Scott Baker | e7144bc | 2019-10-01 14:16:47 -0700 | [diff] [blame] | 1 | // Copyright 2014 The Go Authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style |
| 3 | // license that can be found in the LICENSE file. |
| 4 | |
| 5 | // Package ucd provides a parser for Unicode Character Database files, the |
| 6 | // format of which is defined in http://www.unicode.org/reports/tr44/. See |
| 7 | // http://www.unicode.org/Public/UCD/latest/ucd/ for example files. |
| 8 | // |
| 9 | // It currently does not support substitutions of missing fields. |
| 10 | package ucd // import "golang.org/x/text/internal/ucd" |
| 11 | |
| 12 | import ( |
| 13 | "bufio" |
| 14 | "errors" |
| 15 | "fmt" |
| 16 | "io" |
| 17 | "log" |
| 18 | "regexp" |
| 19 | "strconv" |
| 20 | "strings" |
| 21 | ) |
| 22 | |
| 23 | // UnicodeData.txt fields. |
| 24 | const ( |
| 25 | CodePoint = iota |
| 26 | Name |
| 27 | GeneralCategory |
| 28 | CanonicalCombiningClass |
| 29 | BidiClass |
| 30 | DecompMapping |
| 31 | DecimalValue |
| 32 | DigitValue |
| 33 | NumericValue |
| 34 | BidiMirrored |
| 35 | Unicode1Name |
| 36 | ISOComment |
| 37 | SimpleUppercaseMapping |
| 38 | SimpleLowercaseMapping |
| 39 | SimpleTitlecaseMapping |
| 40 | ) |
| 41 | |
| 42 | // Parse calls f for each entry in the given reader of a UCD file. It will close |
| 43 | // the reader upon return. It will call log.Fatal if any error occurred. |
| 44 | // |
| 45 | // This implements the most common usage pattern of using Parser. |
| 46 | func Parse(r io.ReadCloser, f func(p *Parser)) { |
| 47 | defer r.Close() |
| 48 | |
| 49 | p := New(r) |
| 50 | for p.Next() { |
| 51 | f(p) |
| 52 | } |
| 53 | if err := p.Err(); err != nil { |
| 54 | r.Close() // os.Exit will cause defers not to be called. |
| 55 | log.Fatal(err) |
| 56 | } |
| 57 | } |
| 58 | |
| 59 | // An Option is used to configure a Parser. |
| 60 | type Option func(p *Parser) |
| 61 | |
| 62 | func keepRanges(p *Parser) { |
| 63 | p.keepRanges = true |
| 64 | } |
| 65 | |
| 66 | var ( |
| 67 | // KeepRanges prevents the expansion of ranges. The raw ranges can be |
| 68 | // obtained by calling Range(0) on the parser. |
| 69 | KeepRanges Option = keepRanges |
| 70 | ) |
| 71 | |
| 72 | // The Part option register a handler for lines starting with a '@'. The text |
| 73 | // after a '@' is available as the first field. Comments are handled as usual. |
| 74 | func Part(f func(p *Parser)) Option { |
| 75 | return func(p *Parser) { |
| 76 | p.partHandler = f |
| 77 | } |
| 78 | } |
| 79 | |
| 80 | // The CommentHandler option passes comments that are on a line by itself to |
| 81 | // a given handler. |
| 82 | func CommentHandler(f func(s string)) Option { |
| 83 | return func(p *Parser) { |
| 84 | p.commentHandler = f |
| 85 | } |
| 86 | } |
| 87 | |
| 88 | // A Parser parses Unicode Character Database (UCD) files. |
| 89 | type Parser struct { |
| 90 | scanner *bufio.Scanner |
| 91 | |
| 92 | keepRanges bool // Don't expand rune ranges in field 0. |
| 93 | |
| 94 | err error |
| 95 | comment string |
