32. The World Wide Web

Next to email, the most useful service on the Internet is the World Wide Web (often written "WWW" or "Web"). It is a giant network of hypertext documents and services, and it keeps growing by the instant--anyone with an Internet-connected computer can read anything on the Web, and anyone can publish to the Web. It could well be the world's largest public repository of information.

This chapter describes tools for accessing and using the Web. It also describes tools for writing text files in HTML ("HyperText Markup Language"), the native document format of the Web.

32.1 Browsing the Web		Netscape's famous Web browser.
32.2 Viewing an Image from the Web		Viewing images from the Web.
32.3 Reading Text from the Web		Reading text from the Web.
32.4 Browsing the Web in Emacs
32.5 Getting Files from the Web		Getting files from the Web.
32.6 Writing HTML
32.7 More Web Browsers and Tools		More Web tools to try.

32.1 Browsing the Web

@sf{Debian}: `mozilla'
@sf{Debian}: `skipstone'
@sf{WWW}: http://www.mozilla.org/
@sf{WWW}: http://galeon.sourceforge.net/
@sf{WWW}: http://www.muhri.net/skipstone/

The version of Netscape's browser which had been released as free, open source software (see section What's Open Source?) in 1998 to much fanfare is called Mozilla.(40) When first released, the Mozilla application was a "developer's only" release, but as of this writing it is finally reaching a state where it is ready for general use.

Once the Mozilla browser has been installed, run it in X either by typing mozilla in a shell or by selecting it from a menu in the usual fashion, as dictated by your window manager.

Like most graphical Web browsers, its use is fairly self-explanatory; type a URL in the Location dialog box to open that URL, and left-click on a link to follow it, replacing the contents of the browser's main window with the contents of that link. One nice feature for Emacs fans is that you can use Emacs-style keystrokes for cursor movement in Mozilla's dialog boxes (see section Basic Emacs Editing Keys).

A typical Mozilla window looks like this:

(In this example, the URL http://slashdot.org/ is loaded.)

A criticism of the earlier Netscape Navigator programs is that the browser is a bloated application: it contained its own email client, its own Usenet newsreader, and other functions that are not necessary when one wants to simply browse the Web. Since Mozilla is free software, anyone can take out these excess parts to make a slimmer, faster, smaller application--and that is what some have done. Two of these projects, Galeon and Skipstone, show some promise; see the above URLs for their home pages.(41)

The following recipes will help you get the most out of using a graphical Web browser in Linux.

NOTE: Mozilla development is moving very rapidly these days, and while Mozilla is continually improving at a fantastic rate, some of these recipes may not work as described with the version you have.

Another way to browse the Web is to use Emacs (see section Browsing the Web in Emacs); more alternative browsers are listed in More Web Browsers and Tools.

32.1.1 Maintaining a List of Visited Web Sites		Keeping a history of your browsing.
32.1.2 Opening a URL from a Script		Running Mozilla from a script.
32.1.3 Mozilla Browsing Tips		Tips for using Mozilla.

32.1.1 Maintaining a List of Visited Web Sites

@sf{Debian}: `browser-history'
@sf{WWW}: http://www.inria.fr/koala/colas/browser-history/

You start it in the background, and each time you visit a URL in a Web browser (as of this writing, works with the Netscape, Arena, and Amaya browsers), it writes the name and URL to its current history log, which you can view at any time.

• To start browser-history every time you start X, put the following line in your `.xsession' file:

  browser-history &

The browser history logs are kept in a hidden directory called `.browser-history' in your home directory. The current history log is always called `history-log.html'; it's an HTML file that you can view in a Web browser.

• To view the current history log with lynx, type:

  $ lynx ~/.browser-history/history-log.html RET

Past history logs have the year, month, and week appended to their name, and they are compressed (see section Compressed Files). After uncompressing them, you can view them just as you would view the current log (if you are viewing them in Mozilla, you don't even need to uncompress them--it handles this automagically.)

You can also use zgrep to search through your old browser history logs. The logs keep the URL and title of each site you visit, so you can search for either--then when someone asks, "Remember that good article about such-and-such?" you can do a zgrep on the files in your `~/.browser-history' directory to find it.

