NAME

sitescooper - download news from web sites and convert it automatically into one of several formats suitable for viewing on a Palm handheld.


SYNOPSIS

sitescooper [options] [ [-site sitename] ...]

sitescooper [options] [-sites sitename ...]

sitescooper [options] [-name nm] [-levels n] [-storyurl regexp]
[-set sitefileparam value] url [...]

Options: [-debug] [-refresh] [-fullrefresh] [-config file] [-install dir] [-instapp app] [-dump] [-dumpprc] [-nowrite] [-nodates] [-quiet] [-admin cmd] [-nolinkrewrite] [-stdout-to file] [-badcache] [-keep-tmps] [-fromcache] [-noheaders] [-nofooters] [-outputtemplate file.tmpl] [-grep] [-profile file.nhp] [-profiles file.nhp file2.nhp ...] [-filename template] [-prctitle template] [-parallel] [-disc] [-limit numkbytes] [-maxlinks numlinks] [-maxstories numstories] [-timeout numsecs]

[-text | -html | -mhtml | -doc | -plucker | -mplucker | -isilo | -misilo | -misilox | -richreader | -pipe fmt command] [-bw | -color] [-maxcolors n] [-cvtargs args_for_converter]


DESCRIPTION

This script, in conjunction with its configuration file and its set of site files, will download news stories from several top news sites into text format and/or onto your Palm handheld (with the aid of the makedoc/MakeDocW or iSilo utilities).

Alternatively URLs can be supplied on the command line, in which case those URLs will be downloaded and converted using a reasonable set of default settings.

HTTP and local files, using the file:/// protocol, are both supported.

Multiple types of sites are supported:

1-level sites, where the text to be converted is all present on one page (such as Slashdot, Linux Weekly News, BluesNews, NTKnow, Ars Technica);

2-level sites, where the text to be converted is linked to from a Table of Contents page (such as Wired News, BBC News, and I, Cringely);

3-level sites, where the text to be converted is linked to from a Table of Contents page, which in turned is linked to from a list of issues page (such as PalmPower).

In addition sites that post news as items on one big page, such as Slashdot, Ars Technica, and BluesNews, are supported using diff.

Note that at this moment in time, the URLs-on-the-command-line invocation format does not support 2- or 3-level sites.

The script is portable to most UNIX variants that support perl, as well as the Win32 platform (tested with ActivePerl 5.00502 build 509).

sitescooper maintains a cache in its temporary directory; files are kept in this cache for a week at most. Ditto for the text output directory (set with TextSaveDir in the built-in configuration).

If a password is required for the site, and the current sitescooper session is interactive, the user will be prompted for the username and password. This authentication token will be saved for later use. This way a site that requires login can be set up as a .site -- just log in once, and your password is saved for future non-interactive runs.

Note however that the encryption used to hide the password in the sitescooper configuration is pretty transparent; I recommend that rather than using your own username and password to log in to passworded sites, a dedicated, sitescooper account is used instead.


OPTIONS

-refresh
Refresh all links -- ignore the already_seen file, do not diff pages, and always fetch links. If a cached page is available, it will be used.

-fullrefresh
Refresh all links -- ignore the already_seen file, do not diff pages, and always fetch links, even if they are available in the cache.

-config file
Read the configuration from file instead of using the built-in one.

-limit numkbytes
Set the limit for output file size to numkbytes kilobytes, instead of the default 200K. A limit of 0 means unlimited, any amount of output.

-maxlinks numlinks
Stop retrieving web pages after numlinks have been traversed. This is not used to specify how ``deep'' a site should be scooped -- it is the number of links followed in total.

-maxstories numstories
Stop retrieving web pages after numstories stories have been retrieved.

-timeout numsecs
Time out connection attempts after numsecs. The default is 10*60, 10 minutes.

-install dir
The directory to save PDB files to once they've been converted, in order to have them installed to your Palm handheld.

-instapp app
The application to run to install PDB files onto your Palm, once they've been converted.

