src/etc/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
  27
  28 preamble = '''// Copyright 2012-2015 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 "src/etc/unicode.py", do not edit directly
  39
  40 #![allow(missing_docs, non_upper_case_globals, non_snake_case)]
  41 '''
  42
  43 # Mapping taken from Table 12 from:
  44 # http://www.unicode.org/reports/tr44/#General_Category_Values
  45 expanded_categories = {
  46     'Lu': ['LC', 'L'], 'Ll': ['LC', 'L'], 'Lt': ['LC', 'L'],
  47     'Lm': ['L'], 'Lo': ['L'],
  48     'Mn': ['M'], 'Mc': ['M'], 'Me': ['M'],
  49     'Nd': ['N'], 'Nl': ['N'], 'No': ['No'],
  50     'Pc': ['P'], 'Pd': ['P'], 'Ps': ['P'], 'Pe': ['P'],
  51     'Pi': ['P'], 'Pf': ['P'], 'Po': ['P'],
  52     'Sm': ['S'], 'Sc': ['S'], 'Sk': ['S'], 'So': ['S'],
  53     'Zs': ['Z'], 'Zl': ['Z'], 'Zp': ['Z'],
  54     'Cc': ['C'], 'Cf': ['C'], 'Cs': ['C'], 'Co': ['C'], 'Cn': ['C'],
  55 }
  56
  57 # these are the surrogate codepoints, which are not valid rust characters
  58 surrogate_codepoints = (0xd800, 0xdfff)
  59
  60 def fetch(f):
  61     if not os.path.exists(os.path.basename(f)):
  62         os.system("curl -O http://www.unicode.org/Public/UNIDATA/%s"
  63                   % f)
  64
  65     if not os.path.exists(os.path.basename(f)):
  66         sys.stderr.write("cannot load %s" % f)
  67         exit(1)
  68
  69 def is_surrogate(n):
  70     return surrogate_codepoints[0] <= n <= surrogate_codepoints[1]
  71
  72 def load_unicode_data(f):
  73     fetch(f)
  74     gencats = {}
  75     to_lower = {}
  76     to_upper = {}
  77     to_title = {}
  78     combines = {}
  79     canon_decomp = {}
  80     compat_decomp = {}
  81
  82     udict = {};
  83     range_start = -1;
  84     for line in fileinput.input(f):
  85         data = line.split(';');
  86         if len(data) != 15:
  87             continue
  88         cp = int(data[0], 16);
  89         if is_surrogate(cp):
  90             continue
  91         if range_start >= 0:
  92             for i in xrange(range_start, cp):
  93                 udict[i] = data;
  94             range_start = -1;
  95         if data[1].endswith(", First>"):
  96             range_start = cp;
  97             continue;
  98         udict[cp] = data;
  99
 100     for code in udict:
 101         [code_org, name, gencat, combine, bidi,
 102          decomp, deci, digit, num, mirror,
 103          old, iso, upcase, lowcase, titlecase ] = udict[code];
 104
 105         # generate char to char direct common and simple conversions
 106         # uppercase to lowercase
 107         if lowcase != "" and code_org != lowcase:
 108             to_lower[code] = (int(lowcase, 16), 0, 0)
 109
 110         # lowercase to uppercase
 111         if upcase != "" and code_org != upcase:
 112             to_upper[code] = (int(upcase, 16), 0, 0)
 113
 114         # title case
 115         if titlecase.strip() != "" and code_org != titlecase:
 116             to_title[code] = (int(titlecase, 16), 0, 0)
 117
 118         # store decomposition, if given
 119         if decomp != "":
 120             if decomp.startswith('<'):
 121                 seq = []
 122                 for i in decomp.split()[1:]:
 123                     seq.append(int(i, 16))
 124                 compat_decomp[code] = seq
 125             else:
 126                 seq = []
 127                 for i in decomp.split():
 128                     seq.append(int(i, 16))
 129                 canon_decomp[code] = seq
 130
 131         # place letter in categories as appropriate
 132         for cat in [gencat, "Assigned"] + expanded_categories.get(gencat, []):
 133             if cat not in gencats:
 134                 gencats[cat] = []
 135             gencats[cat].append(code)
 136
 137         # record combining class, if any
 138         if combine != "0":
