src/libcore/unicode/unicode.py

   1 #!/usr/bin/env python
   2 #
   3 # Copyright 2011-2013 The Rust Project Developers. See the COPYRIGHT
   4 # file at the top-level directory of this distribution and at
   5 # http://rust-lang.org/COPYRIGHT.
   6 #
   7 # Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
   8 # http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
   9 # <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
  10 # option. This file may not be copied, modified, or distributed
  11 # except according to those terms.
  12
  13 # This script uses the following Unicode tables:
  14 # - DerivedCoreProperties.txt
  15 # - DerivedNormalizationProps.txt
  16 # - EastAsianWidth.txt
  17 # - auxiliary/GraphemeBreakProperty.txt
  18 # - PropList.txt
  19 # - ReadMe.txt
  20 # - Scripts.txt
  21 # - UnicodeData.txt
  22 #
  23 # Since this should not require frequent updates, we just store this
  24 # out-of-line and check the unicode.rs file into git.
  25
  26 import fileinput, re, os, sys, operator, math
  27
  28 preamble = '''// Copyright 2012-2016 The Rust Project Developers. See the COPYRIGHT
  29 // file at the top-level directory of this distribution and at
  30 // http://rust-lang.org/COPYRIGHT.
  31 //
  32 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
  33 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
  34 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
  35 // option. This file may not be copied, modified, or distributed
  36 // except according to those terms.
  37
  38 // NOTE: The following code was generated by "./unicode.py", do not edit directly
  39
  40 #![allow(missing_docs, non_upper_case_globals, non_snake_case)]
  41
  42 use unicode::version::UnicodeVersion;
  43 use unicode::bool_trie::{BoolTrie, SmallBoolTrie};
  44 '''
  45
  46 # Mapping taken from Table 12 from:
  47 # http://www.unicode.org/reports/tr44/#General_Category_Values
  48 expanded_categories = {
  49     'Lu': ['LC', 'L'], 'Ll': ['LC', 'L'], 'Lt': ['LC', 'L'],
  50     'Lm': ['L'], 'Lo': ['L'],
  51     'Mn': ['M'], 'Mc': ['M'], 'Me': ['M'],
  52     'Nd': ['N'], 'Nl': ['N'], 'No': ['No'],
  53     'Pc': ['P'], 'Pd': ['P'], 'Ps': ['P'], 'Pe': ['P'],
  54     'Pi': ['P'], 'Pf': ['P'], 'Po': ['P'],
  55     'Sm': ['S'], 'Sc': ['S'], 'Sk': ['S'], 'So': ['S'],
  56     'Zs': ['Z'], 'Zl': ['Z'], 'Zp': ['Z'],
  57     'Cc': ['C'], 'Cf': ['C'], 'Cs': ['C'], 'Co': ['C'], 'Cn': ['C'],
  58 }
  59
  60 # these are the surrogate codepoints, which are not valid rust characters
  61 surrogate_codepoints = (0xd800, 0xdfff)
  62
  63 def fetch(f):
  64     if not os.path.exists(os.path.basename(f)):
  65         os.system("curl -O http://www.unicode.org/Public/UNIDATA/%s"
  66                   % f)
  67
  68     if not os.path.exists(os.path.basename(f)):
  69         sys.stderr.write("cannot load %s" % f)
  70         exit(1)
  71
  72 def is_surrogate(n):
  73     return surrogate_codepoints[0] <= n <= surrogate_codepoints[1]
  74
  75 def load_unicode_data(f):
  76     fetch(f)
  77     gencats = {}
  78     to_lower = {}
  79     to_upper = {}
  80     to_title = {}
  81     combines = {}
  82     canon_decomp = {}
  83     compat_decomp = {}
  84
  85     udict = {}
  86     range_start = -1
  87     for line in fileinput.input(f):
  88         data = line.split(';')
  89         if len(data) != 15:
  90             continue
  91         cp = int(data[0], 16)
  92         if is_surrogate(cp):
  93             continue
  94         if range_start >= 0:
  95             for i in range(range_start, cp):
  96                 udict[i] = data
  97             range_start = -1
  98         if data[1].endswith(", First>"):
  99             range_start = cp
 100             continue
 101         udict[cp] = data
 102
 103     for code in udict:
 104         (code_org, name, gencat, combine, bidi,
