Improve CSV and tabular

[lgpl/argeo-commons.git] / basic / runtime / org.argeo.basic.nodeps / src / main / java / org / argeo / util / CsvParser.java

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 ca81bc98305700e3be83f5ebac44c02f8718f40b..c20f4e07c684c90902cef9fd473e16410286d64f 100644 (file)
--- 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
@@ -29,12 +29,12 @@ public abstract class CsvParser {
         * 
         * @param lineNumber
         *            the current line number, starts at 1 (the header, if header
-        *            processing is enabled, the first lien otherwise)
+        *            processing is enabled, the first line otherwise)
         * @param header
         *            the read-only header or null if {@link #setNoHeader(Boolean)}
         *            is true (default is false)
         * @param tokens
-        *            the parse tokens
+        *            the parsed tokens
         */
        protected abstract void processLine(Integer lineNumber,
                        List<String> header, List<String> tokens);
@@ -55,6 +55,9 @@ public abstract class CsvParser {
                                StringBuffer currStr = new StringBuffer("");
                                Boolean wasInquote = false;
                                while (parseLine(headerStr, header, currStr, wasInquote)) {
+                                       headerStr = reader.readLine();
+                                       if (headerStr == null)
+                                               break;
                                        wasInquote = true;
                                }
                                header = Collections.unmodifiableList(header);
@@ -111,7 +114,7 @@ public abstract class CsvParser {
                        StringBuffer currStr, Boolean wasInquote) {
                // List<String> tokens = new ArrayList<String>();
 
-               //System.out.println("#LINE: " + str);
+               // System.out.println("#LINE: " + str);
 
                if (wasInquote)
                        currStr.append('\n');
@@ -124,8 +127,13 @@ public abstract class CsvParser {
                        if (c == separator) {
                                if (!inQuote) {
                                        tokens.add(currStr.toString());
-                                       //System.out.println("# TOKEN: " + currStr);
+                                       // System.out.println("# TOKEN: " + currStr);
                                        currStr.delete(0, currStr.length());
+                               } else {
+                                       // we don't remove separator that are in a quoted substring
+                                       // System.out
+                                       // .println("IN QUOTE, got a separator: [" + c + "]");
+                                       currStr.append(c);
                                }
                        } else if (c == quote) {
                                if (inQuote && (i + 1) < arr.length && arr[i + 1] == quote) {
@@ -142,7 +150,7 @@ public abstract class CsvParser {
 
                if (!inQuote) {
                        tokens.add(currStr.toString());
-                       //System.out.println("# TOKEN: " + currStr);
+                       // System.out.println("# TOKEN: " + currStr);
                }
                // if (inQuote)
                // throw new ArgeoException("Missing quote at the end of the line "