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 line 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 parsed 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                 parse(in, null);
  44         }
  45
  46         public synchronized void parse(InputStream in, String encoding) {
  47                 BufferedReader reader = null;
  48                 Integer lineCount = 0;
  49                 try {
  50                         if (encoding == null)
  51                                 reader = new BufferedReader(new InputStreamReader(in));
  52                         else
  53                                 reader = new BufferedReader(new InputStreamReader(in, encoding));
  54                         List<String> header = null;
  55                         if (!noHeader) {
  56                                 String headerStr = reader.readLine();
  57                                 if (headerStr == null)// empty file
  58                                         return;
  59                                 lineCount++;
  60                                 header = new ArrayList<String>();
  61                                 StringBuffer currStr = new StringBuffer("");
  62                                 Boolean wasInquote = false;
  63                                 while (parseLine(headerStr, header, currStr, wasInquote)) {
  64                                         headerStr = reader.readLine();
  65                                         if (headerStr == null)
  66                                                 break;
  67                                         wasInquote = true;
  68                                 }
  69                                 header = Collections.unmodifiableList(header);
  70                         }
  71
  72                         String line = null;
  73                         lines: while ((line = reader.readLine()) != null) {
  74                                 lineCount++;
  75                                 List<String> tokens = new ArrayList<String>();
  76                                 StringBuffer currStr = new StringBuffer("");
  77                                 Boolean wasInquote = false;
  78                                 while (parseLine(line, tokens, currStr, wasInquote)) {
  79                                         line = reader.readLine();
  80                                         if (line == null)
  81                                                 break;
  82                                         wasInquote = true;
  83                                 }
  84                                 if (!noHeader && strictLineAsLongAsHeader) {
  85                                         int headerSize = header.size();
  86                                         int tokenSize = tokens.size();
  87                                         if (tokenSize == 1 && line.trim().equals(""))
  88                                                 continue lines;// empty line
  89                                         if (headerSize != tokenSize) {
  90                                                 throw new ArgeoException("Token size " + tokenSize
  91                                                                 + " is different from header size "
  92                                                                 + headerSize + " at line " + lineCount
  93                                                                 + ", line: " + line + ", header: " + header
  94                                                                 + ", tokens: " + tokens);
  95                                         }
  96                                 }
  97                                 processLine(lineCount, header, tokens);
  98                         }
  99                 } catch (ArgeoException e) {
 100                         throw e;
 101                 } catch (IOException e) {
 102                         throw new ArgeoException("Cannot parse CSV file (line: "
 103                                         + lineCount + ")", e);
 104                 } finally {
 105                         if (reader != null)
 106                                 try {
 107                                         reader.close();
 108                                 } catch (Exception e2) {
 109                                         // silent
 110                                 }
 111                 }
 112         }
 113
 114         /**
 115          * Parses a line character by character for performance purpose
 116          *
 117          * @return whether to continue parsing this line
 118          */
 119         protected Boolean parseLine(String str, List<String> tokens,
 120                         StringBuffer currStr, Boolean wasInquote) {
 121                 // List<String> tokens = new ArrayList<String>();
 122
 123                 // System.out.println("#LINE: " + str);
 124
 125                 if (wasInquote)
 126                         currStr.append('\n');
 127
 128                 char[] arr = str.toCharArray();
 129                 boolean inQuote = wasInquote;
 130                 // StringBuffer currStr = new StringBuffer("");
 131                 for (int i = 0; i < arr.length; i++) {
 132                         char c = arr[i];
 133                         if (c == separator) {
 134                                 if (!inQuote) {
 135                                         tokens.add(currStr.toString());
 136                                         // System.out.println("# TOKEN: " + currStr);
 137                                         currStr.delete(0, currStr.length());
 138                                 } else {
 139                                         // we don't remove separator that are in a quoted substring
 140                                         // System.out
 141                                         // .println("IN QUOTE, got a separator: [" + c + "]");
 142                                         currStr.append(c);
 143                                 }
 144                         } else if (c == quote) {
 145                                 if (inQuote && (i + 1) < arr.length && arr[i + 1] == quote) {
 146                                         // case of double quote
 147                                         currStr.append(quote);
 148                                         i++;
 149                                 } else {// standard
 150                                         inQuote = inQuote ? false : true;
 151                                 }
 152                         } else {
 153                                 currStr.append(c);
 154                         }
 155                 }
 156
 157                 if (!inQuote) {
 158                         tokens.add(currStr.toString());
 159                         // System.out.println("# TOKEN: " + currStr);
 160                 }
 161                 // if (inQuote)
 162                 // throw new ArgeoException("Missing quote at the end of the line "
 163                 // + str + " (parsed: " + tokens + ")");
 164                 if (inQuote)
 165                         return true;
 166                 else
 167                         return false;
 168                 // return tokens;
 169         }
 170
 171         public char getSeparator() {
 172                 return separator;
 173         }
 174
 175         public synchronized void setSeparator(char separator) {
 176                 this.separator = separator;
 177         }
 178
 179         public char getQuote() {
 180                 return quote;
 181         }
 182
 183         public synchronized void setQuote(char quote) {
 184                 this.quote = quote;
 185         }
 186
 187         public Boolean getNoHeader() {
 188                 return noHeader;
 189         }
 190
 191         public synchronized void setNoHeader(Boolean noHeader) {
 192                 this.noHeader = noHeader;
 193         }
 194
 195         public Boolean getStrictLineAsLongAsHeader() {
 196                 return strictLineAsLongAsHeader;
 197         }
 198
 199         public synchronized void setStrictLineAsLongAsHeader(
 200                         Boolean strictLineAsLongAsHeader) {
 201                 this.strictLineAsLongAsHeader = strictLineAsLongAsHeader;
 202         }
 203
 204 }