util/CsvParser.java

   1 package org.argeo.util;
   2
   3 import java.io.BufferedReader;
   4 import java.io.IOException;
   5 import java.io.InputStream;
   6 import java.io.InputStreamReader;
   7 import java.util.ArrayList;
   8 import java.util.Collections;
   9 import java.util.List;
  10
  11 import org.argeo.ArgeoException;
  12
  13 /**
  14  * Parses a CSV file interpreting the first line as a header. The
  15  * {@link #parse(InputStream)} method and the setters are synchronized so that
  16  * the object cannot be modified when parsing.
  17  */
  18 public abstract class CsvParser {
  19         private char separator = ',';
  20         private char quote = '\"';
  21
  22         private Boolean noHeader = false;
  23         private Boolean strictLineAsLongAsHeader = true;
  24
  25         /**
  26          * Actually process a parsed line. If
  27          * {@link #setStrictLineAsLongAsHeader(Boolean)} is true (default) the
  28          * header and the tokens are guaranteed to have the same size.
  29          *
  30          * @param lineNumber
  31          *            the current line number, starts at 1 (the header, if header
  32          *            processing is enabled, the first lien otherwise)
  33          * @param header
  34          *            the read-only header or null if {@link #setNoHeader(Boolean)}
  35          *            is true (default is false)
  36          * @param tokens
  37          *            the parse tokens
  38          */
  39         protected abstract void processLine(Integer lineNumber,
  40                         List<String> header, List<String> tokens);
  41
  42         public synchronized void parse(InputStream in) {
  43                 BufferedReader reader = null;
  44                 Integer lineCount = 0;
  45                 try {
  46                         reader = new BufferedReader(new InputStreamReader(in));
  47
  48                         List<String> header = null;
  49                         if (!noHeader) {
  50                                 String headerStr = reader.readLine();
  51                                 if (headerStr == null)// empty file
  52                                         return;
  53                                 lineCount++;
  54                                 header = Collections.unmodifiableList(parseLine(headerStr));
  55                         }
  56
  57                         String line = null;
  58                         lines: while ((line = reader.readLine()) != null) {
  59                                 lineCount++;
  60                                 List<String> tokens = parseLine(line);
  61                                 if (!noHeader && strictLineAsLongAsHeader) {
  62                                         int headerSize = header.size();
  63                                         int tokenSize = tokens.size();
  64                                         if (tokenSize == 1 && line.trim().equals(""))
  65                                                 continue lines;// empty line
  66                                         if (headerSize != tokenSize) {
  67                                                 throw new ArgeoException("Token size " + tokenSize
  68                                                                 + " is different from header size "
  69                                                                 + headerSize + " at line " + lineCount
  70                                                                 + ", line: " + line + ", header: " + header
  71                                                                 + ", tokens: " + tokens);
  72                                         }
  73                                 }
  74                                 processLine(lineCount, header, tokens);
  75                         }
  76                 } catch (ArgeoException e) {
  77                         throw e;
  78                 } catch (IOException e) {
  79                         throw new ArgeoException("Cannot parse CSV file (line: "
  80                                         + lineCount + ")", e);
  81                 } finally {
  82                         if (reader != null)
  83                                 try {
  84                                         reader.close();
  85                                 } catch (Exception e2) {
  86                                         // silent
  87                                 }
  88                 }
  89         }
  90
  91         /** Parses a line character by character for performance purpose */
  92         protected List<String> parseLine(String str) {
  93                 List<String> tokens = new ArrayList<String>();
  94
  95                 char[] arr = str.toCharArray();
  96                 boolean inQuote = false;
  97                 StringBuffer currStr = new StringBuffer("");
  98                 for (int i = 0; i < arr.length; i++) {
  99                         char c = arr[i];
 100                         if (c == separator) {
 101                                 if (!inQuote) {
 102                                         tokens.add(currStr.toString());
 103                                         currStr = new StringBuffer("");
 104                                 }
 105                         } else if (c == quote) {
 106                                 if (inQuote && (i + 1) < arr.length && arr[i + 1] == quote) {
 107                                         // case of double quote
 108                                         currStr.append(quote);
 109                                         i++;
 110                                 } else {// standard
 111                                         inQuote = inQuote ? false : true;
 112                                 }
 113                         } else {
 114                                 currStr.append(c);
 115                         }
 116                 }
 117                 tokens.add(currStr.toString());
 118                 if (inQuote)
 119                         throw new ArgeoException("Missing quote at the end of the line "
 120                                         + str + " (parsed: " + tokens + ")");
 121                 return tokens;
 122         }
 123
 124         public char getSeparator() {
 125                 return separator;
 126         }
 127
 128         public synchronized void setSeparator(char separator) {
 129                 this.separator = separator;
 130         }
 131
 132         public char getQuote() {
 133                 return quote;
 134         }
 135
 136         public synchronized void setQuote(char quote) {
 137                 this.quote = quote;
 138         }
 139
 140         public Boolean getNoHeader() {
 141                 return noHeader;
 142         }
 143
 144         public synchronized void setNoHeader(Boolean noHeader) {
 145                 this.noHeader = noHeader;
 146         }
 147
 148         public Boolean getStrictLineAsLongAsHeader() {
 149                 return strictLineAsLongAsHeader;
 150         }
 151
 152         public synchronized void setStrictLineAsLongAsHeader(
 153                         Boolean strictLineAsLongAsHeader) {
 154                 this.strictLineAsLongAsHeader = strictLineAsLongAsHeader;
 155         }
 156
 157 }