• To find any URLs from the list of those you visited in the year 2000 whose titles contain the word `Confessions', type:

  $ zgrep Confessions ~/.browser-history/history-log-2000* RET

This command searches all your logs from the year 2000 for the text `Confessions' in it, and outputs those lines.

NOTE: For more about zgrep, see Matching Lines in Compressed Files.

32.1.2 Opening a URL from a Script

To open a Web page in Mozilla from a shell script, use the `-remote' option followed by the text `'openURL(URL)'', where URL is the URL to open.

• To open the URL http://www.drudgereport.com/ in Mozilla from a shell script, use the following line:

  mozilla -remote 'openURL(http://www.drudgereport.com/)'

32.1.3 Mozilla Browsing Tips

The following tips make Web browsing with Mozilla easier and more efficient.

• Many users disable Java and JavaScript altogether; most Web sites don't require their use, and they often introduce security problems or have other pernicious effects on your browsing. Just say no.

• Disabling the automatic loading of images can help if you are on a slow connection; the broken-image icons take some getting used to, but you'll be surprised at how much more quickly pages will load! If you need to see a page's images, just left-click on the Load Images button. You can also right-click on the broken-image icon of the image you want to load and select Open this Image.

• Right-click on an image to save it to a file; you will be given a choice to either open the image in the browser window or save it to a file.

• To open a link in another browser window, middle-click on the link. Opening multiple links in their own windows saves time when you are doing a lot of "power browsing."

• If a site forces links to open in a new window, and you don't want to do that, right-click on the link you want to open, and choose Open this Link; the link will open in the current browser window.

• To go back to the last URL you visited, type ALT-@leftarrow, and to go forward to the next URL in your history, type ALT-@rightarrow. (These keys may not have the desired effect in some window managers; if they don't work for you, try using the CTRL key instead of the ALT key.) [GNU INFO BUG: any <> in the preceding line should be the <- and/or -> arrow keys.]

• If your visited-URL history on the Go menu is very large, and earlier URLs are truncated, you can still visit them by doing this: left-click one of the lowest entries on the menu, and visit that; then, left-click on the Home button. This eliminates all the URLs in the history list that are more recent than the page you'd just visited, but all of the old pages will be back in the list.

• To open your bookmarks file in a new window, type ALT-b.

• To open a new Mozilla window, ALT-n (it's often useful to have several windows open at once).

32.2 Viewing an Image from the Web

@sf{Debian}: `imagemagick'
@sf{WWW}: ftp://ftp.wizards.dupont.com/pub/ImageMagick/

• To view the image at ftp://garbo.uwasa.fi/garbo-gifs/garbo01.gif, type:

  $ display ftp://garbo.uwasa.fi/garbo-gifs/garbo01.gif RET

NOTE: When viewing the image, you can use all of the image manipulation commands that display supports, including resizing and changing the magnification of the image. For more information about display, see Viewing an Image in X.

32.3 Reading Text from the Web

@sf{Debian}: `lynx'
@sf{WWW}: http://lynx.browser.org/

Type lynx to start it--if a "start page" is defined, it will load. The start page is defined in `/etc/lynx.cfg', and can be a URL pointing to a file on the local system or to an address on the Web; you need superuser privileges to edit this file. On Debian systems, the start page comes defined as the Debian home page, http://www.debian.org/ (but you can change this, of course; many experienced users write their own start page, containing links to frequently-visited URLs, and save it as a local file in their home directory tree).

To open a URL, give the URL as an argument.

• To view the URL http://lycaeum.org/, type:

  $ lynx http://lycaeum.org/ RET

When in lynx, the following keyboard commands work:

The following are some recipes for using lynx.

NOTE: Emacs users might want to use the `-emacskeys' option when starting lynx; it enables you to use Emacs-style keystrokes for cursor movement (see section Basic Emacs Editing Keys).

32.3.1 Perusing Text from the Web		Perusing text from the Web.
32.3.2 Viewing a Site That Requires Authorization		Browsing sites which require logging in.
32.3.3 Options Available while Browsing Text		Lynx startup options.

32.3.1 Perusing Text from the Web

To peruse just the text of an article that's on the Web, output the text of the URL using lynx with the `-dump' option. This dumps the text of the given URL to the standard output, and you can pipe this to less for perusal, or use redirection to save it to a file.