| 96 | field []string |
| 97 | // parsedRange is needed in case Range(0) is called more than once for one |
| 98 | // field. In some cases this requires scanning ahead. |
| 99 | line int |
| 100 | parsedRange bool |
| 101 | rangeStart, rangeEnd rune |
| 102 | |
| 103 | partHandler func(p *Parser) |
| 104 | commentHandler func(s string) |
| 105 | } |
| 106 | |
| 107 | func (p *Parser) setError(err error, msg string) { |
| 108 | if p.err == nil && err != nil { |
| 109 | if msg == "" { |
| 110 | p.err = fmt.Errorf("ucd:line:%d: %v", p.line, err) |
| 111 | } else { |
| 112 | p.err = fmt.Errorf("ucd:line:%d:%s: %v", p.line, msg, err) |
| 113 | } |
| 114 | } |
| 115 | } |
| 116 | |
| 117 | func (p *Parser) getField(i int) string { |
| 118 | if i >= len(p.field) { |
| 119 | return "" |
| 120 | } |
| 121 | return p.field[i] |
| 122 | } |
| 123 | |
| 124 | // Err returns a non-nil error if any error occurred during parsing. |
| 125 | func (p *Parser) Err() error { |
| 126 | return p.err |
| 127 | } |
| 128 | |
| 129 | // New returns a Parser for the given Reader. |
| 130 | func New(r io.Reader, o ...Option) *Parser { |
| 131 | p := &Parser{ |
| 132 | scanner: bufio.NewScanner(r), |
| 133 | } |
| 134 | for _, f := range o { |
| 135 | f(p) |
| 136 | } |
| 137 | return p |
| 138 | } |
| 139 | |
| 140 | // Next parses the next line in the file. It returns true if a line was parsed |
| 141 | // and false if it reached the end of the file. |
| 142 | func (p *Parser) Next() bool { |
| 143 | if !p.keepRanges && p.rangeStart < p.rangeEnd { |
| 144 | p.rangeStart++ |
| 145 | return true |
| 146 | } |
| 147 | p.comment = "" |
| 148 | p.field = p.field[:0] |
| 149 | p.parsedRange = false |
| 150 | |
| 151 | for p.scanner.Scan() && p.err == nil { |
| 152 | p.line++ |
| 153 | s := p.scanner.Text() |
| 154 | if s == "" { |
| 155 | continue |
| 156 | } |
| 157 | if s[0] == '#' { |
| 158 | if p.commentHandler != nil { |
| 159 | p.commentHandler(strings.TrimSpace(s[1:])) |
| 160 | } |
| 161 | continue |
| 162 | } |
| 163 | |
| 164 | // Parse line |
| 165 | if i := strings.IndexByte(s, '#'); i != -1 { |
| 166 | p.comment = strings.TrimSpace(s[i+1:]) |
| 167 | s = s[:i] |
| 168 | } |
| 169 | if s[0] == '@' { |
| 170 | if p.partHandler != nil { |
| 171 | p.field = append(p.field, strings.TrimSpace(s[1:])) |
| 172 | p.partHandler(p) |
| 173 | p.field = p.field[:0] |
| 174 | } |
| 175 | p.comment = "" |
| 176 | continue |
| 177 | } |
| 178 | for { |
| 179 | i := strings.IndexByte(s, ';') |
| 180 | if i == -1 { |
| 181 | p.field = append(p.field, strings.TrimSpace(s)) |
| 182 | break |
| 183 | } |
| 184 | p.field = append(p.field, strings.TrimSpace(s[:i])) |
| 185 | s = s[i+1:] |
| 186 | } |
| 187 | if !p.keepRanges { |
| 188 | p.rangeStart, p.rangeEnd = p.getRange(0) |
| 189 | } |
| 190 | return true |
| 191 | } |
| 192 | p.setError(p.scanner.Err(), "scanner failed") |
| 193 | return false |
| 194 | } |
| 195 | |
| 196 | func parseRune(b string) (rune, error) { |
| 197 | if len(b) > 2 && b[0] == 'U' && b[1] == '+' { |
