Changeset 7562131

Timestamp:

05/07/08 15:19:14 (16 years ago)

Author:

obrebski <obrebski@…>

Branches:

master, help

Children:

a7b254c

Parents:

f5d3b20

git-author:

obrebski <obrebski@…> (05/07/08 15:19:14)

git-committer:

obrebski <obrebski@…> (05/07/08 15:19:14)

Message:

mar obsluguje opcje morph, tags i umie czytac z configa

git-svn-id: ​svn://atos.wmid.amu.edu.pl/utt@39 e293616e-ec6a-49c2-aa92-f4a8b91c5d16

File:

• app/src/mar/mar (modified) (6 diffs)

app/src/mar/mar

@@ -13 +13 @@
 use strict;
 use Getopt::Long;
+use File::HomeDir;
 
 use attr;
+
+
+my $LIB_DIR="/usr/local/lib/utt";
+my $systemconfigfile='/usr/local/etc/utt/mar.conf';
+my $userconfigfile=home()."/.utt/mar.conf";
 
 Getopt::Long::Configure('no_ignore_case_always');
@@ -26 +32 @@
 my $eos="seg(EOS)";
 my $explicit_space=0;
-
-#this is our help function to  cut the re to get another tag
-#it takes only one argument which is our patern (after m4 processing)
-#returns: the first root-level brace with content
-sub cutRe
-{
- my $i = 0;
- my $level = 0;
- my $text = $_[0];
- my $temp;
- for( $i =0; $i < (length $text);$i++)
- {
-         $temp = substr($text, $i,1);
-         if( $temp eq "(")
-         {#we have an opening
-                 $level++;
-         }
-         elsif ( $temp eq ")")
-         {#we close
-                 $level--;
-         }
-         if ( $level == 0)
-         { 
-                $temp  = substr($text,0,$i+1);
-                last;
-         }
- }
- $temp;
-}
-
-#the same function as above althought it returns everything after the
-#first root level brace
-sub restRe
-{
- my $i = 0;
- my $level = 0;
- my $text = $_[0];
- my $temp;
- for( $i =0; $i < (length $text);$i++)
- {
-         $temp = substr($text, $i,1);
-         if( $temp eq "(")
-         {#we have an opening
-                 $level++;
-         }
-         elsif ( $temp eq ")")
-         {#we close
-                 $level--;
-         }
-         if ( $level == 0)
-         { #we cut everything in the begining
-                $temp  = substr($text,$i+1);
-                last;
-         }
- }
- $temp;
-}
+my $morfield='lem';
+my $tags=0;
+
+#read configuration files###########################
+my $file;
+foreach $file ($systemconfigfile, $userconfigfile){
+  if(open(CONFIG, $file)){
+        while (<CONFIG>) {
+                chomp;                  
+                s/#.*//;                
+                s/^\s+//;               
+                s/\s+$//;               
+                next unless length;     
+                my ($name, $value) = split(/\s*=\s*/, $_, 2);
+                if(($name eq "pattern")or($name eq "e")){
+                        $pattern=$value;
+                }
+                elsif($name eq "eos"){
+                        $eos=$value;
+                }
+                elsif($name eq "macros"){
+                        $macrofile=$value;
+                }
+                elsif($name eq "tags"){
+                        $tags=$value;
+                }
+                elsif($name eq "morph"){
+                        $morfield=$value;
+                }
+                elsif($name eq "command"){
+                        $command=1;
+                }
+                elsif($name eq "action"){
+                        $action=$value;
+                }
+                elsif($name eq "space"){
+                        $explicit_space=1;
+                }
+                elsif(($name eq "help")or($name eq "h")){
+                        $help=1;
+                }
+        
+        } 
+        close CONFIG;
+  }
+}
+#########################################################
 
 GetOptions("pattern|e=s" => \$pattern,
@@ -93 +89 @@
            "space|s" => \$explicit_space
    );
+
+
+
 
 if($help)
@@ -141 +140 @@
 die("$0: no pattern given. Run with -h to get help.\n") unless $pattern || $action !~ /g/;
 
-die("$0: macro file not found") unless -e "terms.m4" and $macrofile="terms.m4";
+die("$0: macro file not found") unless
+    $macrofile or
+    -e "$LIB_DIR/terms.m4" and $macrofile="$LIB_DIR/terms.m4";
 
 my $preproc    = ($action =~ /p/) ? ' fla  | '  : '';
 
 my $postproc   = ($action =~ /P/) ? ' | unfla '  : '';
+
+
+#this is our help function to  cut the re to get another tag
+#it takes only one argument which is our patern (after m4 processing)
+#returns: the first root-level brace with content
+sub cutRe
+{
+ my $i = 0;
+ my $level = 0;
+ my $text = $_[0];
+ my $temp;
+ for( $i =0; $i < (length $text);$i++)
+ {
+         $temp = substr($text, $i,1);
+         if( $temp eq "(")
+         {#we have an opening
+                 $level++;
+         }
+         elsif ( $temp eq ")")
+         {#we close
+                 $level--;
+         }
+         if ( $level == 0)
+         { 
+                $temp  = substr($text,0,$i+1);
+                last;
+         }
+ }
+ $temp;
+}
+
+#the same function as above althought it returns everything after the
+#first root level brace
+sub restRe
+{
+ my $i = 0;
+ my $level = 0;
+ my $text = $_[0];
+ my $temp;
+ for( $i =0; $i < (length $text);$i++)
+ {
+         $temp = substr($text, $i,1);
+         if( $temp eq "(")
+         {#we have an opening
+                 $level++;
+         }
+         elsif ( $temp eq ")")
+         {#we close
+                 $level--;
+         }
+         if ( $level == 0)
+         { #we cut everything in the begining
+                $temp  = substr($text,$i+1);
+                last;
+         }
+ }
+ $temp;
+}
+
 
 #here we are preparing re for extended matching
@@ -186 +246 @@
 $patternmod =~ s/(\{\d*),(\d*\})/\1\\`\\`,''\2/g;
 #print "After m4:".$re."\n";
-my $re = `echo \"$patternmod\" | m4 --define=ENDOFSEGMENT='[[:cntrl:]]' $macrofile - 2>/dev/null`;
+
+my $re = `echo \"$patternmod\" | m4 --define=ENDOFSEGMENT='[[:cntrl:]]' --define=MORFIELD=$morfield $macrofile - 2>/dev/null`;
 
 die("Incorrect pattern (m4).") if $? >> 8;
@@ -195 +256 @@
 # <> expansion
 
-$re =~ s/<([^>]+)>/`echo $1 | .\/terms\.tag2re`/ge;
+$re =~ s/<([^>]+)>/`echo $1 | $tags.tag2re`/ge;
 
 # Perl-like special sequences