• To peruse the text of http://www.sc.edu/fitzgerald/winterd/winter.html, type (all on one line):

$ lynx -dump 
http://www.sc.edu/fitzgerald/winterd/winter.html | less RET

It's an old net convention for italicized words to be displayed in an etext inside underscores like `_this_'; use the `-underscore' option to output any italicized text in this manner.

By default, lynx annotates all the hyperlinks and produces a list of footnoted links at the bottom of the screen. If you don't want them, add the `-nolist' option and just the "pure text" will be returned.

• To output the pure text, with underscores, of the previous URL, and save it to the file `winter_dreams', type (all on one line):

$ lynx -dump -nolist -underscore 
http://www.sc.edu/fitzgerald/winterd/winter.html > winter_dreams RET

You can do other things with the pure text, like pipe it to enscript for setting it in a font for printing.

• To print the pure text, with underscores, of the previous URL in a Times Roman font, type (all on one line):

  $ lynx -dump -nolist -underscore http://www.sc.edu/fitzgerald/winterd/winter.html | enscript -B -f "Times-Roman10" RET

NOTE: To peruse the plain text of a URL with its HTML tags removed and no formatting done to the text, see Converting HTML to Another Format.

32.3.2 Viewing a Site That Requires Authorization

To view a site or Web page that requires registration, use lynx with the `-auth' option, giving as arguments the username and password to use for authorization, separating them by a colon (`:') character.

• To view the URL http://www.nytimes.com/archive/ with a username and password of `cypherpunks', type (all on one line):

  $ lynx -auth=cypherpunks:cypherpunks http://www.nytimes.com/archive/ RET

It's often common to combine this with the options for saving to a file, so that you can retrieve an annotated text copy of a file from a site that normally requires registration.

• To save the URL http://www.nytimes.com/archive/ as an annotated text file, `mynews', type (all on one line):

  $ lynx -dump -number_links -auth=cypherpunks:cypherpunks http://www.nytimes.com/archive/ > mynews RET

NOTE: The username and password argument you give on the command line will be recorded in your shell history log (see section Command History), and it will be visible to other users on the system should they look to see what processes you're running (see section Listing All of a User's Processes).

32.3.3 Options Available while Browsing Text

The following table describes some of the command-line options lynx takes.

32.4 Browsing the Web in Emacs

@sf{Debian}: `w3-el-e20'
@sf{WWW}: ftp://ftp.cs.indiana.edu/pub/elisp/w3/

• To start W3 in Emacs, type:

  M-x w3 RET

To open a URL in a new buffer, type C-o and, in the minibuffer, give the URL to open (leaving this blank visits the Emacs/W3 home page). Middle-click a link to follow it, opening the URL in a new buffer.

• To open the URL http://gnuscape.org/, type:

  C-o http://gnuscape.org/ RET

• To open the URL of the Emacs/W3 home page, type:

  C-o RET

The preceding example opens the Emacs/W3 home page in a buffer of its own:

The following table describes some of the various special W3 commands.

NOTE: If you get serious about using Emacs/W3, you'll almost certainly want to run the XEmacs flavor of Emacs--as of this writing, GNU Emacs cannot display images.

32.5 Getting Files from the Web

@sf{Debian}: `wget'
@sf{WWW}: http://www.wget.org/

The following subsections contain recipes for using wget to retrieve information from the Web. See Info file `wget.info', node `Examples', for more examples of things you can do with wget.

NOTE: To retrieve an HTML file from the Web and save it as formatted text, use lynx instead--see Perusing Text from the Web.

32.5.1 Saving a URL to a File		Getting a URL.
32.5.2 Archiving an Entire Web Site		Archiving a site.
32.5.3 Archiving Part of a Web Site		Archiving part of a site.
32.5.4 Reading the Headers of a Web Page		Looking at Web server headers.

32.5.1 Saving a URL to a File

To download a single file from the Web, give the URL of the file as an argument to wget.

• For example, to download ftp://ftp.neuron.net/pub/spiral/septembr.mp3 to a file, type:

  $ wget ftp://ftp.neuron.net/pub/spiral/septembr.mp3 RET

This command reads a given URL, writing its contents to a file with the same name as the original, `septembr.mp3', in the current working directory.

