blob: 302f1940aaf487f4d8a87913ceed8194716365cc [file] [log] [blame]
William Kurkianea869482019-04-09 15:16:11 -04001// Copyright 2013 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// +build ignore
6
7// Language tag table generator.
8// Data read from the web.
9
10package main
11
12import (
13 "bufio"
14 "flag"
15 "fmt"
16 "io"
17 "io/ioutil"
18 "log"
19 "math"
20 "reflect"
21 "regexp"
22 "sort"
23 "strconv"
24 "strings"
25
26 "golang.org/x/text/internal/gen"
27 "golang.org/x/text/internal/tag"
28 "golang.org/x/text/unicode/cldr"
29)
30
31var (
32 test = flag.Bool("test",
33 false,
34 "test existing tables; can be used to compare web data with package data.")
35 outputFile = flag.String("output",
36 "tables.go",
37 "output file for generated tables")
38)
39
40var comment = []string{
41 `
42lang holds an alphabetically sorted list of ISO-639 language identifiers.
43All entries are 4 bytes. The index of the identifier (divided by 4) is the language tag.
44For 2-byte language identifiers, the two successive bytes have the following meaning:
45 - if the first letter of the 2- and 3-letter ISO codes are the same:
46 the second and third letter of the 3-letter ISO code.
47 - otherwise: a 0 and a by 2 bits right-shifted index into altLangISO3.
48For 3-byte language identifiers the 4th byte is 0.`,
49 `
50langNoIndex is a bit vector of all 3-letter language codes that are not used as an index
51in lookup tables. The language ids for these language codes are derived directly
52from the letters and are not consecutive.`,
53 `
54altLangISO3 holds an alphabetically sorted list of 3-letter language code alternatives
55to 2-letter language codes that cannot be derived using the method described above.
56Each 3-letter code is followed by its 1-byte langID.`,
57 `
58altLangIndex is used to convert indexes in altLangISO3 to langIDs.`,
59 `
60langAliasMap maps langIDs to their suggested replacements.`,
61 `
62script is an alphabetically sorted list of ISO 15924 codes. The index
63of the script in the string, divided by 4, is the internal scriptID.`,
64 `
65isoRegionOffset needs to be added to the index of regionISO to obtain the regionID
66for 2-letter ISO codes. (The first isoRegionOffset regionIDs are reserved for
67the UN.M49 codes used for groups.)`,
68 `
69regionISO holds a list of alphabetically sorted 2-letter ISO region codes.
70Each 2-letter codes is followed by two bytes with the following meaning:
71 - [A-Z}{2}: the first letter of the 2-letter code plus these two
72 letters form the 3-letter ISO code.
73 - 0, n: index into altRegionISO3.`,
74 `
75regionTypes defines the status of a region for various standards.`,
76 `
77m49 maps regionIDs to UN.M49 codes. The first isoRegionOffset entries are
78codes indicating collections of regions.`,
79 `
80m49Index gives indexes into fromM49 based on the three most significant bits
81of a 10-bit UN.M49 code. To search an UN.M49 code in fromM49, search in
82 fromM49[m49Index[msb39(code)]:m49Index[msb3(code)+1]]
83for an entry where the first 7 bits match the 7 lsb of the UN.M49 code.
84The region code is stored in the 9 lsb of the indexed value.`,
85 `
86fromM49 contains entries to map UN.M49 codes to regions. See m49Index for details.`,
87 `
88altRegionISO3 holds a list of 3-letter region codes that cannot be
89mapped to 2-letter codes using the default algorithm. This is a short list.`,
90 `
91altRegionIDs holds a list of regionIDs the positions of which match those
92of the 3-letter ISO codes in altRegionISO3.`,
93 `
94variantNumSpecialized is the number of specialized variants in variants.`,
95 `
96suppressScript is an index from langID to the dominant script for that language,
97if it exists. If a script is given, it should be suppressed from the language tag.`,
98 `
99likelyLang is a lookup table, indexed by langID, for the most likely
100scripts and regions given incomplete information. If more entries exist for a
101given language, region and script are the index and size respectively
102of the list in likelyLangList.`,
103 `
104likelyLangList holds lists info associated with likelyLang.`,
105 `
106likelyRegion is a lookup table, indexed by regionID, for the most likely
107languages and scripts given incomplete information. If more entries exist
108for a given regionID, lang and script are the index and size respectively
109of the list in likelyRegionList.
110TODO: exclude containers and user-definable regions from the list.`,
111 `
112likelyRegionList holds lists info associated with likelyRegion.`,
113 `
114likelyScript is a lookup table, indexed by scriptID, for the most likely
115languages and regions given a script.`,
116 `
117matchLang holds pairs of langIDs of base languages that are typically
118mutually intelligible. Each pair is associated with a confidence and
119whether the intelligibility goes one or both ways.`,
120 `
121matchScript holds pairs of scriptIDs where readers of one script
122can typically also read the other. Each is associated with a confidence.`,
123 `
124nRegionGroups is the number of region groups.`,
125 `
126regionInclusion maps region identifiers to sets of regions in regionInclusionBits,
127where each set holds all groupings that are directly connected in a region
128containment graph.`,
129 `
130regionInclusionBits is an array of bit vectors where every vector represents
131a set of region groupings. These sets are used to compute the distance
132between two regions for the purpose of language matching.`,
133 `
134regionInclusionNext marks, for each entry in regionInclusionBits, the set of
135all groups that are reachable from the groups set in the respective entry.`,
136}
137
138// TODO: consider changing some of these structures to tries. This can reduce
139// memory, but may increase the need for memory allocations. This could be
140// mitigated if we can piggyback on language tags for common cases.
141
142func failOnError(e error) {
143 if e != nil {
144 log.Panic(e)
145 }
146}
147
148type setType int
149
150const (
151 Indexed setType = 1 + iota // all elements must be of same size
152 Linear
153)
154
155type stringSet struct {
156 s []string
157 sorted, frozen bool
158
159 // We often need to update values after the creation of an index is completed.
160 // We include a convenience map for keeping track of this.
161 update map[string]string
162 typ setType // used for checking.
163}
164
165func (ss *stringSet) clone() stringSet {
166 c := *ss
167 c.s = append([]string(nil), c.s...)
168 return c
169}
170
171func (ss *stringSet) setType(t setType) {
172 if ss.typ != t && ss.typ != 0 {
173 log.Panicf("type %d cannot be assigned as it was already %d", t, ss.typ)
174 }
175}
176
177// parse parses a whitespace-separated string and initializes ss with its
178// components.
179func (ss *stringSet) parse(s string) {
180 scan := bufio.NewScanner(strings.NewReader(s))
181 scan.Split(bufio.ScanWords)
182 for scan.Scan() {
183 ss.add(scan.Text())
184 }
185}
186
187func (ss *stringSet) assertChangeable() {
188 if ss.frozen {
189 log.Panic("attempt to modify a frozen stringSet")
190 }
191}
192
193func (ss *stringSet) add(s string) {
194 ss.assertChangeable()
195 ss.s = append(ss.s, s)
196 ss.sorted = ss.frozen
197}
198
199func (ss *stringSet) freeze() {
200 ss.compact()
201 ss.frozen = true
202}
203
204func (ss *stringSet) compact() {
205 if ss.sorted {
206 return
207 }
208 a := ss.s
209 sort.Strings(a)
210 k := 0
211 for i := 1; i < len(a); i++ {
212 if a[k] != a[i] {
213 a[k+1] = a[i]
214 k++
215 }
216 }
217 ss.s = a[:k+1]
218 ss.sorted = ss.frozen
219}
220
221type funcSorter struct {
222 fn func(a, b string) bool
223 sort.StringSlice
224}
225
226func (s funcSorter) Less(i, j int) bool {
227 return s.fn(s.StringSlice[i], s.StringSlice[j])
228}
229
230func (ss *stringSet) sortFunc(f func(a, b string) bool) {
231 ss.compact()
232 sort.Sort(funcSorter{f, sort.StringSlice(ss.s)})
233}
234
235func (ss *stringSet) remove(s string) {
236 ss.assertChangeable()
237 if i, ok := ss.find(s); ok {
238 copy(ss.s[i:], ss.s[i+1:])
239 ss.s = ss.s[:len(ss.s)-1]
240 }
241}
242
243func (ss *stringSet) replace(ol, nu string) {
244 ss.s[ss.index(ol)] = nu
245 ss.sorted = ss.frozen
246}
247
248func (ss *stringSet) index(s string) int {
249 ss.setType(Indexed)
250 i, ok := ss.find(s)
251 if !ok {
252 if i < len(ss.s) {
253 log.Panicf("find: item %q is not in list. Closest match is %q.", s, ss.s[i])
254 }
255 log.Panicf("find: item %q is not in list", s)
256
257 }
258 return i
259}
260
261func (ss *stringSet) find(s string) (int, bool) {
262 ss.compact()
263 i := sort.SearchStrings(ss.s, s)
264 return i, i != len(ss.s) && ss.s[i] == s
265}
266
267func (ss *stringSet) slice() []string {
268 ss.compact()
269 return ss.s
270}
271
272func (ss *stringSet) updateLater(v, key string) {
273 if ss.update == nil {
274 ss.update = map[string]string{}
275 }
276 ss.update[v] = key
277}
278
279// join joins the string and ensures that all entries are of the same length.
