ex2/mapreduce/CSVSplitter.java

   1 // Created as a template for Advanced Database Systems 2019\r
   2 \r
   3 import java.util.ArrayList;\r
   4 \r
   5 public class CSVSplitter {\r
   6         public static String[] split(String input){\r
   7                 ArrayList<String> output = new ArrayList<String>();\r
   8                 int start = 0;\r
   9                 boolean insideQuote = false;\r
  10 \r
  11                 for (int current = 0; current < input.length();current++){\r
  12                         char c = input.charAt(current);\r
  13                         switch (c){\r
  14                         case ',':\r
  15                                 if (!insideQuote) {\r
  16                                         output.add(input.substring(start,current));\r
  17                                         start = current +1;\r
  18                                 }\r
  19                                 break;\r
  20                         case '"':\r
  21                                 if (current == start) {\r
  22                                         insideQuote = true;\r
  23                                 } else if ( (current < input.length() - 1) && input.charAt(current+1) == '"'){ // escaped quote\r
  24                                         current = current + 1; // skip ahead\r
  25                                         continue;\r
  26                                 } else if ( (current < input.length() - 1) && input.charAt(current+1) == ',') {\r
  27                                         insideQuote = false;\r
  28                                 }\r
  29                                 break;\r
  30                         }\r
  31                 }\r
  32                 output.add(input.substring(start));\r
  33                 return output.toArray( new String[output.size()]);\r
  34         }\r
  35 \r
  36         //This version does not consider quote when deciding splits\r
  37         public static String[] splitNoQuote(String input){\r
  38                   ArrayList<String> output = new ArrayList<String>();\r
  39                         int start = 0;\r
  40                         boolean insideQuote = false;\r
  41 \r
  42                         for (int current = 0; current < input.length();current++){\r
  43                                 char c = input.charAt(current);\r
  44                                 switch (c){\r
  45                                 case ',':\r
  46                                         if (!insideQuote) {\r
  47                                                 output.add(input.substring(start,current));\r
  48                                                 start = current +1;\r
  49                                         }\r
  50                                         break;\r
  51 \r
  52                                 }\r
  53                         }\r
  54                         output.add(input.substring(start));\r
  55                         return output.toArray( new String[output.size()]);\r
  56         }\r
  57 }\r