| Index: trunk/wikistats/dammit.lt/DammitSyncFiles.pl |
| — | — | @@ -1,206 +0,0 @@ |
| 2 | | -#!/usr/bin/perl |
| 3 | | - |
| 4 | | -# 27 April 2010 renamed from WikiStatsDammitSync.pl |
| 5 | | - |
| 6 | | - use Time::Local ; |
| 7 | | - use Archive::Tar; |
| 8 | | - |
| 9 | | - $tar = Archive::Tar->new; |
| 10 | | - |
| 11 | | - $| = 1; # flush screen output |
| 12 | | - |
| 13 | | - $maxdaysago = 40; # do not download files more than this ago |
| 14 | | - |
| 15 | | - if (-e "a_dammit.lt_index.html") # test |
| 16 | | - { $file_html = "a_dammit.lt_index.html" ; } |
| 17 | | - else |
| 18 | | - { |
| 19 | | - open LOG, '>>', "/a/dammit.lt/WikiStatsDammitSync.log" ; |
| 20 | | - |
| 21 | | - $file_html = "/a/dammit.lt/index.html" ; |
| 22 | | - unlink $file_html ; |
| 23 | | - $cmd = "wget -O $file_html http://dammit.lt/wikistats/" ; |
| 24 | | - $result = `$cmd` ; |
| 25 | | - if ($result == 0) |
| 26 | | - { $result = "OK" ; } |
| 27 | | - &Log ("Cmd '$cmd' -> $result \n\n") ; |
| 28 | | - |
| 29 | | - if (! -e $file_html) { &Abort ("File $file_html not found") ; } |
| 30 | | - if (-s $file_html == 0) { &Abort ("File $file_html empty") ; } |
| 31 | | - } |
| 32 | | - |
| 33 | | - $timestart = time ; |
| 34 | | - |
| 35 | | - chdir "/a/dammit.lt/projectcounts" ; |
| 36 | | - $cmd = `pwd` ; |
| 37 | | - &Log ("Cmd '$cmd'\n") ; |
| 38 | | - $result = `$cmd` ; |
| 39 | | - print "$result\n" ; |
| 40 | | - |
| 41 | | - open HTML,'<',$file_html ; |
| 42 | | - while ($line = <HTML>) |
| 43 | | - { |
| 44 | | - if ($line =~ /<title>/) |
| 45 | | - { |
| 46 | | - $subdir = "" ; |
| 47 | | - if ($line =~ /archive/) |
| 48 | | - { |
| 49 | | - $line =~ s/^.*?\/wikistats\/// ; |
| 50 | | - $line =~ s/<.*$// ; |
| 51 | | - chomp $line ; |
| 52 | | - $subdir = $line ; |
| 53 | | - } |
| 54 | | - &Log ("Subdir = '$subdir'\n") ; |
| 55 | | - next ; |
| 56 | | - } |
| 57 | | - |
| 58 | | - if ($line !~ /application\/octet-stream/) { next ; } |
| 59 | | - |
| 60 | | - ($file = $line) =~ s/^.*?a href=\"([^"]+)\".*$/$1/s ; |
| 61 | | - ($date = $line) =~ s/^.*?class=\"m\">([^<]+)<.*$/$1/s ; |
| 62 | | - ($date,$time) = split (' ', $date) ; |
| 63 | | - |
| 64 | | - if ($file =~ /^pagecounts/) |
| 65 | | - { |
| 66 | | - $yy = substr ($file,11,4) ; |
| 67 | | - $mm = substr ($file,15,2) ; |
| 68 | | - $dd = substr ($file,17,2) ; |
| 69 | | - $daysago = int ((time - timegm(0,0,0,$dd,$mm-1,$yy-1900)) / (24 * 60 * 60)) ; |
| 70 | | - |
| 71 | | - print "$file: $daysago days ago\n" ; |
| 72 | | - if ($daysago > $maxdaysago) { next ; } |
| 73 | | - |
| 74 | | - # $path_7z = "/a/dammit.lt/pagecounts/$yy\-$mm/pagecounts\-$yy$mm$dd\_fdt.7z" ; |
| 75 | | - # if (-e $path_7z) { print "exists\n" ; next ; } |
| 76 | | - |
| 77 | | - $path = "/a/dammit.lt/pagecounts/$yy\-$mm/pagecounts\-$yy$mm$dd\_h" ; |
