javazic-1.8: Remove HST etc. from zones and links (RHEL-59542)

Previously, some of the zones were only removed from the zones
map, not the map of links.  Upstream tzdata added a link involving
HST in commit a0b09c0230089252acf2eb0f1ba922e99f7f4a03
("Mark CET, CST6CDT etc. as obsolescent").

During file output, fail with an error if a region cannot be
found, instead of creating a corrupt file.

The zone removal code in the javazic compiler is different.
It handles only GMT-related zones and already applies the
removal (actually, skipping during parsing) to both Zone
and Link records.

Resolves: RHEL-59542

ZoneTest.java

@@ -0,0 +1,26 @@
+/* Smoke test to ensure that tzdb.data can be loaded. */
+
+import java.time.zone.ZoneRulesProvider;
+import java.util.Calendar;
+import java.util.GregorianCalendar;
+import java.util.Locale;
+import java.util.Set;
+import java.util.TimeZone;
+
+public class ZoneTest {
+    public static void main(String[] args) {
+        // This is what failed in OpenJDK's build.tools.cldrconverter.
+        new GregorianCalendar(TimeZone.getTimeZone("America/Los_Angeles"),
+            Locale.US).get(Calendar.YEAR);
+
+        // In some OpenJDK versions, this exercises a different parser.
+        Set<String> available = ZoneRulesProvider.getAvailableZoneIds();
+        boolean errors = false;
+        if (available.contains("ROC"))
+            System.out.println("error: ROC zone is present");
+        if (!available.contains("America/New_York"))
+            System.out.println("error: America/New_York is missing");
+        if (errors)
+            System.exit(1);
+    }
+}

javazic-harden-links.patch

@@ -0,0 +1,76 @@
+diff -ur tzdata-2024b.orig/javazic-1.8/build/tools/tzdb/TzdbZoneRulesCompiler.java tzdata-2024b/javazic-1.8/build/tools/tzdb/TzdbZoneRulesCompiler.java
+--- tzdata-2024b.orig/javazic-1.8/build/tools/tzdb/TzdbZoneRulesCompiler.java	2014-04-22 19:46:49.000000000 +0200
++++ tzdata-2024b/javazic-1.8/build/tools/tzdb/TzdbZoneRulesCompiler.java	2024-09-20 21:10:12.748483767 +0200
+@@ -248,7 +248,7 @@
+             // link version-region-rules
+             out.writeShort(builtZones.size());
+             for (Map.Entry<String, ZoneRules> entry : builtZones.entrySet()) {
+-                 int regionIndex = Arrays.binarySearch(regionArray, entry.getKey());
++                 int regionIndex = findRegionIndex(regionArray, entry.getKey());
+                  int rulesIndex = rulesList.indexOf(entry.getValue());
+                  out.writeShort(regionIndex);
+                  out.writeShort(rulesIndex);
+@@ -256,8 +256,8 @@
+             // alias-region
+             out.writeShort(links.size());
+             for (Map.Entry<String, String> entry : links.entrySet()) {
+-                 int aliasIndex = Arrays.binarySearch(regionArray, entry.getKey());
+-                 int regionIndex = Arrays.binarySearch(regionArray, entry.getValue());
++                 int aliasIndex = findRegionIndex(regionArray, entry.getKey());
++                 int regionIndex = findRegionIndex(regionArray, entry.getValue());
+                  out.writeShort(aliasIndex);
+                  out.writeShort(regionIndex);
+             }
+@@ -269,6 +269,14 @@
+         }
+     }
+ 
++    private static int findRegionIndex(String[] regionArray, String region) {
++        int index = Arrays.binarySearch(regionArray, region);
++        if (index < 0) {
++            throw new IllegalArgumentException("Unknown region: " + region);
++        }
++        return index;
++    }
++
+     private static final Pattern YEAR = Pattern.compile("(?i)(?<min>min)|(?<max>max)|(?<only>only)|(?<year>[0-9]+)");
+     private static final Pattern MONTH = Pattern.compile("(?i)(jan)|(feb)|(mar)|(apr)|(may)|(jun)|(jul)|(aug)|(sep)|(oct)|(nov)|(dec)");
+     private static final Matcher DOW = Pattern.compile("(?i)(mon)|(tue)|(wed)|(thu)|(fri)|(sat)|(sun)").matcher("");
+@@ -607,22 +615,20 @@
+             }
+             builtZones.put(aliasId, realRules);
+         }
+-        // remove UTC and GMT
+-        // builtZones.remove("UTC");
+-        // builtZones.remove("GMT");
+-        // builtZones.remove("GMT0");
+-        builtZones.remove("GMT+0");
+-        builtZones.remove("GMT-0");
+-        links.remove("GMT+0");
+-        links.remove("GMT-0");
+-        // remove ROC, which is not supported in j.u.tz
+-        builtZones.remove("ROC");
+-        links.remove("ROC");
+-        // remove EST, HST and MST. They are supported via
+-        // the short-id mapping
+-        builtZones.remove("EST");
+-        builtZones.remove("HST");
+-        builtZones.remove("MST");
++
++        List<String> zonesToRemove = Arrays.asList(
++            // remove UTC and GMT
++            "GMT+0",
++            "GMT-0",
++            // remove ROC, which is not supported in j.u.tz
++            "ROC",
++            // remove EST, HST and MST. They are supported via
++            // the short-id mapping
++            "EST",
++            "HST",
++            "MST");
++        builtZones.keySet().removeAll(zonesToRemove);
++        links.keySet().removeAll(zonesToRemove);
+     }
+ 
+     /**
+