If you interrupt a download before it's finished, the contents of the file you were retrieving will contain only the portion of the file wget retrieved until it was interrupted. Use wget with the `-c' option to resume the download from the point it left off.

• To resume download of the URL from the previous example, type:

  $ wget -c ftp://ftp.neuron.net/pub/spiral/septembr.mp3 RET

NOTE: In order for the `-c' option to have the desired effect, you should run wget from the same directory as it was run previously, where that partially-retrieved file should still exist.

32.5.2 Archiving an Entire Web Site

To archive a single Web site, use the `-m' ("mirror") option, which saves files with the exact timestamp of the original, if possible, and sets the "recursive retrieval" option to download everything. To specify the number of retries to use when an error occurs in retrieval, use the `-t' option with a numeric argument---`-t3' is usually good for safely retrieving across the net; use `-t0' to specify an infinite number of retries, good for when a network connection is really bad but you really want to archive something, regardless of how long it takes. Finally, use the `-o' with a file name as an argument to write a progress log to the file--examining it can be useful in the event that something goes wrong during the archiving; once the archival process is complete and you've determined that it was successful, you can delete the log file.

• To mirror the Web site at http://www.bloofga.org/, giving up to three retries for retrieving files and putting error messages in a log file called `mirror.log', type:

  $ wget -m -t3 http://www.bloofga.org/ -o mirror.log RET

This command makes an archive of the Web site at `www.bloofga.org' in a subdirectory called `www.bloofga.org' in the current directory. Log messages are written to a file in the current directory called `mirror.log'.

To continue an archive that you've left off, use the `-nc' ("no clobber") option; it doesn't retrieve files that have already been downloaded. For this option to work the way you want it to, be sure that you are in the same directory that you were in when you originally began archiving the site.

• To continue an interrupted mirror of the Web site at http://www.bloofga.org/ and make sure that existing files are not downloaded, giving up to three retries for retrieval of files and putting error messages in a log file called `mirror.log', type:

  $ wget -nc -m -t3 http://www.bloofga.org/ -o mirror.log RET

32.5.3 Archiving Part of a Web Site

To archive only part of a Web site--such as, say, a user's home page--use the `-I' option followed by a list of the absolute path names of the directories to archive; all other directories on the site are ignored.

• To archive the Web site at http://dougal.bris.ac.uk/~mbt/, only archiving the `/~mbt' directory, and writing log messages to a file called `uk.log', type:

  $ wget -m -t3 -I /~mbt http://dougal.bris.ac.uk/~mbt/ -o uk.log RET

This command archives all files on the http://dougal.bris.ac.uk/~mbt/ Web site whose directory names begin with `/~mbt'.

To only get files in a given directory, use the `-r' and `-l1' options (the `-l' option specifies the number of levels to descend from the given level). To only download files in a given directory, combine these options with the `--no-parent' option, which specifies not to ascend to the parent directory.

Use the `-A' option to specify the exact file name extensions to accept--for example, use `-A txt,text,tex' to only download files whose names end with `.txt', `.text', and `.tex' extensions. The `-R' option works similarly, but specifies the file extensions to reject and not download.

• To download only the files ending in a `.gz' extension and only in the given directory `/~rjh/indiepop-l/download/' at `monash.edu.au', type:

  $ wget -m -r -l1 --no-parent -A.gz http://monash.edu.au/~rjh/indiepop-l/download/ RET

32.5.4 Reading the Headers of a Web Page

All Web servers output special headers at the beginning of page requests, but you normally don't see them when you retrieve a URL with a Web browser. These headers contain information such as the current system date of the Web server host and the name and version of the Web server and operating system software.

Use the `-S' option with wget to output these headers when retrieving files; headers are output to standard output, or to the log file, if used.

• To retrieve the file at http://slashdot.org/ and output the headers, type:

  $ wget -S http://slashdot.org/ RET

This command writes the server response headers to standard output and saves the contents of http://slashdot.org/ to a file in the current directory whose name is the same as the original file.

32.6 Writing HTML

@sf{Debian}: `bluefish'
@sf{WWW}: http://bluefish.openoffice.nl/

Many people swear by Bluefish, a full-featured, user-friendly HTML editor for X.

