Ian Hodson | 2ee91b4 | 2012-05-14 12:29:36 +0100 | [diff] [blame] | 1 | # Copyright 2008 The RE2 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 | """Parser for Unicode data files (as distributed by unicode.org).""" |
| 6 | |
| 7 | import os |
| 8 | import re |
| 9 | import urllib2 |
| 10 | |
| 11 | # Directory or URL where Unicode tables reside. |
| 12 | _UNICODE_DIR = "http://www.unicode.org/Public/6.0.0/ucd" |
| 13 | |
| 14 | # Largest valid Unicode code value. |
| 15 | _RUNE_MAX = 0x10FFFF |
| 16 | |
| 17 | |
| 18 | class Error(Exception): |
| 19 | """Unicode error base class.""" |
| 20 | |
| 21 | |
| 22 | class InputError(Error): |
| 23 | """Unicode input error class. Raised on invalid input.""" |
| 24 | |
| 25 | |
| 26 | def _UInt(s): |
| 27 | """Converts string to Unicode code point ('263A' => 0x263a). |
| 28 | |
| 29 | Args: |
| 30 | s: string to convert |
| 31 | |
| 32 | Returns: |
| 33 | Unicode code point |
| 34 | |
| 35 | Raises: |
| 36 | InputError: the string is not a valid Unicode value. |
| 37 | """ |
| 38 | |
| 39 | try: |
| 40 | v = int(s, 16) |
| 41 | except ValueError: |
| 42 | v = -1 |
| 43 | if len(s) < 4 or len(s) > 6 or v < 0 or v > _RUNE_MAX: |
| 44 | raise InputError("invalid Unicode value %s" % (s,)) |
| 45 | return v |
| 46 | |
| 47 | |
| 48 | def _URange(s): |
| 49 | """Converts string to Unicode range. |
| 50 | |
| 51 | '0001..0003' => [1, 2, 3]. |
| 52 | '0001' => [1]. |
| 53 | |
| 54 | Args: |
| 55 | s: string to convert |
| 56 | |
| 57 | Returns: |
| 58 | Unicode range |
| 59 | |
| 60 | Raises: |
| 61 | InputError: the string is not a valid Unicode range. |
| 62 | """ |
| 63 | a = s.split("..") |
| 64 | if len(a) == 1: |
| 65 | return [_UInt(a[0])] |
| 66 | if len(a) == 2: |
| 67 | lo = _UInt(a[0]) |
| 68 | hi = _UInt(a[1]) |
| 69 | if lo < hi: |
| 70 | return range(lo, hi + 1) |
| 71 | raise InputError("invalid Unicode range %s" % (s,)) |
| 72 | |
| 73 | |
| 74 | def _UStr(v): |
| 75 | """Converts Unicode code point to hex string. |
| 76 | |
| 77 | 0x263a => '0x263A'. |
| 78 | |
| 79 | Args: |
| 80 | v: code point to convert |
| 81 | |
| 82 | Returns: |
| 83 | Unicode string |
| 84 | |
| 85 | Raises: |
| 86 | InputError: the argument is not a valid Unicode value. |
| 87 | """ |
| 88 | if v < 0 or v > _RUNE_MAX: |
| 89 | raise InputError("invalid Unicode value %s" % (v,)) |
| 90 | return "0x%04X" % (v,) |
| 91 | |
| 92 | |
| 93 | def _ParseContinue(s): |
| 94 | """Parses a Unicode continuation field. |
| 95 | |
| 96 | These are of the form '<Name, First>' or '<Name, Last>'. |
| 97 | Instead of giving an explicit range in a single table entry, |
| 98 | some Unicode tables use two entries, one for the first |
| 99 | code value in the range and one for the last. |
| 100 | The first entry's description is '<Name, First>' instead of 'Name' |
| 101 | and the second is '<Name, Last>'. |
| 102 | |
| 103 | '<Name, First>' => ('Name', 'First') |
| 104 | '<Name, Last>' => ('Name', 'Last') |
| 105 | 'Anything else' => ('Anything else', None) |
| 106 | |
| 107 | Args: |
| 108 | s: continuation field string |
| 109 | |
| 110 | Returns: |
| 111 | pair: name and ('First', 'Last', or None) |
| 112 | """ |
| 113 | |
| 114 | match = re.match("<(.*), (First|Last)>", s) |
| 115 | if match is not None: |
