Move Trino RE2J fork to a new package

There are situations where downstream dependencies require com.google.re2j:1.7 and we pull in both dependencies due to a dependence on Trino parts.

These conflicts cannot be resolved if both dependencies use the same package.

java/com/google/re2j/CharClass.java → java/io/trino/re2j/CharClass.java

@@ -5,7 +5,7 @@
 // Original Go source here:
 // http://code.google.com/p/go/source/browse/src/pkg/regexp/syntax/parse.go
 
-package com.google.re2j;
+package io.trino.re2j;
 
 /**
  * A "builder"-style helper class for manipulating character classes

java/com/google/re2j/CharGroup.java → java/io/trino/re2j/CharGroup.java

@@ -1,7 +1,7 @@
 // GENERATED BY make_perl_groups.pl; DO NOT EDIT.
 // make_perl_groups.pl >perl_groups.go
 
-package com.google.re2j;
+package io.trino.re2j;
 
 import java.util.HashMap;
 

java/com/google/re2j/Compiler.java → java/io/trino/re2j/Compiler.java

@@ -5,19 +5,19 @@
 // Original Go source here:
 // http://code.google.com/p/go/source/browse/src/pkg/regexp/syntax/compile.go
 
-package com.google.re2j;
+package io.trino.re2j;
 
 import java.util.LinkedList;
 import java.util.List;
 
-import static com.google.re2j.Inst.Op.BYTE;
-import static com.google.re2j.Inst.Op.BYTE1;
-import static com.google.re2j.RE2.FOLD_CASE;
-import static com.google.re2j.Unicode.RUNE_SELF;
-import static com.google.re2j.Unicode.UTF_MAX;
-import static com.google.re2j.Unicode.codePointToUtf8;
-import static com.google.re2j.Unicode.maxRune;
-import static com.google.re2j.Unicode.simpleFold;
+import static io.trino.re2j.Inst.Op.BYTE;
+import static io.trino.re2j.Inst.Op.BYTE1;
+import static io.trino.re2j.RE2.FOLD_CASE;
+import static io.trino.re2j.Unicode.RUNE_SELF;
+import static io.trino.re2j.Unicode.UTF_MAX;
+import static io.trino.re2j.Unicode.codePointToUtf8;
+import static io.trino.re2j.Unicode.maxRune;
+import static io.trino.re2j.Unicode.simpleFold;
 
 /**
  * Compiler from {@code Regexp} (RE2 abstract syntax) to {@code RE2} (compiled regular expression).

java/com/google/re2j/DFA.java → java/io/trino/re2j/DFA.java

@@ -5,26 +5,20 @@
 // Original RE2 source here:
 // https://github.com/google/re2/blob/master/re2/dfa.cc
 
-package com.google.re2j;
-
-import com.google.re2j.RE2.MatchKind;
+package io.trino.re2j;
 
 import java.util.concurrent.ConcurrentHashMap;
 import java.util.concurrent.atomic.AtomicInteger;
 
-import static com.google.re2j.DFAState.DEAD_STATE;
-import static com.google.re2j.Inst.Op.EMPTY_WIDTH;
-import static com.google.re2j.MachineInput.EOF;
-import static com.google.re2j.RE2.MatchKind.FIRST_MATCH;
-import static com.google.re2j.RE2.MatchKind.LONGEST_MATCH;
-import static com.google.re2j.Utils.EMPTY_BEGIN_LINE;
-import static com.google.re2j.Utils.EMPTY_BEGIN_TEXT;
-import static com.google.re2j.Utils.EMPTY_END_LINE;
-import static com.google.re2j.Utils.EMPTY_END_TEXT;
-import static com.google.re2j.Utils.EMPTY_NO_WORD_BOUNDARY;
-import static com.google.re2j.Utils.EMPTY_WORD_BOUNDARY;
-import static com.google.re2j.Utils.isRuneStart;
-import static com.google.re2j.Utils.isWordByte;
+import static io.trino.re2j.MachineInput.EOF;
+import static io.trino.re2j.Utils.EMPTY_BEGIN_LINE;
+import static io.trino.re2j.Utils.EMPTY_BEGIN_TEXT;
+import static io.trino.re2j.Utils.EMPTY_END_LINE;
+import static io.trino.re2j.Utils.EMPTY_END_TEXT;
+import static io.trino.re2j.Utils.EMPTY_NO_WORD_BOUNDARY;
+import static io.trino.re2j.Utils.EMPTY_WORD_BOUNDARY;
+import static io.trino.re2j.Utils.isRuneStart;
+import static io.trino.re2j.Utils.isWordByte;
 import static java.util.Arrays.sort;
 
 class DFA {
@@ -51,7 +45,7 @@ private DFATooManyStatesException() {
   private static final int START_PARAMS_CACHE_SIZE = 1 << 13;
   private static final int START_PARAMS_CACHE_SHIFT = 12;
 