| 78 | | - if ((-e "$path.7z") || (-e "$path.zip") || (-e "$path.bz2") || (-e "$path.gz")) |
| 79 | | - { print "$path.[7z|zip|bz2|gz] exists\n" ; next ; } |
| 80 | | - else |
| 81 | | - { print "$path.[7z|zip|bz2|gz] new -> download\n" ; } |
| 82 | | - } |
| 83 | | - |
| 84 | | - # if ($file =~ /^projectcounts/) |
| 85 | | - # { |
| 86 | | - # $yy = substr ($file,14,4) ; |
| 87 | | - # $mm = substr ($file,18,2) ; |
| 88 | | - # $dd = substr ($file,20,2) ; |
| 89 | | - # $daysago = int ((time - timegm(0,0,0,$dd,$mm-1,$yy-1900)) / (24 * 60 * 60)) ; |
| 90 | | - # if ($daysago > $maxdaysago) { next ; } |
| 91 | | - # } |
| 92 | | - |
| 93 | | - |
| 94 | | - $yy = substr ($date,0,4) ; |
| 95 | | - $mm = substr ($date,5,3) ; |
| 96 | | - $dd = substr ($date,9,2) ; |
| 97 | | - $hh = substr ($time,0,2) ; |
| 98 | | - $nn = substr ($time,3,2) ; |
| 99 | | - $ss = substr ($time,6,2) ; |
| 100 | | - |
| 101 | | - if ($mm eq 'Jan') { $mm = 1 ; } |
| 102 | | - elsif ($mm eq 'Feb') { $mm = 2 ; } |
| 103 | | - elsif ($mm eq 'Mar') { $mm = 3 ; } |
| 104 | | - elsif ($mm eq 'Apr') { $mm = 4 ; } |
| 105 | | - elsif ($mm eq 'May') { $mm = 5 ; } |
| 106 | | - elsif ($mm eq 'Jun') { $mm = 6 ; } |
| 107 | | - elsif ($mm eq 'Jul') { $mm = 7 ; } |
| 108 | | - elsif ($mm eq 'Aug') { $mm = 8 ; } |
| 109 | | - elsif ($mm eq 'Sep') { $mm = 9 ; } |
| 110 | | - elsif ($mm eq 'Oct') { $mm = 10 ; } |
| 111 | | - elsif ($mm eq 'Nov') { $mm = 11 ; } |
| 112 | | - elsif ($mm eq 'Dec') { $mm = 12 ; } |
| 113 | | - else { &Abort ("Invalid month '$mm' in file date $date $time") ; } |
| 114 | | - |
| 115 | | - $date2 = sprintf ("%02d%02d%02d%02d%02d.%02d", ($yy-2000), $mm, $dd, $hh, $nn, $ss) ; |
| 116 | | - |
| 117 | | - if ($file =~ /^(?:page|project)counts-2/) |
| 118 | | - { |
| 119 | | - |
| 120 | | - if ($file =~ /^pagecounts/) |
| 121 | | - { $path = "/a/dammit.lt/pagecounts/$file" ; } |
| 122 | | - else |
| 123 | | - { $path = "/a/dammit.lt/projectcounts/$file" ; } |
| 124 | | - |
| 125 | | - if (-e $path) |
| 126 | | - { |
| 127 | | - &Log ("File $path exists\n") ; |
| 128 | | - if (-s $path == 0) |
| 129 | | - { |
| 130 | | - &Log ("File $path empty -> overwrite\n") ; |
| 131 | | - unlink $path ; |
| 132 | | - } |
| 133 | | - else { next ; } |
| 134 | | - } |
| 135 | | - |
| 136 | | - if ($file =~ /^projectcounts/) |
| 137 | | - { |
| 138 | | - $tar_file = "/a/dammit.lt/projectcounts/projectcounts-$yy.tar" ; |
| 139 | | - if (-e $tar_file) |
| 140 | | - { |
| 141 | | - if ($tar_file ne $tar_file_prev) |
| 142 | | - { |
| 143 | | - &Log ("\nRead tar file $tar_file\n") ; |
| 144 | | - $tar->read($tar_file); |
| 145 | | - $tar_file_prev = $tar_file ; |
| 146 | | - } |
| 147 | | - if ($tar->contains_file ($file)) |
| 148 | | - { |
| 149 | | - &Log ("File $file exists in tar file $tar_file\n") ; |
| 150 | | - next ; |
