dkcreatto · February 25, 2012 19:27
diff --git a/BBM_CSV_2_XML.php b/BBM_CSV_2_XML.php
 <?php
 /*
 *BBM to XML version 1.0
 *
 * This script will take CSV file generated by
 * Blackberry Messenger and convert it to well
 * formatted xml document which can be imported
 * into any spreadsheet program such as MS Excel,
 * OpenOffice Calc etc.
 *
 * This script came into existence due to my
 * frustration in backing up chat history which
 * was (luckily) stored as csv files in memory 
 * card when BBM is properly configured.
 *
 *
 * This script is made to be used in php in command
 * line in windows. Although, it can be also used in
 * any other os due to php's portable code structure.
 *
 * To configure php to be used as command line utility
 * follow this URI: http://www.youtube.com/watch?v=AEyzKm2pTBw
 *
 */

 $stime = new DateTime(); //Registers starting time of script. To be used in log.

 if (isset($argv[1])) {
 	echo "\n\nInput Directory path  ".$inputDir 	= 					//Stores and prints on screen Input directory consisting of BBM csv files
 													is_dir($argv[1])	//Checks if first parameter (http://php.net/manual/en/reserved.variables.argv.php) passed to script is a directory
 													? $argv[1] 			//If TRUE, then set that parameter to be Input Directory
 													: dirname($argv[1]);//If FALSE, then get the parent directory path and set it as input directory
 } else { 
 	die("Please provide path to BBM CSV file or the folder containing them.");
 }

 echo "\n\nOutput Directory path ".$outputDir 	= 						//Stores and prints on screen Output directory name where xml and logfile will be stored
 													isset($argv[2]) && is_dir($argv[2]) //checks if second parameter is set and is a directory path.
 													? $argv[2] 			//If TRUE, then set that parameter to be Output Directory
 													: $inputDir;		//If FALSE, then set Input Directory as Output Directory	

 $files = 	is_dir($argv[1]) 			//Checks if first parameter is a directory
 			? glob("$argv[1]\*.csv") 	//If TRUE, get all the files with csv extension and save their full paths in a array
 			: $argv[1];					//If FALSE, deems first parameter as a file and sets it to be processed on
 			
 $bb_xml_file_name = (isset($argv[2]) && !is_dir($argv[2]) && pathinfo("$argv[2]",PATHINFO_EXTENSION) == "xml") 		//Checkes if parameter two is set, if it's not a directory and it's an xml file
 					? pathinfo("$argv[2]",PATHINFO_BASENAME) 	//If TRUE, ie. if para 2 is a full path to file, then get only name of file with extension_loaded
 					: "BBMExport-dk_".date("dmYHi").".xml";		//If FALSE, create a new name for xml file
 $bb_xml_file = fopen("$outputDir/$bb_xml_file_name", "w");  	//Creates or overwrites xml file and stores handle as bb_xml_file object. NOTE: here "w" parameter used for fopen to overwrite exported file each time to avoid duplication.

 $bb_xml_data  ='<?xml version="1.0" encoding="UTF-8" ?>' . "\n";//Creates variable in which we will store xml data with first line being dom declaration
 $bb_xml_data .="<bb_history>\n";								//Adds xml root element in xml data by appending it to above string
 fwrite($bb_xml_file,$bb_xml_data); 								//Stores/writes XML header to xml file

 $sf = array();	//declaring empty array of successfully processed files, so can be used globally later in script
 $ef = array();	//declaring empty array of unsuccessfully processed files, so can be used globally later in script
 $dateLimit = date_create(1/1/2012); //Sets date limit if we just want to capture messages since a specific date only. Experimental. Currently not used.

