|
| 1 | +import java.util.regex.Pattern; |
| 2 | + |
| 3 | +class VerbalExpressions { |
| 4 | + private String prefixes, source, suffixes, pattern = ""; |
| 5 | + int modifiers = Pattern.MULTILINE; |
| 6 | + Pattern p; |
| 7 | + |
| 8 | + public VerbalExpressions() { |
| 9 | + |
| 10 | + } |
| 11 | + |
| 12 | + private String sanitize(String value) { |
| 13 | + if(value != null) return value; |
| 14 | + return Pattern.quote(value); |
| 15 | + } |
| 16 | + |
| 17 | + public VerbalExpressions add(String value) { |
| 18 | + this.source = this.source != null ? this.source + value : value; |
| 19 | + if (this.source != null) { |
| 20 | + this.p = Pattern.compile(this.prefixes + this.source + this.suffixes, this.modifiers); |
| 21 | + this.pattern = this.p.pattern(); |
| 22 | + } |
| 23 | + return this; |
| 24 | + } |
| 25 | + |
| 26 | + public VerbalExpressions startOfLine(boolean enable) { |
| 27 | + this.prefixes = enable ? "^" : ""; |
| 28 | + this.add( "" ); |
| 29 | + return this; |
| 30 | + } |
| 31 | + |
| 32 | + public VerbalExpressions startOfLine() { |
| 33 | + return startOfLine(true); |
| 34 | + } |
| 35 | + |
| 36 | + public VerbalExpressions endOfLine(boolean enable) { |
| 37 | + this.suffixes = enable ? "$" : ""; |
| 38 | + this.add( "" ); |
| 39 | + return this; |
| 40 | + } |
| 41 | + |
| 42 | + public VerbalExpressions endOfLine() { |
| 43 | + return endOfLine(true); |
| 44 | + } |
| 45 | + |
| 46 | + public VerbalExpressions then(String value) { |
| 47 | + value = sanitize(value); |
| 48 | + this.add( "(" + value + ")" ); |
| 49 | + return this; |
| 50 | + } |
| 51 | + |
| 52 | + public VerbalExpressions find(String value) { |
| 53 | + this.then(value); |
| 54 | + return this; |
| 55 | + } |
| 56 | + |
| 57 | + public VerbalExpressions maybe(String value) { |
| 58 | + value = sanitize(value); |
| 59 | + this.add( "(" + value + ")?" ); |
| 60 | + return this; |
| 61 | + } |
| 62 | + |
| 63 | + public VerbalExpressions anything() { |
| 64 | + this.add( "(.*)" ); |
| 65 | + return this; |
| 66 | + } |
| 67 | + |
| 68 | + public VerbalExpressions anythingBut(String value) { |
| 69 | + value = sanitize(value); |
| 70 | + this.add( "([^" + value + "]*)" ); |
| 71 | + return this; |
| 72 | + } |
| 73 | + |
| 74 | + public VerbalExpressions replace(String source, String value) { |
| 75 | + this.add( "" ); |
| 76 | + this.source.replaceAll(pattern,value); |
| 77 | + return this; |
| 78 | + } |
| 79 | + |
| 80 | + public VerbalExpressions lineBreak() { |
| 81 | + this.add( "(\\n|(\\r\\n))" ); |
| 82 | + return this; |
| 83 | + } |
| 84 | + |
| 85 | + public VerbalExpressions br() { |
| 86 | + this.lineBreak(); |
| 87 | + return this; |
| 88 | + } |
| 89 | + |
| 90 | + public VerbalExpressions tab() { |
| 91 | + this.add( "\\t" ); |
| 92 | + return this; |
| 93 | + } |
| 94 | + |
| 95 | + public VerbalExpressions word() { |
| 96 | + this.add( "\\w+" ); |
| 97 | + return this; |
| 98 | + } |
| 99 | + |
| 100 | + public VerbalExpressions anyOf(String value) { |
| 101 | + value = sanitize(value); |
| 102 | + this.add( "[" + value + "]" ); |
| 103 | + return this; |
| 104 | + } |
| 105 | + |
| 106 | + public VerbalExpressions any(String value) { |
| 107 | + this.anyOf(value); |
| 108 | + return this; |
| 109 | + } |
| 110 | + |
| 111 | + public VerbalExpressions range(Object[] args) { |
| 112 | + String value = "["; |
| 113 | + for(int _from = 0; _from < args.length; _from += 2) { |
| 114 | + int _to = _from+1; |
| 115 | + if (args.length <= _to) break; |
| 116 | + int from = Integer.getInteger(sanitize((String)args[_from])); |