| 198 | b = b[2:] |
| 199 | } |
| 200 | x, err := strconv.ParseUint(b, 16, 32) |
| 201 | return rune(x), err |
| 202 | } |
| 203 | |
| 204 | func (p *Parser) parseRune(s string) rune { |
| 205 | x, err := parseRune(s) |
| 206 | p.setError(err, "failed to parse rune") |
| 207 | return x |
| 208 | } |
| 209 | |
| 210 | // Rune parses and returns field i as a rune. |
| 211 | func (p *Parser) Rune(i int) rune { |
| 212 | if i > 0 || p.keepRanges { |
| 213 | return p.parseRune(p.getField(i)) |
| 214 | } |
| 215 | return p.rangeStart |
| 216 | } |
| 217 | |
| 218 | // Runes interprets and returns field i as a sequence of runes. |
| 219 | func (p *Parser) Runes(i int) (runes []rune) { |
| 220 | add := func(s string) { |
| 221 | if s = strings.TrimSpace(s); len(s) > 0 { |
| 222 | runes = append(runes, p.parseRune(s)) |
| 223 | } |
| 224 | } |
| 225 | for b := p.getField(i); ; { |
| 226 | i := strings.IndexByte(b, ' ') |
| 227 | if i == -1 { |
| 228 | add(b) |
| 229 | break |
| 230 | } |
| 231 | add(b[:i]) |
| 232 | b = b[i+1:] |
| 233 | } |
| 234 | return |
| 235 | } |
| 236 | |
| 237 | var ( |
| 238 | errIncorrectLegacyRange = errors.New("ucd: unmatched <* First>") |
| 239 | |
| 240 | // reRange matches one line of a legacy rune range. |
| 241 | reRange = regexp.MustCompile("^([0-9A-F]*);<([^,]*), ([^>]*)>(.*)$") |
| 242 | ) |
| 243 | |
| 244 | // Range parses and returns field i as a rune range. A range is inclusive at |
| 245 | // both ends. If the field only has one rune, first and last will be identical. |
| 246 | // It supports the legacy format for ranges used in UnicodeData.txt. |
| 247 | func (p *Parser) Range(i int) (first, last rune) { |
| 248 | if !p.keepRanges { |
| 249 | return p.rangeStart, p.rangeStart |
| 250 | } |
| 251 | return p.getRange(i) |
| 252 | } |
| 253 | |
| 254 | func (p *Parser) getRange(i int) (first, last rune) { |
| 255 | b := p.getField(i) |
| 256 | if k := strings.Index(b, ".."); k != -1 { |
| 257 | return p.parseRune(b[:k]), p.parseRune(b[k+2:]) |
| 258 | } |
| 259 | // The first field may not be a rune, in which case we may ignore any error |
| 260 | // and set the range as 0..0. |
| 261 | x, err := parseRune(b) |
| 262 | if err != nil { |
| 263 | // Disable range parsing henceforth. This ensures that an error will be |
| 264 | // returned if the user subsequently will try to parse this field as |
| 265 | // a Rune. |
| 266 | p.keepRanges = true |
| 267 | } |
| 268 | // Special case for UnicodeData that was retained for backwards compatibility. |
| 269 | if i == 0 && len(p.field) > 1 && strings.HasSuffix(p.field[1], "First>") { |
| 270 | if p.parsedRange { |
| 271 | return p.rangeStart, p.rangeEnd |
| 272 | } |
| 273 | mf := reRange.FindStringSubmatch(p.scanner.Text()) |
| 274 | p.line++ |
| 275 | if mf == nil || !p.scanner.Scan() { |
| 276 | p.setError(errIncorrectLegacyRange, "") |
| 277 | return x, x |
| 278 | } |
| 279 | // Using Bytes would be more efficient here, but Text is a lot easier |
| 280 | // and this is not a frequent case. |
| 281 | ml := reRange.FindStringSubmatch(p.scanner.Text()) |