 //Basic information to let user know what's happening
 echo "\n\n\nI'm working. Please wait.. ;)\nProcessing ";
 echo is_array($files) ? count($files)." files.\n" : "1 file.\n";
 echo "\n\n****************************************\n\n";

 //If $files is an array containing a list of files then process each file one by one, if it's single file then just process it
 If (is_array($files)) {
 		foreach ($files as $fname) 
 		{
 			if (filesize($fname) > 0) { //Checks if file has any data
 				$bb_r = bb_routine($fname,$bb_xml_file);	//Passes values to main bb_routine function to process file and saves results as an array in $bb_r
 				($bb_r[0] == 1) ? $sf[] = $bb_r[1] : $ef[] = $bb_r[1]; //If $bb_r[0] is 1 then all file was successfully processed, else there was failiure. Stores the successful files in $sf array and unsuccessful ones in $ef array.
 			} 
 			else 
 			{
 				$ef[] = $fname." is zero sized"; //If file has no data then stores message containing filename in $ef array
 			}
 		}
 	}
 	else {
 			(filesize($files) > 0 && pathinfo($files,PATHINFO_EXTENSION) == "csv") ? $bb_r = bb_routine($files,$bb_xml_file) : $ef[] = $files." is zero sized"; //Same as above, here is for if only single file is passed
 	}

 	//MAIN processing function which carries out all the extractions
 	function bb_routine($file_name,$bb_xml_file) {
 	
 		$bb_xml_data = "";	//On each call, ie for each file, clears the previous file's $bb_xml_data so memmory don't get overloaded
 		$handle = is_readable($file_name) ? fopen($file_name,"r") : die("Something went wrong. Error for $file_name.");	//creates a new handle object to read the passed csv file
 		echo pathinfo($file_name, PATHINFO_BASENAME)."\n"; //Prints current file name being processed to screen
 		$input_line = fgets($handle); // Reads the first line of the file, which we will discard
 		$input_line = fgets($handle); //read a line of text from the file
 		$bb_msg_data = '';	//Declaring string variable to store BBM message meta data like date, sender's pin and receiver's pin 
 		$bb_msg_text = '';	//Declaring string variable to store actual BBM message text
 		$i = 0;
 		$bb_msg_count = 0;
 		$first_run = TRUE;
 		
 		while ($input_line) // Check whether we've reached the end of the file, 
 							 // fgets returns FALSE if we have 
 		{	
 		  if (new_bbm_message($input_line)) 
 		  {
 			$bb_msg_count++;
 			if(!$first_run) 
 			{
 				//You now have a complete message, do something with it. 

 				$msg_data = str_getcsv($bb_msg_data,','); //Spliting message data into array, to be stored in field variables

 				$dF = date_create("@".mb_substr($msg_data[0],8,10));
 				date_timezone_set($dF, timezone_open('Asia/Calcutta'));
 				$dateField = "<dateField><date>".date_format($dF,"d-m-Y")."</date><time>".date_format($dF,"H:i:s")."</time></dateField>\n";
 				$sPIN =  "<sender>".$msg_data[1]."</sender>\n";
 				$rPIN =  "<reciever>".$msg_data[2]."</reciever>\n";
 				//$bb_msg_text = '<msgText> '. safe_xml(mb_substr($bb_msg_text,0,-2)) .' </msgText>'."\n";
 				$bb_msg_text = '<msgText><![CDATA[ '. safe_xml(mb_substr($bb_msg_text,0,-2)) .' ]]></msgText>'."\n";

 				//Now lets store whole data into a message container 
 				$msg = "<msg>\n".$dateField.$sPIN.$rPIN.$bb_msg_text."</msg>\r\n";
 				$bb_xml_data .= utf8_encode($msg);

 				$i++;
 			} else { $first_run = FALSE; }

 			// Grab the new messages data
 			$bb_msg_data = mb_substr($input_line,0,43); // Get string with Date,SendersPIN,RecieversPIN up to the ',' after receivers pin.
 			$bb_msg_text = mb_substr($input_line,44);
 		  } 
 		  else 
 		  {
 		  // This is another line of text of the message.
 			$bb_msg_text .= $input_line;
 		  }
 		 
