move content.go generation to internal

2025-06-19 14:43:05 -03:00 · 2017-04-05 18:15:27 +02:00
parent ba22a0a243
commit 03c71a9b93
6 changed files with 0 additions and 0 deletions
--- a/internal/code-generator/generator/test_files/content.gold
+++ b/internal/code-generator/generator/test_files/content.gold
@ -0,0 +1,81 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH github.com/src-d/simple-linguist/cli/slinguist-generate
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: fe8b44ab8a225b1ffa75b983b916ea22fee5b6f7
+
+import (
+	"path/filepath"
+	"regexp"
+	"strings"
+)
+
+func GetLanguageByContent(filename string, content []byte) (lang string, safe bool) {
+	ext := strings.ToLower(filepath.Ext(filename))
+	if fnMatcher, ok := matchers[ext]; ok {
+		lang, safe = fnMatcher(content)
+		return
+	}
+
+	return GetLanguageByExtension(filename)
+}
+
+type languageMatcher func ([]byte) (string, bool)
+
+var matchers = map[string]languageMatcher{
+	".asc": func(i []byte) (string, bool) {
+		if asc_PublicKey_Matcher_0.Match(i) {
+			return "Public Key", true
+		} else if asc_AsciiDoc_Matcher_0.Match(i) {
+			return "AsciiDoc", true
+		} else if asc_AGSScript_Matcher_0.Match(i) {
+			return "AGS Script", true
+		}
+	
+		return OtherLanguage, false
+	},
+	".ms": func(i []byte) (string, bool) {
+		if ms_Groff_Matcher_0.Match(i) {
+			return "Groff", true
+		}
+	
+		return "MAXScript", true
+	},
+	".mod": func(i []byte) (string, bool) {
+		if mod_XML_Matcher_0.Match(i) {
+			return "XML", true
+		} else if mod_ModulaDash2_Matcher_0.Match(i) || mod_ModulaDash2_Matcher_1.Match(i) {
+			return "Modula-2", true
+		}
+	
+		return "Linux Kernel Module", false
+	},
+	".pro": func(i []byte) (string, bool) {
+		if pro_Prolog_Matcher_0.Match(i) {
+			return "Prolog", true
+		} else if pro_INI_Matcher_0.Match(i) {
+			return "INI", true
+		} else if pro_QMake_Matcher_0.Match(i) && pro_QMake_Matcher_1.Match(i) {
+			return "QMake", true
+		} else if pro_IDL_Matcher_0.Match(i) {
+			return "IDL", true
+		}
+	
+		return OtherLanguage, false
+	},
+	}
+
+var (
+    asc_PublicKey_Matcher_0 = regexp.MustCompile(`(?m)^(----[- ]BEGIN|ssh-(rsa|dss)) `)
+    asc_AsciiDoc_Matcher_0 = regexp.MustCompile(`(?m)^[=-]+(\s|\n)|{{[A-Za-z]`)
+    asc_AGSScript_Matcher_0 = regexp.MustCompile(`(?m)^(\/\/.+|((import|export)\s+)?(function|int|float|char)\s+((room|repeatedly|on|game)_)?([A-Za-z]+[A-Za-z_0-9]+)\s*[;\(])`)
+    ms_Groff_Matcher_0 = regexp.MustCompile(`(?mi)^[.'][a-z][a-z](\s|$)`)
+    mod_XML_Matcher_0 = regexp.MustCompile(`(?m)<!ENTITY `)
+    mod_ModulaDash2_Matcher_0 = regexp.MustCompile(`(?mi)^\s*MODULE [\w\.]+;`)
+    mod_ModulaDash2_Matcher_1 = regexp.MustCompile(`(?mi)^\s*END [\w\.]+;`)
+    pro_Prolog_Matcher_0 = regexp.MustCompile(`(?m)^[^#]+:-`)
+    pro_INI_Matcher_0 = regexp.MustCompile(`(?m)last_client=`)
+    pro_QMake_Matcher_0 = regexp.MustCompile(`(?m)HEADERS`)
+    pro_QMake_Matcher_1 = regexp.MustCompile(`(?m)SOURCES`)
+    pro_IDL_Matcher_0 = regexp.MustCompile(`(?m)^\s*function[ \w,]+$`)
+    )
--- a/internal/code-generator/generator/test_files/content.test.go.tmpl