 139             if combine not in combines:
 140                 combines[combine] = []
 141             combines[combine].append(code)
 142
 143     # generate Not_Assigned from Assigned
 144     gencats["Cn"] = gen_unassigned(gencats["Assigned"])
 145     # Assigned is not a real category
 146     del(gencats["Assigned"])
 147     # Other contains Not_Assigned
 148     gencats["C"].extend(gencats["Cn"])
 149     gencats = group_cats(gencats)
 150     combines = to_combines(group_cats(combines))
 151
 152     return (canon_decomp, compat_decomp, gencats, combines, to_upper, to_lower, to_title)
 153
 154 def load_special_casing(f, to_upper, to_lower, to_title):
 155     fetch(f)
 156     for line in fileinput.input(f):
 157         data = line.split('#')[0].split(';')
 158         if len(data) == 5:
 159             code, lower, title, upper, _comment = data
 160         elif len(data) == 6:
 161             code, lower, title, upper, condition, _comment = data
 162             if condition.strip():  # Only keep unconditional mappins
 163                 continue
 164         else:
 165             continue
 166         code = code.strip()
 167         lower = lower.strip()
 168         title = title.strip()
 169         upper = upper.strip()
 170         key = int(code, 16)
 171         for (map_, values) in [(to_lower, lower), (to_upper, upper), (to_title, title)]:
 172             if values != code:
 173                 values = [int(i, 16) for i in values.split()]
 174                 for _ in range(len(values), 3):
 175                     values.append(0)
 176                 assert len(values) == 3
 177                 map_[key] = values
 178
 179 def group_cats(cats):
 180     cats_out = {}
 181     for cat in cats:
 182         cats_out[cat] = group_cat(cats[cat])
 183     return cats_out
 184
 185 def group_cat(cat):
 186     cat_out = []
 187     letters = sorted(set(cat))
 188     cur_start = letters.pop(0)
 189     cur_end = cur_start
 190     for letter in letters:
 191         assert letter > cur_end, \
 192             "cur_end: %s, letter: %s" % (hex(cur_end), hex(letter))
 193         if letter == cur_end + 1:
 194             cur_end = letter
 195         else:
 196             cat_out.append((cur_start, cur_end))
 197             cur_start = cur_end = letter
 198     cat_out.append((cur_start, cur_end))
 199     return cat_out
 200
 201 def ungroup_cat(cat):
 202     cat_out = []
 203     for (lo, hi) in cat:
 204         while lo <= hi:
 205             cat_out.append(lo)
 206             lo += 1
 207     return cat_out
 208
 209 def gen_unassigned(assigned):
 210     assigned = set(assigned)
 211     return ([i for i in range(0, 0xd800) if i not in assigned] +
 212             [i for i in range(0xe000, 0x110000) if i not in assigned])
 213
 214 def to_combines(combs):
 215     combs_out = []
 216     for comb in combs:
 217         for (lo, hi) in combs[comb]:
 218             combs_out.append((lo, hi, comb))
 219     combs_out.sort(key=lambda comb: comb[0])
 220     return combs_out
 221
 222 def format_table_content(f, content, indent):
 223     line = " "*indent
 224     first = True
 225     for chunk in content.split(","):
 226         if len(line) + len(chunk) < 98:
 227             if first:
 228                 line += chunk
 229             else:
 230                 line += ", " + chunk
 231             first = False
 232         else:
 233             f.write(line + ",\n")
 234             line = " "*indent + chunk
 235     f.write(line)
 236
 237 def load_properties(f, interestingprops):
 238     fetch(f)
 239     props = {}
 240     re1 = re.compile("^ *([0-9A-F]+) *; *(\w+)")
 241     re2 = re.compile("^ *([0-9A-F]+)\.\.([0-9A-F]+) *; *(\w+)")
 242
 243     for line in fileinput.input(os.path.basename(f)):
 244         prop = None
 245         d_lo = 0
 246         d_hi = 0
