String out of bound exception about aho-corasick HOT 4 CLOSED

The problem can also be reproduced by changing the caseInsensitive test case to:

@Test
public void caseInsensitive() {
    Trie trie = new Trie().caseInsensitive().onlyWholeWords();
    trie.addKeyword("turning");
    trie.addKeyword("once");
    trie.addKeyword("again");
    trie.addKeyword("börkü");
    Collection<Emit> emits = trie.parseText("TurninG OnCe AgAiN BÖRKÜ");
    assertEquals(4, emits.size()); // Match must not be made
    Iterator<Emit> it = emits.iterator();
    checkEmit(it.next(), 0, 6, "turning");
    checkEmit(it.next(), 8, 11, "once");
    checkEmit(it.next(), 13, 17, "again");
    checkEmit(it.next(), 19, 23, "börkü");
}

Tests in error:
caseInsensitive(org.ahocorasick.trie.TrieTest): String index out of range: 24

from aho-corasick.

diff --git a/src/main/java/org/ahocorasick/trie/Trie.java b/src/main/java/org/ahocorasick/trie/Trie.java
index 0d28c9b..f25490d 100644
--- a/src/main/java/org/ahocorasick/trie/Trie.java
+++ b/src/main/java/org/ahocorasick/trie/Trie.java
@@ -117,7 +117,7 @@ public class Trie {
for (Emit emit : collectedEmits) {
if ((emit.getStart() == 0 ||
!Character.isAlphabetic(searchText.charAt(emit.getStart() - 1))) &&

•            (emit.getEnd() == size ||
  

•            (emit.getEnd() == size - 1 ||
            !Character.isAlphabetic(searchText.charAt(emit.getEnd() + 1)))) {
           continue;
       }
  

(I think :))

from aho-corasick.

Great, I've reproduced the error thanks to your finding. Looking into it right now. Thanks!

from aho-corasick.

The error was in this boundary check:

            if ((emit.getStart() == 0 ||
                 !Character.isAlphabetic(searchText.charAt(emit.getStart() - 1))) &&
                (emit.getEnd() == size ||
                 !Character.isAlphabetic(searchText.charAt(emit.getEnd() + 1)))) {
                continue;
            }

The emit.getEnd() == size does not do a proper boundary check. I've changed it to:

                (emit.getEnd() + 1 == size ||

Will commit and release a new version.

Thanks again for submitting your issue!

from aho-corasick.