[med-svn] r2411 - trunk/community/talks/200808_debconf8

[tille at alioth.debian.org]

Author: tille
Date: 2008-08-15 04:07:26 +0000 (Fri, 15 Aug 2008)
New Revision: 2411

Modified:
   trunk/community/talks/200808_debconf8/author_stats
   trunk/community/talks/200808_debconf8/get-archive-pages
Log:
Move encoding stuff to perl code to have it right into the database


Modified: trunk/community/talks/200808_debconf8/author_stats
===================================================================
--- trunk/community/talks/200808_debconf8/author_stats	2008-08-15 03:17:31 UTC (rev 2410)
+++ trunk/community/talks/200808_debconf8/author_stats	2008-08-15 04:07:26 UTC (rev 2411)
@@ -17,19 +17,21 @@
       
 sed -i -e '/^[-+]\+$/d' -e '/^([0-9]\+ [A-Za-z]\+)$/d' \
        -e 's/[[:space:]]*|[[:space:]]*/\t/g'           \
-       -e 's/è/è/g' -e 's/ö/ö/g' \
        -e 's/®//g'  -e 's/é/e/' \
-       -e 's/á/á/g' -e 's/ñ/ñ/' \
        "$DATFILE"
 
+##       -e 's/á/á/g' -e 's/ñ/ñ/' \
+##       -e 's/è/è/g' -e 's/ö/ö/g' \
+
 ENDCOL=$((NUM+1))
 
 R --no-save <<EOT
 library(plotrix)
 dmstats <- read.table(file='$DATFILE', sep = '\t', fill=TRUE, header=TRUE )
 # png("dmstats.png", width = 800, height = 600)
-textcolor="yellow"
-pdf("${NAME}.pdf", fg=textcolor)
+# textcolor="yellow"
+textcolor="black"
+pdf("${NAME}.pdf", fg=textcolor, width=9, height=7)
 par(col.axis=textcolor,col.main=textcolor)
 dmstats.mat <- as.matrix(dmstats)[,2:$ENDCOL]
 rownames(dmstats.mat) <- dmstats[['year']]

Modified: trunk/community/talks/200808_debconf8/get-archive-pages
===================================================================
--- trunk/community/talks/200808_debconf8/get-archive-pages	2008-08-15 03:17:31 UTC (rev 2410)
+++ trunk/community/talks/200808_debconf8/get-archive-pages	2008-08-15 04:07:26 UTC (rev 2411)
@@ -201,10 +201,6 @@
 			     $line =~ /^\s*<li><em>Message not available<\/em>/ ||
 			     $line =~ /<em>\(continued\)<\/em>\s*$/ ||
 			     $line =~ /^\s*$/) { next ; }
-# @SEP1 = ( '<li><strong>.*href="', '\s*' );
-# @SEP2 = ( '">',                   "\s*$SEPARATOR\s*" );
-# @SEP3 = ( '</a></strong>\s*<em>', '\s*<I>\s*');
-# @SEP4 = ( '</em>',                '\s*');
 			# print "DEBUG: $line\n";
 			if ( $storefiles ) {
 			    print HTMLSNIP "$line\n";
@@ -242,7 +238,8 @@
 					if ( $storefiles ) {
 					    print HTMLSNIP "$subject ; $author\n";
 					}
-					$datain->execute($project, "$year-$month-01", $author, $subject,
+					$datain->execute($project, "$year-$month-01", 
+                                                         HTMLcode2UTF8($author), HTMLcode2UTF8($subject),
 							 "${URL}/${year}/${month}/$msgurl") ;
 					$messagelines++ ;
 				    }
@@ -301,7 +298,8 @@
 # in Debian Edu.  This script cleans up this
 # system("./0fix_ralf_edu");
 $query = "UPDATE listarchive SET author = 'Ralf Gesellensetter' WHERE project = 'edu' AND author LIKE 'Ralf%setter';" ;
-$query = "UPDATE listarchive SET author = 'Vagrant Cascadian'   WHERE project = 'edu' AND author LIKE '%vagrant%';" ;
+$query = $query . "UPDATE listarchive SET author = 'Vagrant Cascadian'      WHERE project = 'edu' AND author LIKE '%vagrant%';" ;
+$query = $query . "UPDATE listarchive SET author = 'Francesco P. Lovergine' WHERE project = 'pkg-grass-general' AND author LIKE 'Francesco%Lovergine';" ;
 
 $daten = $dbh->prepare_cached($query);
 $daten->execute() ;
@@ -312,3 +310,15 @@
 foreach $project (keys %ALLPROJECTS) {
     system("./author_stats $project") ;
 }
+
+sub HTMLcode2UTF8 {
+
+  $_ = $_[0]  ;
+
+  s/&#xE8;/è/g;
+  s/&#xF6;/ö/g;
+  s/&#xE1;/á/g;
+  s/&#xF1;/ñ/g;
+
+  return ($_);
+}