diff options
Diffstat (limited to 'vendor/golang.org/x/text/language/match_test.go')
| -rw-r--r-- | vendor/golang.org/x/text/language/match_test.go | 149 |
1 files changed, 136 insertions, 13 deletions
diff --git a/vendor/golang.org/x/text/language/match_test.go b/vendor/golang.org/x/text/language/match_test.go index 26cc2af2b..4fb56a025 100644 --- a/vendor/golang.org/x/text/language/match_test.go +++ b/vendor/golang.org/x/text/language/match_test.go @@ -8,14 +8,109 @@ import ( "bytes" "flag" "fmt" + "os" + "path" "strings" "testing" "golang.org/x/text/internal/testtext" + "golang.org/x/text/internal/ucd" ) var verbose = flag.Bool("verbose", false, "set to true to print the internal tables of matchers") +func TestCLDRCompliance(t *testing.T) { + r, err := os.Open("testdata/localeMatcherTest.txt") + if err != nil { + t.Fatal(err) + } + ucd.Parse(r, func(p *ucd.Parser) { + name := strings.Replace(path.Join(p.String(0), p.String(1)), " ", "", -1) + if skip[name] { + return + } + t.Run(name, func(t *testing.T) { + supported := makeTagList(p.String(0)) + desired := makeTagList(p.String(1)) + gotCombined, index, _ := NewMatcher(supported).Match(desired...) + + gotMatch := supported[index] + wantMatch := Make(p.String(2)) + if gotMatch != wantMatch { + t.Fatalf("match: got %q; want %q", gotMatch, wantMatch) + } + wantCombined, err := Parse(p.String(3)) + if err == nil && gotCombined != wantCombined { + t.Errorf("combined: got %q; want %q", gotCombined, wantCombined) + } + }) + }) +} + +var skip = map[string]bool{ + // TODO: bugs + // und-<region> is not expanded to the appropriate language. + "en-Hant-TW,und-TW/zh-Hant": true, // match: got "en-Hant-TW"; want "und-TW" + "en-Hant-TW,und-TW/zh": true, // match: got "en-Hant-TW"; want "und-TW" + // Honor the wildcard match. This may only be useful to select non-exact + // stuff. + "mul,af/nl": true, // match: got "af"; want "mul" + + // TODO: include other extensions. + // combined: got "en-GB-u-ca-buddhist-nu-arab"; want "en-GB-fonipa-t-m0-iso-i0-pinyin-u-ca-buddhist-nu-arab" + "und,en-GB-u-sd-gbsct/en-fonipa-u-nu-Arab-ca-buddhist-t-m0-iso-i0-pinyin": true, + + // Inconsistencies with Mark Davis' implementation where it is not clear + // which is better. + + // Go prefers exact matches over less exact preferred ones. + // Preferring desired ones might be better. + "en,de,fr,ja/de-CH,fr": true, // match: got "fr"; want "de" + "en-GB,en,de,fr,ja/de-CH,fr": true, // match: got "fr"; want "de" + "pt-PT,pt-BR,es,es-419/pt-US,pt-PT": true, // match: got "pt-PT"; want "pt-BR" + "pt-PT,pt,es,es-419/pt-US,pt-PT,pt": true, // match: got "pt-PT"; want "pt" + "en,sv/en-GB,sv": true, // match: got "sv"; want "en" + "en-NZ,en-IT/en-US": true, // match: got "en-IT"; want "en-NZ" + + // Inconsistencies in combined. I think the Go approach is more appropriate. + // We could use -u-rg- and -u-va- as alternative. + "und,fr/fr-BE-fonipa": true, // combined: got "fr"; want "fr-BE-fonipa" + "und,fr-CA/fr-BE-fonipa": true, // combined: got "fr-CA"; want "fr-BE-fonipa" + "und,fr-fonupa/fr-BE-fonipa": true, // combined: got "fr-fonupa"; want "fr-BE-fonipa" + "und,no/nn-BE-fonipa": true, // combined: got "no"; want "no-BE-fonipa" + "50,und,fr-CA-fonupa/fr-BE-fonipa": true, // combined: got "fr-CA-fonupa"; want "fr-BE-fonipa" + + // Spec says prefer primary locales. But what is the benefit? Shouldn't + // the developer just not specify the primary locale first in the list? + // TODO: consider adding a SortByPreferredLocale function to ensure tags + // are ordered such that the preferred locale rule is observed. + // TODO: most of these cases are solved by getting rid of the region + // distance tie-breaker rule (see comments there). + "und,es,es-MA,es-MX,es-419/es-EA": true, // match: got "es-MA"; want "es" + "und,es-MA,es,es-419,es-MX/es-EA": true, // match: got "es-MA"; want "es" + "und,en,en-GU,en-IN,en-GB/en-ZA": true, // match: got "en-IN"; want "en-GB" + "und,en,en-GU,en-IN,en-GB/en-VI": true, // match: got "en-GU"; want "en" + "und,en-GU,en,en-GB,en-IN/en-VI": true, // match: got "en-GU"; want "en" + + // Falling back to the default seems more appropriate than falling back + // on a language with the same script. + "50,und,fr-Cyrl-CA-fonupa/fr-BE-fonipa": true, + // match: got "und"; want "fr-Cyrl-CA-fonupa" + // combined: got "und"; want "fr-Cyrl-BE-fonipa" + + // Other interesting cases to test: + // - Should same language or same script have the preference if there is + // usually no understanding of the other script? + // - More specific region in desired may replace enclosing supported. +} + +func makeTagList(s string) (tags []Tag) { + for _, s := range strings.Split(s, ",") { + tags = append(tags, Make(strings.TrimSpace(s))) + } + return tags +} + func TestAddLikelySubtags(t *testing.T) { tests := []struct{ in, out string }{ {"aa", "aa-Latn-ET"}, @@ -164,6 +259,31 @@ func TestMinimize(t *testing.T) { } } +func TestRegionGroups(t *testing.T) { + testCases := []struct { + a, b string + distance uint8 + }{ + {"zh-TW", "zh-HK", 5}, + {"zh-MO", "zh-HK", 4}, + } + for _, tc := range testCases { + a := MustParse(tc.a) + aScript, _ := a.Script() + b := MustParse(tc.b) + bScript, _ := b.Script() + + if aScript != bScript { + t.Errorf("scripts differ: %q vs %q", aScript, bScript) + continue + } + d := regionGroupDist(a.region, b.region, aScript.scriptID, a.lang) + if d != tc.distance { + t.Errorf("got %q; want %q", d, tc.distance) + } + } +} + func TestRegionDistance(t *testing.T) { tests := []struct { a, b string @@ -259,17 +379,20 @@ func parseSupported(list string) (out []Tag) { // The test set for TestBestMatch is defined in data_test.go. func TestBestMatch(t *testing.T) { - for i, tt := range matchTests { + for _, tt := range matchTests { supported := parseSupported(tt.supported) - m := newMatcher(supported) + m := newMatcher(supported, nil) if *verbose { fmt.Printf("%s:\n%v\n", tt.comment, m) } for _, tm := range tt.test { - tag, _, conf := m.Match(parseSupported(tm.desired)...) - if tag.String() != tm.match { - t.Errorf("%d:%s: find %s in %q: have %s; want %s (%v)\n", i, tt.comment, tm.desired, tt.supported, tag, tm.match, conf) - } + t.Run(path.Join(tt.comment, tt.supported, tm.desired), func(t *testing.T) { + tag, _, conf := m.Match(parseSupported(tm.desired)...) + if tag.String() != tm.match { + t.Errorf("find %s in %q: have %s; want %s (%v)", tm.desired, tt.supported, tag, tm.match, conf) + } + }) + } } } @@ -352,7 +475,7 @@ var benchWant = [][]Tag{ } func BenchmarkMatch(b *testing.B) { - m := newMatcher(benchHave) + m := newMatcher(benchHave, nil) for i := 0; i < b.N; i++ { for _, want := range benchWant { m.getBest(want...) @@ -362,7 +485,7 @@ func BenchmarkMatch(b *testing.B) { func BenchmarkMatchExact(b *testing.B) { want := mk("en") - m := newMatcher(benchHave) + m := newMatcher(benchHave, nil) for i := 0; i < b.N; i++ { m.getBest(want) } @@ -370,7 +493,7 @@ func BenchmarkMatchExact(b *testing.B) { func BenchmarkMatchAltLanguagePresent(b *testing.B) { want := mk("hr") - m := newMatcher(benchHave) + m := newMatcher(benchHave, nil) for i := 0; i < b.N; i++ { m.getBest(want) } @@ -378,7 +501,7 @@ func BenchmarkMatchAltLanguagePresent(b *testing.B) { func BenchmarkMatchAltLanguageNotPresent(b *testing.B) { want := mk("nn") - m := newMatcher(benchHave) + m := newMatcher(benchHave, nil) for i := 0; i < b.N; i++ { m.getBest(want) } @@ -386,7 +509,7 @@ func BenchmarkMatchAltLanguageNotPresent(b *testing.B) { func BenchmarkMatchAltScriptPresent(b *testing.B) { want := mk("zh-Hant-CN") - m := newMatcher(benchHave) + m := newMatcher(benchHave, nil) for i := 0; i < b.N; i++ { m.getBest(want) } @@ -394,7 +517,7 @@ func BenchmarkMatchAltScriptPresent(b *testing.B) { func BenchmarkMatchAltScriptNotPresent(b *testing.B) { want := mk("fr-Cyrl") - m := newMatcher(benchHave) + m := newMatcher(benchHave, nil) for i := 0; i < b.N; i++ { m.getBest(want) } @@ -402,7 +525,7 @@ func BenchmarkMatchAltScriptNotPresent(b *testing.B) { func BenchmarkMatchLimitedExact(b *testing.B) { want := []Tag{mk("he-NL"), mk("iw-NL")} - m := newMatcher(benchHave) + m := newMatcher(benchHave, nil) for i := 0; i < b.N; i++ { m.getBest(want...) } |