From 24d2e8155618208393100b9ba41ec0e2c656e903 Mon Sep 17 00:00:00 2001
From: Mathieu Baudier <mbaudier@argeo.org>
Date: Sat, 25 Feb 2012 12:47:40 +0000
Subject: [PATCH] Improve CSV parser and writer

git-svn-id: https://svn.argeo.org/commons/trunk@5112 4cfe0d0a-d680-48aa-b62c-e0a02a3f76cc
---
 .../main/java/org/argeo/util/CsvParser.java   | 18 +++++++-
 .../main/java/org/argeo/util/CsvWriter.java   | 41 +++++++++++++------
 2 files changed, 44 insertions(+), 15 deletions(-)

diff --git a/basic/runtime/org.argeo.basic.nodeps/src/main/java/org/argeo/util/CsvParser.java b/basic/runtime/org.argeo.basic.nodeps/src/main/java/org/argeo/util/CsvParser.java
index 3a429f4ad..7e218c55c 100644
--- a/basic/runtime/org.argeo.basic.nodeps/src/main/java/org/argeo/util/CsvParser.java
+++ b/basic/runtime/org.argeo.basic.nodeps/src/main/java/org/argeo/util/CsvParser.java
@@ -71,14 +71,19 @@ public abstract class CsvParser {
 
 			String line = null;
 			lines: while ((line = reader.readLine()) != null) {
+				line = preProcessLine(line);
+				if (line == null) {
+					// skip line
+					continue lines;
+				}
 				lineCount++;
 				List<String> tokens = new ArrayList<String>();
 				StringBuffer currStr = new StringBuffer("");
 				Boolean wasInquote = false;
-				while (parseLine(line, tokens, currStr, wasInquote)) {
+				sublines: while (parseLine(line, tokens, currStr, wasInquote)) {
 					line = reader.readLine();
 					if (line == null)
-						break;
+						break sublines;
 					wasInquote = true;
 				}
 				if (!noHeader && strictLineAsLongAsHeader) {
@@ -111,6 +116,15 @@ public abstract class CsvParser {
 		}
 	}
 
+	/**
+	 * Called before each (logical) line is processed, giving a change to modify
+	 * it (typically for cleaning dirty files). To be overridden, return the
+	 * line unchanged by default. Skip the line if 'null' is returned.
+	 */
+	protected String preProcessLine(String line) {
+		return line;
+	}
+
 	/**
 	 * Parses a line character by character for performance purpose
 	 * 
diff --git a/basic/runtime/org.argeo.basic.nodeps/src/main/java/org/argeo/util/CsvWriter.java b/basic/runtime/org.argeo.basic.nodeps/src/main/java/org/argeo/util/CsvWriter.java
index 85356e4fe..2167af1ad 100644
--- a/basic/runtime/org.argeo.basic.nodeps/src/main/java/org/argeo/util/CsvWriter.java
+++ b/basic/runtime/org.argeo.basic.nodeps/src/main/java/org/argeo/util/CsvWriter.java
@@ -2,7 +2,9 @@ package org.argeo.util;
 
 import java.io.IOException;
 import java.io.OutputStream;
-import java.io.PrintWriter;
+import java.io.OutputStreamWriter;
+import java.io.UnsupportedEncodingException;
+import java.io.Writer;
 import java.util.Iterator;
 import java.util.List;
 
@@ -10,21 +12,33 @@ import org.argeo.ArgeoException;
 
 /** Write in CSV format. */
 public class CsvWriter {
-	private final PrintWriter out;
+	private final Writer out;
 
 	private char separator = ',';
 	private char quote = '\"';
 
 	/**
-	 * Creates a CSV writer. The header will be written immediately to the
-	 * stream.
+	 * Creates a CSV writer.
 	 * 
 	 * @param out
 	 *            the stream to write to. Caller is responsible for closing it.
 	 */
 	public CsvWriter(OutputStream out) {
-		super();
-		this.out = new PrintWriter(out);
+		this.out = new OutputStreamWriter(out);
+	}
+
+	/**
+	 * Creates a CSV writer.
+	 * 
+	 * @param out
+	 *            the stream to write to. Caller is responsible for closing it.
+	 */
+	public CsvWriter(OutputStream out, String encoding) {
+		try {
+			this.out = new OutputStreamWriter(out, encoding);
+		} catch (UnsupportedEncodingException e) {
+			throw new ArgeoException("Cannot initialize CSV writer", e);
+		}
 	}
 
 	/**
@@ -38,9 +52,9 @@ public class CsvWriter {
 			while (it.hasNext()) {
 				writeToken(it.next().toString());
 				if (it.hasNext())
-					out.print(separator);
+					out.write(separator);
 			}
-			out.print('\n');
+			out.write('\n');
 			out.flush();
 		} catch (IOException e) {
 			throw new ArgeoException("Could not write " + tokens, e);
@@ -57,9 +71,9 @@ public class CsvWriter {
 			for (int i = 0; i < tokens.length; i++) {
 				writeToken(tokens[i].toString());
 				if (i != (tokens.length - 1))
-					out.print(separator);
+					out.write(separator);
 			}
-			out.print('\n');
+			out.write('\n');
 			out.flush();
 		} catch (IOException e) {
 			throw new ArgeoException("Could not write " + tokens, e);
@@ -70,6 +84,7 @@ public class CsvWriter {
 		// +2 for possible quotes, another +2 assuming there would be an already
 		// quoted string where quotes needs to be duplicated
 		// another +2 for safety
+		// we don't want to increase buffer size while writing
 		StringBuffer buf = new StringBuffer(token.length() + 6);
 		char[] arr = token.toCharArray();
 		boolean shouldQuote = false;
@@ -92,10 +107,10 @@ public class CsvWriter {
 		}
 
 		if (shouldQuote == true)
-			out.print(quote);
-		out.print(buf.toString());
+			out.write(quote);
+		out.write(buf.toString());
 		if (shouldQuote == true)
-			out.print(quote);
+			out.write(quote);
 	}
 
 	public void setSeparator(char separator) {
-- 
2.30.2