Added browser.loadCSS and browser.htmlParser options.

Switched default parser to htmlparser. Few more tests passing.
jeromegn · Aug 24, 2011 · 2022e1b · 2022e1b
1 parent f230c4d
commit 2022e1b
Show file tree

Hide file tree

Showing 7 changed files with 66 additions and 28 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,6 +1,17 @@
 zombie.js-changelog(7) -- Changelog
 ===================================
 
+### Version 0.10.0  Pending
+
+Added `browser.loadCSS` option.  Set this to load external stylesheets.
+Defaults to `true`.
+
+Added `browser.htmlParser` option.  Tells JSDOM which HTML5 parser to
+use.  Use `null` for the default parser.
+
+Changed the default parser from `HTML5` to `htmlparser`.
+
+
 ### Version 0.9.7  2011-07-28
 
 Fixed: require.paths is deprecated [#158]

diff --git a/package.json b/package.json
@@ -37,7 +37,6 @@
     "node": ">= 0.4.0"
   },
   "dependencies": {
-    "html5": "~0.3.0",
     "jsdom": "~0.2.3",
     "mime": "~1.2.2",
     "websocket-client": "~1.0.0"

diff --git a/spec/browser-spec.coffee b/spec/browser-spec.coffee
@@ -1,5 +1,6 @@
 require "./helpers"
 { vows: vows, assert: assert, zombie: zombie, brains: brains } = require("vows")
+Browser = zombie.Browser
 jsdom = require("jsdom")
 
 brains.get "/static", (req, res)-> res.send """
@@ -85,7 +86,8 @@ brains.get "/soup", (req, res)-> res.send """
   <p>And another
   """
 
-brains.get "/useragent", (req, res)-> res.send "<body>#{req.headers["user-agent"]}</body>"
+brains.get "/useragent", (req, res)->
+  res.send "<html><body>#{req.headers["user-agent"]}</body></html>"
 
 brains.get "/iframe", (req, res)-> res.send """
   <html>
@@ -115,17 +117,17 @@ vows.describe("Browser").addBatch(
     "successful":
       topic: ->
         brains.ready =>
-          browser = new zombie.Browser
+          browser = new Browser
           browser.visit "http://localhost:3003/scripted", =>
             @callback null, arguments
       "should pass three arguments to callback": (args)-> assert.length args, 3
       "should not include an error": (args)-> assert.isNull args[0]
-      "should pass browser to callback": (args)-> assert.ok args[1] instanceof zombie.Browser
+      "should pass browser to callback": (args)-> assert.ok args[1] instanceof Browser
       "should pass status code to callback": (args)-> assert.equal args[2], 200
     "error":
       topic: ->
         brains.ready =>
-          browser = new zombie.Browser
+          browser = new Browser
           browser.visit "http://localhost:3003/missing", =>
             @callback null, arguments
       "should pass single argument to callback": (args)-> assert.length args, 1
@@ -140,14 +142,14 @@ vows.describe("Browser").addBatch(
     "successful":
       topic: ->
         brains.ready =>
-          browser = new zombie.Browser
+          browser = new Browser
           browser.on "loaded", (browser)=> @callback null, browser
           browser.window.location = "http://localhost:3003/"
       "should fire load event": (browser)-> assert.ok browser.visit
     "error":
       topic: ->
         brains.ready =>
-          browser = new zombie.Browser
+          browser = new Browser
           browser.on "error", (err)=> @callback null, err
           browser.window.location = "http://localhost:3003/deadend"
       "should fire onerror event": (err)->
@@ -156,7 +158,7 @@ vows.describe("Browser").addBatch(
     "wait over":
       topic: ->
         brains.ready =>
-          browser = new zombie.Browser
+          browser = new Browser
           browser.on "done", (browser)=> @callback null, browser
           browser.window.location = "http://localhost:3003/"
           browser.wait()
@@ -167,15 +169,17 @@ vows.describe("Browser").addBatch(
       "query text":
         topic: (browser)-> browser
         "should query from document": (browser)-> assert.equal browser.text(".now"), "Walking Aimlessly"
-        "should query from context (exists)": (browser)-> assert.equal browser.text(".now", browser.body), "Walking Aimlessly"
-        "should query from context (unrelated)": (browser)-> assert.equal browser.text(".now", browser.querySelector("#main")), ""
+        "should query from context (exists)": (browser)-> assert.equal browser.text(".now"), "Walking Aimlessly"
+        "should query from context (unrelated)": (browser)->
+          assert.equal browser.text(".now", browser.querySelector("#main")), ""
         "should combine multiple elements": (browser)-> assert.equal browser.text("form label"), "Email Password "
       "query html":
         topic: (browser)-> browser
         "should query from document": (browser)-> assert.equal browser.html(".now"), "<div class=\"now\">Walking Aimlessly</div>"
         "should query from context (exists)": (browser)-> assert.equal browser.html(".now", browser.body), "<div class=\"now\">Walking Aimlessly</div>"
         "should query from context (unrelated)": (browser)-> assert.equal browser.html(".now", browser.querySelector("#main")), ""
-        "should combine multiple elements": (browser)-> assert.equal browser.html("title, #main a"), "<title>The Living</title><a href=\"/dead\">Kill</a>"
+        "should combine multiple elements": (browser)->
+          assert.equal browser.html("title, #main a"), "<title>The Living</title><a href=\"/dead\">Kill</a>"
 
   "click link":
     zombie.wants "http://localhost:3003/living"
@@ -185,25 +189,27 @@ vows.describe("Browser").addBatch(
       "should run all events": (_, browser)-> assert.equal browser.document.title, "The Dead"
       "should return status code": (_, browser, status)-> assert.equal status, 200
 