| 151 | | - } |
| 152 | | - } |
| 153 | | - else |
| 154 | | - { &Log ("Tar file $tar_file not found\n") ; } |
| 155 | | - } |
| 156 | | - |
| 157 | | - &Log ("Write file $path, set date $date2\n") ; |
| 158 | | - |
| 159 | | - $cmd = "wget -a /a/dammit.lt/wget.log -O $path http://dammit.lt/wikistats/$subdir$file" ; |
| 160 | | - $result = `$cmd` ; |
| 161 | | - if ((-e $path) && (-s $path == 0)) |
| 162 | | - { |
| 163 | | - $result = "Empty file -> remove $path" ; |
| 164 | | - unlink $path ; |
| 165 | | - } |
| 166 | | - elsif ($result == 0) |
| 167 | | - { $result = "OK" ; } |
| 168 | | - |
| 169 | | - &Log ("Cmd '$cmd' -> $result \n\n") ; |
| 170 | | - |
| 171 | | - if (-e $path) |
| 172 | | - { |
| 173 | | - `touch $path -t $date2` ; |
| 174 | | - |
| 175 | | - if ($file =~ /^projectcounts/) |
| 176 | | - { |
| 177 | | - $cmd = "tar --append --file=$tar_file $file" ; |
| 178 | | - &Log ("Cmd '$cmd'\n") ; |
| 179 | | - $result = `$cmd` ; |
| 180 | | - print "$result\n" ; |
| 181 | | - unlink $path ; |
| 182 | | - } |
| 183 | | - } |
| 184 | | - } |
| 185 | | - } |
| 186 | | - |
| 187 | | - &Log ("Ready in " . (time - $timestart) . " sec.\n") ; |
| 188 | | - close HTML ; |
| 189 | | - close LOG ; |
| 190 | | - exit ; |
| 191 | | - |
| 192 | | -sub Log |
| 193 | | -{ |
| 194 | | - $msg = shift ; |
| 195 | | - my ($ss, $nn, $hh) = (localtime(time))[0,1,2] ; |
| 196 | | - my $time = sprintf ("%02d:%02d:%02d", $hh, $nn, $ss) ; |
| 197 | | - $msg = "$time $msg" ; |
| 198 | | - print $msg ; |
| 199 | | - print LOG $msg ; |
| 200 | | -} |
| 201 | | - |
| 202 | | -sub Abort |
| 203 | | -{ |
| 204 | | - $msg = shift ; |
| 205 | | - &Log ($msg) ; |
| 206 | | - exit ; |
| 207 | | -} |
| Index: trunk/wikistats/dammit.lt/DammitSyncProjectCounts.pl |
| — | — | @@ -0,0 +1,125 @@ |
| | 2 | +#!/usr/bin/perl |
| | 3 | + |
| | 4 | +# November 2011 this file is much simplified version of DammitSyncFiles.pl |
| | 5 | +# DammitSyncFiles.pl used to wget each day new pagecounts and projectcounts files from dammit.lt/wikistats |
| | 6 | +# These files are now on WMF server. Only copy small projectcounts files and store in tar file. Pagecounts files no longer needs to beed copied. |
| | 7 | +# This provides a compact archive of all files, and allows versioning (after patching certain ranges of files for server underreporting) |
| | 8 | + |
| | 9 | + use Time::Local ; |
| | 10 | + use Archive::Tar; |
| | 11 | + |
| | 12 | + $tar = Archive::Tar->new; |
| | 13 | + |
| | 14 | + $| = 1; # flush screen output |
| | 15 | + |
| | 16 | + $timestart = time ; |
| | 17 | + |
| | 18 | + $dir_tars = "/a/dammit.lt/projectcounts" ; |
| | 19 | + $dir_dumps = "/mnt/data/xmldatadumps/public/other/pagecounts-raw" ; |
| | 20 | + $dir_archive = "/mnt/data/xmldatadumps/public/other/pagecounts-ez/projectcounts" ; |
| | 21 | + |
