GetPageCounts.pl update.

1 files changed, 13 insertions, 15 deletions

diff --git a/tools/GetPageCounts.pl b/tools/GetPageCounts.pl
index 7d9fd9b..5b29f9e 100755
--- a/tools/GetPageCounts.pl
+++ b/tools/GetPageCounts.pl
@@ -4,15 +4,18 @@ use URI::Escape;
 use warnings;
 use strict;
 
-
 # using the traffic stats, get pagecounts for articles
 # count pagecounts of redirect pages towards the main page,
 # report counts for redirect as the counts of their main page
 
+# page list and redirect list generated from GetPages.pl
 my $pagelist = $ARGV[0];
 my $redirectlist = $ARGV[1];
+# traffic stats from file provided by Henrik
 my $trafficlist = $ARGV[2];
 
+# read in page list, store as keys in a "page counts" hash
+# values initally zero, traffic numbers added to them
 open(PAGES,$pagelist) or die;
 my %pagecounts = ();
 while (<PAGES>) {
@@ -21,8 +24,7 @@ while (<PAGES>) {
 }
 close(PAGES);
 
-#print "Done reading pages...\n";
-
+# read in redirect list, store as hash key, value pairs
 open(REDIRECTS,$redirectlist) or die;
 my %redirects = ();
 while (<REDIRECTS>) {
@@ -32,38 +34,34 @@ while (<REDIRECTS>) {
 }
 close(REDIRECTS);
 
-#print "Done reading redirects...\n";
-
-
+# read traffic stats
 open(TRAFFIC,$trafficlist) or die;
 while (<TRAFFIC>) {
   my @data = split;
   my $page = $data[1];
-  $page = uri_unescape($page);
-  $page =~ s/_/ /g;
-  if (exists $redirects{$page}) {
+  $page = uri_unescape($page);          # pages need to be unescaped
+  $page =~ s/_/ /g;                     # and underscores converted
+  if (exists $redirects{$page}) {                       # if redirect, also add count towards main page
     if (exists $pagecounts{$redirects{$page}}) {
       $pagecounts{$redirects{$page}} += $data[2];
     }
   }
-  if (exists $pagecounts{$page}) {
+  if (exists $pagecounts{$page}) {      # add count to this page
     $pagecounts{$page} += $data[2];
   } else {
     #print "$page doesn't exist on page list!\n";
   }
 }
 
+# now output traffic amounts, not ordered
 foreach my $page (keys %pagecounts) {
+  # If redirect, print target page's traffic score
   if (exists $redirects{$page}) {
     if (exists $pagecounts{$redirects{$page}}) {
       print "$pagecounts{$redirects{$page}}\t[[$page]]\n";
-    } else {
-print "ERROR redirect page $page isn't on list??\n";
-    }
+    } 
   } elsif (exists $pagecounts{$page}) {
     print "$pagecounts{$page}\t[[$page]]\n";
-  } else {
-print "ERROR page $page not on list??\n";
   }
 }