regexp for vendored files and directories are generated in vendor_matchers.go

internal/code-generator/assets/utils.go.tmpl

@@ -1,84 +0,0 @@
-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 (
-	"bytes"
-	"path/filepath"
-	"strings"
-
-	"gopkg.in/toqueteos/substring.v1"
-)
-
-func IsAuxiliaryLanguage(lang string) bool {
-	_, ok := auxiliaryLanguages[lang]
-	return ok
-}
-
-func IsConfiguration(path string) bool {
-	lang, _ := GetLanguageByExtension(path)
-	_, is := configurationLanguages[lang]
-
-	return is
-}
-
-func IsDotFile(path string) bool {
-	return strings.HasPrefix(filepath.Base(path), ".")
-}
-
-func IsVendor(path string) bool {
-	return findIndex(path, vendorMatchers) >= 0
-}
-
-func IsDocumentation(path string) bool {
-	return findIndex(path, documentationMatchers) >= 0
-}
-
-func findIndex(path string, matchers substring.StringsMatcher) int {
-	return matchers.MatchIndex(path)
-}
-
-const sniffLen = 8000
-
-//IsBinary detects if data is a binary value based on:
-//http://git.kernel.org/cgit/git/git.git/tree/xdiff-interface.c?id=HEAD#n198
-func IsBinary(data []byte) bool {
-	if len(data) > sniffLen {
-		data = data[:sniffLen]
-	}
-
-	if bytes.IndexByte(data, byte(0)) == -1 {
-		return false
-	}
-
-	return true
-}
-
-var vendorMatchers = substring.Or(
-	{{range $regexp := . -}}
-	substring.Regexp(`{{ $regexp }}`),
-	{{end -}}
-)
-
-var documentationMatchers = substring.Or(
-	substring.Regexp(`^docs?/`),
-	substring.Regexp(`(^|/)[Dd]ocumentation/`),
-	substring.Regexp(`(^|/)javadoc/`),
-	substring.Regexp(`^man/`),
-	substring.Regexp(`^[Ee]xamples/`),
-	substring.Regexp(`(^|/)CHANGE(S|LOG)?(\.|$)`),
-	substring.Regexp(`(^|/)CONTRIBUTING(\.|$)`),
-	substring.Regexp(`(^|/)COPYING(\.|$)`),
-	substring.Regexp(`(^|/)INSTALL(\.|$)`),
-	substring.Regexp(`(^|/)LICEN[CS]E(\.|$)`),
-	substring.Regexp(`(^|/)[Ll]icen[cs]e(\.|$)`),
-	substring.Regexp(`(^|/)README(\.|$)`),
-	substring.Regexp(`(^|/)[Rr]eadme(\.|$)`),
-	substring.Regexp(`^[Ss]amples/`),
-)
-
-var configurationLanguages = map[string]bool{
-	"XML": true, "JSON": true, "TOML": true, "YAML": true, "INI": true, "SQL": true,
-}

internal/code-generator/assets/vendor.go.tmpl

@@ -0,0 +1,13 @@
+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 "gopkg.in/toqueteos/substring.v1"
+
+var vendorMatchers = substring.Or(
+	{{range $regexp := . -}}
+	substring.Regexp(`{{ $regexp }}`),
+	{{end -}}
+)

internal/code-generator/generator/generator_test.go

@@ -13,7 +13,7 @@ const (
 	// FromFile test
 	formatedLangGold    = "test_files/formated_languages.gold"
 	formatedContentGold = "test_files/formated_content.gold"
-	formatedUtilsGold   = "test_files/formated_utils.gold"
+	formatedVendorGold  = "test_files/formated_vendor.gold"
 
 	// Languages test
 	ymlTestFile           = "test_files/languages.test.yml"
@@ -31,9 +31,9 @@ const (
 
 	// Vendor test
 	vendorTestFile     = "test_files/vendor.test.yml"
-	utilsGold         = "test_files/utils.gold"
+	vendorGold         = "test_files/vendor.gold"
-	utilsTestTmplPath = "test_files/utils.test.go.tmpl"
+	vendorTestTmplPath = "test_files/vendor.test.go.tmpl"
-	utilsTestTmplName = "utils.test.go.tmpl"
+	vendorTestTmplName = "vendor.test.go.tmpl"
 	commitVendorTest   = "fe8b44ab8a225b1ffa75b983b916ea22fee5b6f7"
 )
 
@@ -44,7 +44,7 @@ func TestFromFile(t *testing.T) {
 	goldContent, err := ioutil.ReadFile(formatedContentGold)
 	assert.NoError(t, err)
 
-	goldUtils, err := ioutil.ReadFile(formatedUtilsGold)
+	goldVendor, err := ioutil.ReadFile(formatedVendorGold)
 	assert.NoError(t, err)
 
 	outPathLang, err := ioutil.TempFile("/tmp", "generator-test-")
@@ -55,7 +55,7 @@ func TestFromFile(t *testing.T) {
 	assert.NoError(t, err)
 	defer os.Remove(outPathContent.Name())
 
-	outPathUtils, err := ioutil.TempFile("/tmp", "generator-test-")
+	outPathVendor, err := ioutil.TempFile("/tmp", "generator-test-")
 	assert.NoError(t, err)
 	defer os.Remove(outPathContent.Name())
 
