/en.res
-curr/en_AU.res
-curr/en_BE.res
-curr/en_BW.res
-curr/en_BZ.res
-curr/en_CA.res
-curr/en_HK.res
-curr/en_JM.res
-curr/en_MT.res
-curr/en_NA.res
-curr/en_NZ.res
-curr/en_PH.res
-curr/en_PK.res
-curr/en_RH.res
-curr/en_SG.res
-curr/en_TT.res
-curr/en_ZW.res
-curr/es.res
-curr/es_AR.res
-curr/es_CL.res
-curr/es_CO.res
-curr/es_EC.res
-curr/es_GT.res
-curr/es_HN.res
-curr/es_MX.res
-curr/es_PR.res
-curr/es_US.res
-curr/es_UY.res
-curr/fr.res
-curr/fr_CA.res
-curr/fr_LU.res
-curr/it.res
-curr/nl.res
-curr/pl.res
-
-lang/res_index.res
-lang/pool.res
-lang/root.res
-lang/cs.res
-lang/de.res
-lang/de_CH.res
-lang/en.res
-lang/en_RH.res
-lang/en_ZW.res
-lang/es.res
-lang/es_AR.res
-lang/es_CL.res
-lang/fr.res
-lang/it.res
-lang/nl.res
-lang/nl_BE.res
-lang/pl.res
-
-region/res_index.res
-region/pool.res
-region/root.res
-region/cs.res
-region/de.res
-region/de_CH.res
-region/en.res
-region/en_RH.res
-region/en_ZW.res
-region/es.res
-region/es_CL.res
-region/fr.res
-region/it.res
-region/nl.res
-region/nl_BE.res
-region/pl.res
-
-zone/res_index.res
-zone/pool.res
-zone/root.res
-zone/cs.res
-zone/de.res
-zone/en.res
-zone/en_AU.res
-zone/en_CA.res
-zone/en_GB.res
-zone/en_NZ.res
-zone/en_RH.res
-zone/en_ZA.res
-zone/en_ZW.res
-zone/es.res
-zone/fr.res
-zone/fr_CA.res
-zone/it.res
-zone/nl.res
-zone/pl.res
-
-res_index.res
-pool.res
-root.res
-cs.res
-cs_CZ.res
-de.res
-de_AT.res
-de_BE.res
-de_CH.res
-de_DE.res
-de_LI.res
-de_LU.res
-en.res
-en_AU.res
-en_BE.res
-en_BW.res
-en_BZ.res
-en_CA.res
-en_GB.res
-en_HK.res
-en_IE.res
-en_IN.res
-en_JM.res
-en_MH.res
-en_MT.res
-en_NA.res
-en_NZ.res
-en_PH.res
-en_PK.res
-en_RH.res
-en_SG.res
-en_TT.res
-en_US.res
-en_US_POSIX.res
-en_VI.res
-en_ZA.res
-en_ZW.res
-es.res
-es_AR.res
-es_CL.res
-es_CO.res
-es_EC.res
-es_ES.res
-es_GT.res
-es_HN.res
-es_MX.res
-es_PR.res
-es_US.res
-es_UY.res
-fr.res
-fr_BE.res
-fr_CA.res
-fr_CH.res
-fr_FR.res
-fr_LU.res
-fr_MC.res
-it.res
-it_CH.res
-it_IT.res
-nl.res
-nl_BE.res
-nl_NL.res
-pl.res
-pl_PL.res
-
-gsm-03.38-2000.cnv
-ibm-1089_P100-1995.cnv
-ibm-1383_P110-1999.cnv
-ibm-5012_P100-1999.cnv
-ibm-5346_P100-1998.cnv
-ibm-5347_P100-1998.cnv
-ibm-5348_P100-1997.cnv
-ibm-5349_P100-1998.cnv
-ibm-5350_P100-1998.cnv
-ibm-5354_P100-1998.cnv
-ibm-864_X110-1999.cnv
-ibm-878_P100-1996.cnv
-ibm-9005_X110-2007.cnv
-ibm-912_P100-1995.cnv
-ibm-913_P100-2000.cnv
-ibm-914_P100-1995.cnv
-ibm-915_P100-1995.cnv
-ibm-920_P100-1995.cnv
-ibm-921_P100-1995.cnv
-ibm-923_P100-1998.cnv
-ibm-943_P15A-2003.cnv
-ibm-9447_P100-2002.cnv
-ibm-9448_X100-2005.cnv
-ibm-9449_P100-2002.cnv
-ibm-954_P101-2007.cnv
-iso-8859_10-1998.cnv
-iso-8859_11-2001.cnv
-iso-8859_14-1998.cnv
-iso-8859_16-2001.cnv
-macos-0_2-10.2.cnv
-macos-7_3-10.2.cnv
-windows-874-2000.cnv
-windows-936-2000.cnv
-windows-949-2000.cnv
-windows-950-2000.cnv
-noop-cns-11643.cnv
-noop-gb2312_gl.cnv
-noop-iso-ir-165.cnv
-ibm-1168_P100-2002.cnv
-
-cnvalias.icu
-pnames.icu
-supplementalData.res
-numberingSystems.res
-plurals.res
-metaZones.res
-rfc3491.spp
-nfc.nrm
-nfkc.nrm
-timezoneTypes.res
-zoneinfo64.res
-icustd.res
-icuver.res
-keyTypeData.res
-
-
-
-