-site sitename
Limit the run to the site named in the sitename argument. Normally all available sites will be downloaded. To limit the run to 2 or more sites, provide multiple -site arguments like so:
        -site ntk.site -site tbtf.site

-sites sitename [...]
Limit the run to multiple sites; an easier way to specify multiple sites than using the -site argument for each file.

-grep
Use James Brown's NewsHound profile searching code. Any sites that do not contain IgnoreProfiles: 1 will then be searched for the active profiles. Active profiles are loaded from the ProfileDir specified in the sitescooper configuration file, or specified using the -profile or -profiles arguments.

-profile file.nhp
Limit the run to the site named in the file.nhp argument. Normally all available sites will be downloaded. To limit the run to 2 or more sites, provide multiple -profile arguments like so:
        -profile ntk.site -profile tbtf.site

-profiles file.nhp [...]
Limit the run to multiple sites; an easier way to specify multiple sites than using the -profile argument for each file.

-name name
When specifying a URL on the command-line, this provides the name that should be used when installing the site to the Pilot. It acts exactly the same way as the Name: field in a site file.

-levels n
When specifying a URL on the command-line, this indicates how many levels a site has. Not needed when using .site files.

-storyurl regexp
When specifying a URL on the command-line, this indicates the regular expression which links to stories should conform to. Not needed when using .site files.

-doc
Convert the page(s) downloaded into DOC format, with all the articles listed in full, one after the other.

-text
Convert the page(s) downloaded into plain text format, with all the articles listed in full, one after the other.

-html
Convert the page(s) downloaded into HTML format, on one big page, with a table of contents (taken from the site if possible), followed by all the articles one after another.

-mhtml
Convert the page(s) downloaded into HTML format, but retain the multiple-page format. This will create the output in a directory called site_name; in conjunction with the -dump argument, it will output the path of this directory on standard output before exiting.

-plucker
Convert the page(s) downloaded into Plucker format (see http://plucker.gnu-designs.com/ ), on one big page. The page(s) will be displayed with a table of contents (taken from the site if possible), followed by all the articles one after another.

-isilo
Convert the page(s) downloaded into iSilo format (see http://www.isilo.com/ ), on one big page. This is the default. The page(s) will be displayed with a table of contents (taken from the site if possible), followed by all the articles one after another.

-misilo
Convert the page(s) downloaded into iSilo format (see http://www.isilo.com/ ), with one iSilo document per site, with each story on a separate page. The iSilo document will have a table-of-contents page, taken from the site if possible, with each article on a separate page.

-misilox
Convert the page(s) downloaded into iSilo format (see http://www.isilo.com/ ), with one iSilo document per site, with each story on a separate page. This uses the iSiloXC converter. The iSilo document will have a table-of-contents page, taken from the site if possible, with each article on a separate page.

-richreader
Convert the page(s) downloaded into RichReader format using HTML2Doc.exe (see http://users.erols.com/arenakm/palm/RichReader.html ). The page(s) will be displayed with a table of contents (taken from the site if possible), followed by all the articles one after another.

-pipe fmt command
Convert the page(s) downloaded into an arbitrary format, using the command provided. Sitescooper will still rewrite the page(s) according to the fmt argument, which should be one of:
text
Plain text format.

html
HTML in one big page.

mhtml
HTML in multiple pages.

The command argument can contain __SCOOPFILE__, which will be replaced with the filename of the file containing the rewritten pages in the above format, __SYNCFILE__, which will be replaced with a suitable filename in the Palm synchronization folder, and __TITLE__, which will be replaced by the title of the file (generally a string containing the date and site name).

Note that for the -mhtml switch, __SCOOPFILE__ will be replaced with the name of the file containing the table-of-contents page. It's up to the conversion utility to follow the href links to the other files in that directory.

-cvtargs
Arguments for the conversion utility.

-bw
Indicate that the target can display only 2-bit images, black and white only. This is generally the default for iSilo and Plucker.

-color
Indicate that the target can display color images.

-maxcolors n
Indicate that the target supports a maximum of 'n' colors. Plucker will display images better on some Palms if this is used.

-fixlinks
Rewrite links to external sites or unscooped pages as underlined text, to differentiate them from links to scooped pages. This is the default behaviour for most formats apart from -plucker or -mplucker.