 105          decomp, deci, digit, num, mirror,
 106          old, iso, upcase, lowcase, titlecase) = udict[code]
 107
 108         # generate char to char direct common and simple conversions
 109         # uppercase to lowercase
 110         if lowcase != "" and code_org != lowcase:
 111             to_lower[code] = (int(lowcase, 16), 0, 0)
 112
 113         # lowercase to uppercase
 114         if upcase != "" and code_org != upcase:
 115             to_upper[code] = (int(upcase, 16), 0, 0)
 116
 117         # title case
 118         if titlecase.strip() != "" and code_org != titlecase:
 119             to_title[code] = (int(titlecase, 16), 0, 0)
 120
 121         # store decomposition, if given
 122         if decomp != "":
 123             if decomp.startswith('<'):
 124                 seq = []
 125                 for i in decomp.split()[1:]:
 126                     seq.append(int(i, 16))
 127                 compat_decomp[code] = seq
 128             else:
 129                 seq = []
 130                 for i in decomp.split():
 131                     seq.append(int(i, 16))
 132                 canon_decomp[code] = seq
 133
 134         # place letter in categories as appropriate
 135         for cat in [gencat, "Assigned"] + expanded_categories.get(gencat, []):
 136             if cat not in gencats:
 137                 gencats[cat] = []
 138             gencats[cat].append(code)
 139
 140         # record combining class, if any
 141         if combine != "0":
 142             if combine not in combines:
 143                 combines[combine] = []
 144             combines[combine].append(code)
 145
 146     # generate Not_Assigned from Assigned
 147     gencats["Cn"] = gen_unassigned(gencats["Assigned"])
 148     # Assigned is not a real category
 149     del(gencats["Assigned"])
 150     # Other contains Not_Assigned
 151     gencats["C"].extend(gencats["Cn"])
 152     gencats = group_cats(gencats)
 153     combines = to_combines(group_cats(combines))
 154
 155     return (canon_decomp, compat_decomp, gencats, combines, to_upper, to_lower, to_title)
 156
 157 def load_special_casing(f, to_upper, to_lower, to_title):
 158     fetch(f)
 159     for line in fileinput.input(f):
 160         data = line.split('#')[0].split(';')
 161         if len(data) == 5:
 162             code, lower, title, upper, _comment = data
 163         elif len(data) == 6:
 164             code, lower, title, upper, condition, _comment = data
 165             if condition.strip():  # Only keep unconditional mappins
 166                 continue
 167         else:
 168             continue
 169         code = code.strip()
 170         lower = lower.strip()
 171         title = title.strip()
 172         upper = upper.strip()
 173         key = int(code, 16)
 174         for (map_, values) in [(to_lower, lower), (to_upper, upper), (to_title, title)]:
 175             if values != code:
 176                 values = [int(i, 16) for i in values.split()]
 177                 for _ in range(len(values), 3):
 178                     values.append(0)
 179                 assert len(values) == 3
 180                 map_[key] = values
 181
 182 def group_cats(cats):
 183     cats_out = {}
 184     for cat in cats:
 185         cats_out[cat] = group_cat(cats[cat])
 186     return cats_out
 187
 188 def group_cat(cat):
 189     cat_out = []
 190     letters = sorted(set(cat))
 191     cur_start = letters.pop(0)
 192     cur_end = cur_start
 193     for letter in letters:
 194         assert letter > cur_end, \
 195             "cur_end: %s, letter: %s" % (hex(cur_end), hex(letter))
 196         if letter == cur_end + 1:
 197             cur_end = letter
 198         else:
 199             cat_out.append((cur_start, cur_end))
 200             cur_start = cur_end = letter
 201     cat_out.append((cur_start, cur_end))
 202     return cat_out
 203