 247         m = re1.match(line)
 248         if m:
 249             d_lo = m.group(1)
 250             d_hi = m.group(1)
 251             prop = m.group(2)
 252         else:
 253             m = re2.match(line)
 254             if m:
 255                 d_lo = m.group(1)
 256                 d_hi = m.group(2)
 257                 prop = m.group(3)
 258             else:
 259                 continue
 260         if interestingprops and prop not in interestingprops:
 261             continue
 262         d_lo = int(d_lo, 16)
 263         d_hi = int(d_hi, 16)
 264         if prop not in props:
 265             props[prop] = []
 266         props[prop].append((d_lo, d_hi))
 267
 268     # optimize if possible
 269     for prop in props:
 270         props[prop] = group_cat(ungroup_cat(props[prop]))
 271
 272     return props
 273
 274 # load all widths of want_widths, except those in except_cats
 275 def load_east_asian_width(want_widths, except_cats):
 276     f = "EastAsianWidth.txt"
 277     fetch(f)
 278     widths = {}
 279     re1 = re.compile("^([0-9A-F]+);(\w+) +# (\w+)")
 280     re2 = re.compile("^([0-9A-F]+)\.\.([0-9A-F]+);(\w+) +# (\w+)")
 281
 282     for line in fileinput.input(f):
 283         width = None
 284         d_lo = 0
 285         d_hi = 0
 286         cat = None
 287         m = re1.match(line)
 288         if m:
 289             d_lo = m.group(1)
 290             d_hi = m.group(1)
 291             width = m.group(2)
 292             cat = m.group(3)
 293         else:
 294             m = re2.match(line)
 295             if m:
 296                 d_lo = m.group(1)
 297                 d_hi = m.group(2)
 298                 width = m.group(3)
 299                 cat = m.group(4)
 300             else:
 301                 continue
 302         if cat in except_cats or width not in want_widths:
 303             continue
 304         d_lo = int(d_lo, 16)
 305         d_hi = int(d_hi, 16)
 306         if width not in widths:
 307             widths[width] = []
 308         widths[width].append((d_lo, d_hi))
 309     return widths
 310
 311 def escape_char(c):
 312     return "'\\u{%x}'" % c if c != 0 else "'\\0'"
 313
 314 def emit_bsearch_range_table(f):
 315     f.write("""
 316 fn bsearch_range_table(c: char, r: &'static [(char, char)]) -> bool {
 317     use core::cmp::Ordering::{Equal, Less, Greater};
 318     r.binary_search_by(|&(lo, hi)| {
 319          if lo <= c && c <= hi {
 320              Equal
 321          } else if hi < c {
 322              Less
 323          } else {
 324              Greater
 325          }
 326      })
 327      .is_ok()
 328 }\n
 329 """)
 330
 331 def emit_table(f, name, t_data, t_type = "&'static [(char, char)]", is_pub=True,
 332         pfun=lambda x: "(%s,%s)" % (escape_char(x[0]), escape_char(x[1]))):
 333     pub_string = ""
 334     if is_pub:
 335         pub_string = "pub "
 336     f.write("    %sconst %s: %s = &[\n" % (pub_string, name, t_type))
 337     data = ""
 338     first = True
 339     for dat in t_data:
 340         if not first:
 341             data += ","
 342         first = False
 343         data += pfun(dat)
 344     format_table_content(f, data, 8)
 345     f.write("\n    ];\n\n")
 346
 347 def emit_property_module(f, mod, tbl, emit):
 348     f.write("pub mod %s {\n" % mod)
 349     for cat in sorted(emit):
 350         emit_table(f, "%s_table" % cat, tbl[cat])
 351         f.write("    pub fn %s(c: char) -> bool {\n" % cat)
 352         f.write("        super::bsearch_range_table(c, %s_table)\n" % cat)
 353         f.write("    }\n\n")
 354     f.write("}\n\n")
 355
 356 def emit_conversions_module(f, to_upper, to_lower, to_title):
 357     f.write("pub mod conversions {")
 358     f.write("""
 359     use core::cmp::Ordering::{Equal, Less, Greater};
 360     use core::option::Option;
 361     use core::option::Option::{Some, None};