 			// Read in the next line of text.
 			$input_line = fgets($handle);     
 		}
 		
 		
 		//Close current file
 		fclose($handle);
 		
 	if ($i == 0 && $bb_msg_count > 0) { //If file has only one record data won't be converted to record & $i wont increase because of "first run" condition

 		$msg_data = str_getcsv($bb_msg_data,','); //Spliting message data into array, to be stored in field variables

 		$dF = date_create("@".mb_substr($msg_data[0],8,10));
 		date_timezone_set($dF, timezone_open('Asia/Calcutta'));
 		$dateField = "<dateField><date>".date_format($dF,"d-m-Y")."</date><time>".date_format($dF,"H:i:s")."</time></dateField>\n";
 		$sPIN =  "<sender>".$msg_data[1]."</sender>\n";
 		$rPIN =  "<reciever>".$msg_data[2]."</reciever>\n";
 		$bb_msg_text = '<msgText><![CDATA[ '. safe_xml(mb_substr($bb_msg_text,0,-2)) .' ]]></msgText>'."\n";

 		//Now lets store whole data into a message container 
 		$msg = "<msg>\n".$dateField.$sPIN.$rPIN.$bb_msg_text."</msg>\n";
 		$bb_xml_data .= utf8_encode($msg);
 	}
 	//Write captured data to XML file everytime when a record is created in memory.
 	
 		fwrite($bb_xml_file,$bb_xml_data) ? $sf = $file_name : $ef = $file_name;
 		$success = 1;
 		$error = 0;
 		if(isset($sf)) { return array($success, $sf); }
 		if(isset($ef)) { return array($error, $ef); }
 		
 	}
 //BB routine ends here


 	//Function to check if current line is starting of a new record or new line in message. Checks if first 21 characters are combination of YYYYMMDD+Epoch(milliseconds)
 	function new_bbm_message($cln) {
 		if (strlen($cln) > 21 && ($d2 = date_create("@".mb_substr($cln,8,10)))) {


 		$d1 = mb_substr($cln,0,8);

 			date_timezone_set($d2, timezone_open('Asia/Calcutta'));
 			$d3 = $d2;
 			$d2 = date_format($d2,"Ymd");
 			
 			$isDate = ($d1 === $d2) ? TRUE : FALSE;
 			return $isDate;
 		} else {
 		return FALSE;
 		}
 	}
 	


 fwrite($bb_xml_file,"</bb_history>"); //Ends XML file with ending root tag.
 fclose($bb_xml_file); //Closes the XML file link.

 // Escaping ANSI control characters
 	function safe_xml($stri) {
 		
 		$charcode = array();
 		for ($i=0, $j=0; $i<32; $i++) {
 			if ($i != 10 AND $i != 13) { 
 				$j = chr($i);
 				$charcode[$j] = chr(32); 
 				$j++;
 			} 
 		}
 		$newstri = strtr($stri, $charcode);
 		$newstri = utf8_encode($newstri);
 		return $newstri;
 	}


 echo "\nFile operation done...\n\nOutput file: $bb_xml_file_name\nTotal files processed: ". count($files) ."\nSuccess on: ". count($sf) ." files.\nError occured on following: ". count($ef) ." files\n";
 foreach ($ef as $errfile) { echo $errfile."\n";}

 $etime = new DateTime();
 $totaltime = date_diff($stime,$etime);
 echo "\n\n****************************************\n\n";
 $logfile = date("YdmHis").".log"; 
 $logdata = "Input directory: $inputDir\nOutput Directory: $outputDir\nXML File name: $bb_xml_file_name\n\n\nStarted on: ".$stime->format("r")."\nEnded on: ".$etime->format("r")."\nTotal time taken: ".$totaltime->format('%I:%S minutes')."\nSusscessful completed files:\n".implode("\n",$sf)."\n\nFiles with errors: \n".implode("\n",$ef); 
 file_put_contents("$outputDir/$logfile",$logdata,FILE_APPEND);
 echo "\nStarted on: ".$stime->format("r")."\nEnded on: ".$etime->format("r")."\nTotal time taken: ".$totaltime->format('%I:%S minutes')."\nLog file: $logfile";
 die();

