Changeset 8694

Timestamp:

2004-11-29T16:15:13+13:00 (19 years ago)

Author:

kjdon

Message:

added some changes made by Emanuel Dejanu (Simple Words)

Location:

trunk

Files:

• gsdl3/packages/mg/lib/memlib.c (modified) (3 diffs)
• gsdl3/packages/mg/src/text/bool_parser.c (modified) (1 diff)
• gsdl3/packages/mg/src/text/bool_parser.y (modified) (1 diff)
• gsdl3/packages/mg/src/text/mg_passes.c (modified) (2 diffs)
• gsdl3/packages/mg/src/text/words.c (modified) (1 diff)
• gsdl3/packages/mg/src/text/words.h (modified) (1 diff)
• gsdl3/packages/mg/sysfuncs.h (modified) (1 diff)
• gsdl3/src/packages/mg/lib/memlib.c (modified) (3 diffs)
• gsdl3/src/packages/mg/src/text/bool_parser.c (modified) (1 diff)
• gsdl3/src/packages/mg/src/text/bool_parser.y (modified) (1 diff)
• gsdl3/src/packages/mg/src/text/mg_passes.c (modified) (2 diffs)
• gsdl3/src/packages/mg/src/text/words.c (modified) (1 diff)
• gsdl3/src/packages/mg/src/text/words.h (modified) (1 diff)
• gsdl3/src/packages/mg/sysfuncs.h (modified) (1 diff)
• indexers/mg/lib/memlib.c (modified) (3 diffs)
• indexers/mg/src/text/bool_parser.c (modified) (1 diff)
• indexers/mg/src/text/bool_parser.y (modified) (1 diff)
• indexers/mg/src/text/mg_passes.c (modified) (2 diffs)
• indexers/mg/src/text/words.c (modified) (1 diff)
• indexers/mg/src/text/words.h (modified) (1 diff)
• indexers/mg/sysfuncs.h (modified) (1 diff)

trunk/gsdl3/packages/mg/lib/memlib.c

@@ -24 +24 @@
 /*
    $Log$
+   Revision 1.2  2004/11/29 03:15:13  kjdon
+   added some changes made by Emanuel Dejanu (Simple Words)
+
    Revision 1.1  2003/02/20 21:14:16  mdewsnip
    Addition of MG package for search and retrieval
@@ -45 +48 @@
 
 /* Defined as strdup is not an ANSI function */
+/* change the name so we do not have problems with other libs */
 char *
-my_strdup(const char *str)
+my_mg_strdup(const char *str)
 {
   char *ret_str = malloc(strlen(str)+1);
@@ -60 +64 @@
 Free_func Xfree = free;
 
-Strdup_func Xstrdup = my_strdup;
+Strdup_func Xstrdup = my_mg_strdup;

trunk/gsdl3/packages/mg/src/text/bool_parser.c

@@ -1017 +1017 @@
     {
       /* jump over whitespace */
-      while (isspace(*buf_ptr))
-    buf_ptr++;
+      buf_ptr = skipspace(buf_ptr, end);
  
       if (inaword(buf_ptr, end))

trunk/gsdl3/packages/mg/src/text/bool_parser.y

@@ -137 +137 @@
     {
       /* jump over whitespace */
-      while (isspace(*buf_ptr))
-    buf_ptr++;
- 
+      buf_ptr = skipspace(buf_ptr, end);
+
       if (inaword(buf_ptr, end))
     {

trunk/gsdl3/packages/mg/src/text/mg_passes.c

@@ -27 +27 @@
 # include <malloc.h> 
 #endif
-
+#include <stdlib.h>
 #include "memlib.h"
 #include "messages.h"
@@ -44 +44 @@
 /*
    $Log$
+   Revision 1.3  2004/11/29 03:15:13  kjdon
+   added some changes made by Emanuel Dejanu (Simple Words)
+
    Revision 1.2  2004/04/25 23:01:18  kjdon
    added a new -M option to mg_passes, allowing maxnumeric to be altered - made this change to keep gsdl3 mg inline with gsdl2 mg.

