Changeset ac7d970

Timestamp:

04/08/08 13:53:51 (17 years ago)

Author:

pawelk <pawelk@…>

Branches:

master, help

Children:

0214596

Parents:

317d53b

git-author:

pawelk <pawelk@…> (04/08/08 13:53:51)

git-committer:

pawelk <pawelk@…> (04/08/08 13:53:51)

Message:

Uaktualnilismy kora.

git-svn-id: ​svn://atos.wmid.amu.edu.pl/utt@23 e293616e-ec6a-49c2-aa92-f4a8b91c5d16

Location:

app

Files:

• Makefile (modified) (1 diff)
• TODO (modified) (1 diff)
• dist/common/utt_make_config.pl (modified) (2 diffs)
• dist/struktura.txt (modified) (1 diff)
• lib/weights.cor (added)
• src/common/common.h (modified) (1 diff)
• src/kor/Makefile (added)
• src/kor/cmdline_cor.ggo (added)
• src/kor/common_cor.cc (added)
• src/kor/common_cor.h (added)
• src/kor/corlist.cc (added)
• src/kor/corlist.h (added)
• src/kor/corr.cc (added)
• src/kor/corr.hh (added)
• src/kor/main.cc (added)
• src/lib/word.cc (modified) (3 diffs)
• src/lib/word.h (modified) (4 diffs)
• src/tok/tok (modified) (4 diffs)

app/Makefile

@@ -28 +28 @@
 
 # list of components to be included in the distribution
-COMPONENTS = compiledic cor fla gph grp kon kot lem mar rm12 rs12 sen-l sen-nl ser tags tok.l unfla
+COMPONENTS = compiledic cor fla gph grp kon kor kot lem mar rm12 rs12 sen-l sen-nl ser tags tok.l unfla
 # gue nie dziala!
 

app/TODO

@@ -1 +1 @@
 BARDZO WAZNE:
 
-* lem nie obsluguje dlugich wyrazow [TO]
 * gue SIE NIE KOMPILUJE !!!
-* poprawic cora [PK]
 
 WAZNE:

app/dist/common/utt_make_config.pl

@@ -15 +15 @@
 
 conf_cor(catfile($usr_home, 'cor.conf'), $sys_home);
+conf_kor(catfile($usr_home, 'kor.conf'), $sys_home);
 conf_compiledic(catfile($usr_home, 'compiledic.conf'), $sys_home);
 conf_grp(catfile($usr_home, 'grp.conf'), $sys_home);
@@ -99 +100 @@
 }
 
+sub conf_kor() {
+        my $kor_file = shift;
+        my $utthome = shift;
+  open(FILE, ">$kor_file");
+
+  print FILE makeConfigHeader();
+  print FILE "dictionary-home=", abs_path("$utthome/share/utt"), "\n";
+  print FILE "weights=", abs_path("$utthome/lib/utt/weights.cor"), "\n";
+  print FILE "threshold=1.0\n";
+
+  close FILE;
+}
+
 sub conf_grp() {
         my $grp_file = shift;

app/dist/struktura.txt

@@ -44 +44 @@
 /usr/local/lib/utt/ser.l.template
 /usr/local/lib/utt/terms.m4
+/usr/local/lib/utt/weights.cor
 
 /usr/local/share/doc/utt/FAQ

app/src/common/common.h

@@ -4 +4 @@
 #include <stdio.h>
 #include <ctype.h>
+#include <string.h>
+#include <stdlib.h>
 
 #include "../lib/const.h"

app/src/lib/word.cc

@@ -23 +23 @@
 
 //---------------------------------------------------------------------------
-int Word::cmp_w(Word a, Word b) {
+bool Word::cmp_w(Word a, Word b) {
   return (a.w_suf() > b.w_suf());
 }
@@ -124 +124 @@
     tab.resize(tab.size()*2);
   
-  tab[cnt].form(fo);
-  tab[cnt].w_suf(0.0);
+
+  Word o;
+  o.form(fo);
+  o.w_suf(0.0);
+  tab.push_back(o);
+//  tab[cnt].form(fo);
+//  tab[cnt].w_suf(0.0);
+
 
   //  if(cnt<MAX-1) {
@@ -132 +138 @@
     tab[cnt]->w_suf(0.0);
     tab[cnt]->w_pref(0.0);*/
+    return cnt++;
+    //  }
+    //return -1;
+}
+
+//---------------------------------------------------------------------------
+
+ //TYMCZASOWO TAK(DLA CORA)
+int Words::add(const char* fo, float weight)
+{
+  int i = find(fo);
+  if(i!=-1) {
+    return i;
+  }
+  
+  if (cnt>=tab.capacity()-1)
+    tab.resize(tab.size()*2);
+  
+  Word o;
+  o.form(fo);
+  o.w_suf(weight);
+  tab.push_back(o);
+//  tab[cnt].form(fo);
+//  tab[cnt].w_suf(weight);
+  
     return cnt++;
     //  }

app/src/lib/word.h

@@ -12 +12 @@
 
 using namespace std;
+
+
 
 class Word
@@ -37 +39 @@
   //  float _w_pref;
 public:
-  static int cmp_w(Word a, Word b);
+  static bool cmp_w(Word a, Word b);
 
   Word() : _len_suf(-1) { *f='\0'; returned=0; };
@@ -76 +78 @@
 
 };
+
 
 inline Word::Word(const Word& word)
@@ -119 +122 @@
   void clear() { cnt=0; tab.clear(); }
   int add(const char* fo);
+  int add(const char* fo, float weight);
   int add(const char* fo, const char* des);
 

app/src/tok/tok

@@ -11 +11 @@
 use File::HomeDir;
 
+my $max_form_length = 50;
+
 my $interactive=0;
 my $help;
@@ -21 +23 @@
 my $file;
 foreach $file ($systemconfigfile, $userconfigfile){
-        if(open(CONFIG, $configfile1)){
+        if(open(CONFIG, $file)){
                 while (<CONFIG>) {
                         chomp;
@@ -75 +77 @@
 #       | [^[:print:]]   (?{seg("B",$&)})
 
+sub min {
+  my ($val1, $val2) = @_;
+  if($val1 < $val2) {
+        return $val1;
+  }
+  else {
+        return $val2;
+  }
+}
+
 
 sub seg
 {
     my ($tag,$match) = @_;
-    my $len=length $match;
-    printf "%04d %02d %s ", $offset, $len, $tag;
+    my $length = length $match;
+    my $idx = 0;
+    while($idx < $length) {
+        my $l = min $max_form_length, $length - $idx;
+        my $m = substr $match, $idx, $l;
+    
+        printf "%04d %02d %s ", $offset + $idx, $l, $tag;
     if($tag eq 'S')
     {
-        for(my $i=0; $i<$len; ++$i)
+        for(my $i=0; $i<$l; ++$i)
         {
-            my $c = substr $match, $i, 1;
+            my $c = substr $m, $i, 1;
             print '_' if $c eq ' ';
             print '\n' if $c eq "\n";
@@ -95 +112 @@
     elsif($tag eq 'B')
     {
-        printf "\\x%02X", ord($match);
+        printf "\\x%02X", ord($m);
     }
     else
     {
-        print $match;
+        print $m;
     }
     print "\n";
-    $offset += $len;
-}
+    $idx += $l;
+    } # while($idx < $length)
+    $offset += $length;
+} #sub seg
+