hnj_hyphen_hyphword: fix buffer overflow (#13)

* Fix buffer overflow

* hnj_hyphen_hyphword: rewrite, improve overflow checking

* hnj_hyphen_hyphword: add test to detect overflows

example.c

@@ -67,7 +67,7 @@ main(int argc, char** argv)
     char *hyphens;
     char *lcword;
     char *hyphword;
-    char hword[BUFSIZE * 2];
+    char *hword;
     int arg = 1;
     int optd = 1;
     int optn = 0;
@@ -151,12 +151,16 @@ main(int argc, char** argv)
        rep = NULL;
        pos = NULL;
        cut = NULL;
+        
+       /* set minimum required output buffer size (2 * word_size) */
+       hword = (char *) malloc((n-1)*2);
        hword[0] = '\0';
 
        if ((!optd && hnj_hyphen_hyphenate(dict, lcword, n-1, hyphens)) ||
 	    (optd && hnj_hyphen_hyphenate2(dict, lcword, n-1, hyphens, hword, &rep, &pos, &cut))) {
              free(hyphens);
              free(lcword);
+             free(hword);
              fprintf(stderr, "hyphenation error\n");
              exit(1);
        }
@@ -202,6 +206,7 @@ main(int argc, char** argv)
       }
       free(hyphens);
       free(lcword);
+      free(hword);
     }
 
     fclose(wtclst);

hyphen.c

@@ -39,6 +39,7 @@
 #include <stdlib.h> /* for NULL, malloc */
 #include <stdio.h>  /* for fprintf */
 #include <string.h> /* for strdup */
+#include <limits.h> /* for INT_MAX */
 
 #ifdef UNX
 #include <unistd.h> /* for exit */
@@ -1073,23 +1074,41 @@ int hnj_hyphen_norm(const char *word, int word_size, char * hyphens,
 }
 
 /* get the word with all possible hyphenations (output: hyphword) */
-void hnj_hyphen_hyphword(const char * word, int l, const char * hyphens, 
+void hnj_hyphen_hyphword(const char * word, int word_size, const char * hyphens,
     char * hyphword, char *** rep, int ** pos, int ** cut)
 {
-  int hyphenslen = l + 5;
+  
+  if (word_size <= 0 || word_size > INT_MAX / 2) {
+    hyphword[0] = '\0';
+    return;
+  }
+  
+  /* hyphword buffer size must be at least 2 * l */
+  int hyphword_size = 2 * word_size - 1;
 
-  int i, j;
-  for (i = 0, j = 0; i < l; i++, j++) {
-    if (hyphens[i]&1) {
-      hyphword[j] = word[i];
-      if (*rep && *pos && *cut && (*rep)[i]) {
-        size_t offset = j - (*pos)[i] + 1;
-        strncpy(hyphword + offset, (*rep)[i], hyphenslen - offset - 1);
-        hyphword[hyphenslen-1] = '\0';
-        j += strlen((*rep)[i]) - (*pos)[i];
+  int nonstandard = 0;
+  if (*rep && *pos && *cut) {
+    nonstandard = 1;
+  }
+
+  int i;
+  int j = 0;
+  for (i = 0; i < word_size && j < hyphword_size; i++) {
+    hyphword[j++] = word[i];
+    if (hyphens[i]&1 && j < hyphword_size) {
+      if (nonstandard && (*rep)[i] && j >= (*pos)[i]) {
+        /* non-standard */
+        j -= (*pos)[i];
+        char *s = (*rep)[i];
+        while (*s && j < hyphword_size) {
+          hyphword[j++] = *s++;
+        }
         i += (*cut)[i] - (*pos)[i];
-      } else hyphword[++j] = '=';
-    } else hyphword[j] = word[i];
+      } else {
+        /* standard */
+        hyphword[j++] = '=';
+      }
+    }
   }
   hyphword[j] = '\0';
 }

tests/Makefile.am

@@ -10,6 +10,7 @@ alt3.test \
 alt4.test \
 alt5.test \
 alt6.test \
+alt7.test \
 compound.test \
 compound2.test \
 compound3.test \
@@ -46,6 +47,9 @@ alt5.word \
 alt6.hyph \
 alt6.pat \
 alt6.word \
+alt7.hyph \
+alt7.pat \
+alt7.word \
 alt.hyph \
 alt.pat \
 alt.word \
@@ -100,6 +104,7 @@ alt3.test \
 alt4.test \
 alt5.test \
 alt6.test \
+alt7.test \
 alt.test \
 basealt2.test \
 basealt.test \

tests/alt7.hyph

@@ -0,0 +1,2 @@
+b=bbbbbbbbbbbbbbbbbbbbbba=a=a=a=a=a=a=a=a=a=a=a=a=a=a=a
+b=bbbbbbbbbbbbbbbbbbbbbba=b=bbbbbbbbbbbbbbbbbbbbbba=b=b

tests/alt7.pat

@@ -0,0 +1,7 @@
+ISO8859-1
+LEFTHYPHENMIN 1
+RIGHTHYPHENMIN 1
+% Check for buffer overflows
+.a1b.
+a9a
+ab3a/b=bbbbbbbbbbbbbbbbbbbbbb,1,2

tests/alt7.test

@@ -0,0 +1,4 @@
+#!/bin/sh
+DIR="`dirname $0`"
+NAME="`basename $0 .test`"
+$DIR/test.sh $NAME.pat $NAME.word $NAME.hyph

tests/alt7.word

@@ -0,0 +1,2 @@
+abaaaaaaaaaaaaaaaaaaaaaaaaaa
+abaabaababababababababababab