280func (ss *stringSet) join() string {
281 ss.setType(Indexed)
282 n := len(ss.s[0])
283 for _, s := range ss.s {
284 if len(s) != n {
285 log.Panicf("join: not all entries are of the same length: %q", s)
286 }
287 }
288 ss.s = append(ss.s, strings.Repeat("\xff", n))
289 return strings.Join(ss.s, "")
290}
291
292// ianaEntry holds information for an entry in the IANA Language Subtag Repository.
293// All types use the same entry.
294// See http://tools.ietf.org/html/bcp47#section-5.1 for a description of the various
295// fields.
296type ianaEntry struct {
297 typ string
298 description []string
299 scope string
300 added string
301 preferred string
302 deprecated string
303 suppressScript string
304 macro string
305 prefix []string
306}
307
308type builder struct {
309 w *gen.CodeWriter
310 hw io.Writer // MultiWriter for w and w.Hash
311 data *cldr.CLDR
312 supp *cldr.SupplementalData
313
314 // indices
315 locale stringSet // common locales
316 lang stringSet // canonical language ids (2 or 3 letter ISO codes) with data
317 langNoIndex stringSet // 3-letter ISO codes with no associated data
318 script stringSet // 4-letter ISO codes
319 region stringSet // 2-letter ISO or 3-digit UN M49 codes
320 variant stringSet // 4-8-alphanumeric variant code.
321
322 // Region codes that are groups with their corresponding group IDs.
323 groups map[int]index
324
325 // langInfo
326 registry map[string]*ianaEntry
327}
328
329type index uint
330
331func newBuilder(w *gen.CodeWriter) *builder {
332 r := gen.OpenCLDRCoreZip()
333 defer r.Close()
334 d := &cldr.Decoder{}
335 data, err := d.DecodeZip(r)
336 failOnError(err)
337 b := builder{
338 w: w,
339 hw: io.MultiWriter(w, w.Hash),
340 data: data,
341 supp: data.Supplemental(),
342 }
343 b.parseRegistry()
344 return &b
345}
346
347func (b *builder) parseRegistry() {
348 r := gen.OpenIANAFile("assignments/language-subtag-registry")
349 defer r.Close()
350 b.registry = make(map[string]*ianaEntry)
351
352 scan := bufio.NewScanner(r)
353 scan.Split(bufio.ScanWords)
354 var record *ianaEntry
355 for more := scan.Scan(); more; {
356 key := scan.Text()
357 more = scan.Scan()
358 value := scan.Text()
359 switch key {
360 case "Type:":
361 record = &ianaEntry{typ: value}
362 case "Subtag:", "Tag:":
363 if s := strings.SplitN(value, "..", 2); len(s) > 1 {
364 for a := s[0]; a <= s[1]; a = inc(a) {
365 b.addToRegistry(a, record)
366 }
367 } else {
368 b.addToRegistry(value, record)
369 }
370 case "Suppress-Script:":
371 record.suppressScript = value
372 case "Added:":
373 record.added = value
374 case "Deprecated:":
375 record.deprecated = value
376 case "Macrolanguage:":
377 record.macro = value
378 case "Preferred-Value:":
379 record.preferred = value
380 case "Prefix:":
381 record.prefix = append(record.prefix, value)
382 case "Scope:":
383 record.scope = value
384 case "Description:":
385 buf := []byte(value)
386 for more = scan.Scan(); more; more = scan.Scan() {
387 b := scan.Bytes()
388 if b[0] == '%' || b[len(b)-1] == ':' {
389 break
390 }
391 buf = append(buf, ' ')
392 buf = append(buf, b...)
393 }
394 record.description = append(record.description, string(buf))
395 continue
396 default:
397 continue
398 }
399 more = scan.Scan()
400 }
401 if scan.Err() != nil {
402 log.Panic(scan.Err())
403 }
404}
405
406func (b *builder) addToRegistry(key string, entry *ianaEntry) {
407 if info, ok := b.registry[key]; ok {
408 if info.typ != "language" || entry.typ != "extlang" {
409 log.Fatalf("parseRegistry: tag %q already exists", key)
410 }
411 } else {
412 b.registry[key] = entry
413 }
414}
415
416var commentIndex = make(map[string]string)
417
418func init() {
419 for _, s := range comment {
420 key := strings.TrimSpace(strings.SplitN(s, " ", 2)[0])
421 commentIndex[key] = s
422 }
423}
424
425func (b *builder) comment(name string) {
426 if s := commentIndex[name]; len(s) > 0 {
427 b.w.WriteComment(s)
428 } else {
429 fmt.Fprintln(b.w)
430 }
431}
432
433func (b *builder) pf(f string, x ...interface{}) {
434 fmt.Fprintf(b.hw, f, x...)
435 fmt.Fprint(b.hw, "\n")
436}
437
438func (b *builder) p(x ...interface{}) {
439 fmt.Fprintln(b.hw, x...)
440}
441
442func (b *builder) addSize(s int) {
443 b.w.Size += s
444 b.pf("// Size: %d bytes", s)
445}
446
447func (b *builder) writeConst(name string, x interface{}) {
448 b.comment(name)
449 b.w.WriteConst(name, x)
450}
451
452// writeConsts computes f(v) for all v in values and writes the results
453// as constants named _v to a single constant block.
454func (b *builder) writeConsts(f func(string) int, values ...string) {
455 b.pf("const (")
456 for _, v := range values {
457 b.pf("\t_%s = %v", v, f(v))
458 }
459 b.pf(")")
460}
461
462// writeType writes the type of the given value, which must be a struct.
463func (b *builder) writeType(value interface{}) {
464 b.comment(reflect.TypeOf(value).Name())
465 b.w.WriteType(value)
466}
467
468func (b *builder) writeSlice(name string, ss interface{}) {
469 b.writeSliceAddSize(name, 0, ss)
470}
471
472func (b *builder) writeSliceAddSize(name string, extraSize int, ss interface{}) {
473 b.comment(name)
474 b.w.Size += extraSize
475 v := reflect.ValueOf(ss)
476 t := v.Type().Elem()
477 b.pf("// Size: %d bytes, %d elements", v.Len()*int(t.Size())+extraSize, v.Len())
478
479 fmt.Fprintf(b.w, "var %s = ", name)
480 b.w.WriteArray(ss)
481 b.p()
482}
483
484type fromTo struct {
485 from, to uint16
486}
487
488func (b *builder) writeSortedMap(name string, ss *stringSet, index func(s string) uint16) {
489 ss.sortFunc(func(a, b string) bool {
490 return index(a) < index(b)
491 })
492 m := []fromTo{}
493 for _, s := range ss.s {
494 m = append(m, fromTo{index(s), index(ss.update[s])})
495 }
496 b.writeSlice(name, m)
497}
498
499const base = 'z' - 'a' + 1
500
501func strToInt(s string) uint {
502 v := uint(0)
503 for i := 0; i < len(s); i++ {
504 v *= base
505 v += uint(s[i] - 'a')
506 }
507 return v
508}
509
510// converts the given integer to the original ASCII string passed to strToInt.