 ?>
	<?php
	/*
	*BBM to XML version 1.0
	*
	* This script will take CSV file generated by
	* Blackberry Messenger and convert it to well
	* formatted xml document which can be imported
	* into any spreadsheet program such as MS Excel,
	* OpenOffice Calc etc.
	*
	* This script came into existence due to my
	* frustration in backing up chat history which
	* was (luckily) stored as csv files in memory
	* card when BBM is properly configured.
	*
	*
	* This script is made to be used in php in command
	* line in windows. Although, it can be also used in
	* any other os due to php's portable code structure.
	*
	* To configure php to be used as command line utility
	* follow this URI: http://www.youtube.com/watch?v=AEyzKm2pTBw
	*
	*/

	$stime = new DateTime(); //Registers starting time of script. To be used in log.

	if (isset($argv[1])) {
	echo "\n\nInput Directory path ".$inputDir = //Stores and prints on screen Input directory consisting of BBM csv files
	is_dir($argv[1]) //Checks if first parameter (http://php.net/manual/en/reserved.variables.argv.php) passed to script is a directory
	? $argv[1] //If TRUE, then set that parameter to be Input Directory
	: dirname($argv[1]);//If FALSE, then get the parent directory path and set it as input directory
	} else {
	die("Please provide path to BBM CSV file or the folder containing them.");
	}

	echo "\n\nOutput Directory path ".$outputDir = //Stores and prints on screen Output directory name where xml and logfile will be stored
	isset($argv[2]) && is_dir($argv[2]) //checks if second parameter is set and is a directory path.
	? $argv[2] //If TRUE, then set that parameter to be Output Directory
	: $inputDir; //If FALSE, then set Input Directory as Output Directory

	$files = is_dir($argv[1]) //Checks if first parameter is a directory
	? glob("$argv[1]\*.csv") //If TRUE, get all the files with csv extension and save their full paths in a array
	: $argv[1]; //If FALSE, deems first parameter as a file and sets it to be processed on

	$bb_xml_file_name = (isset($argv[2]) && !is_dir($argv[2]) && pathinfo("$argv[2]",PATHINFO_EXTENSION) == "xml") //Checkes if parameter two is set, if it's not a directory and it's an xml file
	? pathinfo("$argv[2]",PATHINFO_BASENAME) //If TRUE, ie. if para 2 is a full path to file, then get only name of file with extension_loaded
	: "BBMExport-dk_".date("dmYHi").".xml"; //If FALSE, create a new name for xml file
	$bb_xml_file = fopen("$outputDir/$bb_xml_file_name", "w"); //Creates or overwrites xml file and stores handle as bb_xml_file object. NOTE: here "w" parameter used for fopen to overwrite exported file each time to avoid duplication.

	$bb_xml_data ='<?xml version="1.0" encoding="UTF-8" ?>' . "\n";//Creates variable in which we will store xml data with first line being dom declaration
	$bb_xml_data .="<bb_history>\n"; //Adds xml root element in xml data by appending it to above string
	fwrite($bb_xml_file,$bb_xml_data); //Stores/writes XML header to xml file

	$sf = array(); //declaring empty array of successfully processed files, so can be used globally later in script
	$ef = array(); //declaring empty array of unsuccessfully processed files, so can be used globally later in script
	$dateLimit = date_create(1/1/2012); //Sets date limit if we just want to capture messages since a specific date only. Experimental. Currently not used.

