- Improve parsing. Now uses 100% Regex Matching

- Clean up functions. Created new ones in the process.

src/biotree/FileProcessor.java

@@ -12,9 +12,8 @@ import java.util.regex.Pattern;
  * 
  * Notes:
  * - Processes one row at a time
- * - Press enter for the next row. 
  * - Parsed data is in string format
- * - Data will be converted to the appropriate type in a toADT() function
+ * - TOO MANY FUNCTION CALLS?
  */
 public class FileProcessor {
 	private static String path = "src/occurrence.csv";
@@ -36,8 +35,9 @@ public class FileProcessor {
 	}
 	
 	/**
+	 * Initialize Processing.
 	 *  Reads file at path. 
-	 *  Calls parse() automatically
+	 *  Calls parse() for each line
 	 */
 	private static void initProcessing() {
 		FileReader fr;
@@ -52,7 +52,7 @@ public class FileProcessor {
 			
 			br.readLine();	// Reads Past Field Names
 			while ((currentLine = br.readLine()) != null) {
-				System.out.println(currentLine);
+				//System.out.println(currentLine); //Testing ONLY for checking one line at a time
 				parse(currentLine);
 				s.nextLine(); //Testing ONLY for checking one line at a time
 			}
@@ -67,80 +67,87 @@ public class FileProcessor {
 	}
 	
 	/**
-	 * Parses out data from string
+	 * Parses data from string
+	 *  Calls createObject on successful regex matching
 	 * 
 	 * @param currentLine, a line/row of data
+	 * @throws IOException
 	 */
-	// TODO: Fix spahgetti code where taxonId needs to be converted to a string when it is received from biotree.
 	private static void parse(String currentLine) throws IOException {
-		String[] splitLine = currentLine.split(",");
-		String eventId = null, occurId, taxonId = null, individualCount, latitude, longitude, year=null, month= null, day=null;
-		
-		occurId = splitLine[3];
-		individualCount = splitLine[4];
-		
-		Pattern patternEventId = Pattern.compile("OP_ID (\\d+)");
-		Matcher matchEventId = patternEventId.matcher(splitLine[7]);
+		/* Regex Pattern Grouping Guide
+		 * 	Retrieve String Groups with: matchEventId.group(x):
+		 *  group 0: full matched string
+		 *  group 1: occurrenceId
+		 *  group 2: individualCount
+		 *  group 3: eventId
+		 *  group 4: year
+		 *  group 5: month
+		 *  group 6: date
+		 *  group 7: latitude
+		 *  group 8: longitude
+		 *  group 9: taxonId
+		 *  group 10: Scientific Name
+		 */
+		Pattern patternEventId = Pattern.compile("([^,]+)?,([^,]+)?,[^,]+,[^,]+,[^,]+OP_ID (\\d+)?,(\\d+)?-(\\d+)?-(\\d+)?,(\\d+.\\d+)?,(-\\d+.\\d+)?,[^,\\d]+(\\d+)?,([^,]+)?");
+		Matcher matchEventId = patternEventId.matcher(currentLine);
+
 		if(matchEventId.find()) {
-			eventId = matchEventId.group(1);
-		}
-		else {
-			// TODO: Throw Exception?
-			System.out.println("Could not parse eventId. String may be unique or missing.");
-		}
-		
-		Pattern patternDate = Pattern.compile("(\\d+)-(\\d+)-(\\d+)");
-		Matcher matchDate = patternDate.matcher(splitLine[8]);
-		if(matchDate.find()) {
-			year = matchDate.group(1);
-			month = matchDate.group(2);
-			day = matchDate.group(3);
+			createObjects(matchEventId);
 		}
 		else {
 			// TODO: Throw Exception?
-			System.out.println("Could not parse eventId. String may be unique or missing.");
-		}
-		latitude = splitLine[9];
-		longitude = splitLine[10];
-		
-		Pattern patternTaxId = Pattern.compile(":(\\d+)");
-		Matcher matchTaxId = patternTaxId.matcher(splitLine[11]);
-		if(matchTaxId.find()) {
-			taxonId = matchTaxId.group(1);
-		}
-		else {
-			// TODO: Throw Exception?
-			System.out.println("Could not parse TaxId. String may be unique or missing.");
-		}
+			System.out.println("Regex Matching Failed.");
+		}		
 		