| | 22 | + if (! -d $dir_tars) |
| | 23 | + { &Abort ("Folder not found: '$dir_tars'\n") ; } |
| | 24 | + if (! -d $dir_dumps) |
| | 25 | + { &Abort ("Folder not found: '$dir_dumps'\n") ; } |
| | 26 | + |
| | 27 | + ($month,$year) = (gmtime(time))[4,5]; |
| | 28 | + $year += 1900; |
| | 29 | + $month ++ ; |
| | 30 | + $this_month = sprintf ("%04d/%04d-%02d", $year, $year, $month) ; |
| | 31 | + $month -- ; |
| | 32 | + if ($month == 0) |
| | 33 | + { $month = 12 ; $year -- ; } |
| | 34 | + $prev_month = sprintf ("%04d/%04d-%02d", $year, $year, $month) ; |
| | 35 | + |
| | 36 | + &GetProjectCounts ($prev_month) ; |
| | 37 | + &GetProjectCounts ($this_month) ; |
| | 38 | + &ArchiveTars ; |
| | 39 | + |
| | 40 | + &Log ("Ready in " . (time - $timestart) . " sec.\n") ; |
| | 41 | + exit ; |
| | 42 | + |
| | 43 | +sub GetProjectCounts |
| | 44 | +{ |
| | 45 | + my ($yyyy_yyyy_mm) = @_ ; |
| | 46 | + my $year = substr ($yyyy_yyyy_mm,0,4) ; |
| | 47 | + my $month = substr ($yyyy_yyyy_mm,10,2) ; |
| | 48 | + |
| | 49 | + print "GetProjectCounts for $year - $month\n" ; |
| | 50 | + |
| | 51 | + $tar_file = "$dir_tars/projectcounts-$year.tar" ; |
| | 52 | + |
| | 53 | + if (-e $tar_file) |
| | 54 | + { |
| | 55 | + if ($tar_file ne $tar_file_prev) |
| | 56 | + { |
| | 57 | + &Log ("\nRead tar file $tar_file\n") ; |
| | 58 | + $tar->read($tar_file); |
| | 59 | + $tar_file_prev = $tar_file ; |
| | 60 | + } |
| | 61 | + } |
| | 62 | + else |
| | 63 | + { &Log ("Tar file $tar_file not found\n") ; } |
| | 64 | + |
| | 65 | + my $dir_files = "$dir_dumps/$year/$year-$month" ; |
| | 66 | + if (! -d $dir_files) |
| | 67 | + { &Abort ("Folder not found: '$dir_files'") ; } |
| | 68 | + |
| | 69 | + chdir $dir_files || &Abort ("Could not change to dir '$dir_files'") ; |
| | 70 | + |
| | 71 | + @files = <*>; |
| | 72 | + foreach $file (sort @files) |
| | 73 | + { |
| | 74 | + next if ! -e $file ; |
| | 75 | + next if $file !~ /^projectcounts/ ; |
| | 76 | + &GetFile ($tar_file,$dir_files, $file) ; |
| | 77 | + $last_file_added = $file ; |
| | 78 | + } |
| | 79 | +} |
| | 80 | + |
| | 81 | +sub GetFile |
| | 82 | +{ |
| | 83 | + my ($tar_file, $dir_files, $file) = @_ ; |
| | 84 | + |
| | 85 | + if ($tar->contains_file ($file)) |
| | 86 | + { |
| | 87 | + # &Log ("File $file exists in tar file $tar_file\n") ; |
| | 88 | + return ; |
| | 89 | + } |
| | 90 | + |
| | 91 | + &Log ("Add new file $file to $tar_file\n") ; |
| | 92 | + |
| | 93 | + $cmd = "tar --append --file=$tar_file $file" ; |
| | 94 | + &Log ("Cmd '$cmd'\n") ; |
| | 95 | + $result = `$cmd` ; |
| | 96 | + print "$result\n" ; |
| | 97 | +} |
| | 98 | + |
| | 99 | +sub ArchiveTars |
| | 100 | +{ |
| | 101 | + $cmd = "rsync -av $dir_tars/projectcounts-20??.tar $dir_archive" ; |
| | 102 | + &Log ("Cmd '$cmd'\n") ; |
| | 103 | + $result = `$cmd` ; |