 362     use core::result::Result::{Ok, Err};
 363
 364     pub fn to_lower(c: char) -> [char; 3] {
 365         match bsearch_case_table(c, to_lowercase_table) {
 366           None        => [c, '\\0', '\\0'],
 367           Some(index) => to_lowercase_table[index].1
 368         }
 369     }
 370
 371     pub fn to_upper(c: char) -> [char; 3] {
 372         match bsearch_case_table(c, to_uppercase_table) {
 373             None        => [c, '\\0', '\\0'],
 374             Some(index) => to_uppercase_table[index].1
 375         }
 376     }
 377
 378     fn bsearch_case_table(c: char, table: &'static [(char, [char; 3])]) -> Option<usize> {
 379         match table.binary_search_by(|&(key, _)| {
 380             if c == key { Equal }
 381             else if key < c { Less }
 382             else { Greater }
 383         }) {
 384             Ok(i) => Some(i),
 385             Err(_) => None,
 386         }
 387     }
 388
 389 """)
 390     t_type = "&'static [(char, [char; 3])]"
 391     pfun = lambda x: "(%s,[%s,%s,%s])" % (
 392         escape_char(x[0]), escape_char(x[1][0]), escape_char(x[1][1]), escape_char(x[1][2]))
 393     emit_table(f, "to_lowercase_table",
 394         sorted(to_lower.iteritems(), key=operator.itemgetter(0)),
 395         is_pub=False, t_type = t_type, pfun=pfun)
 396     emit_table(f, "to_uppercase_table",
 397         sorted(to_upper.iteritems(), key=operator.itemgetter(0)),
 398         is_pub=False, t_type = t_type, pfun=pfun)
 399     f.write("}\n\n")
 400
 401 def emit_charwidth_module(f, width_table):
 402     f.write("pub mod charwidth {\n")
 403     f.write("    use core::option::Option;\n")
 404     f.write("    use core::option::Option::{Some, None};\n")
 405     f.write("    use core::result::Result::{Ok, Err};\n")
 406     f.write("""
 407     fn bsearch_range_value_table(c: char, is_cjk: bool, r: &'static [(char, char, u8, u8)]) -> u8 {
 408         use core::cmp::Ordering::{Equal, Less, Greater};
 409         match r.binary_search_by(|&(lo, hi, _, _)| {
 410             if lo <= c && c <= hi { Equal }
 411             else if hi < c { Less }
 412             else { Greater }
 413         }) {
 414             Ok(idx) => {
 415                 let (_, _, r_ncjk, r_cjk) = r[idx];
 416                 if is_cjk { r_cjk } else { r_ncjk }
 417             }
 418             Err(_) => 1
 419         }
 420     }
 421 """)
 422
 423     f.write("""
 424     pub fn width(c: char, is_cjk: bool) -> Option<usize> {
 425         match c as usize {
 426             _c @ 0 => Some(0),          // null is zero width
 427             cu if cu < 0x20 => None,    // control sequences have no width
 428             cu if cu < 0x7F => Some(1), // ASCII
 429             cu if cu < 0xA0 => None,    // more control sequences
 430             _ => Some(bsearch_range_value_table(c, is_cjk, charwidth_table) as usize)
 431         }
 432     }
 433
 434 """)
 435
 436     f.write("    // character width table. Based on Markus Kuhn's free wcwidth() implementation,\n")
 437     f.write("    //     http://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c\n")
 438     emit_table(f, "charwidth_table", width_table, "&'static [(char, char, u8, u8)]", is_pub=False,
 439             pfun=lambda x: "(%s,%s,%s,%s)" % (escape_char(x[0]), escape_char(x[1]), x[2], x[3]))
 440     f.write("}\n\n")
 441
 442 def emit_norm_module(f, canon, compat, combine, norm_props):
 443     canon_keys = canon.keys()
 444     canon_keys.sort()
 445
 446     compat_keys = compat.keys()
 447     compat_keys.sort()
 448
 449     canon_comp = {}
 450     comp_exclusions = norm_props["Full_Composition_Exclusion"]
 451     for char in canon_keys:
 452         if True in map(lambda (lo, hi): lo <= char <= hi, comp_exclusions):
 453             continue
 454         decomp = canon[char]
 455         if len(decomp) == 2:
 456             if not canon_comp.has_key(decomp[0]):
 457                 canon_comp[decomp[0]] = []
 458             canon_comp[decomp[0]].append( (decomp[1], char) )
 459     canon_comp_keys = canon_comp.keys()
 460     canon_comp_keys.sort()
 461
 462 def remove_from_wtable(wtable, val):
 463     wtable_out = []
 464     while wtable:
 465         if wtable[0][1] < val:
 466             wtable_out.append(wtable.pop(0))
 467         elif wtable[0][0] > val:
 468             break
 469         else:
 470             (wt_lo, wt_hi, width, width_cjk) = wtable.pop(0)
 471             if wt_lo == wt_hi == val:
 472                 continue
 473             elif wt_lo == val:
 474                 wtable_out.append((wt_lo+1, wt_hi, width, width_cjk))
 475             elif wt_hi == val:
 476                 wtable_out.append((wt_lo, wt_hi-1, width, width_cjk))
 477             else:
 478                 wtable_out.append((wt_lo, val-1, width, width_cjk))
 479                 wtable_out.append((val+1, wt_hi, width, width_cjk))
 480     if wtable:
 481         wtable_out.extend(wtable)
 482     return wtable_out
 483
 484
 485
 486 def optimize_width_table(wtable):
 487     wtable_out = []
 488     w_this = wtable.pop(0)
 489     while wtable:
 490         if w_this[1] == wtable[0][0] - 1 and w_this[2:3] == wtable[0][2:3]:
 491             w_tmp = wtable.pop(0)
 492             w_this = (w_this[0], w_tmp[1], w_tmp[2], w_tmp[3])
 493         else:
 494             wtable_out.append(w_this)
 495             w_this = wtable.pop(0)
 496     wtable_out.append(w_this)
 497     return wtable_out
 498
 499 if __name__ == "__main__":
 500     r = "tables.rs"
 501     if os.path.exists(r):
 502         os.remove(r)
 503     with open(r, "w") as rf:
 504         # write the file's preamble
 505         rf.write(preamble)
 506
 507         # download and parse all the data
 508         fetch("ReadMe.txt")
 509         with open("ReadMe.txt") as readme:
 510             pattern = "for Version (\d+)\.(\d+)\.(\d+) of the Unicode"
 511             unicode_version = re.search(pattern, readme.read()).groups()
 512         rf.write("""
 513 /// The version of [Unicode](http://www.unicode.org/)
 514 /// that the unicode parts of `CharExt` and `UnicodeStrPrelude` traits are based on.
 515 pub const UNICODE_VERSION: (u64, u64, u64) = (%s, %s, %s);
 516 """ % unicode_version)
 517         (canon_decomp, compat_decomp, gencats, combines,
 518                 to_upper, to_lower, to_title) = load_unicode_data("UnicodeData.txt")
 519         load_special_casing("SpecialCasing.txt", to_upper, to_lower, to_title)
 520         want_derived = ["XID_Start", "XID_Continue", "Alphabetic", "Lowercase", "Uppercase",
 521                         "Cased", "Case_Ignorable"]
 522         derived = load_properties("DerivedCoreProperties.txt", want_derived)
 523         scripts = load_properties("Scripts.txt", [])
 524         props = load_properties("PropList.txt",
 525                 ["White_Space", "Join_Control", "Noncharacter_Code_Point"])
 526         norm_props = load_properties("DerivedNormalizationProps.txt",
 527                      ["Full_Composition_Exclusion"])
 528
 529         # bsearch_range_table is used in all the property modules below
 530         emit_bsearch_range_table(rf)
 531
 532         # category tables
 533         for (name, cat, pfuns) in ("general_category", gencats, ["N", "Cc"]), \
 534                                   ("derived_property", derived, want_derived), \
 535                                   ("property", props, ["White_Space"]):
 536             emit_property_module(rf, name, cat, pfuns)
 537
 538         # normalizations and conversions module
 539         emit_norm_module(rf, canon_decomp, compat_decomp, combines, norm_props)
 540         emit_conversions_module(rf, to_upper, to_lower, to_title)