+		// Testing ONLY Print lines
+		System.out.println("Full String:" + matchEventId.group(0));
+		System.out.println("Occurence Id:" + matchEventId.group(1));
+		System.out.println("Ind. Count:" + matchEventId.group(2));
+		System.out.println("event Id:" + matchEventId.group(3));
+		System.out.println("Year:" + matchEventId.group(4));
+		System.out.println("Month:" + matchEventId.group(5));
+		System.out.println("Day:" + matchEventId.group(6));
+		System.out.println("lat:" + matchEventId.group(7));
+		System.out.println("long:" + matchEventId.group(8));
+		System.out.println("tax Id:" + matchEventId.group(9));
+		System.out.println("Scientific Name:" + matchEventId.group(10));
+	}
+	
+	/**
+	 * Calls BioTree's processRecord and another method to create a Record
+	 * 
+	 * @param matchEventId
+	 */
+	public static void createObjects(Matcher matchEventId) {
 		// Call BioTree
-		if(taxonId != null) {
-			BioTree.processRecord(Integer.parseInt(taxonId));
+		if(matchEventId.group(9) != null) {
+			BioTree.processRecord(Integer.parseInt(matchEventId.group(9)));
+			// Create Record Object
+			createRecord(Integer.parseInt(matchEventId.group(3)), matchEventId.group(1), Integer.parseInt(matchEventId.group(9)), Integer.parseInt(matchEventId.group(2)), Float.parseFloat(matchEventId.group(7)), Float.parseFloat(matchEventId.group(8)), Integer.parseInt(matchEventId.group(4)), Integer.parseInt(matchEventId.group(5)), Integer.parseInt(matchEventId.group(6)));
 		}
-		else if (splitLine[12] != "NA") {
+		else if (matchEventId.group(10) != "NA") {
 			try{
-				taxonId = Integer.toString(BioTree.processRecord(splitLine[12]));
+				createRecord(Integer.parseInt(matchEventId.group(3)), matchEventId.group(1), BioTree.processRecord(matchEventId.group(10)), Integer.parseInt(matchEventId.group(2)), Float.parseFloat(matchEventId.group(7)), Float.parseFloat(matchEventId.group(8)), Integer.parseInt(matchEventId.group(4)), Integer.parseInt(matchEventId.group(5)), Integer.parseInt(matchEventId.group(6)));
 			} catch(IOException e) {
-				System.out.println("No Taxon ID or Scientific Name");
+				System.out.println("No Taxon ID or Scientific Name. OR createRecord Error");
 			}
 		}		
-		
-		// Create Record Object
-		createRecord(Integer.parseInt(eventId), occurId, Integer.parseInt(taxonId), Integer.parseInt(individualCount), Float.parseFloat(latitude), Float.parseFloat(longitude), Integer.parseInt(year), Integer.parseInt(month), Integer.parseInt(day));
-		
-		
-		// Testing ONLY Print lines
-//		System.out.println("Occurence Id:" + occurId);
-//		System.out.println("Ind. Count:" + individualCount);
-//		System.out.println("event Id:" + eventId);
-//		System.out.println("Year:" + year);
-//		System.out.println("Month:" + month);
-//		System.out.println("Day:" + day);
-//		System.out.println("lat:" + latitude);
-//		System.out.println("long:" + longitude);
-//		System.out.println("tax Id:" + taxonId);
 	}
 	
+	/**
+	 * Create a Record Object
+	 * 
+	 * @param eventId
+	 * @param occurId
+	 * @param taxonId
+	 * @param individualCount
+	 * @param latitude
+	 * @param longitude
+	 * @param year
+	 * @param month
+	 * @param day
+	 * @return
+	 */
 	public static Record createRecord(int eventId, String occurId, int taxonId, int individualCount, float latitude, float longitude, int year, int month, int day) {
 		return new Record(eventId, occurId, taxonId, individualCount, latitude, longitude, year, month, day);
 	}