+++ b/internal/code-generator/generator/test_files/content.test.go.tmpl
@ -0,0 +1,51 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH github.com/src-d/simple-linguist/cli/slinguist-generate
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: {{ getCommit }}
+
+import (
+	"path/filepath"
+	"regexp"
+	"strings"
+)
+
+func GetLanguageByContent(filename string, content []byte) (lang string, safe bool) {
+	ext := strings.ToLower(filepath.Ext(filename))
+	if fnMatcher, ok := matchers[ext]; ok {
+		lang, safe = fnMatcher(content)
+		return
+	}
+
+	return GetLanguageByExtension(filename)
+}
+
+type languageMatcher func ([]byte) (string, bool)
+
+var matchers = map[string]languageMatcher{
+	{{ range $index, $disambiguator := . -}}
+	{{ printf "%q" $disambiguator.Extension }}: func(i []byte) (string, bool) {
+		{{ range $i, $language := $disambiguator.Languages -}}
+
+		{{- if not (avoidLanguage $language) }}
+		{{- if gt (len $language.Heuristics) 0 }}
+		{{- if gt $i 0 }} else {{ end -}} 
+		if {{- range $j, $heuristic := $language.Heuristics }} {{ $heuristic.Name }}.Match(i) 
+		{{- if lt $j (len $language.LogicRelations) }} {{index $language.LogicRelations $j}} {{- end -}} {{ end }} {
+			return {{ printf "%q" $language.Language }}, true
+		}
+		
+		{{- end -}}
+		{{- end -}}
+		{{- end}}
+	
+		return {{ returnLanguage $disambiguator.Languages }}, {{ safeLanguage $disambiguator.Languages }}
+	},
+	{{ end -}}
+}
+
+var (
+    {{ range $index, $heuristic := getAllHeuristics . -}}
+    {{ $heuristic.Name }} = regexp.MustCompile(`{{ $heuristic.Regexp }}`)
+    {{ end -}}
+)
--- a/internal/code-generator/generator/test_files/formated_content.gold
+++ b/internal/code-generator/generator/test_files/formated_content.gold
@ -0,0 +1,81 @@
+package slinguist
+
+// CODE GENERATED AUTOMATICALLY WITH github.com/src-d/simple-linguist/cli/slinguist-generate
+// THIS FILE SHOULD NOT BE EDITED BY HAND
+// Extracted from github/linguist commit: fe8b44ab8a225b1ffa75b983b916ea22fee5b6f7
+
+import (
+	"path/filepath"
+	"regexp"
+	"strings"
+)
+
+func GetLanguageByContent(filename string, content []byte) (lang string, safe bool) {
+	ext := strings.ToLower(filepath.Ext(filename))
+	if fnMatcher, ok := matchers[ext]; ok {
+		lang, safe = fnMatcher(content)
+		return
+	}
+
+	return GetLanguageByExtension(filename)
+}
+
+type languageMatcher func([]byte) (string, bool)
+
+var matchers = map[string]languageMatcher{
+	".asc": func(i []byte) (string, bool) {
+		if asc_PublicKey_Matcher_0.Match(i) {
+			return "Public Key", true
+		} else if asc_AsciiDoc_Matcher_0.Match(i) {
+			return "AsciiDoc", true
+		} else if asc_AGSScript_Matcher_0.Match(i) {
+			return "AGS Script", true
+		}
+
+		return OtherLanguage, false
+	},
+	".ms": func(i []byte) (string, bool) {
+		if ms_Groff_Matcher_0.Match(i) {
+			return "Groff", true
+		}
+
+		return "MAXScript", true
+	},
+	".mod": func(i []byte) (string, bool) {
+		if mod_XML_Matcher_0.Match(i) {
+			return "XML", true
+		} else if mod_ModulaDash2_Matcher_0.Match(i) || mod_ModulaDash2_Matcher_1.Match(i) {
+			return "Modula-2", true
+		}
+
+		return "Linux Kernel Module", false
