WebResourceProvider

• Download demo project - 78 Kb

Introduction

This article describes WebResourceProvider, a simple yet powerful framework for retrieving useful information from publicly available web services. I use the term "web service" in a generic, non-Microsoft sense, to mean information providers such as:

• search engines like Google and Altavista
• online stock quote providers such as Yahoo, CNN and Quicken
• weather information providers such as Weather.com, AccuWeather and CNN
• online translation services like Google, Babelfish and FreeTranslation
• etc.

The demo application included with this article shows how you can easily create objects to get:

• stock quotes
• the weather for a US zip code
• the list of locations served by a US zip code
• the translation of a piece of text
• the list of broken links on an HTML page
• the list of top posters at CodeProject

A Word of Caution

Before you use WebResourceProvider to write the next killer app, be aware that there are legal and ethical issues regarding the use of information obtained from other sources. In particular, the terms of service (TOS) of content providers such as Yahoo, CNN, etc. clearly state what you can and cannot do with information retrieved from their sites. Even if you write a web resource provider for personal use only, you should take into consideration any undue stress that your object may put on a web server. The CodeProjectTopPosters example in the demo won't let you get at more than the top 40 CodeProject posters. Further, it pauses between multiple accesses to the CodeProject server in order to not overload it.

How it Works

WebResourceProvider works by initializing itself, constructing a URL to be retrieved, downloading the resource, and extracting useful information from the downloaded content. The process is repeated until no more data needs to be downloaded.

You use WebResourceProvider by deriving your own resource provider class from it, and overriding any of these virtual methods (shown in red in the diagram on the right):

• init
• constructUrl
• isPost()
• getPostData()
• parseContent()
• moreAvailable()

WebResourceProvider provides an assortment of methods to help parse downloaded content. They are:

Method		Purpose
at		Checks whether current location is at a string
atExact		Case sensitive version of at()
skipTo		Advances current location to next occurence of a string
skipToExact		Case sensitive version of skipTo()
skipBackTo		Retreats current location to previous occurence of a string
skipBackToExact		Case sensitive version of skipBackTo()
extractTo		Extracts text from current location to the start of a string
extractToExact		Case sensitive version of extractTo()
extractToEnd		Extracts text from current location to end of content
getIndex		Returns current location
getLinks		Returns HREF and IMG links in content
resetIndex		Sets current location to start of content
replaceEvery		Replaces every occurence of a string in content with another
removeComments		Removes comments from content
removeScripts		Removes scripts from content
removeEnclosingAnchorTag		Removes anchor tag enclosing a string
removeEnclosingQuotes		Removes quotes enclosing a string
removeHtml		Removes HTML from a string
trim		Removes leading and trailing whitespace from a string

Sample Resource Providers

Here are screenshots of the sample resource providers in action.

The QuoteProvider object works by posting a request to Yahoo's basic stock quote form and parsing the returned information.

The ZipCodeDecoder object works by posting a request to the USPS zip code association form and parsing the returned information.

The WeatherProvider object works by posting a request to CNN's weather form and parsing the returned information.

The Translator object works by posting a request to Google's translation engine and parsing the returned information. The request includes the translation mode. The sample performs a reverse translation and presents it along with the original text for comparison purposes.

The LinkChecker object is a thin layer above the WebResourceProvider class. It delegates the job of determining the document and image links at a URL to the base class. LinkChecker::getLinks() does its best to determine the links on a page, but because of the large number of ways a link can be specified, this method may miss a few links. The LinkChecker demo uses WebResourceProvider::urlExists() to check whether a link is valid. Click here to see a screenshot of the LinkChecker demo run against the CodeProject home page. Keep up the good work, Chris!

Using WebResourceProvider

To use WebResourceProvider do the following:

1. Build the WebResourceProvider_Lib project.
2. Modify your application's project to look for header files and libraries in the WebResourceProvider_Lib project area.
3. Derive an object from WebResourceProvider. You'll need to #include WebResourceProvider.h in your derived class' header file.
4. Override the constructUrl() method in your derived class. This method specifies the URL to be downloaded.
5. Override the parse() method in your derived class. This method extracts information from the downloaded content and stores it in the derived class' member variables.
6. Optionally override other WebResourceProvider virtual methods. See the source code of the sample resource providers included in the demo project for examples.
7. Link your application with WebResourceProvider.lib.

Acknowledgement

WebResourceProvider uses the following code written by others:

• WebGrab by Chris Maunder
• AmHttpUtilities by Anders Molin
• Case-Insensitive String Replace by Uwe Keim

A Call for Interesting WebResourceProviders!

This is an invitation to the CP community to come up with interesting and useful web resource providers. Let your imagination (and coding prowess) flow! Please post your cool WebResourceProvider derived classes at CodeProject.

Revision History

• 23 Mar 2007
  Updated parsing logic in WeatherProvider, ZipCodeProvider and Translator modules.
• 7 Oct 2006
  Bug Fix: Updated parsing logic in Translator module.
• 7 Aug 2002
  Bug Fix: Fixed computation error in extractToEnd().
  Bug Fix: Added missing call to init() in fetchResource().
  Added methods skipBackTo() and skipBackToExact.
• 8 May 2002
  Added methods urlExists(), getLinks(), removeComments(), removeScripts(), findNoCase() and findStringInArray().
  Modified at(), skipTo(), and extractTo() to be case insensitive. Added case sensitive analogs atExact(), skipToExact() and extractToExact().
  Added LinkChecker sample to demo app.
  Fixed a bug in parse() that caused the fetch status to be ignored.
  Speeded up ZipCodeDecoder sample object.
• 30 April 2002
  Corrected control flow image.
• 29 April 2002
  Initial version of article.