@@ -92,12 +92,12 @@ func TestFromFile(t *testing.T) {
 		{
 			name:        "TestFromFile_Vendor",
 			fileToParse: vendorTestFile,
-			outPath:     outPathUtils.Name(),
+			outPath:     outPathVendor.Name(),
-			tmplPath:    utilsTestTmplPath,
+			tmplPath:    vendorTestTmplPath,
-			tmplName:    utilsTestTmplName,
+			tmplName:    vendorTestTmplName,
 			commit:      commitVendorTest,
 			generate:    Vendor,
-			wantOut:     goldUtils,
+			wantOut:     goldVendor,
 		},
 	}
 
@@ -181,7 +181,7 @@ func TestHeuristics(t *testing.T) {
 }
 
 func TestVendor(t *testing.T) {
-	gold, err := ioutil.ReadFile(utilsGold)
+	gold, err := ioutil.ReadFile(vendorGold)
 	assert.NoError(t, err)
 
 	input, err := ioutil.ReadFile(vendorTestFile)
@@ -198,8 +198,8 @@ func TestVendor(t *testing.T) {
 		{
 			name:     "TestVendor",
 			input:    input,
-			tmplPath: utilsTestTmplPath,
+			tmplPath: vendorTestTmplPath,
-			tmplName: utilsTestTmplName,
+			tmplName: vendorTestTmplName,
 			commit:   commitVendorTest,
 			wantOut:  gold,
 		},

internal/code-generator/generator/test_files/formated_utils.gold

@@ -1,95 +0,0 @@
-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 (
-	"bytes"
-	"path/filepath"
-	"strings"
-
-	"gopkg.in/toqueteos/substring.v1"
-)
-
-func IsAuxiliaryLanguage(lang string) bool {
-	_, ok := auxiliaryLanguages[lang]
-	return ok
-}
-
-func IsConfiguration(path string) bool {
-	lang, _ := GetLanguageByExtension(path)
-	_, is := configurationLanguages[lang]
-
-	return is
-}
-
-func IsDotFile(path string) bool {
-	return strings.HasPrefix(filepath.Base(path), ".")
-}
-
-func IsVendor(path string) bool {
-	return findIndex(path, vendorMatchers) >= 0
-}
-
-func IsDocumentation(path string) bool {
-	return findIndex(path, documentationMatchers) >= 0
-}
-
-func findIndex(path string, matchers substring.StringsMatcher) int {
-	return matchers.MatchIndex(path)
-}
-
-const sniffLen = 8000
-
-//IsBinary detects if data is a binary value based on:
-//http://git.kernel.org/cgit/git/git.git/tree/xdiff-interface.c?id=HEAD#n198
-func IsBinary(data []byte) bool {
-	if len(data) > sniffLen {
-		data = data[:sniffLen]
-	}
-
-	if bytes.IndexByte(data, byte(0)) == -1 {
-		return false
-	}
-
-	return true
-}
-
-var vendorMatchers = substring.Or(
-	substring.Regexp(`(^|/)cache/`),
-	substring.Regexp(`^[Dd]ependencies/`),
-	substring.Regexp(`(^|/)dist/`),
-	substring.Regexp(`^deps/`),
-	substring.Regexp(`^tools/`),
-	substring.Regexp(`(^|/)configure$`),
-	substring.Regexp(`(^|/)config.guess$`),
-	substring.Regexp(`(^|/)config.sub$`),
-	substring.Regexp(`(^|/)aclocal.m4`),
-	substring.Regexp(`(^|/)libtool.m4`),
-	substring.Regexp(`(^|/)ltoptions.m4`),
-	substring.Regexp(`(^|/)ltsugar.m4`),
-	substring.Regexp(`(^|/)ltversion.m4`),
-	substring.Regexp(`(^|/)lt~obsolete.m4`),
-)
-
-var documentationMatchers = substring.Or(
-	substring.Regexp(`^docs?/`),
-	substring.Regexp(`(^|/)[Dd]ocumentation/`),
-	substring.Regexp(`(^|/)javadoc/`),
-	substring.Regexp(`^man/`),
-	substring.Regexp(`^[Ee]xamples/`),
-	substring.Regexp(`(^|/)CHANGE(S|LOG)?(\.|$)`),
-	substring.Regexp(`(^|/)CONTRIBUTING(\.|$)`),
-	substring.Regexp(`(^|/)COPYING(\.|$)`),
-	substring.Regexp(`(^|/)INSTALL(\.|$)`),
-	substring.Regexp(`(^|/)LICEN[CS]E(\.|$)`),
-	substring.Regexp(`(^|/)[Ll]icen[cs]e(\.|$)`),
-	substring.Regexp(`(^|/)README(\.|$)`),
-	substring.Regexp(`(^|/)[Rr]eadme(\.|$)`),
-	substring.Regexp(`^[Ss]amples/`),
-)
-
-var configurationLanguages = map[string]bool{
-	"XML": true, "JSON": true, "TOML": true, "YAML": true, "INI": true, "SQL": true,
-}

