frhyme

commit f0f352ea98d49111239f3daf143cbf7025a18a40
parent 2d217822c497c984747bf44760318ae27ce0880b
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Tue, 13 Mar 2012 11:57:47 +0100

improve lexique tools

Diffstat:
lexique/lexique_prepare.sh
 | 
4
+---
lexique/lexique_retrieve.sh
 | 
4
++--
lexique/subst.pl
 | 
28
+++++++++++++++-------------

3 files changed, 18 insertions(+), 18 deletions(-)
diff --git a/lexique/lexique_prepare.sh b/lexique/lexique_prepare.sh
@@ -2,7 +2,5 @@
 
 # Prepare the Lexique file for use with frhyme
 
-cd "$( dirname "$0" )"
-
-awk '{print $1, $2}' | iconv -f latin1 -t utf8 | ./subst.pl
+cut -f 1,2 | uniq
 
diff --git a/lexique/lexique_retrieve.sh b/lexique/lexique_retrieve.sh
@@ -7,6 +7,6 @@ FILE="Lexique371/Bases+Scripts/Lexique3.txt"
 cd "$( dirname "$0" )"
 
 wget $URL
-unzip $ZIP $FILE
-cat $FILE | ./prepare_lexique.sh
+unzip -qq $ZIP $FILE
+cat $FILE | ./lexique_fix.sh | ./lexique_prepare.sh
 
diff --git a/lexique/subst.pl b/lexique/subst.pl
@@ -2,23 +2,23 @@
 
 # This file fixes Lexique's pronunciation info from the home-grown
 # format described in
-# http://www.lexique.org/outils/Manuel_Lexique.htm#_Toc108519023 to the
-# X-SAMPA standard
+# http://www.lexique.org/outils/Manuel_Lexique.htm#_Toc108519023 to a
+# variation of the X-SAMPA standard
 
 
 sub subst {
   my $a = shift;
   # substitutions to apply
   my @s = (
-    ["§", "O~"],
-    ["@", "A~"],
-    ["1", "E~"],
-    ["5", "9~"],
-    ["°", "@"],
-    ["3", "@"],
-    ["H", "8"],
-    ["N", "J"],
-    ["G", "N"],
+    ['§', '$'],
+    ['@', '#'],
+    ['1', '('],
+    ['5', ')'],
+    ['°', '@'],
+    ['3', '@'],
+    ['H', '8'],
+    ['N', 'J'],
+    ['G', 'N'],
   );
   foreach my $t (@s) {
     $a =~ s/${$t}[0]/${$t}[1]/g
@@ -28,9 +28,11 @@ sub subst {
 
 while (<>) {
   chop;
-  if (/^(.*) ([^ ]*)$/) {
+  if (/^([^\t]*)\t([^\t]*)(.*)$/) {
     my $repl = subst $2;
-    print "$1 $repl\n";
+    print "$1\t$repl$3\n";
+  } else {
+    die "Cannot process line: $_\n";
   }
 }