511// len(s) must match the number of characters obtained.
512func intToStr(v uint, s []byte) {
513 for i := len(s) - 1; i >= 0; i-- {
514 s[i] = byte(v%base) + 'a'
515 v /= base
516 }
517}
518
519func (b *builder) writeBitVector(name string, ss []string) {
520 vec := make([]uint8, int(math.Ceil(math.Pow(base, float64(len(ss[0])))/8)))
521 for _, s := range ss {
522 v := strToInt(s)
523 vec[v/8] |= 1 << (v % 8)
524 }
525 b.writeSlice(name, vec)
526}
527
528// TODO: convert this type into a list or two-stage trie.
529func (b *builder) writeMapFunc(name string, m map[string]string, f func(string) uint16) {
530 b.comment(name)
531 v := reflect.ValueOf(m)
532 sz := v.Len() * (2 + int(v.Type().Key().Size()))
533 for _, k := range m {
534 sz += len(k)
535 }
536 b.addSize(sz)
537 keys := []string{}
538 b.pf(`var %s = map[string]uint16{`, name)
539 for k := range m {
540 keys = append(keys, k)
541 }
542 sort.Strings(keys)
543 for _, k := range keys {
544 b.pf("\t%q: %v,", k, f(m[k]))
545 }
546 b.p("}")
547}
548
549func (b *builder) writeMap(name string, m interface{}) {
550 b.comment(name)
551 v := reflect.ValueOf(m)
552 sz := v.Len() * (2 + int(v.Type().Key().Size()) + int(v.Type().Elem().Size()))
553 b.addSize(sz)
554 f := strings.FieldsFunc(fmt.Sprintf("%#v", m), func(r rune) bool {
555 return strings.IndexRune("{}, ", r) != -1
556 })
557 sort.Strings(f[1:])
558 b.pf(`var %s = %s{`, name, f[0])
559 for _, kv := range f[1:] {
560 b.pf("\t%s,", kv)
561 }
562 b.p("}")
563}
564
565func (b *builder) langIndex(s string) uint16 {
566 if s == "und" {
567 return 0
568 }
569 if i, ok := b.lang.find(s); ok {
570 return uint16(i)
571 }
572 return uint16(strToInt(s)) + uint16(len(b.lang.s))
573}
574
575// inc advances the string to its lexicographical successor.
576func inc(s string) string {
577 const maxTagLength = 4
578 var buf [maxTagLength]byte
579 intToStr(strToInt(strings.ToLower(s))+1, buf[:len(s)])
580 for i := 0; i < len(s); i++ {
581 if s[i] <= 'Z' {
582 buf[i] -= 'a' - 'A'
583 }
584 }
585 return string(buf[:len(s)])
586}
587
588func (b *builder) parseIndices() {
589 meta := b.supp.Metadata
590
591 for k, v := range b.registry {
592 var ss *stringSet
593 switch v.typ {
594 case "language":
595 if len(k) == 2 || v.suppressScript != "" || v.scope == "special" {
596 b.lang.add(k)
597 continue
598 } else {
599 ss = &b.langNoIndex
600 }
601 case "region":
602 ss = &b.region
603 case "script":
604 ss = &b.script
605 case "variant":
606 ss = &b.variant
607 default:
608 continue
609 }
610 ss.add(k)
611 }
612 // Include any language for which there is data.
613 for _, lang := range b.data.Locales() {
614 if x := b.data.RawLDML(lang); false ||
615 x.LocaleDisplayNames != nil ||
616 x.Characters != nil ||
617 x.Delimiters != nil ||
618 x.Measurement != nil ||
619 x.Dates != nil ||
620 x.Numbers != nil ||
621 x.Units != nil ||
622 x.ListPatterns != nil ||
623 x.Collations != nil ||
624 x.Segmentations != nil ||
625 x.Rbnf != nil ||
626 x.Annotations != nil ||
627 x.Metadata != nil {
628
629 from := strings.Split(lang, "_")
630 if lang := from[0]; lang != "root" {
631 b.lang.add(lang)
632 }
633 }
634 }
635 // Include locales for plural rules, which uses a different structure.
636 for _, plurals := range b.data.Supplemental().Plurals {
637 for _, rules := range plurals.PluralRules {
638 for _, lang := range strings.Split(rules.Locales, " ") {
639 if lang = strings.Split(lang, "_")[0]; lang != "root" {
640 b.lang.add(lang)
641 }
642 }
643 }
644 }
645 // Include languages in likely subtags.
646 for _, m := range b.supp.LikelySubtags.LikelySubtag {
647 from := strings.Split(m.From, "_")
648 b.lang.add(from[0])
649 }
650 // Include ISO-639 alpha-3 bibliographic entries.
651 for _, a := range meta.Alias.LanguageAlias {
652 if a.Reason == "bibliographic" {
653 b.langNoIndex.add(a.Type)
654 }
655 }
656 // Include regions in territoryAlias (not all are in the IANA registry!)
657 for _, reg := range b.supp.Metadata.Alias.TerritoryAlias {
658 if len(reg.Type) == 2 {
659 b.region.add(reg.Type)
660 }
661 }
662
663 for _, s := range b.lang.s {
664 if len(s) == 3 {
665 b.langNoIndex.remove(s)
666 }
667 }
668 b.writeConst("numLanguages", len(b.lang.slice())+len(b.langNoIndex.slice()))
669 b.writeConst("numScripts", len(b.script.slice()))
670 b.writeConst("numRegions", len(b.region.slice()))
671
672 // Add dummy codes at the start of each list to represent "unspecified".
673 b.lang.add("---")
674 b.script.add("----")
675 b.region.add("---")
676
677 // common locales
678 b.locale.parse(meta.DefaultContent.Locales)
679}
680
681// TODO: region inclusion data will probably not be use used in future matchers.
682
683func (b *builder) computeRegionGroups() {
684 b.groups = make(map[int]index)
685
686 // Create group indices.
687 for i := 1; b.region.s[i][0] < 'A'; i++ { // Base M49 indices on regionID.
688 b.groups[i] = index(len(b.groups))
689 }
690 for _, g := range b.supp.TerritoryContainment.Group {
691 // Skip UN and EURO zone as they are flattening the containment
692 // relationship.
693 if g.Type == "EZ" || g.Type == "UN" {
694 continue
695 }
696 group := b.region.index(g.Type)
697 if _, ok := b.groups[group]; !ok {
698 b.groups[group] = index(len(b.groups))
699 }
700 }
701 if len(b.groups) > 64 {
702 log.Fatalf("only 64 groups supported, found %d", len(b.groups))
703 }
704 b.writeConst("nRegionGroups", len(b.groups))
705}
706
707var langConsts = []string{
708 "af", "am", "ar", "az", "bg", "bn", "ca", "cs", "da", "de", "el", "en", "es",
709 "et", "fa", "fi", "fil", "fr", "gu", "he", "hi", "hr", "hu", "hy", "id", "is",
710 "it", "ja", "ka", "kk", "km", "kn", "ko", "ky", "lo", "lt", "lv", "mk", "ml",
711 "mn", "mo", "mr", "ms", "mul", "my", "nb", "ne", "nl", "no", "pa", "pl", "pt",
712 "ro", "ru", "sh", "si", "sk", "sl", "sq", "sr", "sv", "sw", "ta", "te", "th",
713 "tl", "tn", "tr", "uk", "ur", "uz", "vi", "zh", "zu",
714
715 // constants for grandfathered tags (if not already defined)
716 "jbo", "ami", "bnn", "hak", "tlh", "lb", "nv", "pwn", "tao", "tay", "tsu",
717 "nn", "sfb", "vgt", "sgg", "cmn", "nan", "hsn",
718}
719
720// writeLanguage generates all tables needed for language canonicalization.