+	},
+	".pro": func(i []byte) (string, bool) {
+		if pro_Prolog_Matcher_0.Match(i) {
+			return "Prolog", true
+		} else if pro_INI_Matcher_0.Match(i) {
+			return "INI", true
+		} else if pro_QMake_Matcher_0.Match(i) && pro_QMake_Matcher_1.Match(i) {
+			return "QMake", true
+		} else if pro_IDL_Matcher_0.Match(i) {
+			return "IDL", true
+		}
+
+		return OtherLanguage, false
+	},
+}
+
+var (
+	asc_PublicKey_Matcher_0   = regexp.MustCompile(`(?m)^(----[- ]BEGIN|ssh-(rsa|dss)) `)
+	asc_AsciiDoc_Matcher_0    = regexp.MustCompile(`(?m)^[=-]+(\s|\n)|{{[A-Za-z]`)
+	asc_AGSScript_Matcher_0   = regexp.MustCompile(`(?m)^(\/\/.+|((import|export)\s+)?(function|int|float|char)\s+((room|repeatedly|on|game)_)?([A-Za-z]+[A-Za-z_0-9]+)\s*[;\(])`)
+	ms_Groff_Matcher_0        = regexp.MustCompile(`(?mi)^[.'][a-z][a-z](\s|$)`)
+	mod_XML_Matcher_0         = regexp.MustCompile(`(?m)<!ENTITY `)
+	mod_ModulaDash2_Matcher_0 = regexp.MustCompile(`(?mi)^\s*MODULE [\w\.]+;`)
+	mod_ModulaDash2_Matcher_1 = regexp.MustCompile(`(?mi)^\s*END [\w\.]+;`)
+	pro_Prolog_Matcher_0      = regexp.MustCompile(`(?m)^[^#]+:-`)
+	pro_INI_Matcher_0         = regexp.MustCompile(`(?m)last_client=`)
+	pro_QMake_Matcher_0       = regexp.MustCompile(`(?m)HEADERS`)
+	pro_QMake_Matcher_1       = regexp.MustCompile(`(?m)SOURCES`)
+	pro_IDL_Matcher_0         = regexp.MustCompile(`(?m)^\s*function[ \w,]+$`)
+)
--- a/internal/code-generator/generator/test_files/heuristics.test.rb
+++ b/internal/code-generator/generator/test_files/heuristics.test.rb
@ -0,0 +1,44 @@
+    # Common heuristics
+    ObjectiveCRegex = /^\s*(@(interface|class|protocol|property|end|synchronised|selector|implementation)\b|#import\s+.+\.h[">])/
+
+    disambiguate ".asc" do |data|
+      if /^(----[- ]BEGIN|ssh-(rsa|dss)) /.match(data)
+        Language["Public Key"]
+      elsif /^[=-]+(\s|\n)|{{[A-Za-z]/.match(data)
+        Language["AsciiDoc"]
+      elsif /^(\/\/.+|((import|export)\s+)?(function|int|float|char)\s+((room|repeatedly|on|game)_)?([A-Za-z]+[A-Za-z_0-9]+)\s*[;\(])/.match(data)
+        Language["AGS Script"]
+      end
+    end
+
+    disambiguate ".ms" do |data|
+      if /^[.'][a-z][a-z](\s|$)/i.match(data)
+        Language["Groff"]
+      elsif /(?<!\S)\.(include|globa?l)\s/.match(data) || /(?<!\/\*)(\A|\n)\s*\.[A-Za-z]/.match(data.gsub(/"([^\\"]|\\.)*"|'([^\\']|\\.)*'|\\\s*(?:--.*)?\n/, ""))
+        Language["GAS"]
+      else
+        Language["MAXScript"]
+      end
+    end
+
+    disambiguate ".mod" do |data|
+      if data.include?('<!ENTITY ')
+        Language["XML"]
+      elsif /^\s*MODULE [\w\.]+;/i.match(data) || /^\s*END [\w\.]+;/i.match(data)
+        Language["Modula-2"]
+      else
+        [Language["Linux Kernel Module"], Language["AMPL"]]
+      end
+    end
+
+    disambiguate ".pro" do |data|
+      if /^[^#]+:-/.match(data)
+        Language["Prolog"]
+      elsif data.include?("last_client=")
+        Language["INI"]
+      elsif data.include?("HEADERS") && data.include?("SOURCES")
+        Language["QMake"]
+      elsif /^\s*function[ \w,]+$/.match(data)
+        Language["IDL"]
+      end
+    end