Showing summed invocations

New simpler preprocessed format (v5)

2 files changed, 120 insertions, 215 deletions

diff --git a/lib/CallgrindPreprocessor.php b/lib/CallgrindPreprocessor.php
index 92afa37..7beddd5 100644
--- a/lib/CallgrindPreprocessor.php
+++ b/lib/CallgrindPreprocessor.php
@@ -1,53 +1,41 @@
 <?php
 /*
 
-Preprocessed file format v4:
+Preprocessed file format v5:
 
-file_contents: version_number header_address function_count function_addresses functions subcalls headers
+file_contents: version_number header_address function_count function_addresses functions headers
 version_number: number
 header_address: number
 function_count: number
 function_addresses: {number}
-functions: {total_self_cost total_inclusive_self_cost total_call_cost invocation_count invocations file_name function_name}
-total_self_cost: number
-total_inclusive_self_cost: number
-total_call_cost: number
+functions: {summed_self_cost summed_inclusive_cost invocation_count called_from_count call_information file_name function_name}
+summed_self_cost: number
+summed_inclusive_cost: number
+summed_call_cost: number
 invocation_count: number
-invocations:  self_cost inclusive_self_cost called_from subcall_count subcall_address 
+called_from_count: number
+call_information: {function_number line call_count summed_call_cost}
+function_number: number
+line: number
+call_count: number
+summed_call_cost: number
 file_name: string_newline
 function_name: string_newline
-self_cost: number
-inclusive_self_cost: number
-called_from: call_cost function invocation line_number
-call_cost: number
-function: number
-invocation: number
-line_number: number
-subcall_count: number
-subcall_address: number
-subcalls: {function invocation line_number call_cost}
 headers: {string_newline}
 string_newline: any string terminated by a newline character
 number: unsigned long (always 32 bit, little endian byte order)
 
 */
 
-
-
 class CallgrindPreprocessor{
-	const FILE_FORMAT_VERSION = 4;
+	const FILE_FORMAT_VERSION = 5;
 
 	const NR_FORMAT = 'V';
 	const NR_SIZE = 4;
-	const INVOCATION_LENGTH = 8;
-	const SUBCALL_LENGTH = 4;
 
 	const ENTRY_POINT = '{main}';
 
 	private $in, $out;
-	private $nextFuncNr = 0;
-	private $functions = array();
-	private $headers = array();
 	
 	function __construct($inFile, $outFile){
 		$this->in = fopen($inFile, 'rb');
@@ -56,153 +44,84 @@ class CallgrindPreprocessor{
 	
 	
 	function parse(){
-		/*
-		* Pass 1
-		* Find function information and count function invocations
-		*/
-		while(($line = fgets($this->in))){
-			if(substr($line,0,3)==='fl='){
-				list($function) = fscanf($this->in,"fn=%s");
-				if(!isset($this->functions[$function])){
-					// New function
-					$this->functions[$function] = array('filename'=>substr(trim($line),3), 'invocations'=>1,'nr'=>$this->nextFuncNr++,'stack'=>array(),'count'=>0,'selfCost'=>0,'inclusiveSelfCost'=>0,'callCost'=>0);
-				} else {
-					$this->functions[$function]['invocations']++;
-				}
-			} else if(strpos($line,': ')!==false){
-				$this->headerFound($line);
-			}
-		}
-		// Write function information
-		$functionCount = sizeof($this->functions);
-		fwrite($this->out,pack(self::NR_FORMAT.'*',self::FILE_FORMAT_VERSION,0,$functionCount));
-		// Position where function addresses must be written
-		$funcAddrMark = ftell($this->out);
-		fseek($this->out,self::NR_SIZE*$functionCount, SEEK_CUR);
-		foreach($this->functions as $function=>&$data){
-			$data['position'] = ftell($this->out);
-			fwrite($this->out,pack(self::NR_FORMAT.'*',0,0,0,$data['invocations']));
-			$data['nextInvocationPosition'] = ftell($this->out);
-			fseek($this->out,self::NR_SIZE*self::INVOCATION_LENGTH*$data['invocations'], SEEK_CUR);
-			fwrite($this->out,$data['filename']."\n".$function."\n");			
-		}
-		$callAddress = ftell($this->out);
-		fseek($this->out,$funcAddrMark,SEEK_SET);
-		foreach($this->functions as $func){
-			fwrite($this->out,pack(self::NR_FORMAT,$func['position']));
-		}
-		fseek($this->out,$callAddress,SEEK_SET);
-		rewind($this->in);
+		// Make local for faster access
+		$in = $this->in;
+		$out = $this->out;
 		
