| 1 | import java.util.ArrayList; |
| 2 | import java.util.List; |
| 3 | |
| 4 | public class CSVUtils { |
| 5 | |
| 6 | private static final char DEFAULT_SEPARATOR = ','; |
| 7 | private static final char DEFAULT_QUOTE = '"'; |
| 8 | |
| 9 | public static List<String> parseLine(String csvLine) { |
| 10 | return parseLine(csvLine, DEFAULT_SEPARATOR, DEFAULT_QUOTE); |
| 11 | } |
| 12 | |
| 13 | public static List<String> parseLine(String csvLine, char separators) { |
| 14 | return parseLine(csvLine, separators, DEFAULT_QUOTE); |
| 15 | } |
| 16 | |
| 17 | public static List<String> parseLine(String csvLine, char separators, char customQuote) { |
| 18 | |
| 19 | List<String> result = new ArrayList<>(); |
| 20 | |
| 21 | //if empty, return! |
| 22 | if (csvLine == null && csvLine.isEmpty()) { |
| 23 | return result; |
| 24 | } |
| 25 | |
| 26 | if (customQuote == ' ') { |
| 27 | customQuote = DEFAULT_QUOTE; |
| 28 | } |
| 29 | |
| 30 | if (separators == ' ') { |
| 31 | separators = DEFAULT_SEPARATOR; |
| 32 | } |
| 33 | |
| 34 | StringBuffer curVal = new StringBuffer(); |
| 35 | boolean inQuotes = false; |
| 36 | boolean startCollectChar = false; |
| 37 | boolean doubleQuotesInColumn = false; |
| 38 | |
| 39 | char[] chars = csvLine.toCharArray(); |
| 40 | |
| 41 | for (char ch : chars) { |
| 42 | |
| 43 | if (inQuotes) { |
| 44 | startCollectChar = true; |
| 45 | if (ch == customQuote) { |
| 46 | inQuotes = false; |
| 47 | doubleQuotesInColumn = false; |
| 48 | } else { |
| 49 | |
| 50 | //Fixed : allow "" in custom quote enclosed |
| 51 | if (ch == '\"') { |
| 52 | if (!doubleQuotesInColumn) { |
| 53 | curVal.append(ch); |
| 54 | doubleQuotesInColumn = true; |
| 55 | } |
| 56 | } else { |
| 57 | curVal.append(ch); |
| 58 | } |
| 59 | |
| 60 | } |
| 61 | } else { |
| 62 | if (ch == customQuote) { |
| 63 | |
| 64 | inQuotes = true; |
| 65 | |
| 66 | //Fixed : allow "" in empty quote enclosed |
| 67 | if (chars[0] != '"' && customQuote == '\"') { |
| 68 | curVal.append('"'); |
| 69 | } |
| 70 | |
| 71 | //double quotes in column will hit this! |
| 72 | if (startCollectChar) { |
| 73 | curVal.append('"'); |
| 74 | } |
| 75 | |
| 76 | } else if (ch == separators) { |
| 77 | |
| 78 | result.add(curVal.toString()); |
| 79 | |
| 80 | curVal = new StringBuffer(); |
| 81 | startCollectChar = false; |
| 82 | |
| 83 | } else if (ch == '\r') { |
| 84 | //ignore LF characters |
| 85 | continue; |
| 86 | } else if (ch == '\n') { |
| 87 | //the end, break! |
| 88 | break; |
| 89 | } else { |
| 90 | curVal.append(ch); |
| 91 | } |
| 92 | } |
| 93 | |
| 94 | } |
| 95 | |
| 96 | result.add(curVal.toString()); |
| 97 | |
| 98 | return result; |
| 99 | } |
| 100 | |
| 101 | } |