721func (b *builder) writeLanguage() {
722 meta := b.supp.Metadata
723
724 b.writeConst("nonCanonicalUnd", b.lang.index("und"))
725 b.writeConsts(func(s string) int { return int(b.langIndex(s)) }, langConsts...)
726 b.writeConst("langPrivateStart", b.langIndex("qaa"))
727 b.writeConst("langPrivateEnd", b.langIndex("qtz"))
728
729 // Get language codes that need to be mapped (overlong 3-letter codes,
730 // deprecated 2-letter codes, legacy and grandfathered tags.)
731 langAliasMap := stringSet{}
732 aliasTypeMap := map[string]langAliasType{}
733
734 // altLangISO3 get the alternative ISO3 names that need to be mapped.
735 altLangISO3 := stringSet{}
736 // Add dummy start to avoid the use of index 0.
737 altLangISO3.add("---")
738 altLangISO3.updateLater("---", "aa")
739
740 lang := b.lang.clone()
741 for _, a := range meta.Alias.LanguageAlias {
742 if a.Replacement == "" {
743 a.Replacement = "und"
744 }
745 // TODO: support mapping to tags
746 repl := strings.SplitN(a.Replacement, "_", 2)[0]
747 if a.Reason == "overlong" {
748 if len(a.Replacement) == 2 && len(a.Type) == 3 {
749 lang.updateLater(a.Replacement, a.Type)
750 }
751 } else if len(a.Type) <= 3 {
752 switch a.Reason {
753 case "macrolanguage":
754 aliasTypeMap[a.Type] = langMacro
755 case "deprecated":
756 // handled elsewhere
757 continue
758 case "bibliographic", "legacy":
759 if a.Type == "no" {
760 continue
761 }
762 aliasTypeMap[a.Type] = langLegacy
763 default:
764 log.Fatalf("new %s alias: %s", a.Reason, a.Type)
765 }
766 langAliasMap.add(a.Type)
767 langAliasMap.updateLater(a.Type, repl)
768 }
769 }
770 // Manually add the mapping of "nb" (Norwegian) to its macro language.
771 // This can be removed if CLDR adopts this change.
772 langAliasMap.add("nb")
773 langAliasMap.updateLater("nb", "no")
774 aliasTypeMap["nb"] = langMacro
775
776 for k, v := range b.registry {
777 // Also add deprecated values for 3-letter ISO codes, which CLDR omits.
778 if v.typ == "language" && v.deprecated != "" && v.preferred != "" {
779 langAliasMap.add(k)
780 langAliasMap.updateLater(k, v.preferred)
781 aliasTypeMap[k] = langDeprecated
782 }
783 }
784 // Fix CLDR mappings.
785 lang.updateLater("tl", "tgl")
786 lang.updateLater("sh", "hbs")
787 lang.updateLater("mo", "mol")
788 lang.updateLater("no", "nor")
789 lang.updateLater("tw", "twi")
790 lang.updateLater("nb", "nob")
791 lang.updateLater("ak", "aka")
792 lang.updateLater("bh", "bih")
793
794 // Ensure that each 2-letter code is matched with a 3-letter code.
795 for _, v := range lang.s[1:] {
796 s, ok := lang.update[v]
797 if !ok {
798 if s, ok = lang.update[langAliasMap.update[v]]; !ok {
799 continue
800 }
801 lang.update[v] = s
802 }
803 if v[0] != s[0] {
804 altLangISO3.add(s)
805 altLangISO3.updateLater(s, v)
806 }
807 }
808
809 // Complete canonicalized language tags.
810 lang.freeze()
811 for i, v := range lang.s {
812 // We can avoid these manual entries by using the IANA registry directly.
813 // Seems easier to update the list manually, as changes are rare.
814 // The panic in this loop will trigger if we miss an entry.
815 add := ""
816 if s, ok := lang.update[v]; ok {
817 if s[0] == v[0] {
818 add = s[1:]
819 } else {
820 add = string([]byte{0, byte(altLangISO3.index(s))})
821 }
822 } else if len(v) == 3 {
823 add = "\x00"
824 } else {
825 log.Panicf("no data for long form of %q", v)
826 }
827 lang.s[i] += add
828 }
829 b.writeConst("lang", tag.Index(lang.join()))
830
831 b.writeConst("langNoIndexOffset", len(b.lang.s))
832
833 // space of all valid 3-letter language identifiers.
834 b.writeBitVector("langNoIndex", b.langNoIndex.slice())
835
836 altLangIndex := []uint16{}
837 for i, s := range altLangISO3.slice() {
838 altLangISO3.s[i] += string([]byte{byte(len(altLangIndex))})
839 if i > 0 {
840 idx := b.lang.index(altLangISO3.update[s])
841 altLangIndex = append(altLangIndex, uint16(idx))
842 }
843 }
844 b.writeConst("altLangISO3", tag.Index(altLangISO3.join()))
845 b.writeSlice("altLangIndex", altLangIndex)
846
847 b.writeSortedMap("langAliasMap", &langAliasMap, b.langIndex)
848 types := make([]langAliasType, len(langAliasMap.s))
849 for i, s := range langAliasMap.s {
850 types[i] = aliasTypeMap[s]
851 }
852 b.writeSlice("langAliasTypes", types)
853}
854
855var scriptConsts = []string{
856 "Latn", "Hani", "Hans", "Hant", "Qaaa", "Qaai", "Qabx", "Zinh", "Zyyy",
857 "Zzzz",
858}
859
860func (b *builder) writeScript() {
861 b.writeConsts(b.script.index, scriptConsts...)
862 b.writeConst("script", tag.Index(b.script.join()))
863
864 supp := make([]uint8, len(b.lang.slice()))
865 for i, v := range b.lang.slice()[1:] {
866 if sc := b.registry[v].suppressScript; sc != "" {
867 supp[i+1] = uint8(b.script.index(sc))
868 }
869 }
870 b.writeSlice("suppressScript", supp)
871
872 // There is only one deprecated script in CLDR. This value is hard-coded.
873 // We check here if the code must be updated.
874 for _, a := range b.supp.Metadata.Alias.ScriptAlias {
875 if a.Type != "Qaai" {
876 log.Panicf("unexpected deprecated stript %q", a.Type)
877 }
878 }
879}
880
881func parseM49(s string) int16 {
882 if len(s) == 0 {
883 return 0
884 }
885 v, err := strconv.ParseUint(s, 10, 10)
886 failOnError(err)
887 return int16(v)
888}
889
890var regionConsts = []string{
891 "001", "419", "BR", "CA", "ES", "GB", "MD", "PT", "UK", "US",
892 "ZZ", "XA", "XC", "XK", // Unofficial tag for Kosovo.
893}
894
895func (b *builder) writeRegion() {
896 b.writeConsts(b.region.index, regionConsts...)
897
898 isoOffset := b.region.index("AA")
899 m49map := make([]int16, len(b.region.slice()))
900 fromM49map := make(map[int16]int)
901 altRegionISO3 := ""
902 altRegionIDs := []uint16{}
903
904 b.writeConst("isoRegionOffset", isoOffset)
905
906 // 2-letter region lookup and mapping to numeric codes.
907 regionISO := b.region.clone()
908 regionISO.s = regionISO.s[isoOffset:]
909 regionISO.sorted = false
910
911 regionTypes := make([]byte, len(b.region.s))
912
913 // Is the region valid BCP 47?
914 for s, e := range b.registry {
915 if len(s) == 2 && s == strings.ToUpper(s) {
916 i := b.region.index(s)
917 for _, d := range e.description {
918 if strings.Contains(d, "Private use") {
919 regionTypes[i] = iso3166UserAssigned
920 }
921 }
922 regionTypes[i] |= bcp47Region
923 }
924 }
925
926 // Is the region a valid ccTLD?
