|
| 1 | +package io.github.quafadas.scautable |
| 2 | + |
| 3 | +import munit.FunSuite |
| 4 | + |
| 5 | +class CSVParserSuite extends FunSuite: |
| 6 | + |
| 7 | + test("parseLine should handle simple unquoted fields") { |
| 8 | + val line = "field1,field2,field3" |
| 9 | + val result = CSVParser.parseLine(line) |
| 10 | + assertEquals(result, List("field1", "field2", "field3")) |
| 11 | + } |
| 12 | + |
| 13 | + test("parseLine should handle quoted fields") { |
| 14 | + val line = "\"field1\",\"field2\",\"field3\"" |
| 15 | + val result = CSVParser.parseLine(line) |
| 16 | + assertEquals(result, List("field1", "field2", "field3")) |
| 17 | + } |
| 18 | + |
| 19 | + test("parseLine should handle mixed quoted and unquoted fields") { |
| 20 | + val line = "field1,\"field2\",field3" |
| 21 | + val result = CSVParser.parseLine(line) |
| 22 | + assertEquals(result, List("field1", "field2", "field3")) |
| 23 | + } |
| 24 | + |
| 25 | + test("parseLine should handle fields with commas inside quotes") { |
| 26 | + val line = "field1,\"field2,with,commas\",field3" |
| 27 | + val result = CSVParser.parseLine(line) |
| 28 | + assertEquals(result, List("field1", "field2,with,commas", "field3")) |
| 29 | + } |
| 30 | + |
| 31 | + test("parseLine should handle RFC 4180 compliant double-quote escaping") { |
| 32 | + // RFC 4180: quotes inside quoted fields are escaped by doubling them |
| 33 | + val line = "\"this is my \"\"Test String\"\"\"" |
| 34 | + val result = CSVParser.parseLine(line) |
| 35 | + assertEquals(result, List("this is my \"Test String\"")) |
| 36 | + } |
| 37 | + |
| 38 | + test("parseLine should handle backslash-escaped quotes") { |
| 39 | + // Common alternative: quotes escaped with backslashes |
| 40 | + val line = "\"this is my \\\"Test String\\\"\"" |
| 41 | + val result = CSVParser.parseLine(line) |
| 42 | + assertEquals(result, List("this is my \"Test String\"")) |
| 43 | + } |
| 44 | + |
| 45 | + test("parseLine should handle multiple fields with RFC 4180 escaping") { |
| 46 | + val line = "field1,\"field2 with \"\"quotes\"\"\",field3" |
| 47 | + val result = CSVParser.parseLine(line) |
| 48 | + assertEquals(result, List("field1", "field2 with \"quotes\"", "field3")) |
| 49 | + } |
| 50 | + |
| 51 | + test("parseLine should handle multiple fields with backslash escaping") { |
| 52 | + val line = "field1,\"field2 with \\\"quotes\\\"\",field3" |
| 53 | + val result = CSVParser.parseLine(line) |
| 54 | + assertEquals(result, List("field1", "field2 with \"quotes\"", "field3")) |
| 55 | + } |
| 56 | + |
| 57 | + test("parseLine should handle empty fields") { |
| 58 | + val line = "field1,,field3" |
| 59 | + val result = CSVParser.parseLine(line) |
| 60 | + assertEquals(result, List("field1", "", "field3")) |
| 61 | + } |
| 62 | + |
| 63 | + test("parseLine should handle empty quoted fields") { |
| 64 | + val line = "field1,\"\",field3" |
| 65 | + val result = CSVParser.parseLine(line) |
| 66 | + assertEquals(result, List("field1", "", "field3")) |
| 67 | + } |
| 68 | + |
| 69 | + test("parseLine should handle newlines inside quoted fields") { |
| 70 | + val line = "field1,\"field2\nwith\nnewlines\",field3" |
| 71 | + val result = CSVParser.parseLine(line) |
| 72 | + assertEquals(result, List("field1", "field2\nwith\nnewlines", "field3")) |
| 73 | + } |
| 74 | + |
| 75 | + test("parseLine should handle custom delimiter") { |
| 76 | + val line = "field1;field2;field3" |
| 77 | + val result = CSVParser.parseLine(line, delimiter = ';') |
| 78 | + assertEquals(result, List("field1", "field2", "field3")) |
| 79 | + } |
| 80 | + |
| 81 | + test("parseLine should handle custom quote character") { |
| 82 | + val line = "field1,'field2 with ''spaces''',field3" |
| 83 | + val result = CSVParser.parseLine(line, quote = '\'') |
| 84 | + assertEquals(result, List("field1", "field2 with 'spaces'", "field3")) |
| 85 | + } |
| 86 | + |
| 87 | + test("parseLine should handle complex RFC 4180 case") { |
| 88 | + // Complex case with multiple escaped quotes |
| 89 | + val line = "\"He said \"\"Hello, \"\"World\"\"!\"\"\"" |