| 116 | return match.groups() |
| 117 | return (s, None) |
| 118 | |
| 119 | |
| 120 | def ReadUnicodeTable(filename, nfields, doline): |
| 121 | """Generic Unicode table text file reader. |
| 122 | |
| 123 | The reader takes care of stripping out comments and also |
| 124 | parsing the two different ways that the Unicode tables specify |
| 125 | code ranges (using the .. notation and splitting the range across |
| 126 | multiple lines). |
| 127 | |
| 128 | Each non-comment line in the table is expected to have the given |
| 129 | number of fields. The first field is known to be the Unicode value |
| 130 | and the second field its description. |
| 131 | |
| 132 | The reader calls doline(codes, fields) for each entry in the table. |
| 133 | If fn raises an exception, the reader prints that exception, |
| 134 | prefixed with the file name and line number, and continues |
| 135 | processing the file. When done with the file, the reader re-raises |
| 136 | the first exception encountered during the file. |
| 137 | |
| 138 | Arguments: |
| 139 | filename: the Unicode data file to read, or a file-like object. |
| 140 | nfields: the number of expected fields per line in that file. |
| 141 | doline: the function to call for each table entry. |
| 142 | |
| 143 | Raises: |
| 144 | InputError: nfields is invalid (must be >= 2). |
| 145 | """ |
| 146 | |
| 147 | if nfields < 2: |
| 148 | raise InputError("invalid number of fields %d" % (nfields,)) |
| 149 | |
| 150 | if type(filename) == str: |
| 151 | if filename.startswith("http://"): |
| 152 | fil = urllib2.urlopen(filename) |
| 153 | else: |
| 154 | fil = open(filename, "r") |
| 155 | else: |
| 156 | fil = filename |
| 157 | |
| 158 | first = None # first code in multiline range |
| 159 | expect_last = None # tag expected for "Last" line in multiline range |
| 160 | lineno = 0 # current line number |
| 161 | for line in fil: |
| 162 | lineno += 1 |
| 163 | try: |
| 164 | # Chop # comments and white space; ignore empty lines. |
| 165 | sharp = line.find("#") |
| 166 | if sharp >= 0: |
| 167 | line = line[:sharp] |
| 168 | line = line.strip() |
| 169 | if not line: |
| 170 | continue |
| 171 | |
| 172 | # Split fields on ";", chop more white space. |
| 173 | # Must have the expected number of fields. |
| 174 | fields = [s.strip() for s in line.split(";")] |
| 175 | if len(fields) != nfields: |
| 176 | raise InputError("wrong number of fields %d %d - %s" % |
| 177 | (len(fields), nfields, line)) |
| 178 | |
| 179 | # The Unicode text files have two different ways |
| 180 | # to list a Unicode range. Either the first field is |
| 181 | # itself a range (0000..FFFF), or the range is split |
| 182 | # across two lines, with the second field noting |
| 183 | # the continuation. |
| 184 | codes = _URange(fields[0]) |
| 185 | (name, cont) = _ParseContinue(fields[1]) |
| 186 | |
| 187 | if expect_last is not None: |
| 188 | # If the last line gave the First code in a range, |
| 189 | # this one had better give the Last one. |
| 190 | if (len(codes) != 1 or codes[0] <= first or |
| 191 | cont != "Last" or name != expect_last): |
| 192 | raise InputError("expected Last line for %s" % |
| 193 | (expect_last,)) |
| 194 | codes = range(first, codes[0] + 1) |
| 195 | first = None |