tzdata.spec

@@ -3,7 +3,7 @@ Name: tzdata
 Version: 2024b
 %define tzdata_version 2024b
 %define tzcode_version 2024b
-Release: 1%{?dist}
+Release: 2%{?dist}
 License: Public Domain
 Group: System Environment/Base
 URL: https://www.iana.org/time-zones
@@ -30,12 +30,14 @@ Summary: Timezone data for Java
 Group: System Environment/Base
 Source3: javazic.tar.gz
 Source4: javazic-1.8-37392f2f5d59.tar.xz
+Source5: ZoneTest.java
 Patch100: javazic-fixup.patch
 Patch101: rebase-01.patch
 Patch102: rebase-02.patch
 Patch103: 7090844.patch
 Patch104: 7133138.patch
 Patch105: 8051641.patch
+Patch106: javazic-harden-links.patch
 
 %description java
 This package contains timezone information for use by Java runtimes.
@@ -90,6 +92,7 @@ popd
 
 tar xf %{SOURCE4}
 %patch105
+%patch106 -p1
 
 echo "%{name}%{tzdata_version}" >> VERSION
 
@@ -136,6 +139,20 @@ install -p -m 644 tzdb.dat $RPM_BUILD_ROOT%{_datadir}/javazi-1.8/
 %check
 echo ============TESTING===============
 /usr/bin/env LANG=C make -k VALIDATE=':' check && true
+
+# Create a custom JAVA_HOME, where we can replace tzdb.dat with the
+# one just built, for testing.
+system_java_home=$(dirname $(readlink -f $(which java)))/..
+mkdir -p java_home
+cp -Lr $system_java_home/* java_home/.
+for tzdb in $(find java_home -name tzdb.dat) ; do
+    rm $tzdb
+    cp $RPM_BUILD_ROOT%{_datadir}/javazi-1.8/tzdb.dat $tzdb
+done
+# Compile the smoke test and run it.
+cp %{SOURCE5} .
+javac ZoneTest.java
+java_home/bin/java ZoneTest
 echo ============END TESTING===========
 
 %files
@@ -153,6 +170,9 @@ echo ============END TESTING===========
 %{_datadir}/javazi-1.8
 
 %changelog
+* Fri Sep 20 2024 Florian Weimer <fweimer@redhat.com> - 2024b-2
+- Harden against links to removed zones (RHEL-59542)
+
 * Wed Sep 11 2024 Patsy Griffin <patsy@redhat.com> - 2024b-1
 - Update to tzdata-2024b
   - Improve historical data for Mexico, Mongolia, and Portugal.