Changeset 11287

Timestamp:

05/09/08 04:22:04 (5 years ago)

Author:

stomita

Message:

lang/javascript/jsAutoPageScraper: fixed nextLink, xpath-javascript load for non-native support browsers

Files:

• lang/javascript/jsAutoPageScraper/trunk/js/jsAutoPageScraper.js (modified) (8 diffs)

lang/javascript/jsAutoPageScraper/trunk/js/jsAutoPageScraper.js

@@ -1 +1 @@
 /**
- * jsAutoPageScraper 0.2
+ * jsAutoPageScraper 0.3
  *
  * (c) mashmatrix, inc.
@@ -52 +52 @@
  */
 function createIframe(url, callback, scope) {
-  // ugly W/A to prevent cache...
-  if (url == window.location.href) {
-    url += url.indexOf('?') >= 0 ? '&' : '?';
-  }
   var iframeEl = windoc.createElement('iframe');
   iframeEl.style.width = iframeEl.style.height = 0;
   iframeEl.style.visibility = 'hidden';
   iframeEl.style.position = 'absolute';
-  iframeEl.src = url;
+  iframeEl.src = "about:blank";
+
+  getBodyEl().appendChild(iframeEl);
 
   addEvent(iframeEl, 'load', function() {
@@ -67 +65 @@
   });
 
-  getBodyEl().appendChild(iframeEl);
+  iframeEl.contentWindow.location.href = url;
 
 }
@@ -100 +98 @@
     var scriptEl = doc.createElement('script');
     scriptEl.type = 'text/javascript';
-    scriptEl.src = xpathLibUrl;
+    scriptEl.src = xpathLibUrl + '?useNative=false';
     doc.documentElement.appendChild(scriptEl);
     pollXPathLibLoadStatus();
@@ -198 +196 @@
      .skip(5)
      .scrape(10, callback2)
-     .wait();
+     .force();
 
  function callback1(records) { ... }
@@ -269 +267 @@
 
   /**
-   * Refresh queued commands and wait next command by keeping current cursor and resources.
+   * Wait next command by keeping current cursor and resources.
    *
+   * @param {Number} msec - msec to wait for next fetch (optional)
    * @returns Self instance
    * @type JSAutoPageScraper
    */
-  wait : function() {
-    this.force();
+  wait : function(msec) {
+    var _this = this;
+    if (msec) {
+      var cmd = {
+        method : 'wait', 
+        args : [ msec, function() { _this.force() } ]
+      }
+      this.cmdQueue.push(cmd);
+    }
     return this;
   }
@@ -347 +353 @@
 
   /**
+   * Wait scraping
+   * 
+   * @param {Number} msec - Waiting time (in milliseconds)
+   * @param {Function} callback - Callbacked when timeouted.
+   */
+  wait : function(msec, callback) {
+    window.setTimeout(callback, msec);
+  }
+  ,
+
+  /**
    * Clear and reset fetching states.
    */
@@ -394 +411 @@
     } else if (this.paragraphs.length == 0) {
       delete this.paragraphs;
-      this.nextUrl = queryXPathElement(this.doc, this.siteinfo.nextLink);
+      var nextLink = queryXPathElement(this.doc, this.siteinfo.nextLink);
+      this.currUrl = nextLink && nextLink.href;
       this.cleanupIframe();
-      if (!this.nextUrl) {
+      if (!this.currUrl) {
         this.onFinish();
       } else {