 204 def ungroup_cat(cat):
 205     cat_out = []
 206     for (lo, hi) in cat:
 207         while lo <= hi:
 208             cat_out.append(lo)
 209             lo += 1
 210     return cat_out
 211
 212 def gen_unassigned(assigned):
 213     assigned = set(assigned)
 214     return ([i for i in range(0, 0xd800) if i not in assigned] +
 215             [i for i in range(0xe000, 0x110000) if i not in assigned])
 216
 217 def to_combines(combs):
 218     combs_out = []
 219     for comb in combs:
 220         for (lo, hi) in combs[comb]:
 221             combs_out.append((lo, hi, comb))
 222     combs_out.sort(key=lambda comb: comb[0])
 223     return combs_out
 224
 225 def format_table_content(f, content, indent):
 226     line = " "*indent
 227     first = True
 228     for chunk in content.split(","):
 229         if len(line) + len(chunk) < 98:
 230             if first:
 231                 line += chunk
 232             else:
 233                 line += ", " + chunk
 234             first = False
 235         else:
 236             f.write(line + ",\n")
 237             line = " "*indent + chunk
 238     f.write(line)
 239
 240 def load_properties(f, interestingprops):
 241     fetch(f)
 242     props = {}
 243     re1 = re.compile("^ *([0-9A-F]+) *; *(\w+)")
 244     re2 = re.compile("^ *([0-9A-F]+)\.\.([0-9A-F]+) *; *(\w+)")
 245
 246     for line in fileinput.input(os.path.basename(f)):
 247         prop = None
 248         d_lo = 0
 249         d_hi = 0
 250         m = re1.match(line)
 251         if m:
 252             d_lo = m.group(1)
 253             d_hi = m.group(1)
 254             prop = m.group(2)
 255         else:
 256             m = re2.match(line)
 257             if m:
 258                 d_lo = m.group(1)
 259                 d_hi = m.group(2)
 260                 prop = m.group(3)
 261             else:
 262                 continue
 263         if interestingprops and prop not in interestingprops:
 264             continue
 265         d_lo = int(d_lo, 16)
 266         d_hi = int(d_hi, 16)
 267         if prop not in props:
 268             props[prop] = []
 269         props[prop].append((d_lo, d_hi))
 270
 271     # optimize if possible
 272     for prop in props:
 273         props[prop] = group_cat(ungroup_cat(props[prop]))
 274
 275     return props
 276
 277 def escape_char(c):
 278     return "'\\u{%x}'" % c if c != 0 else "'\\0'"
 279
 280 def emit_table(f, name, t_data, t_type = "&[(char, char)]", is_pub=True,
 281         pfun=lambda x: "(%s,%s)" % (escape_char(x[0]), escape_char(x[1]))):
 282     pub_string = ""
 283     if is_pub:
 284         pub_string = "pub "
 285     f.write("    %sconst %s: %s = &[\n" % (pub_string, name, t_type))
 286     data = ""
 287     first = True
 288     for dat in t_data:
 289         if not first:
 290             data += ","
 291         first = False
 292         data += pfun(dat)
 293     format_table_content(f, data, 8)
 294     f.write("\n    ];\n\n")
 295
 296 def compute_trie(rawdata, chunksize):
 297     root = []
 298     childmap = {}
 299     child_data = []
 300     for i in range(len(rawdata) // chunksize):
 301         data = rawdata[i * chunksize: (i + 1) * chunksize]
 302         child = '|'.join(map(str, data))
 303         if child not in childmap:
 304             childmap[child] = len(childmap)
 305             child_data.extend(data)
 306         root.append(childmap[child])
 307     return (root, child_data)
 308
 309 def emit_bool_trie(f, name, t_data, is_pub=True):
 310     CHUNK = 64
 311     rawdata = [False] * 0x110000
 312     for (lo, hi) in t_data:
 313         for cp in range(lo, hi + 1):
 314             rawdata[cp] = True
 315
 316     # convert to bitmap chunks of 64 bits each
 317     chunks = []
 318     for i in range(0x110000 // CHUNK):
 319         chunk = 0
 320         for j in range(64):
 321             if rawdata[i * 64 + j]:
 322                 chunk |= 1 << j
 323         chunks.append(chunk)
 324
 325     pub_string = ""
 326     if is_pub:
 327         pub_string = "pub "
 328     f.write("    %sconst %s: &super::BoolTrie = &super::BoolTrie {\n" % (pub_string, name))
 329     f.write("        r1: [\n")
 330     data = ','.join('0x%016x' % chunk for chunk in chunks[0:0x800 // CHUNK])
 331     format_table_content(f, data, 12)
 332     f.write("\n        ],\n")
 333
 334     # 0x800..0x10000 trie
 335     (r2, r3) = compute_trie(chunks[0x800 // CHUNK : 0x10000 // CHUNK], 64 // CHUNK)
 336     f.write("        r2: [\n")
 337     data = ','.join(str(node) for node in r2)
 338     format_table_content(f, data, 12)
 339     f.write("\n        ],\n")
 340     f.write("        r3: &[\n")
 341     data = ','.join('0x%016x' % chunk for chunk in r3)
 342     format_table_content(f, data, 12)
 343     f.write("\n        ],\n")
 344
 345     # 0x10000..0x110000 trie
 346     (mid, r6) = compute_trie(chunks[0x10000 // CHUNK : 0x110000 // CHUNK], 64 // CHUNK)
 347     (r4, r5) = compute_trie(mid, 64)
 348     f.write("        r4: [\n")
 349     data = ','.join(str(node) for node in r4)
 350     format_table_content(f, data, 12)
 351     f.write("\n        ],\n")
 352     f.write("        r5: &[\n")
 353     data = ','.join(str(node) for node in r5)
 354     format_table_content(f, data, 12)
 355     f.write("\n        ],\n")
 356     f.write("        r6: &[\n")
 357     data = ','.join('0x%016x' % chunk for chunk in r6)
 358     format_table_content(f, data, 12)
 359     f.write("\n        ],\n")
 360
 361     f.write("    };\n\n")
 362
 363 def emit_small_bool_trie(f, name, t_data, is_pub=True):
 364     last_chunk = max(hi // 64 for (lo, hi) in t_data)
 365     n_chunks = last_chunk + 1
 366     chunks = [0] * n_chunks
 367     for (lo, hi) in t_data:
 368         for cp in range(lo, hi + 1):
 369             if cp // 64 >= len(chunks):
 370                 print(cp, cp // 64, len(chunks), lo, hi)
 371             chunks[cp // 64] |= 1 << (cp & 63)
 372
 373     pub_string = ""
 374     if is_pub:
 375         pub_string = "pub "
 376     f.write("    %sconst %s: &super::SmallBoolTrie = &super::SmallBoolTrie {\n"
 377             % (pub_string, name))
 378
 379     (r1, r2) = compute_trie(chunks, 1)
 380
 381     f.write("        r1: &[\n")
 382     data = ','.join(str(node) for node in r1)
 383     format_table_content(f, data, 12)
 384     f.write("\n        ],\n")
 385
 386     f.write("        r2: &[\n")
 387     data = ','.join('0x%016x' % node for node in r2)
 388     format_table_content(f, data, 12)
 389     f.write("\n        ],\n")
 390
 391     f.write("    };\n\n")
 392
 393 def emit_property_module(f, mod, tbl, emit):
 394     f.write("pub mod %s {\n" % mod)
 395     for cat in sorted(emit):
 396         if cat in ["Cc", "White_Space", "Pattern_White_Space"]:
 397             emit_small_bool_trie(f, "%s_table" % cat, tbl[cat])
 398             f.write("    pub fn %s(c: char) -> bool {\n" % cat)
 399             f.write("        %s_table.lookup(c)\n" % cat)
 400             f.write("    }\n\n")
 401         else:
 402             emit_bool_trie(f, "%s_table" % cat, tbl[cat])
 403             f.write("    pub fn %s(c: char) -> bool {\n" % cat)
 404             f.write("        %s_table.lookup(c)\n" % cat)
 405             f.write("    }\n\n")
 406     f.write("}\n\n")
 407
 408 def emit_conversions_module(f, to_upper, to_lower, to_title):
 409     f.write("pub mod conversions {")
 410     f.write("""
 411     pub fn to_lower(c: char) -> [char; 3] {
 412         match bsearch_case_table(c, to_lowercase_table) {
 413             None        => [c, '\\0', '\\0'],
 414             Some(index) => to_lowercase_table[index].1,
 415         }
 416     }
 417
 418     pub fn to_upper(c: char) -> [char; 3] {
 419         match bsearch_case_table(c, to_uppercase_table) {