	//Basic information to let user know what's happening
	echo "\n\n\nI'm working. Please wait.. ;)\nProcessing ";
	echo is_array($files) ? count($files)." files.\n" : "1 file.\n";
	echo "\n\n****************************************\n\n";

	//If $files is an array containing a list of files then process each file one by one, if it's single file then just process it
	If (is_array($files)) {
	foreach ($files as $fname)
	{
	if (filesize($fname) > 0) { //Checks if file has any data
	$bb_r = bb_routine($fname,$bb_xml_file); //Passes values to main bb_routine function to process file and saves results as an array in $bb_r
	($bb_r[0] == 1) ? $sf[] = $bb_r[1] : $ef[] = $bb_r[1]; //If $bb_r[0] is 1 then all file was successfully processed, else there was failiure. Stores the successful files in $sf array and unsuccessful ones in $ef array.
	}
	else
	{
	$ef[] = $fname." is zero sized"; //If file has no data then stores message containing filename in $ef array
	}
	}
	}
	else {
	(filesize($files) > 0 && pathinfo($files,PATHINFO_EXTENSION) == "csv") ? $bb_r = bb_routine($files,$bb_xml_file) : $ef[] = $files." is zero sized"; //Same as above, here is for if only single file is passed
	}

	//MAIN processing function which carries out all the extractions
	function bb_routine($file_name,$bb_xml_file) {

	$bb_xml_data = ""; //On each call, ie for each file, clears the previous file's $bb_xml_data so memmory don't get overloaded
	$handle = is_readable($file_name) ? fopen($file_name,"r") : die("Something went wrong. Error for $file_name."); //creates a new handle object to read the passed csv file
	echo pathinfo($file_name, PATHINFO_BASENAME)."\n"; //Prints current file name being processed to screen
	$input_line = fgets($handle); // Reads the first line of the file, which we will discard
	$input_line = fgets($handle); //read a line of text from the file
	$bb_msg_data = ''; //Declaring string variable to store BBM message meta data like date, sender's pin and receiver's pin
	$bb_msg_text = ''; //Declaring string variable to store actual BBM message text
	$i = 0;
	$bb_msg_count = 0;
	$first_run = TRUE;

	while ($input_line) // Check whether we've reached the end of the file,
	// fgets returns FALSE if we have
	{
	if (new_bbm_message($input_line))
	{
	$bb_msg_count++;
	if(!$first_run)
	{
	//You now have a complete message, do something with it.

	$msg_data = str_getcsv($bb_msg_data,','); //Spliting message data into array, to be stored in field variables

	$dF = date_create("@".mb_substr($msg_data[0],8,10));
	date_timezone_set($dF, timezone_open('Asia/Calcutta'));
	$dateField = "<dateField><date>".date_format($dF,"d-m-Y")."</date><time>".date_format($dF,"H:i:s")."</time></dateField>\n";
	$sPIN = "<sender>".$msg_data[1]."</sender>\n";
	$rPIN = "<reciever>".$msg_data[2]."</reciever>\n";
	//$bb_msg_text = '<msgText> '. safe_xml(mb_substr($bb_msg_text,0,-2)) .' </msgText>'."\n";
	$bb_msg_text = '<msgText><![CDATA[ '. safe_xml(mb_substr($bb_msg_text,0,-2)) .' ]]></msgText>'."\n";

	//Now lets store whole data into a message container
	$msg = "<msg>\n".$dateField.$sPIN.$rPIN.$bb_msg_text."</msg>\r\n";
	$bb_xml_data .= utf8_encode($msg);

	$i++;
	} else { $first_run = FALSE; }

	// Grab the new messages data
	$bb_msg_data = mb_substr($input_line,0,43); // Get string with Date,SendersPIN,RecieversPIN up to the ',' after receivers pin.
	$bb_msg_text = mb_substr($input_line,44);
	}
	else
	{
	// This is another line of text of the message.
	$bb_msg_text .= $input_line;
	}