Emacs (see section Emacs) has a major mode to facilitate the editing of HTML files; to start this mode in a buffer, type:

M-x html-mode RET

The features of HTML mode include the insertion of "skeleton" constructs.

The help text for the HTML mode function includes a very short HTML authoring tutorial--view the documentation on this function to display the tutorial.

• To read a short HTML tutorial in Emacs, type:

  C-h f html-mode RET

NOTE: When you're editing an HTML file in an Emacs buffer, you can open the same file in a Web browser in another window--Web browsers only read and don't write the HTML files they open, so you can view the rendered document in the browser as you create it in Emacs. When you make and save a change in the Emacs buffer, reload the file in the browser to see your changes take effect immediately.

32.6.1 Adding Parameters to Image Tags		Adding params to image tags.
32.6.2 Converting HTML to Another Format		Converting HTML to text.
32.6.3 Validating an HTML File		Validating HTML.

32.6.1 Adding Parameters to Image Tags

@sf{Debian}: `imgsizer'
@sf{WWW}: http://www.tuxedo.org/~esr/software.html#imgsizer

Use imgsizer to automatically determine the proper values and insert them into an HTML file. Give the name of the HTML file to fix as an argument.

• To add `HEIGHT' and `WIDTH' parameters to the file `index.html', type:

  $ imgsizer index.html RET

32.6.2 Converting HTML to Another Format

@sf{Debian}: `unhtml'
@sf{Debian}: `html2ps'
@sf{WWW}: http://dragon.acadiau.ca/~013639s/
@sf{WWW}: http://www.tdb.uu.se/~jan/html2ps.html

To simply remove the HTML formatting from text, use unhtml. It reads from the standard input (or a specified file name), and it writes its output to standard output.

• To peruse the file `index.html' with its HTML tags removed, type:

  $ unhtml index.html | less RET

• To remove the HTML tags from the file `index.html' and write the output to a file called `index.txt', type:

  $ unhtml index.html > index.txt RET

When you remove the HTML tags from a file with unhtml, no further formatting is done to the text. Furthermore, it only works on files, and not on URLs themselves.

Use lynx to save an HTML file or a URL as a formatted text file, so that the resultant text looks like the original HTML when viewed in lynx. It can also preserve italics and hyperlink information in the original HTML. See section Perusing Text from the Web.

One thing you can do with this lynx output is pipe it to tools for spacing text, and then send that to enscript for setting in a font. This is useful for printing a Web page in typescript "manuscript" form, with images and graphics removed and text set double-spaced in a Courier font.

• To print a copy of the URL http://example.com/essay/ in typescript manuscript form, type:

  $ lynx -dump -underscore -nolist http://example.com/essay/ | pr -d | enscript -B RET

NOTE: In some cases, you might want to edit the file before you print it, such as when a Web page contains text navigation bars or other text that you'd want to remove before you turn it into a manuscript. In such a case, you'd pipe the lynx output to a file, edit the file, and then use pr on the file and pipe that output to enscript for printing.

Finally, you can use html2ps to convert an HTML file to PostScript; this is useful when you want to print a Web page with all its graphics and images, or when you want to convert all or part of a Web site into PDF. Give the URLs or file names of the HTML files to convert as options. Use the `-u' option to underline the anchor text of hypertext links, and specify a file name to write to as an argument to the `-o' option. The defaults are to not underline links, and to write to the standard output.

• To print a PostScript copy of the document at the URL http://example.com/essay/ to the default printer, type:

  $ html2ps http://example.com/essay/ | lpr RET

• To write a copy of the document at the URL http://example.com/essay/ to a PostScript file `submission.ps' with all hypertext links underlined, type:

  $ html2ps -u -o submission.ps http://example.com/essay/ RET

32.6.3 Validating an HTML File

@sf{Debian}: `weblint'
@sf{WWW}: http://www.weblint.org/

• To validate the HTML in the file `index.html', type:

  $ weblint index.html RET

32.7 More Web Browsers and Tools

Surprisingly, there are not nearly as many Web browsers for Linux as there are text editors--or even text viewers. This remains true for any operating system, and I have often pondered why this is; perhaps "browsing the Web," a fairly recent activity in itself, may soon be obsoleted by Web readers and other tools. In any event, the following lists other browsers that are currently available for Linux systems.