NAME

WebFetch - Perl module to download and save information from the Web

SYNOPSIS

use WebFetch;

DESCRIPTION

The WebFetch module is a general framework for downloading and saving information from the web, and for display on the web. It requires another module to inherit it and fill in the specifics of what and how to download. WebFetch provides a generalized interface for saving to a file while keeping the previous version as a backup. This is expected to be used for periodically-updated information which is run as a cron job.

INSTALLATION

After unpacking and the module sources from the tar file, run

perl Makefile.PL

make

make install

Or from a CPAN shell you can simply type "install WebFetch" and it will download, build and install it for you.

If you need help setting up a separate area to install the modules (i.e. if you don't have write permission where perl keeps its modules) then see the Perl FAQ.

To begin using the WebFetch modules, you will need to test your fetch operations manually, put them into a crontab, and then use server-side include (SSI) or a similar server configuration to include the files in a live web page.

MANUALLY TESTING A FETCH OPERATION

Select a directory which will be the storage area for files created by WebFetch. This is an important administrative decision - keep the volatile automatically-generated files in their own directory so they'll be separated from manually-maintained files.

Choose the specific WebFetch-derived modules that do the work you want. See their particular manual/web pages for details on command-line arguments. Test run them first before committing to a crontab.

SETTING UP CRONTAB ENTRIES

First of all, if you don't have crontab access or don't know what they are, contact your site's system administrator(s). Only local help will do any good on local-configuration issues. No one on the Internet can help. (If you are the administrator for your system, see the crontab(1) and crontab(5) manpages and nearly any book on Unix system administration.)

Since the WebFetch command lines are usually very long, you may prefer to make one or more scripts as front-ends so your crontab entries aren't so huge.

Do not run the crontab entries too often - be a good net.citizen and do your updates no more often than necessary. Popular sites need their users to refrain from making automated requests too often because they add up on an enormous scale on the Internet. Some sites such as Freshmeat prefer no shorter than hourly intervals. Slashdot prefers no shorter than half-hourly intervals. When in doubt, ask the site maintainers what they prefer.

(Then again, there are a very few sites like Yahoo and CNN who don't mind getting the extra hits if you're going to create links to them. Even so, more often than every 20 minutes would still be excessive to the biggest web sites.)

SETTING UP SERVER-SIDE INCLUDES

See the manual for your web server to make sure you have server-side include (SSI) enabled for the files that need it. (It's wasteful to enable it for all your files so be careful.)

When using Apache HTTPD, a line like this will include a WebFetch-generated file:

<!--#include file="fetch/slashdot.html"-->

WebFetch FUNCTIONS

The following function definitions assume $obj is a blessed reference to a module that is derived from (inherits from) WebFetch.

import( "param-name" => "value", ... )
WebFetch::module_register( $module, @capabilities );

This function allows a Perl module to register itself with the WebFetch API as able to perform various capabilities.

For subclasses of WebFetch, it can be called as a class method. __PACKAGE__-&gt;module_register( @capabilities );

For the $module parameter, the Perl module should provide its own name, usually via the __PACKAGE__ string.

The @capabilities array is any number of strings as needed to list the capabilities which the module performs for the WebFetch API. The currently-recognized capabilities are "cmdline", "input" and "output". "config" and "storage" are reserved for future use. The function will save all the capability names that the module provides.

fetch_main

This function is exported into the main package. For all modules which registered with an "input" capability for the requested file format at the time this is called, it will call the run() function on behalf of each of the packages.

eval_wrapper ( $code, $throw_func, [ name => value, ...] )

Do not use the new() function directly from WebFetch.

Use the new function from a derived class, not directly from WebFetch. The WebFetch module itself is just infrastructure for the other modules, and contains none of the details needed to complete any specific fetches.

$obj->init( ... )

This is called from the new function of all WebFetch modules. It takes "name" => "value" pairs which are all placed verbatim as attributes in $obj.

WebFetch::mod_load ( $class )

This specifies a WebFetch module (Perl class) which needs to be loaded. In case of an error, it throws an exception.

WebFetch::run

This function can be called by the main::fetch_main function provided by WebFetch or by another user function. This handles command-line processing for some standard options, calling the module-specific fetch function and WebFetch's $obj->save function to save the contents to one or more files.

The command-line processing for some standard options are as follows:

--dir directory

(required) the directory in which to write output files

--group group

(optional) the group ID to set the output file(s) to

--mode mode

(optional) the file mode (permissions) to set the output file(s) to

--save_file save-file-path

(optional) save a copy of the fetched info in the file named by this parameter. The contents of the file are determined by the --dest_format parameter. If --dest_format isn't defined but only one module has registered a file format for saving, then that will be used by default.

--quiet

(optional) suppress printed warnings for HTTP errors (applies only to modules which use the WebFetch::get() function) in case they are not desired for cron outputs

--debug

(optional) print verbose debugging outputs, only useful for developers adding new WebFetch-based modules or finding/reporting a bug in an existing module

Modules derived from WebFetch may add their own command-line options that WebFetch::run() will use by defining a variable called @Options in the calling module, using the name/value pairs defined in Perl's Getopts::Long module. Derived modules can also add to the command-line usage error message by defining a variable called $Usage with a string of the additional parameters, as they should appear in the usage message.