| 90 | + val result = CSVParser.parseLine(line) |
| 91 | + assertEquals(result, List("He said \"Hello, \"World\"!\"")) |
| 92 | + } |
| 93 | + |
| 94 | + test("parseLine should handle complex backslash escape case") { |
| 95 | + // Complex case with multiple escaped quotes using backslashes |
| 96 | + val line = "\"He said \\\"Hello, \\\"World\\\"!\\\"\"" |
| 97 | + val result = CSVParser.parseLine(line) |
| 98 | + assertEquals(result, List("He said \"Hello, \"World\"!\"")) |
| 99 | + } |
| 100 | + |
| 101 | + test("parseLine should handle mixed escaping in different fields") { |
| 102 | + // One field with RFC 4180 escaping, another with backslash escaping |
| 103 | + val line = "\"field with \"\"RFC escaping\"\"\",\"field with \\\"backslash escaping\\\"\"" |
| 104 | + val result = CSVParser.parseLine(line) |
| 105 | + assertEquals(result, List("field with \"RFC escaping\"", "field with \"backslash escaping\"")) |
| 106 | + } |
| 107 | + |
| 108 | + test("parseLine should handle trailing empty field") { |
| 109 | + val line = "field1,field2," |
| 110 | + val result = CSVParser.parseLine(line) |
| 111 | + assertEquals(result, List("field1", "field2", "")) |
| 112 | + } |
| 113 | + |
| 114 | + test("parseLine should handle leading empty field") { |
| 115 | + val line = ",field2,field3" |
| 116 | + val result = CSVParser.parseLine(line) |
| 117 | + assertEquals(result, List("", "field2", "field3")) |
| 118 | + } |
| 119 | + |
| 120 | + // Tests for full escape character support (ESCAPE specification) |
| 121 | + test("parseLine should handle backslash-escaped linefeeds") { |
| 122 | + val line = "field1,\"field2 with \\n newline\",field3" |
| 123 | + val result = CSVParser.parseLine(line) |
| 124 | + assertEquals(result, List("field1", "field2 with \n newline", "field3")) |
| 125 | + } |
| 126 | + |
| 127 | + test("parseLine should handle backslash-escaped carriage returns") { |
| 128 | + val line = "field1,\"field2 with \\r return\",field3" |
| 129 | + val result = CSVParser.parseLine(line) |
| 130 | + assertEquals(result, List("field1", "field2 with \r return", "field3")) |
| 131 | + } |
| 132 | + |
| 133 | + test("parseLine should handle backslash-escaped delimiters") { |
| 134 | + val line = "field1,\"field2 with \\, comma\",field3" |
| 135 | + val result = CSVParser.parseLine(line) |
| 136 | + assertEquals(result, List("field1", "field2 with , comma", "field3")) |
| 137 | + } |
| 138 | + |
| 139 | + test("parseLine should handle backslash-escaped backslashes") { |
| 140 | + val line = "field1,\"field2 with \\\\ backslash\",field3" |
| 141 | + val result = CSVParser.parseLine(line) |
| 142 | + assertEquals(result, List("field1", "field2 with \\ backslash", "field3")) |
| 143 | + } |
| 144 | + |
| 145 | + test("parseLine should handle custom delimiter with backslash-escaped delimiter") { |
| 146 | + val line = "field1;\"field2 with \\; semicolon\";field3" |
| 147 | + val result = CSVParser.parseLine(line, delimiter = ';') |
| 148 | + assertEquals(result, List("field1", "field2 with ; semicolon", "field3")) |
| 149 | + } |
| 150 | + |
| 151 | + test("parseLine should handle multiple escape sequences in one field") { |
| 152 | + val line = "field1,\"field2 with \\n\\r\\, and \\\" escapes\",field3" |
| 153 | + val result = CSVParser.parseLine(line) |
| 154 | + assertEquals(result, List("field1", "field2 with \n\r, and \" escapes", "field3")) |
| 155 | + } |
| 156 | + |
| 157 | + test("parseLine should handle backslash at end of field (not escaping anything)") { |
| 158 | + val line = "field1,\"field2 ends with \\\\\",field3" |
| 159 | + val result = CSVParser.parseLine(line) |
| 160 | + assertEquals(result, List("field1", "field2 ends with \\", "field3")) |
| 161 | + } |
| 162 | + |
| 163 | + test("parseLine should handle invalid escape sequences by treating backslash literally") { |
| 164 | + val line = "field1,\"field2 with \\z invalid escape\",field3" |
| 165 | + val result = CSVParser.parseLine(line) |
| 166 | + assertEquals(result, List("field1", "field2 with \\z invalid escape", "field3")) |
| 167 | + } |
0 commit comments