-keeplinks
Do not rewrite links to external sites or unscooped pages; leave them pointing outside the current scoop. However, links to other pages that are included in the current scoop, are rewritten to point to the scooped pages instead of the source URL. This is the default for Plucker (-plucker or -mplucker arguments).

-nolinkrewrite
Do not rewrite links on scooped documents -- leave them exactly as they are. This includes even links to other scooped pages. See also -keeplinks).

-dump
Output the page(s) downloaded directly to stdout in text or HTML format, instead of writing them to files and converting each one. This option NO LONGER implies -text, like it used to, so to dump text, use -dump -text.

-dumpprc
Output the page(s) downloaded directly to stdout, in converted format as a PDB file (note: not PRC format!), suitable for installation to a Palm handheld.

-nowrite
Test mode -- do not write to the cache or already_seen file, instead write what would be written normally to a directory called new_cache and a new_already_seen file. This is very handy when writing a new site file.

-badcache
Send some HTTP headers to bypass web caching proxy servers. This is generally useful if a web caching proxy server somewhere between sitescooper and the target site is returning out-of-date files.

-debug
Enable debugging output. This output is in addition to the usual progress messages.

-quiet
Process sites quietly, without printing the usual progress messages to STDERR. Warnings about incorrect site files and system errors will still be output, however.

-admin cmd
Perform an administrative command. This is intended to ease the task of writing scripts which use sitescooper output. The following admin commands are available:
dump-sites
List the sites which would be scooped on a scooping run, and their URLs. Instead of scooping any sites, sitescooper will exit after performing this task. The format is one site per line, with the site file name first, a tab, the site's URL, a tab, the site name, a tab, and the output filename that would be generated without path or extension. For example:

foobar.site http://www.foobar.com/ Foo Bar 1999_01_01_Foo_Bar

journal
Write a journal with dumps of the documents as they pass through the formatting and stripping steps of the scooping process. This is written to a file called journal in the sitescooper temporary directory.

import-cookies file
Import a Netscape cookies file into sitescooper, so that certain sites which require them, can use them. For example, the site economist_full.site requires this. Here's how to import cookies on a UNIX machine:

sitescooper.pl -admin import-cookies ~/.netscape/cookies

and on Windows:

perl sitescooper.pl -admin import-cookies   ``C:\Program Files\Netscape\Users\Default\cookies.txt''

Unfortunately, MS Internet Explorer cookies are currently unsupported. If you wish to write a patch to support them, that'd be great.

-noheaders
Do not attach the sitescooper header (URL, site name, and navigation links) to each page.

-nofooters
Do not attach the sitescooper footer (``copyright retained by original authors'' blurb) to each page.

-outputtemplate file.tmpl
Read the output formatting template from the file file.tmpl. This overrides the settings of the -noheaders and -nofooters flags. See the OUTPUT TEMPLATES section below for details on this.

-fromcache
Do not perform any network access, retrieve everything from the cache or the shared cache.

-filename template
Change the format of output filenames. template contains the following keyword strings, which are substituted as follows:
YYYY
The current year, in 4-digit format.

MM
The current month number (from 01 to 12), in 2-digit format.

Mon
The current month name (from Jan to Dec), in 3-letter format.

DD
The current day of the month (from 01 to 31), in 2-digit format.

Day
The current day of the week (from Sun to Sat), in 3-letter format.

hh
The current hour (from 00 to 23), in 2-digit format.

mm
The current minute (from 00 to 59), in 2-digit format.

Site
The current site's name.

Section
The section of the current site (now obsolete).

The default filename template is YYYY_MM_DD_Site.

-prctitle template
Change the format of the titles of the resulting PDB files. template may contain the same keyword strings as -filename.

The default PDB title template is YYYY-Mon-DD: Site.

-nodates
Do not put the date in the installable file's filename. This allows you to automatically overwrite old files with new ones when you HotSync. It's a compatibility shortcut for -filename Site -prctitle ``Site''.