internal/code-generator/generator/test_files/formated_vendor.gold

@@ -0,0 +1,24 @@
+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 "gopkg.in/toqueteos/substring.v1"
+
+var vendorMatchers = substring.Or(
+	substring.Regexp(`(^|/)cache/`),
+	substring.Regexp(`^[Dd]ependencies/`),
+	substring.Regexp(`(^|/)dist/`),
+	substring.Regexp(`^deps/`),
+	substring.Regexp(`^tools/`),
+	substring.Regexp(`(^|/)configure$`),
+	substring.Regexp(`(^|/)config.guess$`),
+	substring.Regexp(`(^|/)config.sub$`),
+	substring.Regexp(`(^|/)aclocal.m4`),
+	substring.Regexp(`(^|/)libtool.m4`),
+	substring.Regexp(`(^|/)ltoptions.m4`),
+	substring.Regexp(`(^|/)ltsugar.m4`),
+	substring.Regexp(`(^|/)ltversion.m4`),
+	substring.Regexp(`(^|/)lt~obsolete.m4`),
+)

internal/code-generator/generator/test_files/utils.gold

@@ -1,95 +0,0 @@
-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 (
-	"bytes"
-	"path/filepath"
-	"strings"
-
-	"gopkg.in/toqueteos/substring.v1"
-)
-
-func IsAuxiliaryLanguage(lang string) bool {
-	_, ok := auxiliaryLanguages[lang]
-	return ok
-}
-
-func IsConfiguration(path string) bool {
-	lang, _ := GetLanguageByExtension(path)
-	_, is := configurationLanguages[lang]
-
-	return is
-}
-
-func IsDotFile(path string) bool {
-	return strings.HasPrefix(filepath.Base(path), ".")
-}
-
-func IsVendor(path string) bool {
-	return findIndex(path, vendorMatchers) >= 0
-}
-
-func IsDocumentation(path string) bool {
-	return findIndex(path, documentationMatchers) >= 0
-}
-
-func findIndex(path string, matchers substring.StringsMatcher) int {
-	return matchers.MatchIndex(path)
-}
-
-const sniffLen = 8000
-
-//IsBinary detects if data is a binary value based on:
-//http://git.kernel.org/cgit/git/git.git/tree/xdiff-interface.c?id=HEAD#n198
-func IsBinary(data []byte) bool {
-	if len(data) > sniffLen {
-		data = data[:sniffLen]
-	}
-
-	if bytes.IndexByte(data, byte(0)) == -1 {
-		return false
-	}
-
-	return true
-}
-
-var vendorMatchers = substring.Or(
-	substring.Regexp(`(^|/)cache/`),
-	substring.Regexp(`^[Dd]ependencies/`),
-	substring.Regexp(`(^|/)dist/`),
-	substring.Regexp(`^deps/`),
-	substring.Regexp(`^tools/`),
-	substring.Regexp(`(^|/)configure$`),
-	substring.Regexp(`(^|/)config.guess$`),
-	substring.Regexp(`(^|/)config.sub$`),
-	substring.Regexp(`(^|/)aclocal.m4`),
-	substring.Regexp(`(^|/)libtool.m4`),
-	substring.Regexp(`(^|/)ltoptions.m4`),
-	substring.Regexp(`(^|/)ltsugar.m4`),
-	substring.Regexp(`(^|/)ltversion.m4`),
-	substring.Regexp(`(^|/)lt~obsolete.m4`),
-	)
-
-var documentationMatchers = substring.Or(
-	substring.Regexp(`^docs?/`),
-	substring.Regexp(`(^|/)[Dd]ocumentation/`),
-	substring.Regexp(`(^|/)javadoc/`),
-	substring.Regexp(`^man/`),
-	substring.Regexp(`^[Ee]xamples/`),
-	substring.Regexp(`(^|/)CHANGE(S|LOG)?(\.|$)`),
-	substring.Regexp(`(^|/)CONTRIBUTING(\.|$)`),
-	substring.Regexp(`(^|/)COPYING(\.|$)`),
-	substring.Regexp(`(^|/)INSTALL(\.|$)`),
-	substring.Regexp(`(^|/)LICEN[CS]E(\.|$)`),
-	substring.Regexp(`(^|/)[Ll]icen[cs]e(\.|$)`),
-	substring.Regexp(`(^|/)README(\.|$)`),
-	substring.Regexp(`(^|/)[Rr]eadme(\.|$)`),
-	substring.Regexp(`^[Ss]amples/`),
-)
-
-var configurationLanguages = map[string]bool{
-	"XML": true, "JSON": true, "TOML": true, "YAML": true, "INI": true, "SQL": true,
-}