| 282 | if ml == nil || mf[2] != ml[2] || ml[3] != "Last" || mf[4] != ml[4] { |
| 283 | p.setError(errIncorrectLegacyRange, "") |
| 284 | return x, x |
| 285 | } |
| 286 | p.rangeStart, p.rangeEnd = x, p.parseRune(p.scanner.Text()[:len(ml[1])]) |
| 287 | p.parsedRange = true |
| 288 | return p.rangeStart, p.rangeEnd |
| 289 | } |
| 290 | return x, x |
| 291 | } |
| 292 | |
| 293 | // bools recognizes all valid UCD boolean values. |
| 294 | var bools = map[string]bool{ |
| 295 | "": false, |
| 296 | "N": false, |
| 297 | "No": false, |
| 298 | "F": false, |
| 299 | "False": false, |
| 300 | "Y": true, |
| 301 | "Yes": true, |
| 302 | "T": true, |
| 303 | "True": true, |
| 304 | } |
| 305 | |
| 306 | // Bool parses and returns field i as a boolean value. |
| 307 | func (p *Parser) Bool(i int) bool { |
| 308 | f := p.getField(i) |
| 309 | for s, v := range bools { |
| 310 | if f == s { |
| 311 | return v |
| 312 | } |
| 313 | } |
| 314 | p.setError(strconv.ErrSyntax, "error parsing bool") |
| 315 | return false |
| 316 | } |
| 317 | |
| 318 | // Int parses and returns field i as an integer value. |
| 319 | func (p *Parser) Int(i int) int { |
| 320 | x, err := strconv.ParseInt(string(p.getField(i)), 10, 64) |
| 321 | p.setError(err, "error parsing int") |
| 322 | return int(x) |
| 323 | } |
| 324 | |
| 325 | // Uint parses and returns field i as an unsigned integer value. |
| 326 | func (p *Parser) Uint(i int) uint { |
| 327 | x, err := strconv.ParseUint(string(p.getField(i)), 10, 64) |
| 328 | p.setError(err, "error parsing uint") |
| 329 | return uint(x) |
| 330 | } |
| 331 | |
| 332 | // Float parses and returns field i as a decimal value. |
| 333 | func (p *Parser) Float(i int) float64 { |
| 334 | x, err := strconv.ParseFloat(string(p.getField(i)), 64) |
| 335 | p.setError(err, "error parsing float") |
| 336 | return x |
| 337 | } |
| 338 | |
| 339 | // String parses and returns field i as a string value. |
| 340 | func (p *Parser) String(i int) string { |
| 341 | return string(p.getField(i)) |
| 342 | } |
| 343 | |
| 344 | // Strings parses and returns field i as a space-separated list of strings. |
| 345 | func (p *Parser) Strings(i int) []string { |
| 346 | ss := strings.Split(string(p.getField(i)), " ") |
| 347 | for i, s := range ss { |
| 348 | ss[i] = strings.TrimSpace(s) |
| 349 | } |
| 350 | return ss |
| 351 | } |
| 352 | |
| 353 | // Comment returns the comments for the current line. |
| 354 | func (p *Parser) Comment() string { |
| 355 | return string(p.comment) |
| 356 | } |
| 357 | |
| 358 | var errUndefinedEnum = errors.New("ucd: undefined enum value") |
| 359 | |
| 360 | // Enum interprets and returns field i as a value that must be one of the values |
| 361 | // in enum. |
| 362 | func (p *Parser) Enum(i int, enum ...string) string { |
| 363 | f := p.getField(i) |
| 364 | for _, s := range enum { |
| 365 | if f == s { |
| 366 | return s |
| 367 | } |
| 368 | } |
| 369 | p.setError(errUndefinedEnum, "error parsing enum") |
| 370 | return "" |
| 371 | } |