trunk/gsdl3/packages/mg/src/text/words.c

@@ -10 +10 @@
   return 0;
 }
+
+/* Return a the UTF-8 encoded Unicode string with begining
+   unicode spaces skipped. */
+u_char *skipspace(u_char *here, u_char *end)
+{
+  unsigned short c; 
+  int length;
+  while(here != end) {
+    length = parse_utf8_char(here, end, &c);
+    if (length == 0 || !is_unicode_space(c)) break;
+    here += length;
+  }
+  return here;
+}
+
+/* It determines whether a given place in a UTF-8 encoded 
+   Unicode string is a unicode space. */
+int isaspace (const u_char *here, const u_char *end)
+{
+  unsigned short c;
+  if (parse_utf8_char(here, end, &c) > 0) return is_unicode_space(c);
+  return 0;
+}

trunk/gsdl3/packages/mg/src/text/words.h

@@ -81 +81 @@
        is part of a word. */
 
+int isaspace (const u_char *here, const u_char *end);
+        /* It determines whether a given place in a UTF-8 encoded 
+       Unicode string is a unicode space. */
+
+u_char *skipspace(u_char *here, u_char *end);
+        /* Return a the UTF-8 encoded Unicode string with 
+       begining unicode spaces skipped. */ 
 
 /* =========================================================================

trunk/gsdl3/packages/mg/sysfuncs.h

@@ -271 +271 @@
 
 #ifndef _POSIX_VERSION
-#ifdef __MSDOS__
+#if defined(__WIN32__)
+#include <io.h>
+#define lseek _lseek
+#elif defined(__MSDOS__)
 #include <io.h>
 #else

trunk/gsdl3/src/packages/mg/lib/memlib.c

@@ -24 +24 @@
 /*
    $Log$
+   Revision 1.2  2004/11/29 03:15:13  kjdon
+   added some changes made by Emanuel Dejanu (Simple Words)
+
    Revision 1.1  2003/02/20 21:14:16  mdewsnip
    Addition of MG package for search and retrieval
@@ -45 +48 @@
 
 /* Defined as strdup is not an ANSI function */
+/* change the name so we do not have problems with other libs */
 char *
-my_strdup(const char *str)
+my_mg_strdup(const char *str)
 {
   char *ret_str = malloc(strlen(str)+1);
@@ -60 +64 @@
 Free_func Xfree = free;
 
-Strdup_func Xstrdup = my_strdup;
+Strdup_func Xstrdup = my_mg_strdup;

trunk/gsdl3/src/packages/mg/src/text/bool_parser.c

@@ -1017 +1017 @@
     {
       /* jump over whitespace */
-      while (isspace(*buf_ptr))
-    buf_ptr++;
+      buf_ptr = skipspace(buf_ptr, end);
  
       if (inaword(buf_ptr, end))

trunk/gsdl3/src/packages/mg/src/text/bool_parser.y

@@ -137 +137 @@
     {
       /* jump over whitespace */
-      while (isspace(*buf_ptr))
-    buf_ptr++;
- 
+      buf_ptr = skipspace(buf_ptr, end);
+
       if (inaword(buf_ptr, end))
     {

trunk/gsdl3/src/packages/mg/src/text/mg_passes.c

@@ -27 +27 @@
 # include <malloc.h> 
 #endif
-
+#include <stdlib.h>
 #include "memlib.h"
 #include "messages.h"
@@ -44 +44 @@
 /*
    $Log$
+   Revision 1.3  2004/11/29 03:15:13  kjdon
+   added some changes made by Emanuel Dejanu (Simple Words)
+
    Revision 1.2  2004/04/25 23:01:18  kjdon
    added a new -M option to mg_passes, allowing maxnumeric to be altered - made this change to keep gsdl3 mg inline with gsdl2 mg.

trunk/gsdl3/src/packages/mg/src/text/words.c

@@ -10 +10 @@
   return 0;
 }
+
+/* Return a the UTF-8 encoded Unicode string with begining
+   unicode spaces skipped. */
+u_char *skipspace(u_char *here, u_char *end)
+{
+  unsigned short c; 
+  int length;
+  while(here != end) {
+    length = parse_utf8_char(here, end, &c);
+    if (length == 0 || !is_unicode_space(c)) break;
+    here += length;
+  }
+  return here;
+}
+
+/* It determines whether a given place in a UTF-8 encoded 
+   Unicode string is a unicode space. */
+int isaspace (const u_char *here, const u_char *end)
+{
+  unsigned short c;
+  if (parse_utf8_char(here, end, &c) > 0) return is_unicode_space(c);
+  return 0;
+}

trunk/gsdl3/src/packages/mg/src/text/words.h

@@ -81 +81 @@
        is part of a word. */
 
+int isaspace (const u_char *here, const u_char *end);
+        /* It determines whether a given place in a UTF-8 encoded 
+       Unicode string is a unicode space. */
+
+u_char *skipspace(u_char *here, u_char *end);
+        /* Return a the UTF-8 encoded Unicode string with 
+       begining unicode spaces skipped. */ 
 