internal/code-generator/generator/test_files/utils.test.go.tmpl

@@ -1,84 +0,0 @@
-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 (
-	"bytes"
-	"path/filepath"
-	"strings"
-
-	"gopkg.in/toqueteos/substring.v1"
-)
-
-func IsAuxiliaryLanguage(lang string) bool {
-	_, ok := auxiliaryLanguages[lang]
-	return ok
-}
-
-func IsConfiguration(path string) bool {
-	lang, _ := GetLanguageByExtension(path)
-	_, is := configurationLanguages[lang]
-
-	return is
-}
-
-func IsDotFile(path string) bool {
-	return strings.HasPrefix(filepath.Base(path), ".")
-}
-
-func IsVendor(path string) bool {
-	return findIndex(path, vendorMatchers) >= 0
-}
-
-func IsDocumentation(path string) bool {
-	return findIndex(path, documentationMatchers) >= 0
-}
-
-func findIndex(path string, matchers substring.StringsMatcher) int {
-	return matchers.MatchIndex(path)
-}
-
-const sniffLen = 8000
-
-//IsBinary detects if data is a binary value based on:
-//http://git.kernel.org/cgit/git/git.git/tree/xdiff-interface.c?id=HEAD#n198
-func IsBinary(data []byte) bool {
-	if len(data) > sniffLen {
-		data = data[:sniffLen]
-	}
-
-	if bytes.IndexByte(data, byte(0)) == -1 {
-		return false
-	}
-
-	return true
-}
-
-var vendorMatchers = substring.Or(
-	{{range $regexp := . -}}
-	substring.Regexp(`{{ $regexp }}`),
-	{{end -}}
-)
-
-var documentationMatchers = substring.Or(
-	substring.Regexp(`^docs?/`),
-	substring.Regexp(`(^|/)[Dd]ocumentation/`),
-	substring.Regexp(`(^|/)javadoc/`),
-	substring.Regexp(`^man/`),
-	substring.Regexp(`^[Ee]xamples/`),
-	substring.Regexp(`(^|/)CHANGE(S|LOG)?(\.|$)`),
-	substring.Regexp(`(^|/)CONTRIBUTING(\.|$)`),
-	substring.Regexp(`(^|/)COPYING(\.|$)`),
-	substring.Regexp(`(^|/)INSTALL(\.|$)`),
-	substring.Regexp(`(^|/)LICEN[CS]E(\.|$)`),
-	substring.Regexp(`(^|/)[Ll]icen[cs]e(\.|$)`),
-	substring.Regexp(`(^|/)README(\.|$)`),
-	substring.Regexp(`(^|/)[Rr]eadme(\.|$)`),
-	substring.Regexp(`^[Ss]amples/`),
-)
-
-var configurationLanguages = map[string]bool{
-	"XML": true, "JSON": true, "TOML": true, "YAML": true, "INI": true, "SQL": true,
-}

internal/code-generator/generator/test_files/vendor.gold

@@ -0,0 +1,24 @@
+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 "gopkg.in/toqueteos/substring.v1"
+
+var vendorMatchers = substring.Or(
+	substring.Regexp(`(^|/)cache/`),
+	substring.Regexp(`^[Dd]ependencies/`),
+	substring.Regexp(`(^|/)dist/`),
+	substring.Regexp(`^deps/`),
+	substring.Regexp(`^tools/`),
+	substring.Regexp(`(^|/)configure$`),
+	substring.Regexp(`(^|/)config.guess$`),
+	substring.Regexp(`(^|/)config.sub$`),
+	substring.Regexp(`(^|/)aclocal.m4`),
+	substring.Regexp(`(^|/)libtool.m4`),
+	substring.Regexp(`(^|/)ltoptions.m4`),
+	substring.Regexp(`(^|/)ltsugar.m4`),
+	substring.Regexp(`(^|/)ltversion.m4`),
+	substring.Regexp(`(^|/)lt~obsolete.m4`),
+	)

internal/code-generator/generator/test_files/vendor.test.go.tmpl

@@ -0,0 +1,13 @@
+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 "gopkg.in/toqueteos/substring.v1"
+
+var vendorMatchers = substring.Or(
+	{{range $regexp := . -}}
+	substring.Regexp(`{{ $regexp }}`),
+	{{end -}}
+)

internal/code-generator/main.go

@@ -19,9 +19,9 @@ const (
 	contentTmpl     = "content.go.tmpl"
 
 	vendorYAML     = ".linguist/lib/linguist/vendor.yml"
-	utilsFile     = "utils.go"
+	vendorFile     = "vendor_matchers.go"
-	utilsTmplPath = "internal/code-generator/assets/utils.go.tmpl"
+	vendorTmplPath = "internal/code-generator/assets/vendor.go.tmpl"
-	utilsTmpl     = "utils.go.tmpl"
+	vendorTmpl     = "vendor.go.tmpl"
 
 	commitPath = ".git/refs/heads/master"
 )
@@ -40,11 +40,7 @@ func main() {
 		log.Println(err)
 	}
 
-	if err := generator.FromFile(vendorYAML, utilsFile, utilsTmplPath, utilsTmpl, commit, generator.Vendor); err != nil {
+	if err := generator.FromFile(vendorYAML, vendorFile, vendorTmplPath, vendorTmpl, commit, generator.Vendor); err != nil {
-		log.Println(err)
-	}
-
-	if err := generator.FromFile(vendorYAML, utilsFile, utilsTmplPath, utilsTmpl, commit, generator.Vendor); err != nil {
 		log.Println(err)
 	}
 }

utils.go

@@ -1,9 +1,5 @@
 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: dae33dc2b20cddc85d1300435c3be7118a7115a9
-
 import (
 	"bytes"
 	"path/filepath"
@@ -56,155 +52,6 @@ func IsBinary(data []byte) bool {
 	return true
 }
 
