cover all

mipengine · Jun 27, 2017 · 136b77b · 136b77b
1 parent 7b4d94c
commit 136b77b
Show file tree

Hide file tree

Showing 19 changed files with 275 additions and 173 deletions.
diff --git a/.gitignore b/.gitignore
@@ -1,4 +1,5 @@
 node_modules/
+coverage/
 tmp/
 *.log
 *.swp

diff --git a/bin/http.js b/bin/http.js
@@ -2,10 +2,8 @@
 
 const program = require('commander')
 const pkg = require('../package.json')
-const Validator = require('..')
 const path = require('path')
-const http = require('http')
-const url = require('url')
+const mkServer = require('../src/http')
 
 /// /////////////////////////////////////////////////////////////////
 // Initialize CLI
@@ -31,56 +29,8 @@ if (program['conf']) {
   var configPath = path.resolve(process.cwd(), program['conf'])
   config = require(configPath)
 }
-var validator = config ? Validator(config) : Validator()
 
-/// /////////////////////////////////////////////////////////////////
-// Initialize Server
-var routes = {}
-const server = http.createServer((req, res) => {
-  var data = ''
-  req.on('data', (chunk) => {
-    data += chunk
-  })
-  req.on('end', () => {
-    var urlObj = url.parse(req.url, true)
-    var route = routes[urlObj.pathname]
-    if (!route) {
-      res.statusCode = 404
-      res.end()
-    }
-    req.params = urlObj.query
-    req.body = data
-    route(req, res)
-  })
-})
-
-/*
- * Router: /
- */
-var usage = [
-  'Usage:',
-  '    GET / for help',
-  '    POST /validate with HTML body to validate',
-  '    POST /validate?fast=true with HTML body to fast validate'
-].join('\n')
-routes['/'] = function (req, res) {
-  res.end(usage)
-}
-
-/*
- * Router: /validate
- */
-routes['/validate'] = function (req, res) {
-  var fastMode = (req.params.fast === 'true')
-  var result = validator.validate(req.body, {fastMode: fastMode})
-  res.setHeader('Content-Type', 'application/json')
-  res.end(JSON.stringify(result, null, 4))
-}
-
-/// ///////////////////////////////////////////////////////////////
-// Return Server
+var server = mkServer(config)
 server.listen(program['port'], program['host'], function () {
   console.log(`[http] listening to ${program['host']}:${program['port']}`)
 })
-
-module.exports = server
diff --git a/examples/http.sh b/examples/http.sh
@@ -0,0 +1,4 @@
+#!/usr/bin/env bash
+# start http server first: node ./bin/http.js
+
+curl localhost:4444/validate -X POST --data '<html></html>'
diff --git a/src/engine.js b/src/engine.js
@@ -28,7 +28,6 @@ function Engine (rules) {
 }
 
 Engine.prototype.setRules = function (rules) {
-  logger.debug('setting rules %J', rules)
   assert(typeof rules === 'object',
         'rules object expected, but ' + (typeof rules) + ' found')
   this.config = ruleParser.mkConfig(rules)
@@ -214,26 +213,18 @@ module.exports = function (rules) {
 }
 
 function behaveBuggyAsTheCPPVersion (doc, errorGenertor) {
-  try {
-    var head = findFirstTagChild(findFirstTagChild(doc))
-    if (head.tagName !== 'head') {
-      return
+  var head = findFirstTagChild(findFirstTagChild(doc))
+  var noscriptAppeared = false
+
+  head.childNodes.forEach((node) => {
+    if (node.tagName === 'noscript') {
+      noscriptAppeared = true
+    } else if (node.tagName && noscriptAppeared) {
+      // it's a tag after noscript
+      var err = ERR.INVALID_NOSCRIPT
+      errorGenertor(err)
     }
-
-    var noscriptAppeared = false
-    head.childNodes.forEach((node) => {
-      if (node.tagName === 'noscript') {
-        noscriptAppeared = true
-      } else if (node.tagName && noscriptAppeared) {
-        // it's a tag after noscript
-        var err = ERR.INVALID_NOSCRIPT
-        errorGenertor(err)
-      }
-    })
-  } catch (e) {
-    // let it be...
-    console.log('behaveBuggyAsTheCPPVersion error:', e)
-  }
+  })
 }
 
