Fix critical security and data integrity issues in BLM parser

phpblm.php

@@ -2,67 +2,143 @@
 
 /**
  * Parser for Rightmove's lovely BLM files.
+ *
+ * BLM (Bulk Load Media) format structure:
+ * - Section delimited by '#' characters
+ * - Header section contains EOF (End Of Field) and EOR (End Of Record) delimiters
+ * - Definition section lists field names
+ * - Data section contains property records
  */
 
 class phpblm {
 	private $blm;
 	private $header = array();
 	private $def = array();
 	private $data = array();
-
+
+	// Maximum file size: 50MB to prevent memory exhaustion
+	const MAX_FILE_SIZE = 52428800;
+
 	public function __construct($file) {
+		// Validate file path
+		if (!is_string($file) || empty($file)) {
+			throw new InvalidArgumentException('File path must be a non-empty string');
+		}
+
+		if (!file_exists($file)) {
+			throw new InvalidArgumentException("File does not exist: $file");
+		}
+
+		if (!is_readable($file)) {
+			throw new InvalidArgumentException("File is not readable: $file");
+		}
+
+		// Check file size to prevent memory exhaustion
+		$fileSize = filesize($file);
+		if ($fileSize === false) {
+			throw new RuntimeException("Cannot determine file size: $file");
+		}
+
+		if ($fileSize > self::MAX_FILE_SIZE) {
+			throw new RuntimeException("File too large. Maximum size: " . self::MAX_FILE_SIZE . " bytes");
+		}
+
+		// Read file with error handling
 		$this->blm = file_get_contents($file);
+		if ($this->blm === false) {
+			throw new RuntimeException("Failed to read file: $file");
+		}
+
 		$this->splitPieces();
 	}
-	
+
 	// Return specific field from row
 	public function getData($data, $row) {
+		if (!isset($this->data[$row])) {
+			throw new OutOfBoundsException("Row index $row does not exist");
+		}
+
+		if (!isset($this->data[$row][$data])) {
+			throw new OutOfBoundsException("Field '$data' does not exist in row $row");
+		}
+
 		return $this->data[$row][$data];
 	}
-	
+
 	// Return header info
 	public function getHeader($hdr) {
+		if (!isset($this->header[$hdr])) {
+			throw new OutOfBoundsException("Header field '$hdr' does not exist");
+		}
+
 		return $this->header[$hdr];
 	}
-	
+
 	// This will return the actual number of properties, regardless of what the
 	// header might say. To retrieve the header value (if present), use:
 	// $blm->getHeader('Property Count');
 	public function propCount() {
 		return count($this->data);
 	}
-	
+
 	public function properties() {
 		return $this->data;
 	}
-	
+
 	// Splits the BLM data into constituent parts
 	private function splitPieces() {
 		$pieces = explode("#", $this->blm);
-
+
+		// Validate we have enough sections
+		if (count($pieces) < 7) {
+			throw new RuntimeException("Invalid BLM format: insufficient sections (expected at least 7, got " . count($pieces) . ")");
+		}
+
 		// Get the header (includes EOF/EOR stuff)
 		$header = explode("\n", trim($pieces[2]));
 		foreach ($header as $h) {
 			$h = preg_replace("/\'/", "", $h);	// Remove quotes on EOF/EOR
-			$h_pieces = explode(" : ", $h);
-			$this->header[trim($h_pieces[0])] = trim($h_pieces[1]);
+			$h_pieces = explode(" : ", $h, 2);  // Limit to 2 pieces in case value contains ' : '
+
+			// Validate header line format
+			if (count($h_pieces) >= 2) {
+				$this->header[trim($h_pieces[0])] = trim($h_pieces[1]);
+			}
 		}
-
+
+		// Validate required headers exist
+		if (!isset($this->header['EOF'])) {
+			throw new RuntimeException("Invalid BLM format: EOF delimiter not found in header");
+		}
+
+		if (!isset($this->header['EOR'])) {
+			throw new RuntimeException("Invalid BLM format: EOR delimiter not found in header");
+		}
+
 		// Get the definitions
 		$def = explode($this->header['EOF'], trim($pieces[4]));
 		foreach ($def as $d) {
 			$this->def[] = $d;
 		}
-		
+
 		// Get the data
 		$data = explode($this->header['EOR'], trim($pieces[6]));
-		$datac = count($data);
+		$dataCount = count($data);
+
+		for ($i = 0; $i < $dataCount; $i++) {
+			// CRITICAL FIX: Initialize $row_arr for each iteration to prevent data leakage
+			$row_arr = array();
 
-		for ($i=0; $i<$datac; $i++) {
 			$row = explode($this->header['EOF'], trim($data[$i]));
-			for ($j=0; $j<count($row); $j++) {
-				$row_arr[$this->def[$j]] = $row[$j];
+			$rowLength = count($row);
+
+			for ($j = 0; $j < $rowLength; $j++) {
+				// Bounds check: ensure definition exists for this field
+				if (isset($this->def[$j])) {
+					$row_arr[$this->def[$j]] = $row[$j];
+				}
 			}
+
 			$this->data[] = $row_arr;
 		}
 	}