| | 104 | + print "$result\n" ; |
| | 105 | + |
| | 106 | + open LAST, '>', "$dir_archive/most_recent_file.txt" ; |
| | 107 | + print LAST $last_file_added ; |
| | 108 | + close LAST ; |
| | 109 | +} |
| | 110 | + |
| | 111 | +sub Log |
| | 112 | +{ |
| | 113 | + $msg = shift ; |
| | 114 | + my ($ss, $nn, $hh) = (localtime(time))[0,1,2] ; |
| | 115 | + my $time = sprintf ("%02d:%02d:%02d", $hh, $nn, $ss) ; |
| | 116 | + $msg = "$time $msg" ; |
| | 117 | + print $msg ; |
| | 118 | + print LOG $msg ; |
| | 119 | +} |
| | 120 | + |
| | 121 | +sub Abort |
| | 122 | +{ |
| | 123 | + $msg = shift ; |
| | 124 | + &Log ("\nError: $msg\n\n") ; |
| | 125 | + exit ; |
| | 126 | +} |
| Index: trunk/wikistats/dammit.lt/!DammitSyncFiles.pl |
| — | — | @@ -0,0 +1,211 @@ |
| | 2 | +#!/usr/bin/perl |
| | 3 | + |
| | 4 | +# 27 April 2010 renamed from WikiStatsDammitSync.pl |
| | 5 | + |
| | 6 | +# Up till October 2011: copy new files from dammit.lt/wikistats |
| | 7 | +# Starting November 2011: copy new files directly from WMF server |
| | 8 | +# Later maybe don't copy large pagecounts files at all. Rather process these files directly from other server. |
| | 9 | +# For now do copy files to this server until other jobs are ready to accept new location (and not unlinking after compression step etc). |
| | 10 | + |
| | 11 | + use Time::Local ; |
| | 12 | + use Archive::Tar; |
| | 13 | + |
| | 14 | + $tar = Archive::Tar->new; |
| | 15 | + |
| | 16 | + $| = 1; # flush screen output |
| | 17 | + |
| | 18 | + $maxdaysago = 40; # do not download files more than this ago |
| | 19 | + |
| | 20 | + if (-e "a_dammit.lt_index.html") # test |
| | 21 | + { $file_html = "a_dammit.lt_index.html" ; } |
| | 22 | + else |
| | 23 | + { |
| | 24 | + open LOG, '>>', "/a/dammit.lt/WikiStatsDammitSync.log" ; |
| | 25 | + |
| | 26 | + $file_html = "/a/dammit.lt/index.html" ; |
| | 27 | + unlink $file_html ; |
| | 28 | + $cmd = "wget -O $file_html http://dammit.lt/wikistats/" ; |
| | 29 | + $result = `$cmd` ; |
| | 30 | + if ($result == 0) |
| | 31 | + { $result = "OK" ; } |
| | 32 | + &Log ("Cmd '$cmd' -> $result \n\n") ; |
| | 33 | + |
| | 34 | + if (! -e $file_html) { &Abort ("File $file_html not found") ; } |
| | 35 | + if (-s $file_html == 0) { &Abort ("File $file_html empty") ; } |
| | 36 | + } |
| | 37 | + |
| | 38 | + $timestart = time ; |
| | 39 | + |
| | 40 | + chdir "/a/dammit.lt/projectcounts" ; |
| | 41 | + $cmd = `pwd` ; |
| | 42 | + &Log ("Cmd '$cmd'\n") ; |
| | 43 | + $result = `$cmd` ; |
| | 44 | + print "$result\n" ; |
| | 45 | + |
| | 46 | + open HTML,'<',$file_html ; |
| | 47 | + while ($line = <HTML>) |
| | 48 | + { |
| | 49 | + if ($line =~ /<title>/) |
| | 50 | + { |
| | 51 | + $subdir = "" ; |
| | 52 | + if ($line =~ /archive/) |
| | 53 | + { |
| | 54 | + $line =~ s/^.*?\/wikistats\/// ; |