$obj->do_actions

do_actions was added in WebFetch 0.10 as part of the WebFetch Embedding API. Upon entry to this function, $obj must contain the following attributes:

data

is a reference to a hash containing the following three (required) keys:

fields

is a reference to an array containing the names of the fetched data fields in the order they appear in the records of the data array. This is necessary to define what each field is called because any kind of data can be fetched from the web.

wk_names

is a reference to a hash which maps from a key string with a "well-known" (to WebFetch) field type to a field name used in this table. The well-known names are defined as follows:

title

a one-liner banner or title text (plain text, no HTML tags)

url

URL or file path (as appropriate) to the news source

date

a date stamp, which must be program-readable by Perl's Date::Calc module in the Parse_Date() function in order to support timestamp-related comparisons and processing that some users have requested. If the date cannot be parsed by Date::Calc, either translate it when your module captures it, or do not define this "well-known" field because it wouldn't fit the definition. (plain text, no HTML tags)

summary

a paragraph of summary text in HTML

comments

number of comments/replies at the news site (plain text, no HTML tags)

author

a name, handle or login name representing the author of the news item (plain text, no HTML tags)

category

a word or short phrase representing the category, topic or department of the news item (plain text, no HTML tags)

location

a location associated with the news item (plain text, no HTML tags)

The field names for this table are defined in the fields array.

The hash only maps for the fields available in the table. If no field representing a given well-known name is present in the data fields, that well-known name key must not be defined in this hash.

records

an array containing the data records. Each record is itself a reference to an array of strings which are the data fields. This is effectively a two-dimensional array or a table.

Only one table-type set of data is permitted per fetch operation. If more are needed, they should be arranged as separate fetches with different parameters.

actions

is a reference to a hash. The hash keys are names for handler functions. The WebFetch core provides internal handler functions called fmt_handler_html (for HTML output), fmt_handler_xml (for XML output), fmt_handler_wf (for WebFetch::General format), However, WebFetch modules may provide additional format handler functions of their own by prepending "fmt_handler_" to the key string used in the actions array.

The values are array references containing "action specs", which are themselves arrays of parameters that will be passed to the handler functions for generating output in a specific format. There may be more than one entry for a given format if multiple outputs with different parameters are needed.

The presence of values in this field mean that output is to be generated in the specified format. The presence of these would have been chosed by the WebFetch module that created them - possibly by default settings or by a command-line argument that directed a specific output format to be used.

For each valid action spec, a separate "savable" (contents to be placed in a file) will be generated from the contents of the data variable.

The valid (but all optional) keys are

html

the value must be a reference to an array which specifies all the HTML generation (html_gen) operations that will take place upon the data. Each entry in the array is itself an array reference, containing the following parameters for a call to html_gen():

filename

a file name or path string (relative to the WebFetch output directory unless a full path is given) for output of HTML text.

params

a hash reference containing optional name/value parameters for the HTML format handler.

filter_func

(optional) a reference to code that, given a reference to an entry in @{$self->{data}{records}}, returns true (1) or false (0) for whether it will be included in the HTML output. By default, all records are included.

sort_func

(optional) a reference to code that, given references to two entries in @{$self->{data}{records}}, returns the sort comparison value for the order they should be in. By default, no sorting is done and all records (subject to filtering) are accepted in order.

format_func

(optional) a refernce to code that, given a reference to an entry in @{$self->{data}{records}}, stores a savable representation of the string.

Additional valid keys may be created by modules that inherit from WebFetch by supplying a method/function named with "fmt_handler_" preceding the string used for the key. For example, for an "xyz" format, the handler function would be fmt_handler_xyz. The value (the "action spec") of the hash entry must be an array reference. Within that array are "action spec entries", each of which is a reference to an array containing the list of parameters that will be passed verbatim to the fmt_handler_xyz function.

When the format handler function returns, it is expected to have created entries in the $obj->{savables} array (even if they only contain error messages explaining a failure), which will be used by $obj->save() to save the files and print the error messages.

For coding examples, use the fmt_handler_* functions in WebFetch.pm itself.

$obj->fetch

This function must be provided by each derived module to perform the fetch operaton specific to that module. It will be called from new() so you should not call it directly. Your fetch function should extract some data from somewhere and place of it in HTML or other meaningful form in the "savable" array.

TODO: cleanup references to WebFetch 0.09 and 0.10 APIs.

Upon entry to this function, $obj must contain the following attributes:

dir

The name of the directory to save in. (If called from the command-line, this will already have been provided by the required --dir parameter.)

savable

a reference to an array where the "savable" items will be placed by the $obj->fetch function. (You only need to provide an array reference - other WebFetch functions can write to it.)

In WebFetch 0.10 and later, this parameter should no longer be supplied by the fetch function (unless you wish to use 0.09 backward compatibility) because it is filled in by the do_actions after the fetch function is completed based on the data and actions variables that are set in the fetch function. (See below.)

