ICU-10521 Allow tz2icu to merge supplemental zone-region mapping data. Also removed unused tz.alias.

X-SVN-Rev: 34623

.gitattributes

@@ -215,6 +215,7 @@ icu4c/source/tools/makeconv/makeconv.vcxproj.filters -text
 icu4c/source/tools/pkgdata/pkgdata.vcxproj -text
 icu4c/source/tools/pkgdata/pkgdata.vcxproj.filters -text
 icu4c/source/tools/toolutil/toolutil.vcxproj -text
+icu4c/source/tools/tzcode/icuregions -text
 icu4j/build.properties -text
 icu4j/demos/.settings/org.eclipse.core.resources.prefs -text
 icu4j/demos/manifest.stub -text

icu4c/source/tools/tzcode/icuregions

@@ -0,0 +1,11 @@
+######################################################################
+# Copyright (C) 2013, International Business Machines
+# Corporation and others.  All Rights Reserved.
+######################################################################
+# This is an ICU-specific file including zone/region mapping.
+#
+# Each line below indicates zone and its region in the syntax below -
+# <zone_id> <region_code>
+#
+America/Montreal CA
+

icu4c/source/tools/tzcode/tz.alias

@@ -1,56 +0,0 @@
-#
-# !!!! WARNING !!!!!
-#
-# This file is OBSOLLETE and no longer used by ICU tzdata build tool.
-# Use the file 'icuzones' to specify backward compatibility aliases.
-#
-# 2007-04-03 Yoshito Umaoka
-#
-
-######################################################################
-# Copyright (C) 1999-2007, International Business Machines
-# Corporation and others.  All Rights Reserved.
-######################################################################
-# A simple alias list. We use this to retain backward compatibility.
-# For example, ICU has always defined the zone name "PST" to indicate
-# the zone America/Los_Angeles. Unless we continue to have a zone with
-# this ID, legacy code may break.
-# 
-# This list is read by tz2icu to incorporate legacy ICU zone aliases
-# into the ICU system zone data.
-#
-# Format: alias_name unix_name # optional comment
-
-#### Aliases that conflict with Olson compatibility Zone definition
-
-ACT Australia/Darwin
-AET Australia/Sydney
-AGT America/Buenos_Aires
-ART Africa/Cairo
-AST America/Anchorage
-BET America/Sao_Paulo
-BST Asia/Dhaka # spelling changed in 2000h; was Asia/Dacca
-CAT Africa/Harare
-CNT America/St_Johns
-CST America/Chicago
-CTT Asia/Shanghai
-EAT Africa/Addis_Ababa
-ECT Europe/Paris
-# EET Europe/Istanbul # EET is a standard UNIX zone
-#### EST America/New_York # Defined as -05:00
-#### HST Pacific/Honolulu # Defined as -10:00
-IET America/Indianapolis
-IST Asia/Calcutta
-JST Asia/Tokyo
-# MET Asia/Tehran # MET is a standard UNIX zone
-MIT Pacific/Apia
-#### MST America/Denver # Defined as -07:00
-NET Asia/Yerevan
-NST Pacific/Auckland
-PLT Asia/Karachi
-PNT America/Phoenix
-PRT America/Puerto_Rico
-PST America/Los_Angeles
-SST Pacific/Guadalcanal
-# UTC Etc/UTC # Olson LINK
-VST Asia/Saigon

icu4c/source/tools/tzcode/tz2icu.cpp

@@ -1,7 +1,7 @@
 
 /*
 **********************************************************************
-* Copyright (c) 2003-2010, International Business Machines
+* Copyright (c) 2003-2013, International Business Machines
 * Corporation and others.  All Rights Reserved.
 **********************************************************************
 * Author: Alan Liu
@@ -1581,8 +1581,36 @@ int main(int argc, char *argv[]) {
     }
 
     // Create the country map
+    map<string, string> icuRegions;        // ICU's custom zone -> country override
     map<string, set<string> > countryMap;  // country -> set of zones
     map<string, string> reverseCountryMap; // zone -> country
+
+    try {
+        // Read icuregions file to collect ICU's own zone-region mapping data.
+        ifstream frg(ICU_REGIONS);
+        if (frg) {
+            string line;
+            while (getline(frg, line)) {
+                if (line[0] == '#') continue;
+
+                string zone, country;
+                istringstream is(line);
+                is >> zone >> country;
+                if (zone.size() == 0) continue;
+                if (country.size() < 2) {
+                    cerr << "Error: Can't parse " << line << " in " << ICU_REGIONS << endl;
+                    return 1;
+                }
+                icuRegions[zone] = country;
+            }
+        } else {
+            cout << "No custom region map [icuregions]" << endl;
+        }
+    } catch (const exception& error) {
+        cerr << "Error: While reading " << ICU_REGIONS << ": " << error.what() << endl;
+        return 1;
+    }
+
     try {
         ifstream f(zonetab.c_str());
         if (!f) {
@@ -1609,6 +1637,13 @@ int main(int argc, char *argv[]) {
                      << " in " << zonetab << endl;
                 return 1;
             }
+            if (icuRegions.find(zone) != icuRegions.end()) {
+                // Custom override
+                string customCountry = icuRegions[zone];
+                cout << "Region Mapping: custom override for " << zone
+                    << " " << country << " -> " << customCountry << endl;
+                country = customCountry;
+            }
             countryMap[country].insert(zone);
             reverseCountryMap[zone] = country;
             //cerr << (n+1) << ": " << country << " <=> " << zone << endl;
@@ -1621,6 +1656,20 @@ int main(int argc, char *argv[]) {
         return 1;
     }
 
+    // Merge ICU's own zone-region mapping data
+    for (map<string,string>::const_iterator i = icuRegions.begin();
+        i != icuRegions.end(); ++i) {
+        const string& zid(i->first);
+        if (reverseCountryMap.find(zid) != reverseCountryMap.end()) {
+            continue;
+        }
+        cout << "Region Mapping: custom data zone=" << zid
+            << ", region=" << i->second << endl;
+
+        reverseCountryMap[zid] = i->second;
+        countryMap[i->second].insert(zid);
+    }
+
     // Merge ICU aliases into country map.  Don't merge any alias
     // that already has a country map, since that doesn't make sense.
     // E.g.  "Link Europe/Oslo Arctic/Longyearbyen" doesn't mean we

icu4c/source/tools/tzcode/tz2icu.h

@@ -1,6 +1,6 @@
 /*
 **********************************************************************
-* Copyright (c) 2003-2010, International Business Machines
+* Copyright (c) 2003-2013, International Business Machines
 * Corporation and others.  All Rights Reserved.
 **********************************************************************
 * Author: Alan Liu
@@ -31,9 +31,6 @@ typedef unsigned char ICUZoneinfoVersion;
  * zic to communicate final zone data to tz2icu. */
 #define ICU_ZONE_FILE "icu_zone.txt"
 
-/* File containing legacy aliases.  Read by tz2icu. */
-#define ICU_TZ_ALIAS  "tz.alias"
-
 /* Output resource name.  This determines both the file name and the
  * resource name within the file.  That is, the output will be to the
  * file ICU_TZ_RESOURCE ".txt" and the resource within it will be
@@ -41,4 +38,7 @@ typedef unsigned char ICUZoneinfoVersion;
 #define ICU_TZ_RESOURCE_OLD "zoneinfo"
 #define ICU_TZ_RESOURCE "zoneinfo64"
 
+/* File containinng custom zone-region mapping. */
+#define ICU_REGIONS "icuregions"
+
 #endif