 /* =========================================================================

trunk/gsdl3/src/packages/mg/sysfuncs.h

@@ -271 +271 @@
 
 #ifndef _POSIX_VERSION
-#ifdef __MSDOS__
+#if defined(__WIN32__)
+#include <io.h>
+#define lseek _lseek
+#elif defined(__MSDOS__)
 #include <io.h>
 #else

trunk/indexers/mg/lib/memlib.c

@@ -24 +24 @@
 /*
    $Log$
+   Revision 1.2  2004/11/29 03:15:13  kjdon
+   added some changes made by Emanuel Dejanu (Simple Words)
+
    Revision 1.1  2003/02/20 21:14:16  mdewsnip
    Addition of MG package for search and retrieval
@@ -45 +48 @@
 
 /* Defined as strdup is not an ANSI function */
+/* change the name so we do not have problems with other libs */
 char *
-my_strdup(const char *str)
+my_mg_strdup(const char *str)
 {
   char *ret_str = malloc(strlen(str)+1);
@@ -60 +64 @@
 Free_func Xfree = free;
 
-Strdup_func Xstrdup = my_strdup;
+Strdup_func Xstrdup = my_mg_strdup;

trunk/indexers/mg/src/text/bool_parser.c

@@ -1017 +1017 @@
     {
       /* jump over whitespace */
-      while (isspace(*buf_ptr))
-    buf_ptr++;
+      buf_ptr = skipspace(buf_ptr, end);
  
       if (inaword(buf_ptr, end))

trunk/indexers/mg/src/text/bool_parser.y

@@ -137 +137 @@
     {
       /* jump over whitespace */
-      while (isspace(*buf_ptr))
-    buf_ptr++;
- 
+      buf_ptr = skipspace(buf_ptr, end);
+
       if (inaword(buf_ptr, end))
     {

trunk/indexers/mg/src/text/mg_passes.c

@@ -27 +27 @@
 # include <malloc.h> 
 #endif
-
+#include <stdlib.h>
 #include "memlib.h"
 #include "messages.h"
@@ -44 +44 @@
 /*
    $Log$
+   Revision 1.3  2004/11/29 03:15:13  kjdon
+   added some changes made by Emanuel Dejanu (Simple Words)
+
    Revision 1.2  2004/04/25 23:01:18  kjdon
    added a new -M option to mg_passes, allowing maxnumeric to be altered - made this change to keep gsdl3 mg inline with gsdl2 mg.

trunk/indexers/mg/src/text/words.c

@@ -10 +10 @@
   return 0;
 }
+
+/* Return a the UTF-8 encoded Unicode string with begining
+   unicode spaces skipped. */
+u_char *skipspace(u_char *here, u_char *end)
+{
+  unsigned short c; 
+  int length;
+  while(here != end) {
+    length = parse_utf8_char(here, end, &c);
+    if (length == 0 || !is_unicode_space(c)) break;
+    here += length;
+  }
+  return here;
+}
+
+/* It determines whether a given place in a UTF-8 encoded 
+   Unicode string is a unicode space. */
+int isaspace (const u_char *here, const u_char *end)
+{
+  unsigned short c;
+  if (parse_utf8_char(here, end, &c) > 0) return is_unicode_space(c);
+  return 0;
+}

trunk/indexers/mg/src/text/words.h

@@ -81 +81 @@
        is part of a word. */
 
+int isaspace (const u_char *here, const u_char *end);
+        /* It determines whether a given place in a UTF-8 encoded 
+       Unicode string is a unicode space. */
+
+u_char *skipspace(u_char *here, u_char *end);
+        /* Return a the UTF-8 encoded Unicode string with 
+       begining unicode spaces skipped. */ 
 
 /* =========================================================================

trunk/indexers/mg/sysfuncs.h

@@ -271 +271 @@
 
 #ifndef _POSIX_VERSION
-#ifdef __MSDOS__
+#if defined(__WIN32__)
+#include <io.h>
+#define lseek _lseek
+#elif defined(__MSDOS__)
 #include <io.h>
 #else