-		/*
-		* Pass 2
-		* Parse invocations and insert address information in data created above
-		*/
-		while(($line = fgets($this->in))){
+		$nextFuncNr = 0;
+		$functions = array();
+		$headers = array();
+		
+		
+		// Read information into memory
+		while(($line = fgets($in))){
 			if(substr($line,0,3)==='fl='){
-				list($function) = fscanf($this->in,"fn=%s");
-				
-				if($function==self::ENTRY_POINT){
-					// The entry point function has a summary header in the middle of no where. Header has been read above
-					fgets($this->in);					
-					fgets($this->in);
-					fgets($this->in);
-				}
-				
-				$stackSize = sizeof($this->functions[$function]['stack']);
-				$count = $this->functions[$function]['count'];
-				
-				if($stackSize>0 && $this->functions[$function]['stack'][$stackSize-1]['nr']==$count-1)
-					$this->functions[$function]['stack'][$stackSize-1]['nr']++;
-				else
-					$this->functions[$function]['stack'][] = array('min'=>$count, 'nr'=>$count);
-				
-				$this->functions[$function]['count']++;
-				list($lnr, $selfCost) = fscanf($this->in,"%d %d");
-				$inclusiveSelfCost = $selfCost;
-				$this->functions[$function]['selfCost'] += $selfCost;
-				$callAddr = ftell($this->out);
-				$callCount = 0;				
-				
-				
-				$callsPos = ftell($this->in);
-				$functionCalls = array();
-				while(($line=fgets($this->in))!="\n"){
-						if(substr($line,0,4)==='cfn='){
-							$calledFunctionName = substr(trim($line),4);
-							if(!isset($functionCalls[$calledFunctionName]))
-								$functionCalls[$calledFunctionName]=0;
-							$functionCalls[$calledFunctionName]++;
-						}
+				list($function) = fscanf($in,"fn=%s");
+				if(!isset($functions[$function])){
+					$functions[$function] = array('filename'=>substr(trim($line),3), 'invocationCount'=>0,'nr'=>$nextFuncNr++,'count'=>0,'summedSelfCost'=>0,'summedInclusiveCost'=>0,'callInformation'=>array());
+				} 
+				$functions[$function]['invocationCount']++;
+				// Special case for {main} - it contains summary header
+				if(self::ENTRY_POINT == $function){
+					fgets($in);					
+					$headers[] = fgets($in);
+					fgets($in);
 				}
+				// Cost line
+				list($lnr, $cost) = fscanf($in,"%d %d");
+				$functions[$function]['summedSelfCost'] += $cost;
+				$functions[$function]['summedInclusiveCost'] += $cost;				
+			} else if(substr($line,0,4)==='cfn=') {
+				$calledFunctionName = substr(trim($line),4);
+				// Skip call line
+				fgets($in);
+				// Cost line
+				list($lnr, $cost) = fscanf($in,"%d %d");
 				
-				fseek($this->in, $callsPos, SEEK_SET);
-				while(($line=fgets($this->in))!="\n"){
-					if(substr($line,0,4)==='cfn='){
-						// Skip call line
-						fgets($this->in);
-						// Cost line
-						list($lnr, $cost) = fscanf($this->in,"%d %d");
-						$calledFunctionName = substr(trim($line),4);
-						$inclusiveSelfCost += $cost;
-						$this->functions[$calledFunctionName]['callCost'] += $cost;
-						
-						$stackTop = sizeof($this->functions[$calledFunctionName]['stack'])-1;
-
-						$invocationNr = $this->functions[$calledFunctionName]['stack'][$stackTop]['nr']--;
-						if($invocationNr == $this->functions[$calledFunctionName]['stack'][$stackTop]['min'])
-							array_pop($this->functions[$calledFunctionName]['stack']);
-							
-						//echo 'Function '.$function.' calls '.$calledFunctionName.' invocation '.$invocationNr."\n";
-						
-						$here = ftell($this->out);
-						$pos = $this->functions[$calledFunctionName]['position'];
-						// Seek past total selfcost, total invocation cost, total callcost, invocationcount and invocations and self cost and inclusive self cost
-						$pos = $pos+(6+self::INVOCATION_LENGTH*$invocationNr)*self::NR_SIZE;
-						fseek($this->out, $pos, SEEK_SET);
-						fwrite($this->out, pack(self::NR_FORMAT.'*', $cost, $this->functions[$function]['nr'], $this->functions[$function]['count']-1, $lnr));
-						fseek($this->out, $here, SEEK_SET);
-						
-						fwrite($this->out, pack(self::NR_FORMAT.'*',$this->functions[$calledFunctionName]['nr'],$invocationNr,$lnr, $cost));
-						$callCount++;	
-					}
-				}
+				$functions[$function]['summedInclusiveCost'] += $cost;
 				
