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