Ponfish - A Binary-only Newsreader For Unix

See manual.txt for more information.



Keywords:

Usenet Newsreader Linux Free Binaries Decode Unix Newsgroups




manual.txt duplicated below:
----------------------------

Ponfish:  Binary Newsreader for Unix and Windows.*

* Actually, it's more of a newsleecher.


Disclaimer
----------------------------------------------------------------------
Please use and enjoy this newsreader, but I can't take any
responsibility for anything bad that might happen as a result of
using this.  (It's possible, though highly unlikely, that the use of
this program could recursively delete the root directory)


Purpose / Goals
----------------------------------------------------------------------
To write a decent newsreader that I can use in Linux.

My main requirements at the start of this project were:

- Must handle reading 30 million headers worth of news at a time (and
  still be responsive with acceptable memory usage).
- Must work in Windows as well as Linux.
- Must be able to work offline.
- Must be able to transfer the download queue to different machines.
- Must not require any non-core Perl modules.
- Must decode Yenc.


Implementation
----------------------------------------------------------------------
I wanted something useable ASAP.  In under two weeks of part time
programming, I had a stable and useable skeleton.  Functionality
was added as I needed it.

I opted for a road-warrior style of programming, I wanted to get the
job quickly after all.  The code isn't the nicest I've ever written,
but it's very straight forward.  (Actually, I'm almost embarrassed to
release it)

I chose to use a terminal interface which uses a few ugly hacks for
functionality.  This allowed me to keep the design exceedingly simple.
It also meant that it has more or less a command-line interface.

After getting thoughts of releasing this monstrosity onto the general
public, I added help features and fixed as many of the glaring bugs
as possible.

(I actually have a history of writing newsreaders... my first one
written in Perl/Tk worked quite well, but consumed about 100MB of ram
for each 100K articles)


Features:
----------------------------------------------------------------------
* It's free, which always helps.
* Runs in both Linux and Windows.
* Highly memory efficient - reading a newsgroup with over 15 million
  headers on a machine with only 512K of RAM will not force the system
  to it's knees.
* Uses Perl's built in (and very powerful) regular expressions to
  offer never before seen search and destroy functionality.
* Multiple download threads to max out your bandwidth.
* Intelligently managed download queue that downloads the oldest posts
  first, and which persists through crashes. (Why don't other
  newsreaders do this?  It's the most important part of a newsreader!)
* It is very fast.
* It is very powerful.
* It's a terminal program, so you can access it remotely.
* If you're running it in Linux, you will have access to funky colors
  and eventually, perhaps a better (curses based) interface.
* Full non-interactive help.


Tested Environments:
----------------------------------------------------------------------
- Linux (various) / Perl 5.8.8
- Windows 2000 & XP Media Ed / AS Perl 5.8.8


Installation Instructions:
----------------------------------------------------------------------
Unix & Windows:

Unpack into any directory.  Run the programs from this directory.*

* It's extremely beta - do not run a "make install" please.

Requirements:
----------------------------------------------------------------------
You will need to have uudeview installed in the path.  This is
available for both Linux and Windows.  If not, downloaded articles
will not decode for you.  For Windows, make sure you install the
command-line version of uudeview.

You will also need par2repair and rar, as standard tools for repairing
and extracting most usenet binaries.

Bugs/To Do:
----------------------------------------------------------------------
- The save functionality is only partially implemented.  It works, but
the command files are currently not being deleted, and you can't
specify how or where these files are saved.  Try to avoid using
this if you can, clean the command files manually if you do.
- The junk folder (where all the complete command files go) is not
cleaned out.  It will eventually grow huge and will need manual
deletion.
- The manner in which the downloads / decodes happen increase drive
fragmentation.  While not a huge problem, eventually I want to address
this issue.
- The help system is slightly inaccurate.


Quickstart Guide:
----------------------------------------------------------------------

-------------------------- WARNING -----------------------------------

This program was written by a programmer for use by a programmer.
It has no GUI, a scant manual, and minimal help.  Good luck.