-				$this->functions[$function]['inclusiveSelfCost'] += $inclusiveSelfCost;
+				if(!isset($functions[$calledFunctionName]['callInformation'][$function.':'.$lnr]))
+					$functions[$calledFunctionName]['callInformation'][$function.':'.$lnr] = array('functionNr'=>$functions[$function]['nr'],'line'=>$lnr,'callCount'=>0,'summedCallCost'=>0);
+				$functions[$calledFunctionName]['callInformation'][$function.':'.$lnr]['callCount']++;
+				$functions[$calledFunctionName]['callInformation'][$function.':'.$lnr]['summedCallCost'] += $cost;
 				
-				$this->addInvocationInfo($function, $selfCost, $inclusiveSelfCost, $callCount, $callAddr);
+			} else if(strpos($line,': ')!==false){
+				$headers[] = $line;
 			}
 		}
-		
-		foreach($this->functions as $func){
-			fseek($this->out, $func['position'], SEEK_SET);
-			fwrite($this->out,pack(self::NR_FORMAT.'*',$func['selfCost'],$func['inclusiveSelfCost'],$func['callCost']));			
+			
+				
+		// Write output
+		$functionCount = sizeof($functions);
+		fwrite($out, pack(self::NR_FORMAT.'*', self::FILE_FORMAT_VERSION, 0, $functionCount));
+		// Make room for function addresses
+		fseek($out,self::NR_SIZE*$functionCount, SEEK_CUR);
+		$functionAddresses = array();
+		foreach($functions as $functionName => $function){
+			$functionAddresses[] = ftell($out);
+			$calledFromCount = sizeof($function['callInformation']);
+			fwrite($out, pack(self::NR_FORMAT.'*', $function['summedSelfCost'], $function['summedInclusiveCost'], $function['invocationCount'], $calledFromCount));
+			// Write call information
+			foreach($function['callInformation'] as $call){
+				fwrite($out, pack(self::NR_FORMAT.'*', $call['functionNr'], $call['line'], $call['callCount'], $call['summedCallCost']));
+			}
+			fwrite($out, $function['filename']."\n".$functionName."\n");
 		}
-		
-		fseek($this->out,0,SEEK_END);
 		$headersPos = ftell($this->out);
 		// Write headers
-		foreach($this->headers as $header){
-			fwrite($this->out,$header);
+		foreach($headers as $header){
+			fwrite($out,$header);
 		}
-		fseek($this->out,self::NR_SIZE, SEEK_SET);
-		fwrite($this->out, pack('L',$headersPos));
-	}
-	
-	private function addInvocationInfo($function, $selfCost, $inclusiveSelfCost, $subCallCount, $subCallsAddr){
-		$here = ftell($this->out);	
-			
-		$pos = &$this->functions[$function]['nextInvocationPosition'];
-		fseek($this->out, $pos, SEEK_SET);
-		fwrite($this->out, pack(self::NR_FORMAT.'*',$selfCost, $inclusiveSelfCost, 0, -1, 0, 0, $subCallCount, $subCallsAddr));	
-		$pos = ftell($this->out);
-
-		fseek($this->out, $here, SEEK_SET);						
+		
+		// Write addresses
+		fseek($out,self::NR_SIZE, SEEK_SET);
+		fwrite($out, pack(self::NR_FORMAT, $headersPos));
+		// Skip function count
+		fseek($out,self::NR_SIZE, SEEK_CUR);
+		foreach($functionAddresses as $address){
+			fwrite($out, pack(self::NR_FORMAT, $address));			
+		}
+		
 	}
 	
-	
-	private function headerFound($line){
-		$this->headers[] = $line;
-	}
 }
-
diff --git a/lib/Reader.php b/lib/Reader.php
index acefc07..26401b7 100644
--- a/lib/Reader.php
+++ b/lib/Reader.php
@@ -1,14 +1,11 @@
 <?php
-// TODO Error handling
 
 class Reader{
-	const FILE_FORMAT_VERSION = 4;
+	const FILE_FORMAT_VERSION = 5;
 
 	const NR_FORMAT = 'V';
 	const NR_SIZE = 4;
-	const INVOCATION_LENGTH = 8;
-	const SUBCALL_LENGTH = 4;
-	
+	const CALLINFORMATION_LENGTH = 4;
 
 	const ENTRY_POINT = '{main}';
 	
@@ -21,27 +18,7 @@ class Reader{
 			throw new Exception('Error opening file!');
 		$this->init();
 	}
-	
-	private function read($numbers=1){
-		$values = unpack(self::NR_FORMAT.$numbers,fread($this->fp,self::NR_SIZE*$numbers));
-		if($numbers==1)
-			return $values[1];
-		else 
-			return array_values($values); // reindex and return
-	}
-	
-	private function readLine(){
-		$result = fgets($this->fp);
-		if($result)
-			return trim($result);
-		else
-			return $result;
-	}
-	
-	private function seek($offset, $whence=SEEK_SET){
-		return fseek($this->fp, $offset, $whence);
-	}
-	
+		
 	// Read initial information from datafile
 	private function init(){
 		list($version, $this->headersPos, $functionCount) = $this->read(3);
@@ -56,55 +33,45 @@ class Reader{
 