927 r := gen.OpenIANAFile("domains/root/db")
928 defer r.Close()
929
930 buf, err := ioutil.ReadAll(r)
931 failOnError(err)
932 re := regexp.MustCompile(`"/domains/root/db/([a-z]{2}).html"`)
933 for _, m := range re.FindAllSubmatch(buf, -1) {
934 i := b.region.index(strings.ToUpper(string(m[1])))
935 regionTypes[i] |= ccTLD
936 }
937
938 b.writeSlice("regionTypes", regionTypes)
939
940 iso3Set := make(map[string]int)
941 update := func(iso2, iso3 string) {
942 i := regionISO.index(iso2)
943 if j, ok := iso3Set[iso3]; !ok && iso3[0] == iso2[0] {
944 regionISO.s[i] += iso3[1:]
945 iso3Set[iso3] = -1
946 } else {
947 if ok && j >= 0 {
948 regionISO.s[i] += string([]byte{0, byte(j)})
949 } else {
950 iso3Set[iso3] = len(altRegionISO3)
951 regionISO.s[i] += string([]byte{0, byte(len(altRegionISO3))})
952 altRegionISO3 += iso3
953 altRegionIDs = append(altRegionIDs, uint16(isoOffset+i))
954 }
955 }
956 }
957 for _, tc := range b.supp.CodeMappings.TerritoryCodes {
958 i := regionISO.index(tc.Type) + isoOffset
959 if d := m49map[i]; d != 0 {
960 log.Panicf("%s found as a duplicate UN.M49 code of %03d", tc.Numeric, d)
961 }
962 m49 := parseM49(tc.Numeric)
963 m49map[i] = m49
964 if r := fromM49map[m49]; r == 0 {
965 fromM49map[m49] = i
966 } else if r != i {
967 dep := b.registry[regionISO.s[r-isoOffset]].deprecated
968 if t := b.registry[tc.Type]; t != nil && dep != "" && (t.deprecated == "" || t.deprecated > dep) {
969 fromM49map[m49] = i
970 }
971 }
972 }
973 for _, ta := range b.supp.Metadata.Alias.TerritoryAlias {
974 if len(ta.Type) == 3 && ta.Type[0] <= '9' && len(ta.Replacement) == 2 {
975 from := parseM49(ta.Type)
976 if r := fromM49map[from]; r == 0 {
977 fromM49map[from] = regionISO.index(ta.Replacement) + isoOffset
978 }
979 }
980 }
981 for _, tc := range b.supp.CodeMappings.TerritoryCodes {
982 if len(tc.Alpha3) == 3 {
983 update(tc.Type, tc.Alpha3)
984 }
985 }
986 // This entries are not included in territoryCodes. Mostly 3-letter variants
987 // of deleted codes and an entry for QU.
988 for _, m := range []struct{ iso2, iso3 string }{
989 {"CT", "CTE"},
990 {"DY", "DHY"},
991 {"HV", "HVO"},
992 {"JT", "JTN"},
993 {"MI", "MID"},
994 {"NH", "NHB"},
995 {"NQ", "ATN"},
996 {"PC", "PCI"},
997 {"PU", "PUS"},
998 {"PZ", "PCZ"},
999 {"RH", "RHO"},
1000 {"VD", "VDR"},
1001 {"WK", "WAK"},
1002 // These three-letter codes are used for others as well.
1003 {"FQ", "ATF"},
1004 } {
1005 update(m.iso2, m.iso3)
1006 }
1007 for i, s := range regionISO.s {
1008 if len(s) != 4 {
1009 regionISO.s[i] = s + " "
1010 }
1011 }
1012 b.writeConst("regionISO", tag.Index(regionISO.join()))
1013 b.writeConst("altRegionISO3", altRegionISO3)
1014 b.writeSlice("altRegionIDs", altRegionIDs)
1015
1016 // Create list of deprecated regions.
1017 // TODO: consider inserting SF -> FI. Not included by CLDR, but is the only
1018 // Transitionally-reserved mapping not included.
1019 regionOldMap := stringSet{}
1020 // Include regions in territoryAlias (not all are in the IANA registry!)
1021 for _, reg := range b.supp.Metadata.Alias.TerritoryAlias {
1022 if len(reg.Type) == 2 && reg.Reason == "deprecated" && len(reg.Replacement) == 2 {
1023 regionOldMap.add(reg.Type)
1024 regionOldMap.updateLater(reg.Type, reg.Replacement)
1025 i, _ := regionISO.find(reg.Type)
1026 j, _ := regionISO.find(reg.Replacement)
1027 if k := m49map[i+isoOffset]; k == 0 {
1028 m49map[i+isoOffset] = m49map[j+isoOffset]
1029 }
1030 }
1031 }
1032 b.writeSortedMap("regionOldMap", &regionOldMap, func(s string) uint16 {
1033 return uint16(b.region.index(s))
1034 })
1035 // 3-digit region lookup, groupings.
1036 for i := 1; i < isoOffset; i++ {
1037 m := parseM49(b.region.s[i])
1038 m49map[i] = m
1039 fromM49map[m] = i
1040 }
1041 b.writeSlice("m49", m49map)
1042
1043 const (
1044 searchBits = 7
1045 regionBits = 9
1046 )
1047 if len(m49map) >= 1<<regionBits {
1048 log.Fatalf("Maximum number of regions exceeded: %d > %d", len(m49map), 1<<regionBits)
1049 }
1050 m49Index := [9]int16{}
1051 fromM49 := []uint16{}
1052 m49 := []int{}
1053 for k, _ := range fromM49map {
1054 m49 = append(m49, int(k))
1055 }
1056 sort.Ints(m49)
1057 for _, k := range m49[1:] {
1058 val := (k & (1<<searchBits - 1)) << regionBits
1059 fromM49 = append(fromM49, uint16(val|fromM49map[int16(k)]))
1060 m49Index[1:][k>>searchBits] = int16(len(fromM49))
1061 }
1062 b.writeSlice("m49Index", m49Index)
1063 b.writeSlice("fromM49", fromM49)
1064}
1065
1066const (
1067 // TODO: put these lists in regionTypes as user data? Could be used for
1068 // various optimizations and refinements and could be exposed in the API.
1069 iso3166Except = "AC CP DG EA EU FX IC SU TA UK"
1070 iso3166Trans = "AN BU CS NT TP YU ZR" // SF is not in our set of Regions.
1071 // DY and RH are actually not deleted, but indeterminately reserved.
1072 iso3166DelCLDR = "CT DD DY FQ HV JT MI NH NQ PC PU PZ RH VD WK YD"
1073)
1074
1075const (
1076 iso3166UserAssigned = 1 << iota
1077 ccTLD
1078 bcp47Region
1079)
1080
1081func find(list []string, s string) int {
1082 for i, t := range list {
1083 if t == s {
1084 return i
1085 }
1086 }
1087 return -1
1088}
1089
1090// writeVariants generates per-variant information and creates a map from variant
1091// name to index value. We assign index values such that sorting multiple
1092// variants by index value will result in the correct order.
1093// There are two types of variants: specialized and general. Specialized variants
1094// are only applicable to certain language or language-script pairs. Generalized
1095// variants apply to any language. Generalized variants always sort after
1096// specialized variants. We will therefore always assign a higher index value
1097// to a generalized variant than any other variant. Generalized variants are
1098// sorted alphabetically among themselves.
1099// Specialized variants may also sort after other specialized variants. Such
1100// variants will be ordered after any of the variants they may follow.
1101// We assume that if a variant x is followed by a variant y, then for any prefix
1102// p of x, p-x is a prefix of y. This allows us to order tags based on the
1103// maximum of the length of any of its prefixes.
1104// TODO: it is possible to define a set of Prefix values on variants such that
1105// a total order cannot be defined to the point that this algorithm breaks.
