Bug#319369: unmunch not always can extract all words
Package: libmyspell-dev
Version: 1:3.1-11
If you are using a MySpell dictionary (.dic) with a word/affix that
generate more than 5.000 words than the unmunch utility can generate
only the first 5.000 of that.
In the next Italian dictionary there will be a word/affix that generate
more than 16.500 words.
This bug affect only unmunch utility and not MySpell library.
I have also write to the MySpell author.
I attach a patch ... I have exaggerated on it :-)
I have also add two missing cast and used the mychop() of MySpell library.
Ciao
Davide
Debian: Etch
Kernel: Linux localhost.localdomain 2.4.27 i686 GNU/Linux
glibc version: 2.3.2.ds1-22
gcc version: 3.3.5 (Debian 1:3.3.5-13)
--- /mnt/mandrake/src/MySpell/unmunch.c 2003-03-26 14:02:49.000000000 +0100
+++ unmunch.c 2005-07-21 17:26:50.000000000 +0200
@@ -22,8 +22,8 @@
int main(int argc, char** argv)
{
- int i;
- int al, wl;
+ long int i;
+ long int al, wl;
FILE * wrdlst;
FILE * afflst;
@@ -122,14 +122,14 @@
void parse_aff_file(FILE * afflst)
{
- int i, j;
- int numents=0;
+ long int i, j;
+ long int numents=0;
char achar='\0';
short ff=0;
char ft;
struct affent * ptr= NULL;
struct affent * nptr= NULL;
- char * line = malloc(MAX_LN_LEN);
+ char * line = (char *) malloc(MAX_LN_LEN);
while (fgets(line,MAX_LN_LEN,afflst)) {
mychomp(line);
@@ -149,7 +149,7 @@
case 1: { achar = *piece; break; }
case 2: { if (*piece == 'Y') ff = XPRODUCT; break; }
case 3: { numents = atoi(piece);
- ptr = malloc(numents * sizeof(struct affent));
+ ptr = (struct affent *) malloc(numents * sizeof(struct affent));
ptr->achar = achar;
ptr->xpflg = ff;
fprintf(stderr,"parsing %c entries %d\n",achar,numents);
@@ -230,14 +230,14 @@
void encodeit(struct affent * ptr, char * cs)
{
- int nc;
- int neg;
- int grp;
+ long int nc;
+ long int neg;
+ long int grp;
unsigned char c;
- int n;
- int ec;
- int nm;
- int i, j, k;
+ long int n;
+ long int ec;
+ long int nm;
+ long int i, j, k;
unsigned char mbr[MAX_WD_LEN];
/* now clear the conditions array */
@@ -316,13 +316,13 @@
/* add a prefix to word */
-void pfx_add (const char * word, int len, struct affent* ep, int num)
+void pfx_add (const char * word, long int len, struct affent* ep, long int num)
{
struct affent * aent;
- int cond;
- int tlen;
+ long int cond;
+ long int tlen;
unsigned char * cp;
- int i;
+ long int i;
char * pp;
char tword[MAX_WD_LEN];
@@ -361,13 +361,13 @@
/* add a suffix to a word */
-void suf_add (const char * word, int len, struct affent * ep, int num)
+void suf_add (const char * word, long int len, struct affent * ep, long int num)
{
struct affent * aent;
- int tlen;
- int cond;
+ long int tlen;
+ long int cond;
unsigned char * cp;
- int i;
+ long int i;
char tword[MAX_WD_LEN];
char * pp;
@@ -407,12 +407,12 @@
-int expand_rootword(const char * ts, int wl, const char * ap, int al)
+int expand_rootword(const char * ts, long int wl, const char * ap, long int al)
{
- int i;
- int j;
- int nh=0;
- int nwl;
+ long int i;
+ long int j;
+ long int nh=0;
+ long int nwl;
for (i=0; i < numsfx; i++) {
if (strchr(ap,(stable[i].aep)->achar)) {
@@ -492,7 +492,7 @@
void mychomp(char * s)
{
int k = strlen(s);
- if (k > 0) *(s+k-1) = '\0';
+ if ((k > 0) && ((*(s+k-1)=='\r') || (*(s+k-1)=='\n'))) *(s+k-1) = '\0';
if ((k > 1) && (*(s+k-2) == '\r')) *(s+k-2) = '\0';
}
--- unmunch.h 2005-07-21 16:41:32.000000000 +0200
+++ /tmp/a/unmunch.h 2005-07-21 16:17:30.000000000 +0200
@@ -4,10 +4,10 @@
#define MAX_WD_LEN 200
#define MAX_PREFIXES 256
#define MAX_SUFFIXES 256
-#define MAX_WORDS 5000
-
+#define MAX_WORDS 500000
+
#define ROTATE_LEN 5
-
+
#define ROTATE(v,q) \
(v) = ((v) << (q)) | (((v) >> (32 - q)) & ((1 << (q))-1));
@@ -33,7 +33,7 @@
struct affixptr
{
struct affent * aep;
- int num;
+ int num;
};
/* the prefix and suffix table */
@@ -47,7 +47,7 @@
struct affixptr stable[MAX_SUFFIXES];
-int numwords; /* number of words found */
+long int numwords; /* number of words found */
struct dwords
{
char * word;
@@ -63,11 +63,11 @@
void encodeit(struct affent * ptr, char * cs);
-int expand_rootword(const char *, int, const char*, int);
+int expand_rootword(const char *, long int, const char*, long int);
-void pfx_add (const char * word, int len, struct affent* ep, int num);
+void pfx_add (const char * word, long int len, struct affent* ep, long int num);
-void suf_add (const char * word, int len, struct affent * ep, int num);
+void suf_add (const char * word, long int len, struct affent * ep, long int num);
char * mystrsep(char ** stringp, const char delim);
Reply to: