make_unicode_tables.awk is now UnicodeTablesGenerator

UnicodeTablesGenerator uses Unicode data from ICU4J to generate Unicode
tables for consumption by RE2/J. Output is google-java-formatted before
it is written.

No new runtime dependencies are added to RE2/J.

The generator uses ICU4J 4.8.2 which bundles Unicode 6.0.0. This keeps
it compatible with Java 8, which RE2/J targets. Consideration should be
given for how we might upgrade to later Unicode versions without
introducing inconsistencies (e.g. RE2/J matches something that shouldn't
match according to java.lang.Character data).

There are some differences in the generated tables:
  * the new tables do not contain binary property character ranges (e.g.
    ASCII_Hex_digit), as those tables are currently unused in RE2/J.

  * Cc (control) char class now contains NUL (u+0000), this is correct
    and was also the subject of #26.

See https://github.com/google/re2j/files/4725343/diff.txt for a full
list of differences between the old tables and the new.

java/com/google/re2j/Unicode.java

@@ -75,23 +75,6 @@ static boolean isUpper(int r) {
     return is(UnicodeTables.Upper, r);
   }
 
-  // isLower reports whether the rune is a lower case letter.
-  static boolean isLower(int r) {
-    // See comment in isGraphic.
-    if (r <= MAX_LATIN1) {
-      return Character.isLowerCase((char) r);
-    }
-    return is(UnicodeTables.Lower, r);
-  }
-
-  // isTitle reports whether the rune is a title case letter.
-  static boolean isTitle(int r) {
-    if (r <= MAX_LATIN1) {
-      return false;
-    }
-    return is(UnicodeTables.Title, r);
-  }
-
   // isPrint reports whether the rune is printable (Unicode L/M/N/P/S or ' ').
   static boolean isPrint(int r) {
     if (r <= MAX_LATIN1) {

javatests/com/google/re2j/ParserTest.java

@@ -23,6 +23,7 @@
 import java.util.EnumMap;
 import java.util.Map;
 
+import com.google.common.truth.Truth;
 import org.junit.Test;
 
 /**
@@ -286,7 +287,7 @@ public boolean applies(int r) {
   // - Java UTF-16 things.
 
   @Test
-  public void testParseSimple() throws Exception {
+  public void testParseSimple() {
     testParseDump(PARSE_TESTS, TEST_FLAGS);
   }
 
@@ -346,9 +347,7 @@ private void testParseDump(String[][] tests, int flags) {
       try {
         Regexp re = Parser.parse(test[0], flags);
         String d = dump(re);
-        if (!test[1].equals(d)) {
-          fail(String.format("parse/dump of " + test[0] + " expected " + test[1] + ", got " + d));
-        }
+        Truth.assertWithMessage("parse/dump of " + test[0]).that(d).isEqualTo(test[1]);
       } catch (PatternSyntaxException e) {
         throw new RuntimeException("Parsing failed: " + test[0], e);
       }

settings.gradle

@@ -1,3 +1,4 @@
 rootProject.name = 're2j'
 
 include ':benchmarks'
+include ':unicode'

unicode/README.md

@@ -0,0 +1,9 @@
+Utilities for emitting Unicode tables used by RE2J.
+
+To rebuild the Unicode tables, run:
+
+```
+./gradlew :unicode:run -q > java/com/google/re2j/UnicodeTables.java
+```
+
+from the project root directory.

unicode/build.gradle

@@ -0,0 +1,17 @@
+plugins {
+  id 'java'
+  id 'application'
+}
+
+mainClassName = 'com.google.re2j.UnicodeTablesGenerator'
+
+repositories {
+  mavenCentral()
+}
+
+dependencies {
+  compile 'com.google.googlejavaformat:google-java-format:1.0'
+  compile 'com.squareup:javapoet:1.12.1'
+  compile 'com.ibm.icu:icu4j:4.8.2'
+  compile 'com.google.guava:guava:29.0-jre'
+}