	// Read in the next line of text.
	$input_line = fgets($handle);
	}


	//Close current file
	fclose($handle);

	if ($i == 0 && $bb_msg_count > 0) { //If file has only one record data won't be converted to record & $i wont increase because of "first run" condition

	$msg_data = str_getcsv($bb_msg_data,','); //Spliting message data into array, to be stored in field variables

	$dF = date_create("@".mb_substr($msg_data[0],8,10));
	date_timezone_set($dF, timezone_open('Asia/Calcutta'));
	$dateField = "<dateField><date>".date_format($dF,"d-m-Y")."</date><time>".date_format($dF,"H:i:s")."</time></dateField>\n";
	$sPIN = "<sender>".$msg_data[1]."</sender>\n";
	$rPIN = "<reciever>".$msg_data[2]."</reciever>\n";
	$bb_msg_text = '<msgText><![CDATA[ '. safe_xml(mb_substr($bb_msg_text,0,-2)) .' ]]></msgText>'."\n";

	//Now lets store whole data into a message container
	$msg = "<msg>\n".$dateField.$sPIN.$rPIN.$bb_msg_text."</msg>\n";
	$bb_xml_data .= utf8_encode($msg);
	}
	//Write captured data to XML file everytime when a record is created in memory.

	fwrite($bb_xml_file,$bb_xml_data) ? $sf = $file_name : $ef = $file_name;
	$success = 1;
	$error = 0;
	if(isset($sf)) { return array($success, $sf); }
	if(isset($ef)) { return array($error, $ef); }

	}
	//BB routine ends here


	//Function to check if current line is starting of a new record or new line in message. Checks if first 21 characters are combination of YYYYMMDD+Epoch(milliseconds)
	function new_bbm_message($cln) {
	if (strlen($cln) > 21 && ($d2 = date_create("@".mb_substr($cln,8,10)))) {


	$d1 = mb_substr($cln,0,8);

	date_timezone_set($d2, timezone_open('Asia/Calcutta'));
	$d3 = $d2;
	$d2 = date_format($d2,"Ymd");

	$isDate = ($d1 === $d2) ? TRUE : FALSE;
	return $isDate;
	} else {
	return FALSE;
	}
	}



	fwrite($bb_xml_file,"</bb_history>"); //Ends XML file with ending root tag.
	fclose($bb_xml_file); //Closes the XML file link.

	// Escaping ANSI control characters
	function safe_xml($stri) {

	$charcode = array();
	for ($i=0, $j=0; $i<32; $i++) {
	if ($i != 10 AND $i != 13) {
	$j = chr($i);
	$charcode[$j] = chr(32);
	$j++;
	}
	}
	$newstri = strtr($stri, $charcode);
	$newstri = utf8_encode($newstri);
	return $newstri;
	}


	echo "\nFile operation done...\n\nOutput file: $bb_xml_file_name\nTotal files processed: ". count($files) ."\nSuccess on: ". count($sf) ." files.\nError occured on following: ". count($ef) ." files\n";
	foreach ($ef as $errfile) { echo $errfile."\n";}

	$etime = new DateTime();
	$totaltime = date_diff($stime,$etime);
	echo "\n\n****************************************\n\n";
	$logfile = date("YdmHis").".log";
	$logdata = "Input directory: $inputDir\nOutput Directory: $outputDir\nXML File name: $bb_xml_file_name\n\n\nStarted on: ".$stime->format("r")."\nEnded on: ".$etime->format("r")."\nTotal time taken: ".$totaltime->format('%I:%S minutes')."\nSusscessful completed files:\n".implode("\n",$sf)."\n\nFiles with errors: \n".implode("\n",$ef);
	file_put_contents("$outputDir/$logfile",$logdata,FILE_APPEND);
	echo "\nStarted on: ".$stime->format("r")."\nEnded on: ".$etime->format("r")."\nTotal time taken: ".$totaltime->format('%I:%S minutes')."\nLog file: $logfile";
	die();

	?>
No results found