-  private static final StartParams DEAD_START_PARAMS = new StartParams(DEAD_STATE, new boolean[256]);
+  private static final StartParams DEAD_START_PARAMS = new StartParams(DFAState.DEAD_STATE, new boolean[256]);
 
   // Info for the search
   private final Prog prog;
@@ -60,7 +54,7 @@ private DFATooManyStatesException() {
   private final Inst[] instructions;
 
   // Search for longest match like egrep or POSIX or for first match like Perl, PCRE
-  private final MatchKind matchKind;
+  private final RE2.MatchKind matchKind;
 
   // Should input bytes be read forward or backward
   private final boolean runForward;
@@ -74,7 +68,7 @@ private DFATooManyStatesException() {
   private final ConcurrentHashMap<DFAStateKey, DFAState> statesCache;
   private final AtomicInteger availableStates;
 
-  public DFA(Prog prog, MatchKind matchKind, boolean reversed,
+  public DFA(Prog prog, RE2.MatchKind matchKind, boolean reversed,
              ConcurrentHashMap<DFAStateKey, DFAState> statesCache, AtomicInteger availableStates) {
     this.prog = prog;
     this.instructions = prog.getInst();
@@ -85,7 +79,7 @@ public DFA(Prog prog, MatchKind matchKind, boolean reversed,
 
     int progSize = prog.numInst();
     int nMarks = 0;
-    if (matchKind == LONGEST_MATCH) {
+    if (matchKind == RE2.MatchKind.LONGEST_MATCH) {
       nMarks = progSize;
     }
 
@@ -130,7 +124,7 @@ private DFAState workQueueToCachedState(WorkQueue q, int flag) {
         case MATCH:
         case ALT:
           instIndexes[nIndexes++] = instIndex;
-          if (inst.op() == EMPTY_WIDTH) {
+          if (inst.op() == Inst.Op.EMPTY_WIDTH) {
             neededFlags |= inst.arg;
           }
           break;
@@ -153,13 +147,13 @@ private DFAState workQueueToCachedState(WorkQueue q, int flag) {
 
     // No match possibilities
     if (nIndexes == 0 && flag == 0) {
-      return DEAD_STATE;
+      return DFAState.DEAD_STATE;
     }
 
     // If we're in the longest match mode, the state is a sequence of
     // unordered state sets separated by Marks. Sort each set to
     // canonicalize, to reduce the number of distinct sets stored.
-    if (matchKind == LONGEST_MATCH) {
+    if (matchKind == RE2.MatchKind.LONGEST_MATCH) {
       int ip = 0;
       while (ip < nIndexes) {
         int markp = ip;
@@ -322,7 +316,7 @@ private boolean runWorkQueueOnByte(byte b, int flag) {
           break;
         case MATCH:
           isMatch = true;
-          if (matchKind == FIRST_MATCH) {
+          if (matchKind == RE2.MatchKind.FIRST_MATCH) {
             return true;
           }
           break;

java/com/google/re2j/DFAMachine.java → java/io/trino/re2j/DFAMachine.java

@@ -5,18 +5,12 @@
 // Original RE2 source here:
 // https://github.com/google/re2/blob/master/re2/dfa.cc
 
-package com.google.re2j;
-
-import com.google.re2j.RE2.Anchor;
-import com.google.re2j.RE2.MatchKind;
+package io.trino.re2j;
 
 import java.util.concurrent.ConcurrentHashMap;
 import java.util.concurrent.atomic.AtomicInteger;
 
-import static com.google.re2j.DFA.NO_MATCH;
-import static com.google.re2j.RE2.Anchor.ANCHOR_START;
-import static com.google.re2j.RE2.MatchKind.FIRST_MATCH;
-import static com.google.re2j.RE2.MatchKind.LONGEST_MATCH;
+import static io.trino.re2j.DFA.NO_MATCH;
 
 /**
  * A {@link Machine} implementation using a DFA.
@@ -40,14 +34,14 @@ class DFAMachine implements Machine {
       stateCache[i] = new ConcurrentHashMap<>();
     }
 
-    setDfaThreadLocal(LONGEST_MATCH, true);
-    setDfaThreadLocal(LONGEST_MATCH, false);
-    setDfaThreadLocal(FIRST_MATCH, true);
-    setDfaThreadLocal(FIRST_MATCH, false);
+    setDfaThreadLocal(RE2.MatchKind.LONGEST_MATCH, true);
+    setDfaThreadLocal(RE2.MatchKind.LONGEST_MATCH, false);
+    setDfaThreadLocal(RE2.MatchKind.FIRST_MATCH, true);
+    setDfaThreadLocal(RE2.MatchKind.FIRST_MATCH, false);
   }
 
