| Index: trunk/wikistats/squids/SquidCountArchiveProcessLogRecord.pm |
| — | — | @@ -24,6 +24,7 @@ |
| 25 | 25 | return ; |
| 26 | 26 | } |
| 27 | 27 | |
| | 28 | + |
| 28 | 29 | # remember for each squid per hour lowest and highest sequence number and number of events |
| 29 | 30 | # later calc per hour average distance between events = (higest - lowest sequence number) / events - 1 |
| 30 | 31 | # distance between consecutive events that lay in different hour bin are ignored, begligible |
| — | — | @@ -50,6 +51,18 @@ |
| 51 | 52 | $method = $fields [7] ; |
| 52 | 53 | $url = lc ($fields [8]) ; |
| 53 | 54 | |
| | 55 | + if ($url =~ /\.m\.wikipedia.org/) |
| | 56 | + { |
| | 57 | + $url_wikipedia_mobile ++ ; |
| | 58 | + $status_url_wikipedia_mobile {$status} ++ ; |
| | 59 | + $status_mime_url_wikipedia_mobile {"$status,$mime"} ++ ; |
| | 60 | + if ($status eq "TCP_MISS/302") |
| | 61 | + { |
| | 62 | + $redirected_to_mobile ++ ; |
| | 63 | + return ; |
| | 64 | + } |
| | 65 | + } |
| | 66 | + |
| 54 | 67 | $referer = lc ($fields [11]) ; |
| 55 | 68 | $agent = $fields [13] ; |
| 56 | 69 | |
| — | — | @@ -531,7 +544,7 @@ |
| 532 | 545 | if ($os =~ /playstation/io) |
| 533 | 546 | { $version = "NetFront (PlayStation)" ; } |
| 534 | 547 | |
| 535 | | - $clients {"$mobile,$version"}++ ; |
| | 548 | + $clients {"$mobile,$version,$mimecat"}++ ; |
| 536 | 549 | |
| 537 | 550 | $operating_systems =~ s/,/,/go ; |
| 538 | 551 | $operating_systems {"$mobile,$os"} ++ ; |
| — | — | @@ -637,10 +650,17 @@ |
| 638 | 651 | |
| 639 | 652 | $clients_by_wiki {"$mobile,$version,$domain"}++ ; |
| 640 | 653 | |
| | 654 | + # different output use either 'bot=N' or 'M'(anual) / 'bot=Y' or 'B'(ot) |
| 641 | 655 | if ($bot) |
| 642 | | - { $ind_bot = 'bot=Y' ; } |
| | 656 | + { |
| | 657 | + $ind_bot = 'bot=Y' ; |
| | 658 | + $ind_bot2 = 'B' ; |
| | 659 | + } |
| 643 | 660 | else |
| 644 | | - { $ind_bot = 'bot=N' ; } |
| | 661 | + { |
| | 662 | + $ind_bot = 'bot=N' ; |
| | 663 | + $ind_bot2 = 'M' ; |
| | 664 | + } |
| 645 | 665 | |
| 646 | 666 | if (($domain =~ /^\@/) || ($domain =~ /^\*/)) |
| 647 | 667 | { |
| — | — | @@ -686,6 +706,11 @@ |
| 687 | 707 | $time_tt = $time_hh * 60 + $time_mm ; |
| 688 | 708 | $time_tt2 = $time_tt - $time_tt % 15 ; |
| 689 | 709 | $countries_timed {"$ind_bot,$domain,$country,$time_tt2"} ++ ; |
| | 710 | + |
| | 711 | + |
| | 712 | + $time2 = substr ($time,0,19) ; # omit msec |
| | 713 | + $line = "$time2,$client_ip,$domain,$ind_bot2,$mobile,$os,$version,$mimecat\n" ; |
| | 714 | + $gz_csv_views_viz->gzwrite($line) || die "Zlib error writing to $file_csv_views_viz: $gz_csv_views_viz->gzerror\n" ; |
| 690 | 715 | } |
| 691 | 716 | } |
| 692 | 717 | |