----------------------- END OF WARNING -------------------------------


######################################################################
Below is a 'manual' of sorts, but it's getting out of date.  Instead
of maintaining it, I will re-write it when/if the next release comes
out
######################################################################


There are two programs you will need to run: ponfishr (the READER),
and ponfishd (the DOWNLOADER).

Ponfishr is the newsreader client, and ponfishd is the
downloading service.  Ponfishr does not connect to Usenet, it only
issues commands that ponfishd queues up and services.

For the rest of this document, I'll call ponfishr the READER, and
ponfishd the DAEMON.

Run the daemon in one terminal.  Make sure you do not run two
instances of this program or your downloads will likely fail.*

* I will eventually write some code that prevents two instances of
  the daemon running

Run the reader in another terminal.  Make sure the terminal is nice
and large, you'll need all the visible space you can get.  I will
normally maximize the terminal and use a small fixed font to get
around 300 characters width.

The following sections now describe usage within the reader:


--- GETTING HELP:

At any time, you can type "?" or "help" to get a (hopefully)
context-sensitive list of help topics.  You can drill deeper by
issuing a "help TOPIC" to access the help on any particular topic.


--- ADDING NEWS SERVERS:

It's been a while since I've had to add a new server, and I've
actually forgotten how.  I type at the prompt "help", which
brings up a list of topics.  I see the "add_server" command,
which has the alias "add" (it's in brackets).

I decide to see how to use it by typing "help add" and get the
usage:

Usage:

  add_server NAME,SERVER[,USERNAME,PASSWORD,TIMEOUT,CONNECTIONS

At the prompt, I type

> add bogus, news.bogus.com, chowda222, imagoat, 60, 5

Now I can see that the news server "bogus" appears in my list of
servers.

It's time to explain that your username and password are going to
be stored in a plain text file.  If you don't feel this meets your
stringent security standards, please submit a patch to the source
code and I'll be happy to integrate the changes.

I really don't want "bogus", since this was just to illustrage,
so I go to help and figure out that I can remove it like so:

> remove bogus


--- SERVERS LIST: Downloading the newsgroup list:

The first screen you'll see is the SERVERS LIST.  To download header
data for the server(s), you'll need to run the "sync_groups" command:

------------------------------------------------------------------
01. Shaw                (Wed Sep  6 12:46:32 2006)
02. Giganews            (Wed Sep  6 12:31:41 2006)
------------------------------------------------------------------
Comamnd: > sy 1-2

This will queue sync commands for both these servers, which will take
a minute or so to be processed.  To check that what you expect is
working, switch to the terminal you ran the daemon in.  You should
see some activity - or an error message!  Otherwise probably something
is very wrong.


--- COMMANDS: How they work:

The interface this program implements is a list of choices to choose
from, numbered 01, 02, 03, etc, all the way down to the end of the
page.  In general, all commands that work on a selection, such as the
selection of articles to decode, can be given numbers separated by
spaces, or a range.  For example:

Command: > de 1-8 20-30

This will issue decode commands for articles 1 to 8 and 20 to 30.

Command: > de 1 2 3 8 18

This does as expected.


--- SERVERS MENU: Selecting a server to read:

You can read a newsgroup by simply typing the number at the prompt:

------------------------------------------------------------------
01. Shaw                (Wed Sep  6 12:46:32 2006)
02. Giganews            (Wed Sep  6 12:31:41 2006)
------------------------------------------------------------------
Comamnd: > 2


Note that at each menu, depending on where you are (ie: server list,
newsgroup list, articles list), there is a default command that is
invoked by typing a number, numbers, or range.  You can see what that
default command is in the HELP - it is the command with the "*"
as the abbreviation.


--- NEWSGROUPS MENU: Displaying ALL newsgroups:

The next screen will be the NEWSGROUPS MENU.

It defaults to displaying SUBSCRIBED newsgroups, which will be empty
on your first run.  Type the "all" command to view the entire
newsgroup list.

Command: > all


--- NEWSGROUPS MENU: Resizing the text area:

Displayed will probably get the first 20-70 newsgroups.  If you're
running in Linux, the program will use the entire size of the terminal
and will resize automatically when the terminal is resized.  If you're
running in Windows, you have to manually resize the screen using the
"resize" command:

Command: > resize WIDTH HEIGHT


Just substitute the appropriate WIDTH and HEIGHT for your terminal
here.