| | 55 | + $line =~ s/<.*$// ; |
| | 56 | + chomp $line ; |
| | 57 | + $subdir = $line ; |
| | 58 | + } |
| | 59 | + &Log ("Subdir = '$subdir'\n") ; |
| | 60 | + next ; |
| | 61 | + } |
| | 62 | + |
| | 63 | + if ($line !~ /application\/octet-stream/) { next ; } |
| | 64 | + |
| | 65 | + ($file = $line) =~ s/^.*?a href=\"([^"]+)\".*$/$1/s ; |
| | 66 | + ($date = $line) =~ s/^.*?class=\"m\">([^<]+)<.*$/$1/s ; |
| | 67 | + ($date,$time) = split (' ', $date) ; |
| | 68 | + |
| | 69 | + if ($file =~ /^pagecounts/) |
| | 70 | + { |
| | 71 | + $yy = substr ($file,11,4) ; |
| | 72 | + $mm = substr ($file,15,2) ; |
| | 73 | + $dd = substr ($file,17,2) ; |
| | 74 | + $daysago = int ((time - timegm(0,0,0,$dd,$mm-1,$yy-1900)) / (24 * 60 * 60)) ; |
| | 75 | + |
| | 76 | + print "$file: $daysago days ago\n" ; |
| | 77 | + if ($daysago > $maxdaysago) { next ; } |
| | 78 | + |
| | 79 | + # $path_7z = "/a/dammit.lt/pagecounts/$yy\-$mm/pagecounts\-$yy$mm$dd\_fdt.7z" ; |
| | 80 | + # if (-e $path_7z) { print "exists\n" ; next ; } |
| | 81 | + |
| | 82 | + $path = "/a/dammit.lt/pagecounts/$yy\-$mm/pagecounts\-$yy$mm$dd\_h" ; |
| | 83 | + if ((-e "$path.7z") || (-e "$path.zip") || (-e "$path.bz2") || (-e "$path.gz")) |
| | 84 | + { print "$path.[7z|zip|bz2|gz] exists\n" ; next ; } |
| | 85 | + else |
| | 86 | + { print "$path.[7z|zip|bz2|gz] new -> download\n" ; } |
| | 87 | + } |
| | 88 | + |
| | 89 | + # if ($file =~ /^projectcounts/) |
| | 90 | + # { |
| | 91 | + # $yy = substr ($file,14,4) ; |
| | 92 | + # $mm = substr ($file,18,2) ; |
| | 93 | + # $dd = substr ($file,20,2) ; |
| | 94 | + # $daysago = int ((time - timegm(0,0,0,$dd,$mm-1,$yy-1900)) / (24 * 60 * 60)) ; |
| | 95 | + # if ($daysago > $maxdaysago) { next ; } |
| | 96 | + # } |
| | 97 | + |
| | 98 | + |
| | 99 | + $yy = substr ($date,0,4) ; |
| | 100 | + $mm = substr ($date,5,3) ; |
| | 101 | + $dd = substr ($date,9,2) ; |
| | 102 | + $hh = substr ($time,0,2) ; |
| | 103 | + $nn = substr ($time,3,2) ; |
| | 104 | + $ss = substr ($time,6,2) ; |
| | 105 | + |
| | 106 | + if ($mm eq 'Jan') { $mm = 1 ; } |
| | 107 | + elsif ($mm eq 'Feb') { $mm = 2 ; } |
| | 108 | + elsif ($mm eq 'Mar') { $mm = 3 ; } |
| | 109 | + elsif ($mm eq 'Apr') { $mm = 4 ; } |
| | 110 | + elsif ($mm eq 'May') { $mm = 5 ; } |
| | 111 | + elsif ($mm eq 'Jun') { $mm = 6 ; } |
| | 112 | + elsif ($mm eq 'Jul') { $mm = 7 ; } |
| | 113 | + elsif ($mm eq 'Aug') { $mm = 8 ; } |
| | 114 | + elsif ($mm eq 'Sep') { $mm = 9 ; } |
| | 115 | + elsif ($mm eq 'Oct') { $mm = 10 ; } |
| | 116 | + elsif ($mm eq 'Nov') { $mm = 11 ; } |
| | 117 | + elsif ($mm eq 'Dec') { $mm = 12 ; } |
| | 118 | + else { &Abort ("Invalid month '$mm' in file date $date $time") ; } |
| | 119 | + |
| | 120 | + $date2 = sprintf ("%02d%02d%02d%02d%02d.%02d", ($yy-2000), $mm, $dd, $hh, $nn, $ss) ; |
| | 121 | + |
