SYNOPSISnetrik [options] URL or filename
DESCRIPTIONnetrik is an advanced text mode HTML (WWW) browser, that is to say a web browser running on character cell displays (linux console, xterm etc.) -- not unlike w3m, links or lynx.
When invoked with a filename or URL as argument, the specified document is loaded and displayed in interactive mode, so you can start browsing the web or some local HTML repository from there. (Alternatively the document can be just layouted and dumped to the screen, see OPTIONS below.)
When invoked with '-' instead of a file/URL, an HTML document is read from standard input, and displayed just the same.
If some file name/URL is present in the configuration file (~/.netrikrc), you can also start netrik without any non-option arguments; the one from the config file will be loaded then. See netrikrc(5) for details.
When multiple non-option arguments are specified, netrik simply takes the last one, and ignores the others. (This is for the above to work.)
The type of the resource to load is automatically determined from the argument. If it starts with "http://", "ftp://" or "file://", the corresponding type is used. If none of these is specified, netrik tries to guess the type: First it tries to open a local file of the given name, and if that fails, it tries HTTP. Only if both fail, an error is generated. (See EXAMPLES below.)
Local files can be also compressed by gzip or bzip2, which is handled transparently, meaning the ".gz" or ".bz2" extension needn't be given in the URL.
Once in the pager, you can explore the web interactively, as with any other web
browser. The keyboard commands should look familiar if you know vi (or some of
the myriads of programs with vi-like keys) as well as lynx: Use the 'j' and 'k'
keys to move around, the up and down arrow keys (or capital 'J' and 'K') to
select links, and the <return> key to follow links. See PAGER COMMANDS
below for a description of other useful commands.
OPTIONSNote: netrik is still in early development state; options are subject to changes.
When using the pager, this causes a page that contains extremely long
words to be rendered wider than the screen, instead of breaking the
word. Note however that side scrolling isn't implemented yet -- you
won't be able to see the end of the line when using this option... In
dump mode, this option causes usage of the default width of 80 columns
instead of what the terminal definition says. (Words are always broken
in dump mode.)
- Abort on any HTML syntax errors or warnings encountered. A short error description is printed. (This description may not be terribly useful at times...) This mode is primarily intended for HTML debugging. (Note however that netrik may oversee some errors; but most are reported.)
- Do not abort on HTML syntax errors. Error descriptions are printed for every syntax error (or warning), but netrik tries to parse the page anyhow. Workarounds are used for some typical syntax errors (e.g. unescaped '<' or '&' characters); other errors are ignored. After the whole page is loaded, if some error(s) were found, a warning message is printed (according to the severity of the worst encountered bug), and the pager starts after a keypress.
- This mode is identical to --clean-html, except that netrik doesn't pause after loading completes, if only warnings were generated but no real errors were encountered. (i.e. constructs that are discouraged in the standard, but strictly speaking are valid.)
- --broken-html (default)
- This mode is identical to --valid-html, except that netrik also doesn't pause if only simple errors with known workaround were encountered, which probably won't disturb layouting. Usage should be avoided if possible. (The file syntax_error.txt or syntax_error.html in the documentation directory (see SEE ALSO below) explains why.)
In this mode no warning is showm for any syntax errors, even if they might
cause heavily broken layouting. Don't use!
Before displaying (or dumping) the page, some intermediate layouting
stages are shown. (This output is described in the README.) Try it --
it's quite interesting to watch netrik work :-) It can be also
useful to find HTML errors in a page, as it dumps the page while
(This option is not available if compiled with --disable-debug to ./configure)
- Issue a warning when encountering an unknown HTML element or attribute. This is probably only useful for debugging purposes, as there are quite a lot of (legal) HTML facilities netrik doesn't know.
- Just dump the file given as argument to the screen and quit, instead of starting the pager. (The page is layouted correctly.)
- Ignore the "http_proxy" and "HTTP_PROXY" environment variables with --builtin-http. (No effect on wget! See below.)
