X-Git-Url: http://git.polytechnique.org/?a=blobdiff_plain;f=modules%2Ffusionax%2Fformation.pl;h=d10ce65f617213d5fc2d4e9eab46d2e4508fa9c7;hb=0efb08e688d7ba0de0a0fbf2116cd4d10791d598;hp=69a8923ce9b5bde23d2d123e65fdb1999d900650;hpb=023c46fb7f3015fc87744c5a8293342c2e569acd;p=platal.git

diff --git a/modules/fusionax/formation.pl b/modules/fusionax/formation.pl
index 69a8923..d10ce65 100755
--- a/modules/fusionax/formation.pl
+++ b/modules/fusionax/formation.pl
@@ -1632,3 +1632,33 @@ while (<FILE>)
 
 close(FILE);
 close(OUT);
+
+$path = $0;
+$path =~ s/modules\/fusionax\/formation\.pl//;
+$path .= "spool/fusionax/";
+$in = $path . "Formations_MD.txt";
+$out = $path . "Formations_MD_out.txt";
+open(FILE, "<:encoding(UTF-8)", $in) || die ("Formations_MD.txt failed to open.");
+open(OUT, ">:encoding(UTF-8)", $out) || die ("Formations_MD_out.txt failed to open.");
+
+while (<FILE>)
+{
+  # Dates removal.
+  s/\r$//;
+  s/^(FO\t\w{8}\t(\w|\.|'|&| )+)\t.*$/\1/;
+  # Trailing tab, spaces and dot removal.
+  s/(\t| )*$//;
+  s/( \t|\t )/\t/g;
+  s/ +/ /g;
+
+  # Removes diploma
+  s/Doct. de l'Ec. polytechnique//;
+  s/Doct. de l'Ec. Polytechnique//;
+  s/Etudiante en Master de l'Ec. polytechnique//;
+  s/Etudiant en Master de l'Ec. polytechnique//;
+
+  print OUT $_;
+}
+
+close(FILE);
+close(OUT);