 420             None        => [c, '\\0', '\\0'],
 421             Some(index) => to_uppercase_table[index].1,
 422         }
 423     }
 424
 425     fn bsearch_case_table(c: char, table: &[(char, [char; 3])]) -> Option<usize> {
 426         table.binary_search_by(|&(key, _)| key.cmp(&c)).ok()
 427     }
 428
 429 """)
 430     t_type = "&[(char, [char; 3])]"
 431     pfun = lambda x: "(%s,[%s,%s,%s])" % (
 432         escape_char(x[0]), escape_char(x[1][0]), escape_char(x[1][1]), escape_char(x[1][2]))
 433     emit_table(f, "to_lowercase_table",
 434         sorted(to_lower.items(), key=operator.itemgetter(0)),
 435         is_pub=False, t_type = t_type, pfun=pfun)
 436     emit_table(f, "to_uppercase_table",
 437         sorted(to_upper.items(), key=operator.itemgetter(0)),
 438         is_pub=False, t_type = t_type, pfun=pfun)
 439     f.write("}\n\n")
 440
 441 def emit_norm_module(f, canon, compat, combine, norm_props):
 442     canon_keys = sorted(canon.keys())
 443
 444     compat_keys = sorted(compat.keys())
 445
 446     canon_comp = {}
 447     comp_exclusions = norm_props["Full_Composition_Exclusion"]
 448     for char in canon_keys:
 449         if any(lo <= char <= hi for lo, hi in comp_exclusions):
 450             continue
 451         decomp = canon[char]
 452         if len(decomp) == 2:
 453             if decomp[0] not in canon_comp:
 454                 canon_comp[decomp[0]] = []
 455             canon_comp[decomp[0]].append( (decomp[1], char) )
 456     canon_comp_keys = sorted(canon_comp.keys())
 457
 458 if __name__ == "__main__":
 459     r = "tables.rs"
 460     if os.path.exists(r):
 461         os.remove(r)
 462     with open(r, "w") as rf:
 463         # write the file's preamble
 464         rf.write(preamble)
 465
 466         # download and parse all the data
 467         fetch("ReadMe.txt")
 468         with open("ReadMe.txt") as readme:
 469             pattern = "for Version (\d+)\.(\d+)\.(\d+) of the Unicode"
 470             unicode_version = re.search(pattern, readme.read()).groups()
 471         rf.write("""
 472 /// The version of [Unicode](http://www.unicode.org/) that the Unicode parts of
 473 /// `char` and `str` methods are based on.
 474 #[unstable(feature = "unicode_version", issue = "49726")]
 475 pub const UNICODE_VERSION: UnicodeVersion = UnicodeVersion {
 476     major: %s,
 477     minor: %s,
 478     micro: %s,
 479     _priv: (),
 480 };
 481 """ % unicode_version)
 482         (canon_decomp, compat_decomp, gencats, combines,
 483                 to_upper, to_lower, to_title) = load_unicode_data("UnicodeData.txt")
 484         load_special_casing("SpecialCasing.txt", to_upper, to_lower, to_title)
 485         want_derived = ["XID_Start", "XID_Continue", "Alphabetic", "Lowercase", "Uppercase",
 486                         "Cased", "Case_Ignorable"]
 487         derived = load_properties("DerivedCoreProperties.txt", want_derived)
 488         scripts = load_properties("Scripts.txt", [])
 489         props = load_properties("PropList.txt",
 490                 ["White_Space", "Join_Control", "Noncharacter_Code_Point", "Pattern_White_Space"])
 491         norm_props = load_properties("DerivedNormalizationProps.txt",
 492                      ["Full_Composition_Exclusion"])
 493
 494         # category tables
 495         for (name, cat, pfuns) in ("general_category", gencats, ["N", "Cc"]), \
 496                                   ("derived_property", derived, want_derived), \
 497                                   ("property", props, ["White_Space", "Pattern_White_Space"]):
 498             emit_property_module(rf, name, cat, pfuns)
 499
 500         # normalizations and conversions module
 501         emit_norm_module(rf, canon_decomp, compat_decomp, combines, norm_props)
 502         emit_conversions_module(rf, to_upper, to_lower, to_title)