1106// In other words, we cannot guarantee the same order of variants for the
1107// future using the same algorithm or for non-compliant combinations of
1108// variants. For this reason, consider using simple alphabetic sorting
1109// of variants and ignore Prefix restrictions altogether.
1110func (b *builder) writeVariant() {
1111 generalized := stringSet{}
1112 specialized := stringSet{}
1113 specializedExtend := stringSet{}
1114 // Collate the variants by type and check assumptions.
1115 for _, v := range b.variant.slice() {
1116 e := b.registry[v]
1117 if len(e.prefix) == 0 {
1118 generalized.add(v)
1119 continue
1120 }
1121 c := strings.Split(e.prefix[0], "-")
1122 hasScriptOrRegion := false
1123 if len(c) > 1 {
1124 _, hasScriptOrRegion = b.script.find(c[1])
1125 if !hasScriptOrRegion {
1126 _, hasScriptOrRegion = b.region.find(c[1])
1127
1128 }
1129 }
1130 if len(c) == 1 || len(c) == 2 && hasScriptOrRegion {
1131 // Variant is preceded by a language.
1132 specialized.add(v)
1133 continue
1134 }
1135 // Variant is preceded by another variant.
1136 specializedExtend.add(v)
1137 prefix := c[0] + "-"
1138 if hasScriptOrRegion {
1139 prefix += c[1]
1140 }
1141 for _, p := range e.prefix {
1142 // Verify that the prefix minus the last element is a prefix of the
1143 // predecessor element.
1144 i := strings.LastIndex(p, "-")
1145 pred := b.registry[p[i+1:]]
1146 if find(pred.prefix, p[:i]) < 0 {
1147 log.Fatalf("prefix %q for variant %q not consistent with predecessor spec", p, v)
1148 }
1149 // The sorting used below does not work in the general case. It works
1150 // if we assume that variants that may be followed by others only have
1151 // prefixes of the same length. Verify this.
1152 count := strings.Count(p[:i], "-")
1153 for _, q := range pred.prefix {
1154 if c := strings.Count(q, "-"); c != count {
1155 log.Fatalf("variant %q preceding %q has a prefix %q of size %d; want %d", p[i+1:], v, q, c, count)
1156 }
1157 }
1158 if !strings.HasPrefix(p, prefix) {
1159 log.Fatalf("prefix %q of variant %q should start with %q", p, v, prefix)
1160 }
1161 }
1162 }
1163
1164 // Sort extended variants.
1165 a := specializedExtend.s
1166 less := func(v, w string) bool {
1167 // Sort by the maximum number of elements.
1168 maxCount := func(s string) (max int) {
1169 for _, p := range b.registry[s].prefix {
1170 if c := strings.Count(p, "-"); c > max {
1171 max = c
1172 }
1173 }
1174 return
1175 }
1176 if cv, cw := maxCount(v), maxCount(w); cv != cw {
1177 return cv < cw
1178 }
1179 // Sort by name as tie breaker.
1180 return v < w
1181 }
1182 sort.Sort(funcSorter{less, sort.StringSlice(a)})
1183 specializedExtend.frozen = true
1184
1185 // Create index from variant name to index.
1186 variantIndex := make(map[string]uint8)
1187 add := func(s []string) {
1188 for _, v := range s {
1189 variantIndex[v] = uint8(len(variantIndex))
1190 }
1191 }
1192 add(specialized.slice())
1193 add(specializedExtend.s)
1194 numSpecialized := len(variantIndex)
1195 add(generalized.slice())
1196 if n := len(variantIndex); n > 255 {
1197 log.Fatalf("maximum number of variants exceeded: was %d; want <= 255", n)
1198 }
1199 b.writeMap("variantIndex", variantIndex)
1200 b.writeConst("variantNumSpecialized", numSpecialized)
1201}
1202
1203func (b *builder) writeLanguageInfo() {
1204}
1205
1206// writeLikelyData writes tables that are used both for finding parent relations and for
1207// language matching. Each entry contains additional bits to indicate the status of the
1208// data to know when it cannot be used for parent relations.
1209func (b *builder) writeLikelyData() {
1210 const (
1211 isList = 1 << iota
1212 scriptInFrom
1213 regionInFrom
1214 )
1215 type ( // generated types
1216 likelyScriptRegion struct {
1217 region uint16
1218 script uint8
1219 flags uint8
1220 }
1221 likelyLangScript struct {
1222 lang uint16
1223 script uint8
1224 flags uint8
1225 }
1226 likelyLangRegion struct {
1227 lang uint16
1228 region uint16
1229 }
1230 // likelyTag is used for getting likely tags for group regions, where
1231 // the likely region might be a region contained in the group.
1232 likelyTag struct {
1233 lang uint16
1234 region uint16
1235 script uint8
1236 }
1237 )
1238 var ( // generated variables
1239 likelyRegionGroup = make([]likelyTag, len(b.groups))
1240 likelyLang = make([]likelyScriptRegion, len(b.lang.s))
1241 likelyRegion = make([]likelyLangScript, len(b.region.s))
1242 likelyScript = make([]likelyLangRegion, len(b.script.s))
1243 likelyLangList = []likelyScriptRegion{}
1244 likelyRegionList = []likelyLangScript{}
1245 )
1246 type fromTo struct {
1247 from, to []string
1248 }
1249 langToOther := map[int][]fromTo{}
1250 regionToOther := map[int][]fromTo{}
1251 for _, m := range b.supp.LikelySubtags.LikelySubtag {
1252 from := strings.Split(m.From, "_")
1253 to := strings.Split(m.To, "_")
1254 if len(to) != 3 {
1255 log.Fatalf("invalid number of subtags in %q: found %d, want 3", m.To, len(to))
1256 }
1257 if len(from) > 3 {
1258 log.Fatalf("invalid number of subtags: found %d, want 1-3", len(from))
1259 }
1260 if from[0] != to[0] && from[0] != "und" {
1261 log.Fatalf("unexpected language change in expansion: %s -> %s", from, to)
1262 }
1263 if len(from) == 3 {
1264 if from[2] != to[2] {
1265 log.Fatalf("unexpected region change in expansion: %s -> %s", from, to)
1266 }
1267 if from[0] != "und" {
1268 log.Fatalf("unexpected fully specified from tag: %s -> %s", from, to)
1269 }
1270 }
1271 if len(from) == 1 || from[0] != "und" {
1272 id := 0
1273 if from[0] != "und" {
1274 id = b.lang.index(from[0])
1275 }
1276 langToOther[id] = append(langToOther[id], fromTo{from, to})
1277 } else if len(from) == 2 && len(from[1]) == 4 {
1278 sid := b.script.index(from[1])
1279 likelyScript[sid].lang = uint16(b.langIndex(to[0]))
1280 likelyScript[sid].region = uint16(b.region.index(to[2]))
1281 } else {
1282 r := b.region.index(from[len(from)-1])
1283 if id, ok := b.groups[r]; ok {
1284 if from[0] != "und" {
1285 log.Fatalf("region changed unexpectedly: %s -> %s", from, to)
1286 }
1287 likelyRegionGroup[id].lang = uint16(b.langIndex(to[0]))
1288 likelyRegionGroup[id].script = uint8(b.script.index(to[1]))
1289 likelyRegionGroup[id].region = uint16(b.region.index(to[2]))
1290 } else {
1291 regionToOther[r] = append(regionToOther[r], fromTo{from, to})
1292 }
1293 }
1294 }
1295 b.writeType(likelyLangRegion{})
1296 b.writeSlice("likelyScript", likelyScript)
1297
1298 for id := range b.lang.s {
1299 list := langToOther[id]
1300 if len(list) == 1 {
1301 likelyLang[id].region = uint16(b.region.index(list[0].to[2]))
1302 likelyLang[id].script = uint8(b.script.index(list[0].to[1]))
1303 } else if len(list) > 1 {
1304 likelyLang[id].flags = isList
1305 likelyLang[id].region = uint16(len(likelyLangList))
1306 likelyLang[id].script = uint8(len(list))
1307 for _, x := range list {
1308 flags := uint8(0)
1309 if len(x.from) > 1 {
1310 if x.from[1] == x.to[2] {
1311 flags = regionInFrom
1312 } else {
1313 flags = scriptInFrom
1314 }
1315 }
1316 likelyLangList = append(likelyLangList, likelyScriptRegion{
1317 region: uint16(b.region.index(x.to[2])),
1318 script: uint8(b.script.index(x.to[1])),
1319 flags: flags,
1320 })
1321 }
1322 }
1323 }
1324 // TODO: merge suppressScript data with this table.