-var vendorMatchers = substring.Or(
-	substring.Regexp(`(^|/)cache/`),
-	substring.Regexp(`^[Dd]ependencies/`),
-	substring.Regexp(`(^|/)dist/`),
-	substring.Regexp(`^deps/`),
-	substring.Regexp(`^tools/`),
-	substring.Regexp(`(^|/)configure$`),
-	substring.Regexp(`(^|/)config.guess$`),
-	substring.Regexp(`(^|/)config.sub$`),
-	substring.Regexp(`(^|/)aclocal.m4`),
-	substring.Regexp(`(^|/)libtool.m4`),
-	substring.Regexp(`(^|/)ltoptions.m4`),
-	substring.Regexp(`(^|/)ltsugar.m4`),
-	substring.Regexp(`(^|/)ltversion.m4`),
-	substring.Regexp(`(^|/)lt~obsolete.m4`),
-	substring.Regexp(`cpplint.py`),
-	substring.Regexp(`node_modules/`),
-	substring.Regexp(`bower_components/`),
-	substring.Regexp(`^rebar$`),
-	substring.Regexp(`erlang.mk`),
-	substring.Regexp(`Godeps/_workspace/`),
-	substring.Regexp(`.indent.pro`),
-	substring.Regexp(`(\.|-)min\.(js|css)$`),
-	substring.Regexp(`([^\s]*)import\.(css|less|scss|styl)$`),
-	substring.Regexp(`(^|/)bootstrap([^.]*)\.(js|css|less|scss|styl)$`),
-	substring.Regexp(`(^|/)custom\.bootstrap([^\s]*)(js|css|less|scss|styl)$`),
-	substring.Regexp(`(^|/)font-awesome\.(css|less|scss|styl)$`),
-	substring.Regexp(`(^|/)foundation\.(css|less|scss|styl)$`),
-	substring.Regexp(`(^|/)normalize\.(css|less|scss|styl)$`),
-	substring.Regexp(`(^|/)[Bb]ourbon/.*\.(css|less|scss|styl)$`),
-	substring.Regexp(`(^|/)animate\.(css|less|scss|styl)$`),
-	substring.Regexp(`third[-_]?party/`),
-	substring.Regexp(`3rd[-_]?party/`),
-	substring.Regexp(`vendors?/`),
-	substring.Regexp(`extern(al)?/`),
-	substring.Regexp(`(^|/)[Vv]+endor/`),
-	substring.Regexp(`^debian/`),
-	substring.Regexp(`run.n$`),
-	substring.Regexp(`bootstrap-datepicker/`),
-	substring.Regexp(`(^|/)jquery([^.]*)\.js$`),
-	substring.Regexp(`(^|/)jquery\-\d\.\d+(\.\d+)?\.js$`),
-	substring.Regexp(`(^|/)jquery\-ui(\-\d\.\d+(\.\d+)?)?(\.\w+)?\.(js|css)$`),
-	substring.Regexp(`(^|/)jquery\.(ui|effects)\.([^.]*)\.(js|css)$`),
-	substring.Regexp(`jquery.fn.gantt.js`),
-	substring.Regexp(`jquery.fancybox.(js|css)`),
-	substring.Regexp(`fuelux.js`),
-	substring.Regexp(`(^|/)jquery\.fileupload(-\w+)?\.js$`),
-	substring.Regexp(`(^|/)slick\.\w+.js$`),
-	substring.Regexp(`(^|/)Leaflet\.Coordinates-\d+\.\d+\.\d+\.src\.js$`),
-	substring.Regexp(`leaflet.draw-src.js`),
-	substring.Regexp(`leaflet.draw.css`),
-	substring.Regexp(`Control.FullScreen.css`),
-	substring.Regexp(`Control.FullScreen.js`),
-	substring.Regexp(`leaflet.spin.js`),
-	substring.Regexp(`wicket-leaflet.js`),
-	substring.Regexp(`.sublime-project`),
-	substring.Regexp(`.sublime-workspace`),
-	substring.Regexp(`(^|/)prototype(.*)\.js$`),
-	substring.Regexp(`(^|/)effects\.js$`),
-	substring.Regexp(`(^|/)controls\.js$`),
-	substring.Regexp(`(^|/)dragdrop\.js$`),
-	substring.Regexp(`(.*?)\.d\.ts$`),
-	substring.Regexp(`(^|/)mootools([^.]*)\d+\.\d+.\d+([^.]*)\.js$`),
-	substring.Regexp(`(^|/)dojo\.js$`),
-	substring.Regexp(`(^|/)MochiKit\.js$`),
-	substring.Regexp(`(^|/)yahoo-([^.]*)\.js$`),
-	substring.Regexp(`(^|/)yui([^.]*)\.js$`),
-	substring.Regexp(`(^|/)ckeditor\.js$`),
-	substring.Regexp(`(^|/)tiny_mce([^.]*)\.js$`),
-	substring.Regexp(`(^|/)tiny_mce/(langs|plugins|themes|utils)`),
-	substring.Regexp(`(^|/)ace-builds/`),
-	substring.Regexp(`(^|/)fontello(.*?)\.css$`),
-	substring.Regexp(`(^|/)MathJax/`),
-	substring.Regexp(`(^|/)Chart\.js$`),