 function findFirstTagChild (parent) {

diff --git a/src/http.js b/src/http.js
@@ -0,0 +1,50 @@
+const Validator = require('..')
+const url = require('url')
+const http = require('http')
+const usage = [
+  'Usage:',
+  '    GET / for help',
+  '    POST /validate with HTML body to validate',
+  '    POST /validate?fast=true with HTML body to fast validate'
+].join('\n')
+
+function mkServer (rules) {
+  var validator = rules ? Validator(rules) : Validator()
+  var routes = {}
+  var server = mkRouter(routes)
+
+  routes['/'] = function (req, res) {
+    res.end(usage)
+  }
+
+  routes['/validate'] = function (req, res) {
+    var fastMode = (req.params.fast === 'true')
+    var result = validator.validate(req.body, {fastMode: fastMode})
+    res.setHeader('Content-Type', 'application/json')
+    res.end(JSON.stringify(result, null, 4))
+  }
+
+  return server
+}
+
+function mkRouter (routes) {
+  return http.createServer((req, res) => {
+    var data = ''
+    req.on('data', (chunk) => {
+      data += chunk
+    })
+    req.on('end', () => {
+      var urlObj = url.parse(req.url, true)
+      var route = routes[urlObj.pathname]
+      if (!route) {
+        res.statusCode = 404
+        return res.end()
+      }
+      req.params = urlObj.query
+      req.body = data
+      route(req, res)
+    })
+  })
+}
+
+module.exports = mkServer
diff --git a/src/logger.js b/src/logger.js
@@ -39,7 +39,7 @@ function match (root, path) {
  * create logger method with STDOUT instance and trace ID
  * @param {Function} output STDOUT instance
  * @param {String} id trace ID
- * legacy:
+ * legal:
  *      createWith(console.log.bind(console), 'mip:cache:getCache')
  */
 function createWith (output, id) {
@@ -55,7 +55,7 @@ function createWith (output, id) {
  * Pad number to 2-digit.
  * @param {Number} n The number
  * @return {String} 2-digit number string
- * legacy:
+ * legal:
  *      pad(2)      // 02
  *      pad(22)     // 22
  */
@@ -66,7 +66,7 @@ function pad (n) {
 /*
  * Generate a timestamp from current time
  * @return {String} the timestamp string
- * legacy:
+ * legal:
  *      timestamp()    // "2016/09/27-17:31:22"
  */
 function timestamp () {
@@ -88,7 +88,7 @@ var formatRegExp = /%[sdjJ%]/g
  * format arguments to a single string
  * @param {String} f Optional, the format string
  * @return {String} the formated string
- * legacy:
+ * legal:
  *      format('%s', 'foo', 'bar')  // foo bar
  *      format('%d', '200', 'bar')  // 200 bar
  *      format('%j', {foo: 'bar'}, 'bar')
@@ -104,8 +104,7 @@ function format (f) {
   var str = ''
   if (isString(f)) {
     i++
-    str += String(f).replace(formatRegExp, function (x) {
-      if (i >= len) return x
+    str += f.replace(formatRegExp, function (x) {
       switch (x) {
         case '%%':
           return '%'

diff --git a/src/matcher.js b/src/matcher.js
@@ -37,7 +37,7 @@ function matchValue (src, target) {
  * object match: match src with target
  * @param {Object} src the object to match
  * @param {Object} target the object to match with
- * legacy:
+ * legal:
  *      match({
  *          id: 'modal-user'
  *      }, {
@@ -58,7 +58,7 @@ function match (src, target) {
  * attributes match
  * @param {ASTNode} node the node of which attributes will be matched
  * @param {Object} target the attribute list object to match with
- * legacy:
+ * legal:
  *      matchAttrs(node, {
  *          style: 'color:red',
  *          id: '/mip-.+/'
@@ -76,7 +76,7 @@ function matchAttrs (node, target) {
  * match ancestor name
  * @param {ASTNode} node the node of which parent will be matched
  * @param {String} ancestorNodeName string or regex-like string to match with
- * legacy:
+ * legal:
  *      matchAncestor(node, 'form');
  *      matchAncestor(node, '/form|div|section/'
  */
@@ -94,7 +94,7 @@ function matchAncestor (node, ancestorNodeName) {
  * match parent name
  * @param {ASTNode} node the node of which parent will be matched
  * @param {String} parentNodeName string or regex-like string to match with
- * legacy:
+ * legal:
  *      matchParent(node, 'form');
  *      matchParent(node, '/form|div|section/'
  */
@@ -112,7 +112,7 @@ function matchParent (node, parentNodeName) {
  * match descendant node name
  * @param {ASTNode} node the node of which parent will be matched
  * @param {String} descendantNodeName string or regex-like string to match with
- * legacy:
+ * legal:
  *      matchDescendant(node, 'form');
  *      matchDescendant(node, '/form|div|section/'
  */
@@ -127,7 +127,7 @@ function matchDescendant (node, descendantNodeName) {
  * nomatch descendant node name
  * @param {ASTNode} node the node of which parent will be matched
  * @param {String} descendantNodeName string or regex-like string to match with
- * legacy:
+ * legal:
  *      nomatchDescendant(node, 'form');
  *      nomatchDescendant(node, '/form|div|section/'
  */
@@ -141,8 +141,7 @@ function nomatchDescendant (node, descendantNodeName) {
 }
 
