michael@0: ##########################################################################################
michael@0: #
michael@0: # This Source Code Form is subject to the terms of the Mozilla Public
michael@0: # License, v. 2.0. If a copy of the MPL was not distributed with this
michael@0: # file, You can obtain one at http://mozilla.org/MPL/2.0/.
michael@0: 
michael@0: # uncombine.pl : Break up the combined performance run output file into a file per URL
michael@0: #
michael@0: # When viewer or mozilla are run and provided a command line argument of '-f urlfile.txt'
michael@0: # then the file 'urlfile.txt' is read and each line is expected to be a URL. The browser 
michael@0: # then cycles through each url and loads it, waiting for the load to end or a safety 
michael@0: # timeout to occur. In eaiter case the next url is loaded until there are no more and then
michael@0: # the browser exits. The output from the run (stdout) is captured to a file which then
michael@0: # contains all of the performance output we need to build charts of performance numbers
michael@0: # (see average2.pl, along with header.pl and footer.pl).
michael@0: #
michael@0: # ASSUMES file urls are pathed as: file:///S|/Mozilla/Tools/ProfTools/WebSites/URL/FILE
michael@0: #         or if an http URL: http://URL/FILE
michael@0: #  Normally it is expected that local files will be used, however if we should use http
michael@0: #  URLs they can be processed (have not tested to date...)
michael@0: #
michael@0: #  For file urls, the websites tree is assumed to be the one committed to cvs. Installed
michael@0: #  files will be at s:\mozilla\tools\perftools\websites\url\file
michael@0: #  If you have the files in a different location, adjust the part that extracts the url
michael@0: #  (most likely the index into the tokens will change. Search for SENSITIVE in the script)
michael@0: 
michael@0: #------------------------------------------------------------------------------
michael@0: sub debug_print {
michael@0:   foreach $str (@_){
michael@0: #    print( $str );
michael@0:   }
michael@0: }
michael@0: #------------------------------------------------------------------------------
michael@0: 
michael@0: 
michael@0: @ARGV;
michael@0: $dir="Logs\\";
michael@0: $i=0;
michael@0: $TimingBlockBegun=0;
michael@0: $fileURL=0;
michael@0: $httpURL=0;
michael@0: $shellID;
michael@0: $infileName = $ARGV[0];
michael@0: $supportHTTP = 0; # set to 1 if HTTP URLs are to be supported
michael@0: 
michael@0: open(COMBINEFILE, "< $infileName") or die "Unable to open $infileName\n";
michael@0: while(<COMBINEFILE>){
michael@0: 	$url;
michael@0: 	@tokens = split( / /, $_ );
michael@0: 	
michael@0: 	if($TimingBlockBegun == 0) {
michael@0: 		# look for the start of a new block
michael@0: 		if($_ =~ /Timing layout processes on url/){
michael@0: 			debug_print( "Timing begin candidate: $_ \n" );
michael@0: 
michael@0: 			# see if it is a file or http url. 
michael@0: 			# If so, we are starting, otherwise it is probably a chrome url so ignore it
michael@0: 			if( $_ =~ /url: \'file:/ ){
michael@0: 				debug_print( " - file URL\n" );
michael@0: 				$url = $tokens[6];
michael@0: 				$TimingBlockBegun=1;
michael@0: 				$httpURL=0;
michael@0: 				$fileURL=1;
michael@0: 			}
michael@0:       if($supportHTTP > 0) {
michael@0: 			  if( $_ =~ /url: \'http:/  ){
michael@0: 				  debug_print( "http URL\n" );
michael@0: 				  $url = $tokens[6];			### SENSITIVE to installation path
michael@0: 				  $TimingBlockBegun=1;
michael@0: 				  $fileURL=0;
michael@0: 				  $httpURL=1;
michael@0: 			  }
michael@0:       }
michael@0: 
michael@0: 			# if we got a valid block then extract the WebShellID 
michael@0: 			# for matching the end-of-block later
michael@0: 			if($TimingBlockBegun > 0){
michael@0: 				chop($url);
michael@0: 				$shellID = $tokens[8];
michael@0: 				chop( $shellID );
michael@0: 				debug_print( " - WebShellID: $shellID\n");
michael@0: 				@urlParts = split(/\//, $url);
michael@0: 				if($fileURL > 0){
michael@0: 					$urlName = $urlParts[9];	### SENSITIVE to installation path
michael@0:                                     ### eg. 'file:///S|/Mozilla/Tools/performance/layout/WebSites/amazon/index.html'
michael@0: 				} else {
michael@0: 					$urlName = $urlParts[2];	### http://all.of.this.is.the.url.name/index.html
michael@0: 				}
michael@0: 				open(URLFILE, ">$dir$urlName-log"."\.txt") or die "cannot open file $dir$urlName\n";
michael@0: 				print("Breaking out url $url into "."$dir$urlName-log"."\.txt"."\n");
michael@0: 			}
michael@0: 		}
michael@0: 	}
michael@0: 
michael@0: 	if($TimingBlockBegun > 0){
michael@0:     $done=0;
michael@0:     $keepLine=1;
michael@0: 		# Look for end of block:
michael@0: 		#  - Find the line with the "Layout + Page Load" in it...
michael@0: 		if( $_ =~ /Layout \+ Page Load/ ){
michael@0: 			# Match the WebShell ID - if it is a match then our block ended,
michael@0: 			# otherwise it is the end of another block within our block
michael@0: 			$webshellID = "\(webBrowserChrome=".$shellID."\)";
michael@0: 			if( $tokens[6] =~ /$webshellID/ ){
michael@0: 				debug_print( "- WebShellID MATCH: $webshellID $tokens[6]\n" );
michael@0:         $done=1;
michael@0: 			} else {
michael@0:         $keepLine=0; 
michael@0:       }
michael@0: 		}
michael@0:     if($keepLine == 1){
michael@0: 		  # write the line to the file
michael@0: 		  print(URLFILE $_);
michael@0:     }
michael@0:     if($done == 1){
michael@0:  			$TimingBlockBegun=0;
michael@0: 			close(URLFILE);
michael@0:     }
michael@0: 	}
michael@0: 	$i++;
michael@0: }