feat: add strict mode

patrickhulce · patrickhulce · commit 8d6d09dd9c58 · 2017-04-16T09:56:48.000-07:00
Strict mode significantly speeds up runtime and limits false positives by restricting the notion of a used token to entire strings only, i.e. substrings and combinations of strings are not considered used.
diff --git a/lib/css-nuker.js b/lib/css-nuker.js
@@ -6,6 +6,7 @@ const DISABLE_COMMENT_REGEX = /\bnukecss:disable\b/
 class CssNuker {
   constructor(sources, parser, options = {}) {
     this._options = Object.assign({
+      strict: false,
       simple: false,
       amalgamate: false,
       whitelist: undefined,
diff --git a/lib/sources/html-source.js b/lib/sources/html-source.js
@@ -28,16 +28,19 @@ class HtmlSource extends ParsedSource {
       onopentag(name, attributes) {
         const attributeValues = Array.isArray(attributeKeys) ?
           attributeKeys.map(key => attributes[key]) : []
-        attributeValues.concat([name])
+        _(attributeValues)
+          .concat([name])
           .filter(candidate => typeof candidate === 'string')
+          .map(value => value.split(' '))
+          .flatten()
           .forEach(candidate => tokens.add(candidate.toLowerCase()))
       },
       ontext(value) {
         innerText = value
       },
       onclosetag(name) {
         if (name === 'script' && innerText) {
-          children.push(JsSource.from(innerText))
+          children.push(JsSource.from(innerText, options))
         }
       },
       onerror(err) {
diff --git a/lib/sources/js-source.js b/lib/sources/js-source.js
@@ -9,7 +9,9 @@ class JsSource extends ParsedSource {
   }
 
   _findToken(selector) {
-    if (selector.length === 0) {
+    if (this._options.strict) {
+      return false
+    } else if (selector.length === 0) {
       return true
     }
 
diff --git a/lib/sources/parsed-source.js b/lib/sources/parsed-source.js
@@ -16,6 +16,10 @@ class ParsedSource {
   }
 
   _findWholeSelectorInTokens(selector) {
+    if (this._options.strict) {
+      return false
+    }
+
     return this._tokensArray.find(token => ParsedSource.textContains(token, selector))
   }
 
diff --git a/test/sources/html-source.test.js b/test/sources/html-source.test.js
@@ -97,5 +97,27 @@ describe('sources/html-source.js', () => {
         expect(source).to.not.contain('myJsVar')
       })
     })
+
+    context('when options.strict=true', () => {
+      const html = `
+        <html>
+        <div class="first-class second-class"></div>
+        <script>const hello = 'my-class'</script>
+        </html>
+      `
+
+      const source = HtmlSource.from(html, {strict: true})
+
+      it('should find tokens as multiple classes', () => {
+        expect(source).to.contain('first-class')
+        expect(source).to.contain('second-class')
+      })
+
+      it('should find tokens in script', () => {
+        expect(source).to.contain('my-class')
+        expect(source).to.not.contain('my')
+        expect(source).to.not.contain('hello')
+      })
+    })
   })
 })
diff --git a/test/sources/js-source.test.js b/test/sources/js-source.test.js
@@ -87,6 +87,44 @@ describe('sources/js-source.js', () => {
       })
     })
 
+    context('when options.strict=true', () => {
+      const script = `
+        const myVar = 'the-class'
+        const otherVar = 'the-OtHer-cLass'
+        const html = '<div class="inner-class">Content</div>'
+        const dynamicVar = ['fa', 'icon'].join('-')
+        const obj = {inVisible: true, BLOCK__element: 1}
+        obj.additional_class = false
+      `
+
+      const source = JsSource.from(script, {strict: true})
+
+      it('should find tokens as strings', () => {
+        expect(source).to.contain('the-class')
+        expect(source).to.contain('the-other-class')
+        expect(source).to.contain('fa')
+        expect(source).to.contain('tHe-cLaSs')
+      })
+
+      it('should not find tokens within strings', () => {
+        expect(source).to.not.contain('div')
+        expect(source).to.not.contain('inner-class')
+      })
+
+      it('should not find tokens as combinations of strings', () => {
+        expect(source).to.not.contain('fa-icon')
+      })
+
+      it('should find tokens as object keys', () => {
+        expect(source).to.contain('invisible')
+        expect(source).to.contain('block__element')
+      })
+
+      it('should find tokens as object key assignment', () => {
+        expect(source).to.contain('additional_class')
+      })
+    })
+
     context('when script is malformed', () => {
       it('should not throw', () => {
         expect(() => JsSource.from('const foo != "whaaaa')).to.not.throw()

Original file line number	Diff line number	Diff line change
`@@ -9,7 +9,9 @@ class JsSource extends ParsedSource {`
`9`	`9`	`}`
`10`	`10`
`11`	`11`	`_findToken(selector) {`
`12`		`- if (selector.length === 0) {`
	`12`	`+ if (this._options.strict) {`
	`13`	`+ return false`
	`14`	`+ } else if (selector.length === 0) {`
`13`	`15`	`return true`
`14`	`16`	`}`
`15`	`17`
Original file line number	Diff line number	Diff line change
`@@ -16,6 +16,10 @@ class ParsedSource {`
`16`	`16`	`}`
`17`	`17`
`18`	`18`	`_findWholeSelectorInTokens(selector) {`
	`19`	`+ if (this._options.strict) {`
	`20`	`+ return false`
	`21`	`+ }`
	`22`	`+`
`19`	`23`	`return this._tokensArray.find(token => ParsedSource.textContains(token, selector))`
`20`	`24`	`}`
`21`	`25`