| 117 | + int to = Integer.getInteger(sanitize((String)args[_to])); |
| 118 | + |
| 119 | + value += from + "-" + to; |
| 120 | + } |
| 121 | + |
| 122 | + value += "]"; |
| 123 | + |
| 124 | + this.add(value); |
| 125 | + return this; |
| 126 | + } |
| 127 | + |
| 128 | + public VerbalExpressions addModifier(char modifier) { |
| 129 | + switch (modifier) { |
| 130 | + case 'd': |
| 131 | + modifiers |= Pattern.UNIX_LINES; |
| 132 | + break; |
| 133 | + case 'i': |
| 134 | + modifiers |= Pattern.CASE_INSENSITIVE; |
| 135 | + break; |
| 136 | + case 'x': |
| 137 | + modifiers |= Pattern.COMMENTS; |
| 138 | + break; |
| 139 | + case 'm': |
| 140 | + modifiers |= Pattern.MULTILINE; |
| 141 | + break; |
| 142 | + case 's': |
| 143 | + modifiers |= Pattern.DOTALL; |
| 144 | + break; |
| 145 | + case 'u': |
| 146 | + modifiers |= Pattern.UNICODE_CASE; |
| 147 | + break; |
| 148 | + case 'U': |
| 149 | + modifiers |= Pattern.UNICODE_CHARACTER_CLASS; |
| 150 | + break; |
| 151 | + default: |
| 152 | + break; |
| 153 | + } |
| 154 | + |
| 155 | + this.add( "" ); |
| 156 | + return this; |
| 157 | + } |
| 158 | + |
| 159 | + public VerbalExpressions removeModifier(char modifier) { |
| 160 | + switch (modifier) { |
| 161 | + case 'd': |
| 162 | + modifiers ^= Pattern.UNIX_LINES; |
| 163 | + break; |
| 164 | + case 'i': |
| 165 | + modifiers ^= Pattern.CASE_INSENSITIVE; |
| 166 | + break; |
| 167 | + case 'x': |
| 168 | + modifiers ^= Pattern.COMMENTS; |
| 169 | + break; |
| 170 | + case 'm': |
| 171 | + modifiers ^= Pattern.MULTILINE; |
| 172 | + break; |
| 173 | + case 's': |
| 174 | + modifiers ^= Pattern.DOTALL; |
| 175 | + break; |
| 176 | + case 'u': |
| 177 | + modifiers ^= Pattern.UNICODE_CASE; |
| 178 | + break; |
| 179 | + case 'U': |
| 180 | + modifiers ^= Pattern.UNICODE_CHARACTER_CLASS; |
| 181 | + break; |
| 182 | + default: |
| 183 | + break; |
| 184 | + } |
| 185 | + |
| 186 | + this.add( "" ); |
| 187 | + return this; |
| 188 | + } |
| 189 | + |
| 190 | + public VerbalExpressions withAnyCase(boolean enable) { |
| 191 | + if (enable) this.addModifier( 'i' ); |
| 192 | + else this.removeModifier( 'i' ); |
| 193 | + this.add( "" ); |
| 194 | + return this; |
| 195 | + } |
| 196 | + |
| 197 | + public VerbalExpressions withAnyCase() { |
| 198 | + return withAnyCase(true); |
| 199 | + } |
| 200 | + |
| 201 | + public VerbalExpressions searchOneLine(boolean enable) { |
| 202 | + if (enable) this.removeModifier( 'm' ); |
| 203 | + else this.addModifier( 'm' ); |
| 204 | + this.add( "" ); |
| 205 | + return this; |
| 206 | + } |
| 207 | + |
| 208 | + public VerbalExpressions multiple(String value) { |
| 209 | + value = this.sanitize(value); |
| 210 | + switch (value.charAt(0)) { |
| 211 | + case '*': |
| 212 | + case '+': |
| 213 | + break; |
| 214 | + default: |
| 215 | + value += '+'; |
| 216 | + } |
| 217 | + this.add(value); |
| 218 | + return this; |
| 219 | + } |
| 220 | + |
| 221 | + public VerbalExpressions or(String value) { |
| 222 | + if (this.prefixes.indexOf("(") == -1) this.prefixes += "("; |
| 223 | + if (this.suffixes.indexOf(")") == -1) this.suffixes = ")" + this.suffixes; |
| 224 | + |
| 225 | + this.add( ")|(" ); |
| 226 | + if (value != null) this.then(value); |
| 227 | + return this; |
| 228 | + } |
| 229 | + |
| 230 | + public boolean test(String toTest) { |
| 231 | + this.add( "" ); |
| 232 | + return Pattern.matches(this.pattern, toTest); |
| 233 | + } |
| 234 | + |
| 235 | + public String toString() { |
| 236 | + this.add( "" ); |
| 237 | + return this.pattern.toString(); |
| 238 | + } |
| 239 | +} |
0 commit comments