URLs and Links

Lecture Notes for CS349W
Fall Quarter 2008
John Ousterhout

URLs

• URLs: Uniform Resource Locators:
  • Provide names for Web content
  • Tell browser how to fetch a particular piece of Web content
  • Provides detailed information for server on location of content, possibly including additional parameters for generated content.
• Go over example URL:

  http://www.company.com:81/a/b/c.html?name=Alice&age=23#foo
  

  • Scheme (http:): identifies protocol used to fetch the content.
    • http: is the most common scheme; it means use the HTTP protocol, which we will discuss soon.
    • https: is similar to http: except that it uses SSL encryption when communicating with the server for greater security.
    • file: means read a file from the local disk.
    • There are several other schemes, such as ftp:, but they aren't used much (mostly for older protocols that existed before the Web was developed).
  • Host name (//host.company.com): name of a machine running an HTTP server.
  • Server's port number (81): allows multiple servers to run on the same machine. Servers almost always run on port 80, which is the default.
  • Hierarchical portion (/a/b/c.html): identifies a particular piece of content provided by the server.
    • In the simplest case this is just the path name for an HTML file in a Web directory.
    • If it refers to a file name ending in .php, then the file is a PHP script that will regenerate the page each time it is referenced.
    • This information can be interpreted in any way the server wishes; for example, it might have the form className/method, selecting a particular method to invoke in a particular class in the server's code, which will generate the HTML dynamically.
    • When you set up a Web server you provide configuration information that tells how to interpret the hierarchical portion of a URL.
  • Query info (name=Alice&age=23): provides additional parameters that can be used by the server to select dynamic content. For example:

    http://www.company.com/showOrder.php?order=4621047
    

  • Fragment (#foo): selects a particular location in the resulting page (instead of displaying the top of the page, scroll the window so the particular fragment appears at the top). Used on the browser only; not sent to the server.
• URL encoding: mechanism for including special characters in a URL, such as &
  • Can't use such characters as-is (will cause misinterpretation):

    http://host.company.com/showCompanyInfo?name=C&H Sugar
    

  • Any character in a URL other than A-Z, a-z, or any of -_.~ must be represented as %xx, where xx is the hexadecimal value of the character:

    http://host.company.com/showCompanyInfo?name=C%26H%20Sugar
    

  • This is yet another example of quoting, which is an issue whenever information is encoded textually:
    • Some characters in the encoding have special structural significance, while other characters are just data.
    • What if I want to include a special character in my data?
    • Must introduce some quoting mechanism for handling special characters in data; typically this involves additional special characters for the quoting mechanism, such as & in HTML/XML or % in URLs
    • The quoting mechanism better also be able to quote the quote characters.
    • If you ever receive data whose content is uncontrolled (e.g., typed by a user) and you want to incorporate it into a text encoding, you must check the data for special characters and add appropriate quoting.
    • If you forget to quote special characters, unexpected user data could change the structure of the encoded value. Malicious users can capitalize on such mistakes to violate the security of the system. Example: SQL query injection.
• Note: you will also hear the term URI: Uniform Resource Identifier.
  • This term was introduced by standards committees after URL and means almost the same thing (the distinction is hard to remember and not very interesting).
  • Official organizations generally use "URI" and deprecate "URL".
  • However, "URL" is the term that people are most familiar with; I have found that using "URI" instead of "URL" tends to confuse rather than clarify, so I use "URL" even though it isn't technically correct.

Links

• Now that we can name Web content, how do we use the names?
• Loading a page: type the URL into your browser.
• Nested content within a page: for example, images are named with URLs.
• Links: content in a page which, when clicked on, switches the browser display to another page.
• Links are implemented with the <a> tag:

  <a href="http://www.company.com/news.html>Latest News</a>
  

• <a> elements can be used in other ways:
  • Relative URL (makes it easier to reorganize a Web site):

    <a href="movies.html">Movie reviews</a>
    

  • Go to a different place in the same page:

    <a href="#anchor3">...</a>
    

  • Define an anchor point (a position that can be referenced with # notation):

    <a name="anchor3">...</a>
    

Hypertext and the Web

• We take the Web's hypertext capabilities for granted now, but they were controversial when they first came out.
• The basic ideas behind hypertext have been around since the 1960's but there were few practical limitations before the Web (Apple's Hypercard may have been the most widely used one).
• Before the Web, hypertext systems generally required "referential integrity": you couldn't create a link unless the target was known to exist and you couldn't delete a target with a link referring to it.
• Key contribution of the Web: allow broken links! (Don't require referential integrity)
• The hypertext community looked down on the Web in its early days as a step backwards.
• However, referential integrity doesn't scale to systems the size of today's Web; eliminating this requirement made it possible for the Web to scale.
• To build a truly scalable system, have to give up something in the area of integrity/consistency.