| | 122 | + if ($file =~ /^(?:page|project)counts-2/) |
| | 123 | + { |
| | 124 | + |
| | 125 | + if ($file =~ /^pagecounts/) |
| | 126 | + { $path = "/a/dammit.lt/pagecounts/$file" ; } |
| | 127 | + else |
| | 128 | + { $path = "/a/dammit.lt/projectcounts/$file" ; } |
| | 129 | + |
| | 130 | + if (-e $path) |
| | 131 | + { |
| | 132 | + &Log ("File $path exists\n") ; |
| | 133 | + if (-s $path == 0) |
| | 134 | + { |
| | 135 | + &Log ("File $path empty -> overwrite\n") ; |
| | 136 | + unlink $path ; |
| | 137 | + } |
| | 138 | + else { next ; } |
| | 139 | + } |
| | 140 | + |
| | 141 | + if ($file =~ /^projectcounts/) |
| | 142 | + { |
| | 143 | + $tar_file = "/a/dammit.lt/projectcounts/projectcounts-$yy.tar" ; |
| | 144 | + if (-e $tar_file) |
| | 145 | + { |
| | 146 | + if ($tar_file ne $tar_file_prev) |
| | 147 | + { |
| | 148 | + &Log ("\nRead tar file $tar_file\n") ; |
| | 149 | + $tar->read($tar_file); |
| | 150 | + $tar_file_prev = $tar_file ; |
| | 151 | + } |
| | 152 | + if ($tar->contains_file ($file)) |
| | 153 | + { |
| | 154 | + &Log ("File $file exists in tar file $tar_file\n") ; |
| | 155 | + next ; |
| | 156 | + } |
| | 157 | + } |
| | 158 | + else |
| | 159 | + { &Log ("Tar file $tar_file not found\n") ; } |
| | 160 | + } |
| | 161 | + |
| | 162 | + &Log ("Write file $path, set date $date2\n") ; |
| | 163 | + |
| | 164 | + $cmd = "wget -a /a/dammit.lt/wget.log -O $path http://dammit.lt/wikistats/$subdir$file" ; |
| | 165 | + $result = `$cmd` ; |
| | 166 | + if ((-e $path) && (-s $path == 0)) |
| | 167 | + { |
| | 168 | + $result = "Empty file -> remove $path" ; |
| | 169 | + unlink $path ; |
| | 170 | + } |
| | 171 | + elsif ($result == 0) |
| | 172 | + { $result = "OK" ; } |
| | 173 | + |
| | 174 | + &Log ("Cmd '$cmd' -> $result \n\n") ; |
| | 175 | + |
| | 176 | + if (-e $path) |
| | 177 | + { |
| | 178 | + `touch $path -t $date2` ; |
| | 179 | + |
| | 180 | + if ($file =~ /^projectcounts/) |
| | 181 | + { |
| | 182 | + $cmd = "tar --append --file=$tar_file $file" ; |
| | 183 | + &Log ("Cmd '$cmd'\n") ; |
| | 184 | + $result = `$cmd` ; |
| | 185 | + print "$result\n" ; |
| | 186 | + unlink $path ; |
| | 187 | + } |
| | 188 | + } |
| | 189 | + } |
| | 190 | + } |
| | 191 | + |
| | 192 | + &Log ("Ready in " . (time - $timestart) . " sec.\n") ; |
| | 193 | + close HTML ; |
| | 194 | + close LOG ; |
| | 195 | + exit ; |
| | 196 | + |
| | 197 | +sub Log |
| | 198 | +{ |
| | 199 | + $msg = shift ; |
| | 200 | + my ($ss, $nn, $hh) = (localtime(time))[0,1,2] ; |
| | 201 | + my $time = sprintf ("%02d:%02d:%02d", $hh, $nn, $ss) ; |
| | 202 | + $msg = "$time $msg" ; |
| | 203 | + print $msg ; |
| | 204 | + print LOG $msg ; |
| | 205 | +} |
| | 206 | + |
| | 207 | +sub Abort |
| | 208 | +{ |
| | 209 | + $msg = shift ; |
| | 210 | + &Log ($msg) ; |
| | 211 | + exit ; |
| | 212 | +} |