Each entry of the savable array is a hash reference with the following attributes:

file

file name to save in

content

scalar w/ entire text or raw content to write to the file

group

(optional) group setting to apply to file

mode

(optional) file permissions to apply to file

Contents of savable items may be generated directly by derived modules or with WebFetch's html_gen, html_savable or raw_savable functions. These functions will set the group and mode parameters from the object's own settings, which in turn could have originated from the WebFetch command-line if this was called that way.

Note that the fetch functions requirements changed in WebFetch 0.10. The old requirement (0.09 and earlier) is supported for backward compatibility.

In WebFetch 0.09 and earlier, upon exit from this function, the $obj->savable array must contain one entry for each file to be saved. More than one array entry means more than one file to save. The WebFetch infrastructure will save them, retaining backup copies and setting file modes as needed.

Beginning in WebFetch 0.10, the "WebFetch embedding" capability was introduced. In order to do this, the captured data of the fetch function had to be externalized where other Perl routines could access it. So the fetch function now only populates data structures (including code references necessary to process the data.)

Upon exit from the function, the following variables must be set in $obj:

data

is a reference to a hash which will be used by the do_actions function. (See above.)

actions

is a reference to a hash which will be used by the do_actions function. (See above.)

$obj->get

This WebFetch utility function will get a URL and return a reference to a scalar with the retrieved contents. Upon entry to this function, $obj must contain the following attributes:

source

the URL to get

quiet

a flag which, when set to a non-zero (true) value, suppresses printing of HTTP request errors on STDERR

$obj->html_savable( $filename, $content )

In WebFetch 0.10 and later, this should be used only in format handler functions. See do_actions() for details.

This WebFetch utility function stores pre-generated HTML in a new entry in the $obj->{savable} array, for later writing to a file. It's basically a simple wrapper that puts HTML comments warning that it's machine-generated around the provided HTML text. This is generally a good idea so that neophyte webmasters (and you know there are a lot of them in the world :-) will see the warning before trying to manually modify your automatically-generated text.

See $obj->fetch for details on the contents of the savable parameter

$obj->raw_savable( $filename, $content )

In WebFetch 0.10 and later, this should be used only in format handler functions. See do_actions() for details.

This WebFetch utility function stores any raw content and a filename in the $obj->{savable} array, in preparation for writing to that file. (The actual save operation may also automatically include keeping backup files and setting the group and mode of the file.)

See $obj->fetch for details on the contents of the savable parameter

$obj->direct_fetch_savable( $filename, $source )

This should be used only in format handler functions. See do_actions() for details.

This adds a task for the save function to fetch a URL and save it verbatim in a file. This can be used to download links contained in a news feed.

$obj->save

This WebFetch utility function goes through all the entries in the $obj->{savable} array and saves their contents, providing several services such as keeping backup copies, and setting the group and mode of the file, if requested to do so.

If you call a WebFetch-derived module from the command-line run() or fetch_main() functions, this will already be done for you. Otherwise you will need to call it after populating the savable array with one entry per file to save.

Upon entry to this function, $obj must contain the following attributes:

dir

directory to save files in

savable

names and contents for files to save

See $obj->fetch for details on the contents of the savable parameter

WRITING WebFetch-DERIVED MODULES

The easiest way to make a new WebFetch-derived module is to start from the module closest to your fetch operation and modify it. Make sure to change all of the following:

fetch function

The fetch function is the meat of the operation. Get the desired info from a local file or remote site and place the contents that need to be saved in the savable parameter.

module name

Be sure to catch and change them all.

file names

The code and documentation may refer to output files by name.

module parameters

Change the URL, number of links, etc as necessary.

command-line parameters

If you need to add command-line parameters, modify both the @Options and $Usage variables. Don't forget to add documentation for your command-line options and remove old documentation for any you removed.

When adding documentation, if the existing formatting isn't enough for your changes, there's more information about Perl's POD ("plain old documentation") embedded documentation format at http://www.cpan.org/doc/manual/html/pod/perlpod.html

authors

Add yourself as an author if you added any significant functionality. But if you used anyone else's code, retain the existing author credits in any module you modify to make a new one.

Please consider contributing any useful changes back to the WebFetch project at maint@webfetch.org.

ACKNOWLEDGEMENTS

WebFetch was written by Ian Kluft Send patches, bug reports, suggestions and questions to maint@webfetch.org.

LICENSE

WebFetch is Open Source software distributed via the Comprehensive Perl Archive Network (CPAN), a worldwide network of Perl web mirror sites. WebFetch may be copied under the same terms and licensing as Perl itelf.

SEE ALSO

A current copy of the source code and documentation may be found at http://www.webfetch.org/

TODO: fill in these lists

perl(1), WebFetch::Input::PerlStruct, WebFetch::Input::SiteNews, WebFetch::Input::Atom, WebFetch::Input::RSS, WebFetch::Input::Dump, WebFetch::Output::TT, WebFetch::Output::Dump,

5 POD Errors

The following errors were encountered while parsing the POD:

Around line 380:

You forgot a '=back' before '=head2'

Around line 537:

'=item' outside of any '=over'

Around line 938:

=back without =over

Around line 1005:

'=item' outside of any '=over'

Around line 1631:

You forgot a '=back' before '=head2'