Changeset 7228 for trunk/indexers/mg

Timestamp:

2004-04-26T11:01:18+12:00 (20 years ago)

Author:

kjdon

Message:

added a new -M option to mg_passes, allowing maxnumeric to be altered - made this change to keep gsdl3 mg inline with gsdl2 mg.

Location:

trunk/indexers/mg/src/text

Files:

• environment.c (modified) (3 diffs)
• mg_passes.c (modified) (4 diffs)
• words.h (modified) (8 diffs)

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

@@ -24 +24 @@
 /*
    $Log$
+   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.
+
    Revision 1.1  2003/02/20 21:18:23  mdewsnip
    Addition of MG package for search and retrieval
@@ -436 +439 @@
 }
 
+/* ARGSUSED */
+char *
+MaxNumericCons (char *Old, char *New)
+{
+  return (NumberCmp (New, 4, 512));
+}
 
 
@@ -669 +678 @@
   SetEnv ("stem", "on", BooleanCons);  /* [RPAP - Jan 97: Stem Index Change] */
   SetEnv ("term_freq", "off", BooleanCons);  /* [RPAP - Feb 97: Term Frequency] */
+  SetEnv ("maxnumeric", "4", MaxNumericCons);  /* [sjboddie - Jun 2002: Max Numeric word length] */
 }
 

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

@@ -40 +40 @@
 #include "stemmer.h"
 
+#include "words.h"
 
 /*
    $Log$
+   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.
+
    Revision 1.1  2003/02/20 21:18:24  mdewsnip
    Addition of MG package for search and retrieval
@@ -155 +159 @@
 "  %*s [-t trace-point Mb] [-m invf-memory] [-c chunk-limit]\n"
 "  %*s [-n trace-name] [-C comp-stat-size] [-s stem_method]\n"
-"  %*s [-a stemmer] -f doc-collection-name\n";
-
-
-
-
-
-
-
+"  %*s [-a stemmer] [-M max-numeric] -f doc-collection-name\n";
 
 
@@ -493 +490 @@
 
   opterr = 0;
-  while ((ch = getopt (argc, argv, "hC:WGSD123f:d:b:T:I:t:m:N:c:n:s:a:")) != -1)
+  while ((ch = getopt (argc, argv, "hC:WGSD123f:d:b:T:I:t:m:N:c:n:s:a:M:")) != -1)
     {
       switch (ch)
@@ -564 +561 @@
     case 't':
       trace = (unsigned long) (atof (optarg) * 1024 * 1024);
+      break;
+    case 'M':
+      SetEnv ("maxnumeric", optarg, NULL);
       break;
     case 'h':

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

@@ -22 +22 @@
  **************************************************************************/
 
+
 #include "sysfuncs.h"
-
 #include "unitool.h"
-
 
 /*
@@ -51 +50 @@
        of the program, , i.e., leave MAXSTEMLEN alone... */
 
-#define MAXNUMERIC  4
+/*#define MAXNUMERIC    4*/
+
     /* Maximum number of numeric characters permitted in a word.
        This avoids long sequences of numbers creating just one
@@ -100 +100 @@
     register int numeric = 0;                                      \
     unsigned short c;                                              \
+    register int maxnumeric = IntEnv (GetEnv ("maxnumeric"), 4);   \
                                                                    \
     charlength = parse_utf8_char((s_in),(end),&c);                 \
@@ -105 +106 @@
     while (length+charlength <= MAXWORDLEN && charlength > 0 &&    \
        (is_unicode_letter(c) || (is_unicode_digit(c) &&        \
-                     ++numeric <= MAXNUMERIC))) {  \
+                     ++numeric <= maxnumeric))) {  \
       while (charlength-- > 0) {                                   \
         *wptr++ = *(s_in)++; length++;                             \
@@ -197 +198 @@
     register int numeric = 0;                                      \
     unsigned short c;                                              \
+    register int maxnumeric = IntEnv (GetEnv ("maxnumeric"), 4);   \
                                                                    \
     charlength = parse_utf8_char((s_in),(end),&c);                 \
@@ -202 +204 @@
     while (length+charlength <= MAXSTEMLEN && charlength > 0 &&    \
        (is_unicode_letter(c) || (is_unicode_digit(c) &&        \
-                     ++numeric <= MAXNUMERIC))) {  \
+                     ++numeric <= maxnumeric))) {  \
       while (charlength-- > 0) {                                   \
         *wptr++ = *(s_in)++; length++;                             \
@@ -210 +212 @@
     *(Word) = length;                                              \
   }while(0)
-
     /*
 #define PARSE_STEM_WORD(Word, s_in, end)                      \
@@ -385 +386 @@
   } while (0)
     */
-