- Use wget(1) to retrieve pages from a HTTP server, instead of the builtin HTTP handling code. Note that HTTP redirects in most cases cause relative links in the page to be broken when using wget. The builtin HTTP code seems to work good now; using wget shouldn't be necessary. (FTP pages however are always loaded via wget.)
- When jumping to an anchor (following a link with a fragment identifier), the page will be scrolled (if possible) so that the anchor will stand just below the screen top. (In the second line, which is the first line in which links can be activated.) By default, the anchor is at about 1/5 of the screen height below the top.
- Use the arrow keys to move the cursor, instead of the lynx-like navigation used by default. (This is useful for blind users, as it allows using the "flash cursor" keys found on braille displays.)
Assume the terminal has xterm-like attribute handling. (i.e. needs a workaround
to display a bright background color.)
This setting is used automatically if the terminal type ($TERM environment variable) contains the string "xterm", so you only need to set it manually if you have some other terminal that also needs that workaround or if you have set --console in netrikrc(5) and need to override that.
Note that this workaround works *only* on xterm (and maybe some other terminals), but not on linux console, so you can't just set it categorically!
- Assume the terminal doesn't need and understand the xterm workaround for bright background colors. (See above.)
Use the color definitions from colors-dark.c (formerly
colors.alt.c). A black background will be used (even if the terminal uses
a bright background by default!), and a set of foreground colors which look
very nice on black backgound. (But would be unusable on bright background.)
This is the default now.
Use color definitions from colors-bright.c (formerly
colors.default.c). The terminal's default colors will be used for
background and normal text, and an alternative color scheme suitable for bright
background will be used for other text types.
Use this if you have a terminal with bright background (like most xterms), and also want to stick to that in netrik.
Note that this can be used on a terminal with dark background as well; some colors are somewhat hard to read, however.
- Use terminal's default colors even with --dark-background, instead of forcing usage of netrik's default text colors (white on black for normal text). This is useful if you use the default (dark) colors and your terminal has a black background anyways -- forcing the default colors is only a waste of time in this situation.
You can also specify any of these options as default in the netrik config file ~/.netrikrc, see netrikrc(5).
- netrik http://netrik.sourceforge.net/index.html
- Load the netrik web site and start browsing.
- netrik sourceforge.net
- Load a local file named "sourceforge.net" in the current directory, or start browing http://sourceforge.net if no such local file exists.
- netrik file:///usr/local/share/doc/index.html
- Start browsing the netrik HTML documentation.
- ssh [email protected] cat foo.html|netrik -
- Load file "foo.html" from your ssh account on someshell.invalid, and display in builtin pager.
- netrik --dump foo.html
- Layout and dump the file "foo.html" from current directory.
- TERM=ansi netrik --dump foo.html >foo.rtext (bourne shell version)
- Dump (layouted) content of "foo.html" to the file "foo.rtext", which can be viewed on any ANSI compatible color terminal later. (Using "less -R" for example.)
- netrik http://foo.invalid/broken.html --dump --debug 2>&1|less -R
- Examine "http://foo.invalid/broken.html" to find the reason for some HTML error.
- netrik --broken-html freshmeat.net
- Start browsing freashmeat.net, don't halt on noncritical HTML errors.
- netrik --ignore-broken www.cnn.com
Start browsing cnn.com, don't halt on *any* HTML errors. (Expect it to look broken, but that's probably not our fault... If you think it is, please file a bug report.)
PAGER COMMANDSSimilar to vi(1), netrik basically knows two kinds of pager commands. Simple commands (presently all of them are one-letter commands) are executed directly when the corresponding key is pressed. These are all the pager movement commands, plus some more. In the following overview they are represented by just the letter for letter keys, or a symbol of the form <key> for special keys. Upper case letters mean the letter key with <shift>, and letters preceded with '^' mean the letter key with <ctrl>.
The others (presently only two) need to be typed into a command prompt (with readline(3) and all), which is activated by pressing ':', and has to be confirmed by <return>. These are indicated by a ':' before the command name. (Just as they are typed...)