-	substring.Regexp(`(^|/)[Cc]ode[Mm]irror/(\d+\.\d+/)?(lib|mode|theme|addon|keymap|demo)`),
-	substring.Regexp(`(^|/)shBrush([^.]*)\.js$`),
-	substring.Regexp(`(^|/)shCore\.js$`),
-	substring.Regexp(`(^|/)shLegacy\.js$`),
-	substring.Regexp(`(^|/)angular([^.]*)\.js$`),
-	substring.Regexp(`(^|\/)d3(\.v\d+)?([^.]*)\.js$`),
-	substring.Regexp(`(^|/)react(-[^.]*)?\.js$`),
-	substring.Regexp(`(^|/)modernizr\-\d\.\d+(\.\d+)?\.js$`),
-	substring.Regexp(`(^|/)modernizr\.custom\.\d+\.js$`),
-	substring.Regexp(`(^|/)knockout-(\d+\.){3}(debug\.)?js$`),
-	substring.Regexp(`(^|/)docs?/_?(build|themes?|templates?|static)/`),
-	substring.Regexp(`(^|/)admin_media/`),
-	substring.Regexp(`(^|/)env/`),
-	substring.Regexp(`^fabfile\.py$`),
-	substring.Regexp(`^waf$`),
-	substring.Regexp(`^.osx$`),
-	substring.Regexp(`\.xctemplate/`),
-	substring.Regexp(`\.imageset/`),
-	substring.Regexp(`^Carthage/`),
-	substring.Regexp(`^Pods/`),
-	substring.Regexp(`(^|/)Sparkle/`),
-	substring.Regexp(`Crashlytics.framework/`),
-	substring.Regexp(`Fabric.framework/`),
-	substring.Regexp(`BuddyBuildSDK.framework/`),
-	substring.Regexp(`Realm.framework`),
-	substring.Regexp(`RealmSwift.framework`),
-	substring.Regexp(`gitattributes$`),
-	substring.Regexp(`gitignore$`),
-	substring.Regexp(`gitmodules$`),
-	substring.Regexp(`(^|/)gradlew$`),
-	substring.Regexp(`(^|/)gradlew\.bat$`),
-	substring.Regexp(`(^|/)gradle/wrapper/`),
-	substring.Regexp(`-vsdoc\.js$`),
-	substring.Regexp(`\.intellisense\.js$`),
-	substring.Regexp(`(^|/)jquery([^.]*)\.validate(\.unobtrusive)?\.js$`),
-	substring.Regexp(`(^|/)jquery([^.]*)\.unobtrusive\-ajax\.js$`),
-	substring.Regexp(`(^|/)[Mm]icrosoft([Mm]vc)?([Aa]jax|[Vv]alidation)(\.debug)?\.js$`),
-	substring.Regexp(`^[Pp]ackages\/.+\.\d+\/`),
-	substring.Regexp(`(^|/)extjs/.*?\.js$`),
-	substring.Regexp(`(^|/)extjs/.*?\.xml$`),
-	substring.Regexp(`(^|/)extjs/.*?\.txt$`),
-	substring.Regexp(`(^|/)extjs/.*?\.html$`),
-	substring.Regexp(`(^|/)extjs/.*?\.properties$`),
-	substring.Regexp(`(^|/)extjs/.sencha/`),
-	substring.Regexp(`(^|/)extjs/docs/`),
-	substring.Regexp(`(^|/)extjs/builds/`),
-	substring.Regexp(`(^|/)extjs/cmd/`),
-	substring.Regexp(`(^|/)extjs/examples/`),
-	substring.Regexp(`(^|/)extjs/locale/`),
-	substring.Regexp(`(^|/)extjs/packages/`),
-	substring.Regexp(`(^|/)extjs/plugins/`),
-	substring.Regexp(`(^|/)extjs/resources/`),
-	substring.Regexp(`(^|/)extjs/src/`),
-	substring.Regexp(`(^|/)extjs/welcome/`),
-	substring.Regexp(`(^|/)html5shiv\.js$`),
-	substring.Regexp(`^[Tt]ests?/fixtures/`),
-	substring.Regexp(`^[Ss]pecs?/fixtures/`),
-	substring.Regexp(`(^|/)cordova([^.]*)\.js$`),
-	substring.Regexp(`(^|/)cordova\-\d\.\d(\.\d)?\.js$`),
-	substring.Regexp(`foundation(\..*)?\.js$`),
-	substring.Regexp(`^Vagrantfile$`),
-	substring.Regexp(`.[Dd][Ss]_[Ss]tore$`),
-	substring.Regexp(`^vignettes/`),
-	substring.Regexp(`^inst/extdata/`),
-	substring.Regexp(`octicons.css`),
-	substring.Regexp(`sprockets-octicons.scss`),
-	substring.Regexp(`(^|/)activator$`),
-	substring.Regexp(`(^|/)activator\.bat$`),
-	substring.Regexp(`proguard.pro`),
-	substring.Regexp(`proguard-rules.pro`),
-	substring.Regexp(`^puphpet/`),
-	substring.Regexp(`(^|/)\.google_apis/`),
-	substring.Regexp(`^Jenkinsfile$`),
-)
-
 var documentationMatchers = substring.Or(
 	substring.Regexp(`^docs?/`),
 	substring.Regexp(`(^|/)[Dd]ocumentation/`),