 function dfsUntil (node, predict) {
-  var children = node.childNodes || []
-  return predict(node) || children.some(child => dfsUntil(child, predict))
+  return predict(node) || node.childNodes.some(child => dfsUntil(child, predict))
 }
 
 /*
@@ -163,7 +162,7 @@ function createNode (nodeName, attrsObj) {
 
 /*
  * Generate a fingerprint for given nodeName and attributes
- * legacy:
+ * legal:
  *      // returns: <div id="modal">
  *      fingerprintByObject('div', {id: 'modal'});
  */
@@ -191,7 +190,7 @@ function fingerprintByTag (node) {
  * Match tagnames from the given HTML
  * @param {Array} tagNames
  * @param {String} html
- * legacy: matchTagNames(['div', 'head', 'iframe'], '<div><iframe></div>')
+ * legal: matchTagNames(['div', 'head', 'iframe'], '<div><iframe></div>')
  */
 function matchTagNames (tagNames, html) {
   var tagsStr = tagNames.join('|')

diff --git a/src/rule-parser.js b/src/rule-parser.js
@@ -88,10 +88,10 @@ function mkConfig (rules) {
   }
 }
 
-function typedRules (config) {
-  var rules = TYPED_RULES[config]
+function typedRules (type) {
+  var rules = TYPED_RULES[type]
   if (!rules) {
-    throw new Error('rules not found for config ' + config)
+    throw new Error('rules not found for type ' + type)
   }
   return rules
 }

diff --git a/src/validators/disallowed_tag.js b/src/validators/disallowed_tag.js
@@ -22,7 +22,9 @@ function validatePolyfill (error, engine) {
     var tagName = tag.match(/\w+/)
     var rules = _.get(engine.config.nodes, `${tagName}`)
     _.map(rules, rule => {
-      if (!rule.disallow) return
+      if (!rule.disallow) {
+        return
+      }
 
       var err = ERR.DISALLOWED_TAG
       error(err, tag)

diff --git a/src/validators/duplicate_unique_tag.js b/src/validators/duplicate_unique_tag.js
@@ -12,13 +12,11 @@ exports.onBegin = function (error, engine) {
 
   _.forOwn(engine.config.nodes, (rules, ruleName) => {
     _.map(rules, rule => {
-      if (rule.duplicate) {
-        _.map(rule.duplicate, pattern => {
-          var fingerprint = matcher.fingerprintByObject(ruleName, pattern)
-          var hash = fingerprint + rule.id
-          cache[hash] = 0
-        })
-      }
+      _.map(rule.duplicate, pattern => {
+        var fingerprint = matcher.fingerprintByObject(ruleName, pattern)
+        var hash = fingerprint + rule.id
+        cache[hash] = 0
+      })
     })
   })
   validatePolyfill(error, engine)
@@ -27,8 +25,7 @@ exports.onBegin = function (error, engine) {
 exports.onNode = function (node, rule, error) {
   if (!rule.duplicate || _.includes(POLYFILL_TAGS, node.nodeName)) return
 
-  var duplicates = _.isArray(rule.duplicate)
-        ? rule.duplicate : [rule.duplicate]
+  var duplicates = rule.duplicate
 
   _.map(duplicates, pattern => {
     if (!matcher.matchAttrs(node, pattern)) return
@@ -46,7 +43,7 @@ function validatePolyfill (error, engine) {
   POLYFILL_TAGS.forEach(tag => {
     var rules = _.get(engine.config.nodes, `${tag}`)
     _.map(rules, rule => {
-      if (!rule.duplicate) return
+      if (rule.duplicate.length === 0) return
       var matches = matcher.matchTagNames([tag], engine.html)
       if (matches.length > 1) {
         error(ERR.DUPLICATE_UNIQUE_TAG, tag)