Changeset 1700

Timestamp:

Oct 7, 2010, 1:05:25 AM (11 years ago)

Author:

teymour

Message:

Parsing des députés prenant en compte le nouveau design de l'AN

File:

• cpc/trunk/project/batch/depute/parse_depute.pl (modified) (5 diffs)

cpc/trunk/project/batch/depute/parse_depute.pl

@@ -24 +24 @@
     $depute{'Sexe'} = ($depute{'Nom'} =~ /M[ml]/) ? 'F' : 'H';
     $depute{'Nom'} =~ s/^\s*M\S+\s//;
-    while($t = $p->get_tag('u', '/div')) {
+    while($t = $p->get_tag('span', '/div')) {
         return if ($t->[0] eq '/div');
-        $txt = $p->get_text('/u');
+        $txt = $p->get_text('/span');
         $txt =~ /^(\S+)\s*/;
         $e = $1;
@@ -48 +48 @@
 sub contact {
     $p = shift;
-    while($p->get_tag('u', '/div')) {
+    while($p->get_tag('span', '/div')) {
         last if ($t->[0] =~ /^\//);
-        $_ = $p->get_text('/u');
+        $_ = $p->get_text('/span');
         if (/Mél/) {
             $_ = $p->get_text('/li');
@@ -83 +83 @@
 sub mandat {
     $p = shift;
-    while ($t = $p->get_tag('u', '/div')) {
+    while ($t = $p->get_tag('span', '/div')) {
         last if ($t->[0] =~ /^\//);
-        $_ = $p->get_text('/u');
+        $_ = $p->get_text('/span');
         if (/Mandat|Commission|Mission/) {
             $text = $p->get_text('ul');
@@ -151 +151 @@
 }
 
-while($t = $p->get_tag("h1", "img")) {
+while($t = $p->get_tag("h2", "img")) {
     if ($t->[0] eq 'img') {
         if (! $depute{'photo'} && $t->[1]{'src'} =~ /photo/) {
@@ -162 +162 @@
         next;
     }
-    $_ = $p->get_text('/h1');
+    $_ = $p->get_text('/h2');
     if (/Informations générales/) {
         infosgene($p);