1325 b.writeType(likelyScriptRegion{})
1326 b.writeSlice("likelyLang", likelyLang)
1327 b.writeSlice("likelyLangList", likelyLangList)
1328
1329 for id := range b.region.s {
1330 list := regionToOther[id]
1331 if len(list) == 1 {
1332 likelyRegion[id].lang = uint16(b.langIndex(list[0].to[0]))
1333 likelyRegion[id].script = uint8(b.script.index(list[0].to[1]))
1334 if len(list[0].from) > 2 {
1335 likelyRegion[id].flags = scriptInFrom
1336 }
1337 } else if len(list) > 1 {
1338 likelyRegion[id].flags = isList
1339 likelyRegion[id].lang = uint16(len(likelyRegionList))
1340 likelyRegion[id].script = uint8(len(list))
1341 for i, x := range list {
1342 if len(x.from) == 2 && i != 0 || i > 0 && len(x.from) != 3 {
1343 log.Fatalf("unspecified script must be first in list: %v at %d", x.from, i)
1344 }
1345 x := likelyLangScript{
1346 lang: uint16(b.langIndex(x.to[0])),
1347 script: uint8(b.script.index(x.to[1])),
1348 }
1349 if len(list[0].from) > 2 {
1350 x.flags = scriptInFrom
1351 }
1352 likelyRegionList = append(likelyRegionList, x)
1353 }
1354 }
1355 }
1356 b.writeType(likelyLangScript{})
1357 b.writeSlice("likelyRegion", likelyRegion)
1358 b.writeSlice("likelyRegionList", likelyRegionList)
1359
1360 b.writeType(likelyTag{})
1361 b.writeSlice("likelyRegionGroup", likelyRegionGroup)
1362}
1363
1364type mutualIntelligibility struct {
1365 want, have uint16
1366 distance uint8
1367 oneway bool
1368}
1369
1370type scriptIntelligibility struct {
1371 wantLang, haveLang uint16
1372 wantScript, haveScript uint8
1373 distance uint8
1374 // Always oneway
1375}
1376
1377type regionIntelligibility struct {
1378 lang uint16 // compact language id
1379 script uint8 // 0 means any
1380 group uint8 // 0 means any; if bit 7 is set it means inverse
1381 distance uint8
1382 // Always twoway.
1383}
1384
1385// writeMatchData writes tables with languages and scripts for which there is
1386// mutual intelligibility. The data is based on CLDR's languageMatching data.
1387// Note that we use a different algorithm than the one defined by CLDR and that
1388// we slightly modify the data. For example, we convert scores to confidence levels.
1389// We also drop all region-related data as we use a different algorithm to
1390// determine region equivalence.
1391func (b *builder) writeMatchData() {
1392 lm := b.supp.LanguageMatching.LanguageMatches
1393 cldr.MakeSlice(&lm).SelectAnyOf("type", "written_new")
1394
1395 regionHierarchy := map[string][]string{}
1396 for _, g := range b.supp.TerritoryContainment.Group {
1397 regions := strings.Split(g.Contains, " ")
1398 regionHierarchy[g.Type] = append(regionHierarchy[g.Type], regions...)
1399 }
1400 regionToGroups := make([]uint8, len(b.region.s))
1401
1402 idToIndex := map[string]uint8{}
1403 for i, mv := range lm[0].MatchVariable {
1404 if i > 6 {
1405 log.Fatalf("Too many groups: %d", i)
1406 }
1407 idToIndex[mv.Id] = uint8(i + 1)
1408 // TODO: also handle '-'
1409 for _, r := range strings.Split(mv.Value, "+") {
1410 todo := []string{r}
1411 for k := 0; k < len(todo); k++ {
1412 r := todo[k]
1413 regionToGroups[b.region.index(r)] |= 1 << uint8(i)
1414 todo = append(todo, regionHierarchy[r]...)
1415 }
1416 }
1417 }
1418 b.writeSlice("regionToGroups", regionToGroups)
1419
1420 // maps language id to in- and out-of-group region.
1421 paradigmLocales := [][3]uint16{}
1422 locales := strings.Split(lm[0].ParadigmLocales[0].Locales, " ")
1423 for i := 0; i < len(locales); i += 2 {
1424 x := [3]uint16{}
1425 for j := 0; j < 2; j++ {
1426 pc := strings.SplitN(locales[i+j], "-", 2)
1427 x[0] = b.langIndex(pc[0])
1428 if len(pc) == 2 {
1429 x[1+j] = uint16(b.region.index(pc[1]))
1430 }
1431 }
1432 paradigmLocales = append(paradigmLocales, x)
1433 }
1434 b.writeSlice("paradigmLocales", paradigmLocales)
1435
1436 b.writeType(mutualIntelligibility{})
1437 b.writeType(scriptIntelligibility{})
1438 b.writeType(regionIntelligibility{})
1439
1440 matchLang := []mutualIntelligibility{}
1441 matchScript := []scriptIntelligibility{}
1442 matchRegion := []regionIntelligibility{}
1443 // Convert the languageMatch entries in lists keyed by desired language.
1444 for _, m := range lm[0].LanguageMatch {
1445 // Different versions of CLDR use different separators.
1446 desired := strings.Replace(m.Desired, "-", "_", -1)
1447 supported := strings.Replace(m.Supported, "-", "_", -1)
1448 d := strings.Split(desired, "_")
1449 s := strings.Split(supported, "_")
1450 if len(d) != len(s) {
1451 log.Fatalf("not supported: desired=%q; supported=%q", desired, supported)
1452 continue
1453 }
1454 distance, _ := strconv.ParseInt(m.Distance, 10, 8)
1455 switch len(d) {
1456 case 2:
1457 if desired == supported && desired == "*_*" {
1458 continue
1459 }
1460 // language-script pair.
1461 matchScript = append(matchScript, scriptIntelligibility{
1462 wantLang: uint16(b.langIndex(d[0])),
1463 haveLang: uint16(b.langIndex(s[0])),
1464 wantScript: uint8(b.script.index(d[1])),
1465 haveScript: uint8(b.script.index(s[1])),
1466 distance: uint8(distance),
1467 })
1468 if m.Oneway != "true" {
1469 matchScript = append(matchScript, scriptIntelligibility{
1470 wantLang: uint16(b.langIndex(s[0])),
1471 haveLang: uint16(b.langIndex(d[0])),
1472 wantScript: uint8(b.script.index(s[1])),
1473 haveScript: uint8(b.script.index(d[1])),
1474 distance: uint8(distance),
1475 })
1476 }
1477 case 1:
1478 if desired == supported && desired == "*" {
1479 continue
1480 }
1481 if distance == 1 {
1482 // nb == no is already handled by macro mapping. Check there
1483 // really is only this case.
1484 if d[0] != "no" || s[0] != "nb" {
1485 log.Fatalf("unhandled equivalence %s == %s", s[0], d[0])
1486 }
1487 continue
1488 }
1489 // TODO: consider dropping oneway field and just doubling the entry.
1490 matchLang = append(matchLang, mutualIntelligibility{
1491 want: uint16(b.langIndex(d[0])),
1492 have: uint16(b.langIndex(s[0])),
1493 distance: uint8(distance),
1494 oneway: m.Oneway == "true",
1495 })
1496 case 3:
1497 if desired == supported && desired == "*_*_*" {
1498 continue
1499 }
1500 if desired != supported {
1501 // This is now supported by CLDR, but only one case, which
1502 // should already be covered by paradigm locales. For instance,
1503 // test case "und, en, en-GU, en-IN, en-GB ; en-ZA ; en-GB" in
1504 // testdata/CLDRLocaleMatcherTest.txt tests this.