   @Override
-  public boolean match(MachineInput in, int pos, Anchor anchor, int[] submatches) {
+  public boolean match(MachineInput in, int pos, RE2.Anchor anchor, int[] submatches) {
     // Don't ask for the location if we won't use it. SearchDFA can do extra optimizations in that case.
     boolean wantMatchPosition = true;
     if (submatches.length == 0) {
@@ -71,7 +65,7 @@ public boolean match(MachineInput in, int pos, Anchor anchor, int[] submatches)
 
         // SearchDFA gives match end position but we don't know where the match started. Run the
         // regexp backwards from end position to find the longest possible match -- that's where it started.
-        matchStart = searchDFA(in, pos, matchEnd, ANCHOR_START, true, LONGEST_MATCH, true);
+        matchStart = searchDFA(in, pos, matchEnd, RE2.Anchor.ANCHOR_START, true, RE2.MatchKind.LONGEST_MATCH, true);
         if (matchStart == NO_MATCH) {
           throw new IllegalStateException("reverse DFA did not found a match");
         }
@@ -101,7 +95,7 @@ public boolean match(MachineInput in, int pos, Anchor anchor, int[] submatches)
     return true;
   }
 
-  private int searchDFA(MachineInput in, int startPos, int endPos, Anchor anchor, boolean wantMatchPosition, MatchKind matchKind, boolean reversed) {
+  private int searchDFA(MachineInput in, int startPos, int endPos, RE2.Anchor anchor, boolean wantMatchPosition, RE2.MatchKind matchKind, boolean reversed) {
     boolean hasCarat = reversed ? anchor.isAnchorEnd() : anchor.isAnchorStart();
     if (hasCarat && startPos != 0) {
       return NO_MATCH;
@@ -112,7 +106,7 @@ private int searchDFA(MachineInput in, int startPos, int endPos, Anchor anchor,
     boolean endMatch = false;
     if (anchor.isAnchorEnd()) {
       endMatch = true;
-      matchKind = LONGEST_MATCH;
+      matchKind = RE2.MatchKind.LONGEST_MATCH;
     }
 
     // If the caller doesn't care where the match is (just whether one exists),
@@ -121,7 +115,7 @@ private int searchDFA(MachineInput in, int startPos, int endPos, Anchor anchor,
     boolean wantEarliestMatch = false;
     if (!wantMatchPosition && !endMatch) {
       wantEarliestMatch = true;
-      matchKind = LONGEST_MATCH;
+      matchKind = RE2.MatchKind.LONGEST_MATCH;
     }
 
     DFA dfa = getDfa(matchKind, reversed);
@@ -140,17 +134,17 @@ private int searchDFA(MachineInput in, int startPos, int endPos, Anchor anchor,
     return match;
   }
 
-  private DFA getDfa(MatchKind matchKind, boolean reversed) {
+  private DFA getDfa(RE2.MatchKind matchKind, boolean reversed) {
     return dfaCache[dfaKey(matchKind, reversed)].get();
   }
 
-  private int dfaKey(MatchKind matchKind, boolean reversed) {
-    int longestInt = matchKind == LONGEST_MATCH ? 1 : 0;
+  private int dfaKey(RE2.MatchKind matchKind, boolean reversed) {
+    int longestInt = matchKind == RE2.MatchKind.LONGEST_MATCH ? 1 : 0;
     int reversedInt = reversed ? 1 : 0;
     return longestInt | (reversedInt << 1);
   }
 
-  private void setDfaThreadLocal(MatchKind matchKind, boolean reversed) {
+  private void setDfaThreadLocal(RE2.MatchKind matchKind, boolean reversed) {
     int dfaKey = dfaKey(matchKind, reversed);
     Prog prog = reversed ? re2.reverseProg : re2.prog;
     dfaCache[dfaKey] = new ThreadLocal<DFA>() {

java/com/google/re2j/DFAState.java → java/io/trino/re2j/DFAState.java

@@ -5,11 +5,11 @@
 // Original RE2 source here:
 // https://github.com/google/re2/blob/master/re2/dfa.cc
 
-package com.google.re2j;
+package io.trino.re2j;
 