* NOTE: If your terminal supports it, you will see some lines
  highlighted.  This is to make it easier to select the correct menu
  number, as well as to highlight particular headers when reading a
  newsgroup.  Sorry Windows users - I've tried but can't find a way
  to change the colors.

When you're happy with the state of your terminal, continue on...


--- NEWSGROUPS MENU: Searching

The newsgroup list isn't too useful in itself - we need to be able to
narrow down the list, which consists of between 60-120 THOUSAND
newsgroups.  To do this use the filter command by typing a slash "/",
followed by a REGEXP.  If you don't know what a REGEXP is, I refer you
to either the perlre man page or a Regular Expression tutorial on the
internet.  Barring that, just type in part of the name of a newsgroup
you want to search on:

Command: > /multimedia.cartoon


Filtered: 8 groups!
01.  1434915 - alt.binaries.multimedia.cartoons
02.        4 - alt.binaries.multimedia.cartoons.french
03.        1 - alt.binaries.multimedia.cartoons.looneytoons
04.   439860 - alt.binaries.multimedia.cartoons.looneytunes
05.    41259 - alt.binaries.multimedia.cartoons.repost
06.        0 - alt.binaries.multimedia.cartoons.scoobydoo
07.        4 - alt.binaries.multimedia.cartoons.video-games
08.   101179 - alt.binaries.multimedia.cartoons.vintage
Comamnd: >

The filter command will probably be the most used in your arsenal,
which you will use here and in the ARTICLES MENU to search posts.


--- NEWSGROUPS MENU: Downloading headers

Use the sync (sy) command to download headers, for example, to
download the headers for alt.binaries.multimedia.cartoons and
alt.binaries.multimedia.cartoons.repost, do the following:

alt.binaries.multimedia.cartoons
Filtered: 8 groups!
01.  1434915 - alt.binaries.multimedia.cartoons
02.        4 - alt.binaries.multimedia.cartoons.french
03.        1 - alt.binaries.multimedia.cartoons.looneytoons
04.   439860 - alt.binaries.multimedia.cartoons.looneytunes
05.    41259 - alt.binaries.multimedia.cartoons.repost
06.        0 - alt.binaries.multimedia.cartoons.scoobydoo
07.        4 - alt.binaries.multimedia.cartoons.video-games
08.   101179 - alt.binaries.multimedia.cartoons.vintage
Comamnd: > sy 1 4

Note that there is a sync command for servers as well.  They
are named the same and logically do a similar job, but of course
are quite different.  The program knows which menu you are on and
does what you expect.

Remember that the daemon will be doing the networking tasks in the
other terminal.  It may take a while before all headers for a
newsgroup are downloaded.


-- NEWSGROUPS MENU:  Reading a newsgroup:

In the above example, type in the number of a newsgroup you have
downloaded headers for.  Remember there may be a delay between issuing
the command to download headers, and having the headers be
downloaded.  You may, however, read a newsgroup WHILE the articles are
being downloaded - you will only be able to see the ones that have
been downloaded at the time you began reading the newsgroup.

Command: > 1


--- ARTICLES MENU: Functionality:

By now you kind of have an idea of how the newsreader works.  The
in-program help is more up to date than this text file, but here
are a list of commands for you to peruse:

* de RANGE  (ie: de 1-8 12, 19):  Queues the specified articles for
  decode.  You can do fancy things like this:

  de RANGE > DIR

  To have the decoded binaries stored in DIR.

* sa RANGE: Saves a range of articles.

* pr RANGE: Same as "de", only it jumps the decode queue.  It will
  take highest precedence, and will be put to the front of the decode
  queue.

* scan REGEXP: Positions the first article on the page to the first
  article matching REGEXP.

* scan date MMM [DD]: Like scan above, but to a particular month (ie:
  aug, sep), and optional day.

* filter REGEXP: (same as /REGEXP)  Filters out only those articles
  matching REGEXP.

* so sub|date|poster:  Sorts the list by Subject, Date, or Poster.

