frhyme

commit 43512193dc0351a29b61800e133e28793c16e1e2
parent d0ef73f5cc6e0221dc6c90f84b5af43a46dac781
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Sat, 17 Aug 2019 18:25:54 +0200

remove lexique scripts -- moved to a3nm.net/git/lexique

Diffstat:
scripts/lexique/lexique_fix.sh
 | 
8
--------
scripts/lexique/lexique_prepare.sh
 | 
6
------
scripts/lexique/lexique_retrieve.sh
 | 
12
------------
scripts/lexique/subst.pl
 | 
38
--------------------------------------

4 files changed, 0 insertions(+), 64 deletions(-)
diff --git a/scripts/lexique/lexique_fix.sh b/scripts/lexique/lexique_fix.sh
@@ -1,8 +0,0 @@
-#!/bin/bash
-
-# General fixes for lexique
-
-cd "$( dirname "$0" )"
-
-sed 1d | ./subst.pl
-
diff --git a/scripts/lexique/lexique_prepare.sh b/scripts/lexique/lexique_prepare.sh
@@ -1,6 +0,0 @@
-#!/bin/bash
-
-# Prepare the Lexique file for use with frhyme
-
-cut -f 1,2 | uniq
-
diff --git a/scripts/lexique/lexique_retrieve.sh b/scripts/lexique/lexique_retrieve.sh
@@ -1,12 +0,0 @@
-#!/bin/bash
-
-ZIP="Lexique382.zip"
-URL="http://www.lexique.org/databases/Lexique382/$ZIP"
-FILE="Lexique382.tsv"
-
-cd "$( dirname "$0" )"
-
-wget $URL
-unzip -qq $ZIP $FILE
-cat $FILE | ./lexique_fix.sh | ./lexique_prepare.sh
-
diff --git a/scripts/lexique/subst.pl b/scripts/lexique/subst.pl
@@ -1,38 +0,0 @@
-#!/usr/bin/perl
-
-# This file fixes Lexique's pronunciation info from the home-grown
-# format described in
-# http://www.lexique.org/outils/Manuel_Lexique.htm#_Toc108519023 to a
-# variation of the X-SAMPA standard
-
-
-sub subst {
-  my $a = shift;
-  # substitutions to apply
-  my @s = (
-    ['§', '$'],
-    ['@', '#'],
-    ['1', '('],
-    ['5', ')'],
-    ['°', '@'],
-    ['3', '@'],
-    ['H', '8'],
-    ['N', 'J'],
-    ['G', 'N'],
-  );
-  foreach my $t (@s) {
-    $a =~ s/${$t}[0]/${$t}[1]/g
-  }
-  return $a;
-}
-
-while (<>) {
-  chop;
-  if (/^([^\t]*)\t([^\t]*)(.*)$/) {
-    my $repl = subst $2;
-    print "$1\t$repl$3\n";
-  } else {
-    die "Cannot process line: $_\n";
-  }
-}
-