vendor_matchers.go

@@ -0,0 +1,156 @@
+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: dae33dc2b20cddc85d1300435c3be7118a7115a9
+
+import "gopkg.in/toqueteos/substring.v1"
+
+var vendorMatchers = substring.Or(
+	substring.Regexp(`(^|/)cache/`),
+	substring.Regexp(`^[Dd]ependencies/`),
+	substring.Regexp(`(^|/)dist/`),
+	substring.Regexp(`^deps/`),
+	substring.Regexp(`^tools/`),
+	substring.Regexp(`(^|/)configure$`),
+	substring.Regexp(`(^|/)config.guess$`),
+	substring.Regexp(`(^|/)config.sub$`),
+	substring.Regexp(`(^|/)aclocal.m4`),
+	substring.Regexp(`(^|/)libtool.m4`),
+	substring.Regexp(`(^|/)ltoptions.m4`),
+	substring.Regexp(`(^|/)ltsugar.m4`),
+	substring.Regexp(`(^|/)ltversion.m4`),
+	substring.Regexp(`(^|/)lt~obsolete.m4`),
+	substring.Regexp(`cpplint.py`),
+	substring.Regexp(`node_modules/`),
+	substring.Regexp(`bower_components/`),
+	substring.Regexp(`^rebar$`),
+	substring.Regexp(`erlang.mk`),
+	substring.Regexp(`Godeps/_workspace/`),
+	substring.Regexp(`.indent.pro`),
+	substring.Regexp(`(\.|-)min\.(js|css)$`),
+	substring.Regexp(`([^\s]*)import\.(css|less|scss|styl)$`),
+	substring.Regexp(`(^|/)bootstrap([^.]*)\.(js|css|less|scss|styl)$`),
+	substring.Regexp(`(^|/)custom\.bootstrap([^\s]*)(js|css|less|scss|styl)$`),
+	substring.Regexp(`(^|/)font-awesome\.(css|less|scss|styl)$`),
+	substring.Regexp(`(^|/)foundation\.(css|less|scss|styl)$`),
+	substring.Regexp(`(^|/)normalize\.(css|less|scss|styl)$`),
+	substring.Regexp(`(^|/)[Bb]ourbon/.*\.(css|less|scss|styl)$`),
+	substring.Regexp(`(^|/)animate\.(css|less|scss|styl)$`),
+	substring.Regexp(`third[-_]?party/`),
+	substring.Regexp(`3rd[-_]?party/`),
+	substring.Regexp(`vendors?/`),
+	substring.Regexp(`extern(al)?/`),
+	substring.Regexp(`(^|/)[Vv]+endor/`),
+	substring.Regexp(`^debian/`),
+	substring.Regexp(`run.n$`),
+	substring.Regexp(`bootstrap-datepicker/`),
+	substring.Regexp(`(^|/)jquery([^.]*)\.js$`),
+	substring.Regexp(`(^|/)jquery\-\d\.\d+(\.\d+)?\.js$`),
+	substring.Regexp(`(^|/)jquery\-ui(\-\d\.\d+(\.\d+)?)?(\.\w+)?\.(js|css)$`),
+	substring.Regexp(`(^|/)jquery\.(ui|effects)\.([^.]*)\.(js|css)$`),
+	substring.Regexp(`jquery.fn.gantt.js`),
+	substring.Regexp(`jquery.fancybox.(js|css)`),
+	substring.Regexp(`fuelux.js`),
+	substring.Regexp(`(^|/)jquery\.fileupload(-\w+)?\.js$`),
+	substring.Regexp(`(^|/)slick\.\w+.js$`),
+	substring.Regexp(`(^|/)Leaflet\.Coordinates-\d+\.\d+\.\d+\.src\.js$`),
+	substring.Regexp(`leaflet.draw-src.js`),
+	substring.Regexp(`leaflet.draw.css`),
+	substring.Regexp(`Control.FullScreen.css`),
+	substring.Regexp(`Control.FullScreen.js`),
+	substring.Regexp(`leaflet.spin.js`),
+	substring.Regexp(`wicket-leaflet.js`),
+	substring.Regexp(`.sublime-project`),
+	substring.Regexp(`.sublime-workspace`),
+	substring.Regexp(`(^|/)prototype(.*)\.js$`),
+	substring.Regexp(`(^|/)effects\.js$`),
+	substring.Regexp(`(^|/)controls\.js$`),
+	substring.Regexp(`(^|/)dragdrop\.js$`),
+	substring.Regexp(`(.*?)\.d\.ts$`),
+	substring.Regexp(`(^|/)mootools([^.]*)\d+\.\d+.\d+([^.]*)\.js$`),
+	substring.Regexp(`(^|/)dojo\.js$`),
+	substring.Regexp(`(^|/)MochiKit\.js$`),
+	substring.Regexp(`(^|/)yahoo-([^.]*)\.js$`),
+	substring.Regexp(`(^|/)yui([^.]*)\.js$`),
+	substring.Regexp(`(^|/)ckeditor\.js$`),
+	substring.Regexp(`(^|/)tiny_mce([^.]*)\.js$`),
+	substring.Regexp(`(^|/)tiny_mce/(langs|plugins|themes|utils)`),
+	substring.Regexp(`(^|/)ace-builds/`),