1505 if supported != "en_*_GB" {
1506 log.Fatalf("not supported: desired=%q; supported=%q", desired, supported)
1507 }
1508 continue
1509 }
1510 ri := regionIntelligibility{
1511 lang: b.langIndex(d[0]),
1512 distance: uint8(distance),
1513 }
1514 if d[1] != "*" {
1515 ri.script = uint8(b.script.index(d[1]))
1516 }
1517 switch {
1518 case d[2] == "*":
1519 ri.group = 0x80 // not contained in anything
1520 case strings.HasPrefix(d[2], "$!"):
1521 ri.group = 0x80
1522 d[2] = "$" + d[2][len("$!"):]
1523 fallthrough
1524 case strings.HasPrefix(d[2], "$"):
1525 ri.group |= idToIndex[d[2]]
1526 }
1527 matchRegion = append(matchRegion, ri)
1528 default:
1529 log.Fatalf("not supported: desired=%q; supported=%q", desired, supported)
1530 }
1531 }
1532 sort.SliceStable(matchLang, func(i, j int) bool {
1533 return matchLang[i].distance < matchLang[j].distance
1534 })
1535 b.writeSlice("matchLang", matchLang)
1536
1537 sort.SliceStable(matchScript, func(i, j int) bool {
1538 return matchScript[i].distance < matchScript[j].distance
1539 })
1540 b.writeSlice("matchScript", matchScript)
1541
1542 sort.SliceStable(matchRegion, func(i, j int) bool {
1543 return matchRegion[i].distance < matchRegion[j].distance
1544 })
1545 b.writeSlice("matchRegion", matchRegion)
1546}
1547
1548func (b *builder) writeRegionInclusionData() {
1549 var (
1550 // mm holds for each group the set of groups with a distance of 1.
1551 mm = make(map[int][]index)
1552
1553 // containment holds for each group the transitive closure of
1554 // containment of other groups.
1555 containment = make(map[index][]index)
1556 )
1557 for _, g := range b.supp.TerritoryContainment.Group {
1558 // Skip UN and EURO zone as they are flattening the containment
1559 // relationship.
1560 if g.Type == "EZ" || g.Type == "UN" {
1561 continue
1562 }
1563 group := b.region.index(g.Type)
1564 groupIdx := b.groups[group]
1565 for _, mem := range strings.Split(g.Contains, " ") {
1566 r := b.region.index(mem)
1567 mm[r] = append(mm[r], groupIdx)
1568 if g, ok := b.groups[r]; ok {
1569 mm[group] = append(mm[group], g)
1570 containment[groupIdx] = append(containment[groupIdx], g)
1571 }
1572 }
1573 }
1574
1575 regionContainment := make([]uint64, len(b.groups))
1576 for _, g := range b.groups {
1577 l := containment[g]
1578
1579 // Compute the transitive closure of containment.
1580 for i := 0; i < len(l); i++ {
1581 l = append(l, containment[l[i]]...)
1582 }
1583
1584 // Compute the bitmask.
1585 regionContainment[g] = 1 << g
1586 for _, v := range l {
1587 regionContainment[g] |= 1 << v
1588 }
1589 }
1590 b.writeSlice("regionContainment", regionContainment)
1591
1592 regionInclusion := make([]uint8, len(b.region.s))
1593 bvs := make(map[uint64]index)
1594 // Make the first bitvector positions correspond with the groups.
1595 for r, i := range b.groups {
1596 bv := uint64(1 << i)
1597 for _, g := range mm[r] {
1598 bv |= 1 << g
1599 }
1600 bvs[bv] = i
1601 regionInclusion[r] = uint8(bvs[bv])
1602 }
1603 for r := 1; r < len(b.region.s); r++ {
1604 if _, ok := b.groups[r]; !ok {
1605 bv := uint64(0)
1606 for _, g := range mm[r] {
1607 bv |= 1 << g
1608 }
1609 if bv == 0 {
1610 // Pick the world for unspecified regions.
1611 bv = 1 << b.groups[b.region.index("001")]
1612 }
1613 if _, ok := bvs[bv]; !ok {
1614 bvs[bv] = index(len(bvs))
1615 }
1616 regionInclusion[r] = uint8(bvs[bv])
1617 }
1618 }
1619 b.writeSlice("regionInclusion", regionInclusion)
1620 regionInclusionBits := make([]uint64, len(bvs))
1621 for k, v := range bvs {
1622 regionInclusionBits[v] = uint64(k)
1623 }
1624 // Add bit vectors for increasingly large distances until a fixed point is reached.
1625 regionInclusionNext := []uint8{}
1626 for i := 0; i < len(regionInclusionBits); i++ {
1627 bits := regionInclusionBits[i]
1628 next := bits
1629 for i := uint(0); i < uint(len(b.groups)); i++ {
1630 if bits&(1<<i) != 0 {
1631 next |= regionInclusionBits[i]
1632 }
1633 }
1634 if _, ok := bvs[next]; !ok {
1635 bvs[next] = index(len(bvs))
1636 regionInclusionBits = append(regionInclusionBits, next)
1637 }
1638 regionInclusionNext = append(regionInclusionNext, uint8(bvs[next]))
1639 }
1640 b.writeSlice("regionInclusionBits", regionInclusionBits)
1641 b.writeSlice("regionInclusionNext", regionInclusionNext)
1642}
1643
1644type parentRel struct {
1645 lang uint16
1646 script uint8
1647 maxScript uint8
1648 toRegion uint16
1649 fromRegion []uint16
1650}
1651
1652func (b *builder) writeParents() {
1653 b.writeType(parentRel{})
1654
1655 parents := []parentRel{}
1656
1657 // Construct parent overrides.
1658 n := 0
1659 for _, p := range b.data.Supplemental().ParentLocales.ParentLocale {
1660 // Skipping non-standard scripts to root is implemented using addTags.
1661 if p.Parent == "root" {
1662 continue
1663 }
1664
1665 sub := strings.Split(p.Parent, "_")
1666 parent := parentRel{lang: b.langIndex(sub[0])}
1667 if len(sub) == 2 {
1668 // TODO: check that all undefined scripts are indeed Latn in these
1669 // cases.
1670 parent.maxScript = uint8(b.script.index("Latn"))
1671 parent.toRegion = uint16(b.region.index(sub[1]))
1672 } else {
1673 parent.script = uint8(b.script.index(sub[1]))
1674 parent.maxScript = parent.script
1675 parent.toRegion = uint16(b.region.index(sub[2]))
1676 }
1677 for _, c := range strings.Split(p.Locales, " ") {
1678 region := b.region.index(c[strings.LastIndex(c, "_")+1:])
1679 parent.fromRegion = append(parent.fromRegion, uint16(region))
1680 }
1681 parents = append(parents, parent)
1682 n += len(parent.fromRegion)
1683 }
1684 b.writeSliceAddSize("parents", n*2, parents)
1685}
1686
1687func main() {
1688 gen.Init()
1689
1690 gen.Repackage("gen_common.go", "common.go", "language")
1691
1692 w := gen.NewCodeWriter()
1693 defer w.WriteGoFile("tables.go", "language")
1694
1695 fmt.Fprintln(w, `import "golang.org/x/text/internal/tag"`)
1696
1697 b := newBuilder(w)
1698 gen.WriteCLDRVersion(w)
1699
1700 b.parseIndices()
1701 b.writeType(fromTo{})
1702 b.writeLanguage()
1703 b.writeScript()
1704 b.writeRegion()
1705 b.writeVariant()
1706 // TODO: b.writeLocale()
1707 b.computeRegionGroups()
1708 b.writeLikelyData()
1709 b.writeMatchData()
1710 b.writeRegionInclusionData()
1711 b.writeParents()
1712}