MOVEMENT (SCROLLING) COMMANDS
- scroll one line forward
- scroll one line backward
- scroll two lines forward
- scroll two lines backward
- scroll one screen forward
- scroll one screen backward
- scroll one half screen forward
- scroll one half screen backward
- go to page top
go to page end
CURSOR MOVEMENT COMMANDS
- cursor left
- cursor down
- cursor up
- follow selected link (or manipulate form control)
- go to next link, or scroll one line forward (if no more links on screen)
- go to previous link, or scroll one line backward (if no more links on screen)
- go to first link on next line, or scroll one line forward (if no more links on screen)
- go to first link on previous line, or scroll one line backward (if no more links on screen)
- go to first link starting in line
- go to first link in line (different from '^' if there is a link wrapped from previous line)
- go to last link in line
- go to first link on screen
- go to last link on screen
- go to midmost link (first link in second screen half)
- go to next link
- go to previous link
- go to first link on page
activate links by label
PAGE HISTORY COMMANDS
- reload current page
- back to previous page in history
- forward to next page in history (after 'b')
- back to previous site (page before last absolute URL entered/followed)
- forward to next site
- set page mark
- remove page mark
- return to previous page in history marked with 's' (or first page)
forward to next page in history marked with 's' (or last page)
- show link URL
- show absolute link target URL
- show current page URL
- :e URL
- load document "URL" and display it in the pager (URL relative to current page)
- :E URL
- load document "URL" and display it in the pager (absolute URL)
- search for a string in current page
- quit netrik
- ^C (SIGINT)
- Interrupt file/HTTP loading (no effect otherwise)
- ^\ (SIGQUIT)
Immediately quit netrik unconditionally. (Presently, this signal violently terminates netrik; thus no cleanup takes place... Use only in "emergency".)
ENVIRONMENThttp_proxy (the uppercase variant HTTP_PROXY is also recognized, but discouraged) specifies the address of an optional proxy server.
TERM specifies a terminal type for which netrik (actually, ncurses) will produce output.
- Normal exit. (Should occur only if explicitly issued 'q' command, or after whole page has been printed with --dump.)
- Some condition occured that prevents netrik from continuing normal operation. This might be a memory allocation error, failure to open some important file etc. Note that it might also indicate some bug in netrik; if the error message printed doesn't seem to make sense (e.g. a memory allocation error when there is plenty of free RAM available), please report that. (See BUGS below.)
- Operational error: The user induced an error condition that netrik can't handle gracefully (yet), e.g. trying to follow a relative link where no base URL is available. (On a page loaded from standard input, for example.)
- One of the internal sanity checks gave alarm. This is a bug in netrik! Please report it. (See BUGS below.)
Other errors codes shouldn't occur. (Note that error codes above 128 are generated when the program is terminated by a signal, e.g. 139 for SIGSEGV, which usually also indicates a bug, unless the signal was generated by some user action.)
CONFORMING TONetrik knows most of HTML 4.01 and XHTML 1.0. There are several facilities it doesn't recognize yet, though; and some things are layouted incorrect. (Especially space handling is totaly broken.)
It doesn't conform to XHTML also for the reason that it only warns about syntax errors (rather than aborting), and some may even slip through alltogether.
Note that full standards compliance is not a primary development goal; we may ignore known minor incompatibilities, as long as they do not undermine W3C's standardization efforts, and there is more important work to do...
FILES~/.netrikrc: The netrik configuration file.
VERSIONThis manual page documents netrik version 1.16.1.
BUGSLayouting is so strange that it always looks broken...
Layouting is actually broken. (Wrong space handling.)
HTML syntax error messages are penetrating.
The UI is inconsistent.
Netrik is vaporware.
Netrik has less command line switches than ls ;-)
Please report any other problems you find to <[email protected]>. Thanks.
AUTHORNetrik was created and is maintained by Olaf D. Buddenhagen AKA antrik (<[email protected]>), with major contributions from Patrice Neff, Sören Schulze, and others. (For a full listing of all contributors see AUTHORS in the doc directory, see below.)
This man page was created by Patrice Neff and modified by antrik.