+	substring.Regexp(`(^|/)fontello(.*?)\.css$`),
+	substring.Regexp(`(^|/)MathJax/`),
+	substring.Regexp(`(^|/)Chart\.js$`),
+	substring.Regexp(`(^|/)[Cc]ode[Mm]irror/(\d+\.\d+/)?(lib|mode|theme|addon|keymap|demo)`),
+	substring.Regexp(`(^|/)shBrush([^.]*)\.js$`),
+	substring.Regexp(`(^|/)shCore\.js$`),
+	substring.Regexp(`(^|/)shLegacy\.js$`),
+	substring.Regexp(`(^|/)angular([^.]*)\.js$`),
+	substring.Regexp(`(^|\/)d3(\.v\d+)?([^.]*)\.js$`),
+	substring.Regexp(`(^|/)react(-[^.]*)?\.js$`),
+	substring.Regexp(`(^|/)modernizr\-\d\.\d+(\.\d+)?\.js$`),
+	substring.Regexp(`(^|/)modernizr\.custom\.\d+\.js$`),
+	substring.Regexp(`(^|/)knockout-(\d+\.){3}(debug\.)?js$`),
+	substring.Regexp(`(^|/)docs?/_?(build|themes?|templates?|static)/`),
+	substring.Regexp(`(^|/)admin_media/`),
+	substring.Regexp(`(^|/)env/`),
+	substring.Regexp(`^fabfile\.py$`),
+	substring.Regexp(`^waf$`),
+	substring.Regexp(`^.osx$`),
+	substring.Regexp(`\.xctemplate/`),
+	substring.Regexp(`\.imageset/`),
+	substring.Regexp(`^Carthage/`),
+	substring.Regexp(`^Pods/`),
+	substring.Regexp(`(^|/)Sparkle/`),
+	substring.Regexp(`Crashlytics.framework/`),
+	substring.Regexp(`Fabric.framework/`),
+	substring.Regexp(`BuddyBuildSDK.framework/`),
+	substring.Regexp(`Realm.framework`),
+	substring.Regexp(`RealmSwift.framework`),
+	substring.Regexp(`gitattributes$`),
+	substring.Regexp(`gitignore$`),
+	substring.Regexp(`gitmodules$`),
+	substring.Regexp(`(^|/)gradlew$`),
+	substring.Regexp(`(^|/)gradlew\.bat$`),
+	substring.Regexp(`(^|/)gradle/wrapper/`),
+	substring.Regexp(`-vsdoc\.js$`),
+	substring.Regexp(`\.intellisense\.js$`),
+	substring.Regexp(`(^|/)jquery([^.]*)\.validate(\.unobtrusive)?\.js$`),
+	substring.Regexp(`(^|/)jquery([^.]*)\.unobtrusive\-ajax\.js$`),
+	substring.Regexp(`(^|/)[Mm]icrosoft([Mm]vc)?([Aa]jax|[Vv]alidation)(\.debug)?\.js$`),
+	substring.Regexp(`^[Pp]ackages\/.+\.\d+\/`),
+	substring.Regexp(`(^|/)extjs/.*?\.js$`),
+	substring.Regexp(`(^|/)extjs/.*?\.xml$`),
+	substring.Regexp(`(^|/)extjs/.*?\.txt$`),
+	substring.Regexp(`(^|/)extjs/.*?\.html$`),
+	substring.Regexp(`(^|/)extjs/.*?\.properties$`),
+	substring.Regexp(`(^|/)extjs/.sencha/`),
+	substring.Regexp(`(^|/)extjs/docs/`),
+	substring.Regexp(`(^|/)extjs/builds/`),
+	substring.Regexp(`(^|/)extjs/cmd/`),
+	substring.Regexp(`(^|/)extjs/examples/`),
+	substring.Regexp(`(^|/)extjs/locale/`),
+	substring.Regexp(`(^|/)extjs/packages/`),
+	substring.Regexp(`(^|/)extjs/plugins/`),
+	substring.Regexp(`(^|/)extjs/resources/`),
+	substring.Regexp(`(^|/)extjs/src/`),
+	substring.Regexp(`(^|/)extjs/welcome/`),
+	substring.Regexp(`(^|/)html5shiv\.js$`),
+	substring.Regexp(`^[Tt]ests?/fixtures/`),
+	substring.Regexp(`^[Ss]pecs?/fixtures/`),
+	substring.Regexp(`(^|/)cordova([^.]*)\.js$`),
+	substring.Regexp(`(^|/)cordova\-\d\.\d(\.\d)?\.js$`),
+	substring.Regexp(`foundation(\..*)?\.js$`),
+	substring.Regexp(`^Vagrantfile$`),
+	substring.Regexp(`.[Dd][Ss]_[Ss]tore$`),
+	substring.Regexp(`^vignettes/`),
+	substring.Regexp(`^inst/extdata/`),
+	substring.Regexp(`octicons.css`),
+	substring.Regexp(`sprockets-octicons.scss`),
+	substring.Regexp(`(^|/)activator$`),
+	substring.Regexp(`(^|/)activator\.bat$`),
+	substring.Regexp(`proguard.pro`),
+	substring.Regexp(`proguard-rules.pro`),
+	substring.Regexp(`^puphpet/`),
+	substring.Regexp(`(^|/)\.google_apis/`),
+	substring.Regexp(`^Jenkinsfile$`),
+)