| 196 | expect_last = None |
| 197 | fields[0] = "%04X..%04X" % (codes[0], codes[-1]) |
| 198 | fields[1] = name |
| 199 | elif cont == "First": |
| 200 | # Otherwise, if this is the First code in a range, |
| 201 | # remember it and go to the next line. |
| 202 | if len(codes) != 1: |
| 203 | raise InputError("bad First line: range given") |
| 204 | expect_last = name |
| 205 | first = codes[0] |
| 206 | continue |
| 207 | |
| 208 | doline(codes, fields) |
| 209 | |
| 210 | except Exception, e: |
| 211 | print "%s:%d: %s" % (filename, lineno, e) |
| 212 | raise |
| 213 | |
| 214 | if expect_last is not None: |
| 215 | raise InputError("expected Last line for %s; got EOF" % |
| 216 | (expect_last,)) |
| 217 | |
| 218 | |
| 219 | def CaseGroups(unicode_dir=_UNICODE_DIR): |
| 220 | """Returns list of Unicode code groups equivalent under case folding. |
| 221 | |
| 222 | Each group is a sorted list of code points, |
| 223 | and the list of groups is sorted by first code point |
| 224 | in the group. |
| 225 | |
| 226 | Args: |
| 227 | unicode_dir: Unicode data directory |
| 228 | |
| 229 | Returns: |
| 230 | list of Unicode code groups |
| 231 | """ |
| 232 | |
| 233 | # Dict mapping lowercase code point to fold-equivalent group. |
| 234 | togroup = {} |
| 235 | |
| 236 | def DoLine(codes, fields): |
| 237 | """Process single CaseFolding.txt line, updating togroup.""" |
| 238 | (_, foldtype, lower, _) = fields |
| 239 | if foldtype not in ("C", "S"): |
| 240 | return |
| 241 | lower = _UInt(lower) |
| 242 | togroup.setdefault(lower, [lower]).extend(codes) |
| 243 | |
| 244 | ReadUnicodeTable(unicode_dir+"/CaseFolding.txt", 4, DoLine) |
| 245 | |
| 246 | groups = togroup.values() |
| 247 | for g in groups: |
| 248 | g.sort() |
| 249 | groups.sort() |
| 250 | return togroup, groups |
| 251 | |
| 252 | |
| 253 | def Scripts(unicode_dir=_UNICODE_DIR): |
| 254 | """Returns dict mapping script names to code lists. |
| 255 | |
| 256 | Args: |
| 257 | unicode_dir: Unicode data directory |
| 258 | |
| 259 | Returns: |
| 260 | dict mapping script names to code lists |
| 261 | """ |
| 262 | |
| 263 | scripts = {} |
| 264 | |
| 265 | def DoLine(codes, fields): |
| 266 | """Process single Scripts.txt line, updating scripts.""" |
| 267 | (_, name) = fields |
| 268 | scripts.setdefault(name, []).extend(codes) |
| 269 | |
| 270 | ReadUnicodeTable(unicode_dir+"/Scripts.txt", 2, DoLine) |
| 271 | return scripts |
| 272 | |
| 273 | |
| 274 | def Categories(unicode_dir=_UNICODE_DIR): |
| 275 | """Returns dict mapping category names to code lists. |
| 276 | |
| 277 | Args: |
| 278 | unicode_dir: Unicode data directory |
| 279 | |
| 280 | Returns: |
| 281 | dict mapping category names to code lists |
| 282 | """ |
| 283 | |
| 284 | categories = {} |
| 285 | |
| 286 | def DoLine(codes, fields): |
| 287 | """Process single UnicodeData.txt line, updating categories.""" |
| 288 | category = fields[2] |
| 289 | categories.setdefault(category, []).extend(codes) |
| 290 | # Add codes from Lu into L, etc. |
| 291 | if len(category) > 1: |
| 292 | short = category[0] |
| 293 | categories.setdefault(short, []).extend(codes) |
| 294 | |
| 295 | ReadUnicodeTable(unicode_dir+"/UnicodeData.txt", 15, DoLine) |
| 296 | return categories |
| 297 | |