 	function getFunctionInfo($nr, $costFormat = 'absolute'){
 		$this->seek($this->functionPos[$nr]);
-		list($totalSelfCost, $totalInclusiveSelfCost, $totalCallCost, $invocationCount) = $this->read(4);
-		$this->seek(self::NR_SIZE*self::INVOCATION_LENGTH*$invocationCount, SEEK_CUR);
+		
+		list($summedSelfCost, $summedInclusiveCost, $invocationCount, $calledFromCount) = $this->read(4);
+		
+		$this->seek(self::NR_SIZE*self::CALLINFORMATION_LENGTH*$calledFromCount, SEEK_CUR);
 		$file = $this->readLine();
 		$function = $this->readLine();
 
 	   	$result = array(
     	    'file'=>$file, 
    		    'functionName'=>$function, 
-   		    'totalSelfCost'=>$totalSelfCost,
-   		    'totalInclusiveSelfCost'=>$totalInclusiveSelfCost, 
-   		    'totalCallCost'=>$totalCallCost, 
-   		    'invocationCount'=>$invocationCount
+   		    'summedSelfCost'=>$summedSelfCost,
+   		    'summedInclusiveCost'=>$summedInclusiveCost, 
+   		    'invocationCount'=>$invocationCount,
+			'callInfoCount'=>$calledFromCount
    		);            
         if ($costFormat == 'percentual') {
-	        $result['totalSelfCost'] = $this->percentCost($result['totalSelfCost']);
-	        $result['totalInclusiveSelfCost'] = $this->percentCost($result['totalInclusiveSelfCost']);
-	        $result['totalCallCost'] = $this->percentCost($result['totalCallCost']);
+	        $result['summedSelfCost'] = $this->percentCost($result['summedSelfCost']);
+	        $result['summedInclusiveCost'] = $this->percentCost($result['summedInclusiveCost']);
 	    }
 
 		return $result;
 	}
 	
-	function getInvocation($functionNr, $invocationNr, $includeSubCalls, $costFormat = 'absolute'){
-		$this->seek($this->functionPos[$functionNr]+self::NR_SIZE*(self::INVOCATION_LENGTH*$invocationNr+4));
-		$data = $this->read(self::INVOCATION_LENGTH);
+	function getCallInfo($functionNr, $calledFromNr, $costFormat = 'absolute'){
+		// 4 = number of numbers beforea call information
+		$this->seek($this->functionPos[$functionNr]+self::NR_SIZE*(self::CALLINFORMATION_LENGTH*$calledFromNr+4));
+		$data = $this->read(self::CALLINFORMATION_LENGTH);
 
 	    $result = array(
-	        'selfCost'=>$data[0], 
-	        'inclusiveSelfCost'=>$data[1], 
-	        'callCost'=>$data[2], 
-	        'calledFromFunction'=>$data[3], 
-	        'calledFromInvocation'=>$data[4], 
-	        'calledFromLine'=>$data[5], 
+	        'functionNr'=>$data[0], 
+	        'line'=>$data[1], 
+	        'callCount'=>$data[2], 
+	        'summedCallCost'=>$data[3]
 	    );
 		
 		if ($costFormat == 'percentual') {
-	        $result['selfCost'] = $this->percentCost($result['selfCost']);
-	        $result['inclusiveSelfCost'] = $this->percentCost($result['inclusiveSelfCost']);
-	        $result['callCost'] = $this->percentCost($result['callCost']);
+	        $result['summedCallCost'] = $this->percentCost($result['summedCallCost']);
 	    }
 
-		if($includeSubCalls){
-			$result['subCalls'] = array();
-			$this->seek($data[7]);
-			for($i=0;$i<$data[6];$i++){
-				$scData = $this->read(self::SUBCALL_LENGTH);
-				$result['subCalls'][] = array('functionNr'=>$scData[0], 'invocationNr'=>$scData[1], 'line'=>$scData[2], 'cost'=>$scData[3]);
-			}
-		}
 		return $result;
 	}
 	
@@ -130,5 +97,24 @@ class Reader{
 		return number_format($result, 3, '.', '');
 	}
 	
+	private function read($numbers=1){
+		$values = unpack(self::NR_FORMAT.$numbers,fread($this->fp,self::NR_SIZE*$numbers));
+		if($numbers==1)
+			return $values[1];
+		else 
+			return array_values($values); // reindex and return
+	}
+	
+	private function readLine(){
+		$result = fgets($this->fp);
+		if($result)
+			return trim($result);
+		else
+			return $result;
+	}
+	
+	private function seek($offset, $whence=SEEK_SET){
+		return fseek($this->fp, $offset, $whence);
+	}
 	
 }