-import static com.google.re2j.DFA.FLAG_MATCH;
-import static com.google.re2j.DFAState.StateType.DEAD;
-import static com.google.re2j.DFAState.StateType.REGULAR;
+import static io.trino.re2j.DFA.FLAG_MATCH;
+import static io.trino.re2j.DFAState.StateType.DEAD;
+import static io.trino.re2j.DFAState.StateType.REGULAR;
 import static java.lang.System.arraycopy;
 
 final class DFAState {

java/com/google/re2j/DFAStateKey.java → java/io/trino/re2j/DFAStateKey.java

@@ -5,11 +5,11 @@
 // Original RE2 source here:
 // https://github.com/google/re2/blob/master/re2/dfa.cc
 
-package com.google.re2j;
+package io.trino.re2j;
 
 import java.util.Arrays;
 
-import static com.google.re2j.Utils.arrayFirstElementsEqual;
+import static io.trino.re2j.Utils.arrayFirstElementsEqual;
 
 final class DFAStateKey {
   private final int[] instIndexes;

java/com/google/re2j/Inst.java → java/io/trino/re2j/Inst.java

@@ -5,9 +5,9 @@
 // Original Go source here:
 // http://code.google.com/p/go/source/browse/src/pkg/regexp/syntax/prog.go
 
-package com.google.re2j;
+package io.trino.re2j;
 
-import static com.google.re2j.Inst.Op.BYTE;
+import static io.trino.re2j.Inst.Op.BYTE;
 
 /**
  * A single instruction in the regular expression virtual machine.

java/com/google/re2j/Machine.java → java/io/trino/re2j/Machine.java

@@ -5,9 +5,9 @@
 // Original Go source here:
 // http://code.google.com/p/go/source/browse/src/pkg/regexp/exec.go
 
-package com.google.re2j;
+package io.trino.re2j;
 
-import com.google.re2j.RE2.Anchor;
+import io.trino.re2j.RE2.Anchor;
 
 /**
  * A Machine matches an input string of Unicode characters against an RE2 instance.

java/com/google/re2j/MachineInput.java → java/io/trino/re2j/MachineInput.java

@@ -5,7 +5,7 @@
 // Original Go source here:
 // http://code.google.com/p/go/source/browse/src/pkg/regexp/regexp.go
 
-package com.google.re2j;
+package io.trino.re2j;
 
 import io.airlift.slice.Slice;
 

java/com/google/re2j/Matcher.java → java/io/trino/re2j/Matcher.java

@@ -1,17 +1,11 @@
 // Copyright 2010 Google Inc. All Rights Reserved.
 
-package com.google.re2j;
-
-import com.google.re2j.RE2.Anchor;
+package io.trino.re2j;
 
 import io.airlift.slice.DynamicSliceOutput;
 import io.airlift.slice.Slice;
 import io.airlift.slice.SliceOutput;
 
-import static com.google.re2j.RE2.Anchor.ANCHOR_BOTH;
-import static com.google.re2j.RE2.Anchor.ANCHOR_START;
-import static com.google.re2j.RE2.Anchor.UNANCHORED;
-
 /**
  * A stateful iterator that interprets a regex {@code Pattern} on a
  * specific input.  Its interface mimics the JDK 1.4.2
@@ -64,7 +58,7 @@ public final class Matcher {
   private boolean hasGroups;
 
   // The anchor flag to use when repeating the match to find subgroups.
-  private Anchor anchorFlag;
+  private RE2.Anchor anchorFlag;
 
   private Matcher(Pattern pattern) {
     if (pattern == null) {
@@ -236,7 +230,7 @@ private void loadGroup(int group) {
    * @return true if the entire input matches the pattern
    */
   public boolean matches() {
-    return genMatch(0, ANCHOR_BOTH);
+    return genMatch(0, RE2.Anchor.ANCHOR_BOTH);
   }
 
   /**
@@ -246,7 +240,7 @@ public boolean matches() {
    * @return true if the beginning of the input matches the pattern
    */
   public boolean lookingAt() {
-    return genMatch(0, ANCHOR_START);
+    return genMatch(0, RE2.Anchor.ANCHOR_START);
   }
 
   /**
@@ -265,7 +259,7 @@ public boolean find() {
         start++;
       }
     }
-    return genMatch(start, UNANCHORED);
+    return genMatch(start, RE2.Anchor.UNANCHORED);
   }
 
   /**
@@ -283,11 +277,11 @@ public boolean find(int start) {
           "start index out of bounds: " + start);
     }
     reset();
-    return genMatch(start, UNANCHORED);
+    return genMatch(start, RE2.Anchor.UNANCHORED);
   }
 
   /** Helper: does match starting at start, with RE2 anchor flag. */
-  private boolean genMatch(int startByte, Anchor anchor) {
+  private boolean genMatch(int startByte, RE2.Anchor anchor) {
     // TODO(rsc): Is matches/lookingAt supposed to reset the append or input positions?
     // From the JDK docs, looks like no.
     boolean ok = pattern.re2().match(input, startByte,