-  "tag soup":
-    zombie.wants "http://localhost:3003/soup"
-      "should parse to complete HTML": (browser)->
-        assert.ok browser.querySelector("html head")
-        assert.equal browser.text("html body h1"), "Tag soup"
-      "should close tags": (browser)->
-        paras = browser.querySelectorAll("body p").toArray().map((e)-> e.textContent.trim())
-        assert.deepEqual paras, ["One paragraph", "And another"]
+  "tag soup using HTML5 parser":
+    topic: ->
+      browser = new Browser
+      browser.wants "http://localhost:3003/soup", { htmlParser: require("html5").HTML5 }, @callback
+    "should parse to complete HTML": (browser)->
+      assert.ok browser.querySelector("html head")
+      assert.equal browser.text("html body h1"), "Tag soup"
+    "should close tags": (browser)->
+      paras = browser.querySelectorAll("body p").toArray().map((e)-> e.textContent.trim())
+      assert.deepEqual paras, ["One paragraph", "And another"]
 
   "with options":
     topic: ->
-      browser = new zombie.Browser
+      browser = new Browser
       browser.wants "http://localhost:3003/scripted", { runScripts: false }, @callback
     "should set options for the duration of the request": (browser)-> assert.equal browser.document.title, "Whatever"
     "should reset options following the request": (browser)-> assert.isTrue browser.runScripts
 
   "user agent":
     topic: ->
-      browser = new zombie.Browser
+      browser = new Browser
       browser.wants "http://localhost:3003/useragent", @callback
     "should send own version to server": (browser)-> assert.match browser.text("body"), /Zombie.js\/\d\.\d/
     "should be accessible from navigator": (browser)-> assert.match browser.window.navigator.userAgent, /Zombie.js\/\d\.\d/
@@ -239,7 +245,7 @@ vows.describe("Browser").addBatch(
 
   "fork":
     topic: ->
-      browser = new zombie.Browser
+      browser = new Browser
       browser.visit("http://localhost:3003/living")
       browser.wait()
       browser.cookies("www.localhost").update("foo=bar; domain=.localhost")

diff --git a/spec/helpers.coffee b/spec/helpers.coffee
@@ -2,6 +2,7 @@ require.paths.unshift __dirname + "/../node_modules"
 fs = require("fs")
 express = require("express")
 zombie = require("../src/index")
+Browser = zombie.Browser
 debug = process.env.DEBUG || process.env.TRAVIS
 
 
@@ -58,7 +59,8 @@ brains.ready = (callback)->
 zombie.wants = (url, context)->
   topic = context.topic
   context.topic = ->
-    new zombie.Browser().wants url, {}, (err, browser)=>
+    browser = new Browser
+    browser.wants url, {}, (err, browser)=>
       if topic
         try
           value = topic.call(this, browser)
@@ -71,7 +73,7 @@ zombie.wants = (url, context)->
     return
   return context
 
-zombie.Browser.prototype.wants = (url, options, callback)->
+Browser.prototype.wants = (url, options, callback)->
   brains.ready =>
     options.debug = debug
     @visit url, options, (err, browser)=>

diff --git a/src/zombie/browser.coffee b/src/zombie/browser.coffee
@@ -32,12 +32,21 @@ class Browser extends require("events").EventEmitter
     # Options
     # -------
 
-    @OPTIONS = ["debug", "runScripts", "userAgent"]
+    OPTIONS = ["debug", "htmlParser", "loadCSS", "runScripts", "userAgent"]
 
     # ### debug
     #
     # True to have Zombie report what it's doing.
     @debug = false
+    # ### htmlParser
+    #
+    # Which parser to use (null for default). For example:
+    #   zombie.htmlParser = require("html5").HTML5
+    @htmlParser = null
+    # ### loadCSS
+    #
+    # True to load external stylesheets.
+    @loadCSS = true
     # ### runScripts
     #
     # Run scripts included in or loaded from the page. Defaults to true.
@@ -65,7 +74,7 @@ class Browser extends require("events").EventEmitter
     # Sets the browser options.
     if options
       for k,v of options
-        if @OPTIONS.indexOf(k) >= 0
+        if OPTIONS.indexOf(k) >= 0
           @[k] = v
         else
           throw "I don't recognize the option #{k}"

diff --git a/src/zombie/history.coffee b/src/zombie/history.coffee
@@ -71,16 +71,19 @@ class History
           QuerySelector: true
           MutationEvents: "2.0"
           ProcessExternalResources: []
-          FetchExternalResources: ["css", "frame"]
-        parser: require("html5").HTML5
+          FetchExternalResources: ["frame"]
+        parser: browser.htmlParser
         url: URL.format(url)
       if browser.runScripts
         options.features.ProcessExternalResources.push "script"
         options.features.FetchExternalResources.push "script"
+      if browser.loadCSS
+        options.features.FetchExternalResources.push "css"
       document = jsdom.jsdom(null, jsdom.dom.level3.core, options)
       browser.window.document = document
       document.window = document.parentWindow = browser.window
       document.fixQueue()
+      document.fixQuerySelector()
 
       headers = if headers then JSON.parse(JSON.stringify(headers)) else {}
       referer = stack[index-1]?.url

diff --git a/src/zombie/jsdom_patches.coffee b/src/zombie/jsdom_patches.coffee
@@ -117,3 +117,11 @@ core.HTMLDocument.prototype.fixQueue = ->
       item.err = err
       item.data = data
       item.check()
+
+
+# JSDOM el.querySelectorAll selects from the parent.
+Sizzle = require("jsdom/lib/jsdom/selectors/sizzle").Sizzle
+core.HTMLDocument.prototype.fixQuerySelector = ->
+  core.Element.prototype.querySelectorAll = (selector)->
+    new core.NodeList(@ownerDocument, => Sizzle(selector, this))
+