-preload preload_method
Preload pages using the given preload method. Currently supported preload methods are:
lwp
Use the Perl LWP module to load pages. This is the default, and is single-threaded; in other words, each page needs to load fully before the next page can be requested.

fork[n]
Use a number of subprocesses running LWP requests to load pages. This is multi-threaded, and several pages can be loaded at once; however you pay in costs of network bandwidth, CPU time and memory used. The optional n argument instructs sitescooper to use that number of processes; the default n is 4. This is only available on UNIX at the moment.

-disc
Disconnect a PPP connection once the scooping has finished. Currently this code is experimental, and will probably only work on Macintoshes. This is off by default.

-stdout-to file
Redirect the output of sitescooper into the named file. This is needed on Windows NT and 95, where certain combinations of perl and Windows do not seem to support the > operator.

-keep-tmps
Keep temporary files after conversion. Normally the .txt or .html rendition of a site is deleted after conversion; this option keeps it around.


OUTPUT TEMPLATES

You can control exactly what HTML or text is written to the output file using the -outputtemplate argument. This argument takes the name of a file, which is read and parsed to provide replacement templates for sitescooper.

The file is read as a HTML- or XML-style tagged format; so for example the template for the main page in HTML format is read from between the <htmlmainpage> and </htmlmainpage> tags. The templates that can be defined are as follows:

htmlmainpage
The main page, in HTML format; this is used when the -html output format, or one based on it (such as -plucker or -isilo), is used. It is also used for the -mhtml format's main (top-level) page.

htmlsubpage
Sub-page, in HTML format; this is used for the -mhtml output format's sub pages, ie. pages other than the top-level one.

htmlstory
The snippet of HTML encapsulating each story. This is included for each piece of snarfed text, in all HTML files.

textmainpage
The main page, in text format; this is used when the -text output format, or one based on it (such as -doc), is used.

textsubpage
Sub-page, in text format; this is currently unused.

textstory
The snippet of text encapsulating each story. This is included for each piece of snarfed text, in all text-format or DOC-format files.

A sample template file is provided in the file default_templates.html; this may have been installed in the sitescooper install directory, /usr/share/sitescooper, or /usr/local/share/sitescooper. Note that the actual templates used are not loaded from this file; instead they are incorporated inside the sitescooper script, so changing this file will have no effect.


INSTALLATION

To install, edit the script and change the #! line. You may also need to (a) change the Pilot install dir if you plan to use the pilot installation functionality, and (b) edit the other parameters marked with CUSTOMISE in case they need to be customised for your site. They should be set to acceptable defaults (unless I forgot to comment out the proxy server lines I use ;).


EXAMPLES

        sitescooper.pl http://www.ntk.net/

To snarf the ever-cutting NTKnow newsletter.

        sitescooper.pl -refresh -html http://www.ntk.net/

To snarf NTKnow, ignoring any previously-read text, and producing HTML output.

        sitescooper.pl -refresh -html -site site_samples/tech/ntk.site

To snarf NTKnow using the site file provided with the main distribution, producing HTML output.


ENVIRONMENT

sitescooper makes use of the $http_proxy environment variable, if it is set.


AUTHOR

Justin Mason <jm /at/ jmason.org>


COPYRIGHT

Copyright (C) 1999-2000 Justin Mason

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA, or read it on the web at http://www.gnu.org/copyleft/gpl.html .


SCRIPT CATEGORIES

The CPAN script category for this script is Web. See http://www.cpan.org/scripts/ .


PREREQUISITES

File::Find File::Copy File::Path FindBin Carp Cwd URI::URL LWP::UserAgent HTTP::Request::Common HTTP::Date HTML::Entities

All these can be picked up from CPAN at http://www.cpan.org/ . Note that HTML::Entities is actually included in one of the previous packages, so you do not need to install it separately.


COREQUISITES

Win32::TieRegistry will be used, if running on a Win32 platform, to find the Pilot Desktop software's installation directory. Algorithm::Diff to support diffing sites without running an external diff application (this is required on Mac systems).


README

Sitescooper downloads news stories from the web and converts them to Palm handheld iSilo, DOC or text format for later reading on-the-move. Site files and full documentation can be found at http://sitescooper.org/ .