* fresher MMM [DD]: Keeps only articles newer than MMM-DD, where DD is
  optional, and MMM is a 3 character month code.

If you want to replace the original article list, use the ++ and --
prefixes.  Here's an example:

To remove (from memory) all articles that were posted prior to August:

Command: > ++ fresher aug


To remove any articles that do not match REGEXP from memory:

Command: > ++/REGEXP


To remove any articles from memory that DO MATCH REGEXP:

Command: > --/REGEXP


This functionality allows you to narrow down searches incrementally.


--- PREFIXES:

The following prefixes can be used in front of certain commands:

  + - ++ --

The plus means keep, the minus means remove.  As explained weakly
above, there are two lists maintained for the menu at all times - the
main list and the filtered list.  The main list never changes, unless
you use the -- or ++ prefixes.  The rules I can imagine must be
somewhat confusing.  (not for me of course, this is my own invention
after all)

Here's an example to illustrate:

I will first display a list of newsgroups.  Now let's say I am
interested in finding pictures of dogs.  I will do a simple filter,
like so:

> /pet

The result is a list of 560 newsgroups!  Too many to go through
manually, but I know I am interested in pictures, so a binary
newsgroup.  I could filter on the text "binar" now:

> /binar

Now I have a list of 6938 newsgroups!  What is happening?  The main
list contains all 100_000 plus newsgroups.  The display list contains
the current search results.  When I do a filter: "/such-and-such", the
filter works off the main list and populates the display list.

I can use a prefix to force the filter to operate on the display list:

> /pet
> +/binar

This first populates the display list with any newsgroups containing
the text "pet", and then filters that list for any newsgroups
containing the text "binar":

>>>filter<<< (binar)
FILTER: 'binar'
Filtered: 6 groups!
01.       14 - alt.binaries.games.petz
02.  1006824 - alt.binaries.models.petite
03.      126 - alt.binaries.multimedia.erotica.peternorth
04.        0 - alt.binaries.pets.birds
05.       96 - alt.binaries.pictures.erotica.jay-denebeim.and.his.pet.pig
06.      155 - alt.binaries.pictures.petra-verkaik
Comamnd: >


Now let's say I want to do a bunch of searching, and I know I'm ONLY
interested in binary newsgroups.  I can go ahead and remove any
non-binary newsgroups from my main list like so:

> ++/binar

What is actually happening is that the display list is being populated
with any newsgroups matching "binar", and the ++ then tells the
display list to replace the main list.  Now the main list will contain
only the 6938 newsgroups containing the string "binar".

Subsequent searches will be limited to only those newsgroups.

To re-load the main list, type the "all" command, which load all the
newsgroups from the newsgroups file.  (Or the "sub" command, which
will load all the 'subscribed' newsgroups)


--- TECHNIQUES:

I encourage experimenting with the features to find what works for
you.  Here are some tips that might help you get started:


--- TECHNIQUES: Use the "++ fresher DATE" sequence.

Because currently this newsreader does not track what you have or have
not read, you will need this to filter out the old headers.  For
example, after syncing the articles for a newsgroup not read since
February 15th, use the command:

> ++fresher feb 15

To limit the list to articles posted on or after Feb 15.


--- TECHNIQUES: Use pr and ponfishd with the -p option.

If you run the daemon with the -p option (preview), the only executed
newsgroup commands that the daemon will execute are "preview", and
"sync".  This allows you to preview posts / newsgroups quickly and
queue downloads but not have those downloads begin until you restart
the daemon without the -p option.


--- TECHNIQUES: Use the "poster" command.

Use the "poster" command to locate all articles posted by a particular
persona.  When you find things of interest, this will allow you to
find posts by the same person.


--- FURTHER READING:

The single best thing you can do is to read up on regular
expressions.  The perl manpage is a good place to start: at a shell
prompt, type "man perlre".


--- AUTHOR:

Des Lee:	desimus.prime / gmail.com

All comments / suggestions welcome

If you're interested in signing up for a good Usenet provider,
please consider signing up with Giganews via this link:

http://www.giganews.com/?c=gn289495

Or using my referral id: gn289495

(I get credited $15 or $20 for each referee who stays on
for 90 days)