@node Usage
@unnumbered Usage

How @strong{I} use it:

@table @asis

@item Get its source code

@example
$ git clone git://git.stargrave.org/feeder.git
$ cd feeder
@end example

@item Compile @command{feed2mdir} utility

@example
$ ( cd cmd/feed2mdir ; go build )
@end example

@item Create feeds state directories

You can create feeds subdirectories under @file{feeds/} manually:

@example
$ mkdir -p feeds/my_first_feed/@{cur,new,tmp@}
$ echo http://example.com/feed.atom > feeds/my_first_feed/url
@end example

or convert Newsboat @file{urls} file (containing many lines with URLs)
with @file{urls2feeds.zsh} to subdirectories hierarchy:

@example
$ ./urls2feeds.zsh < ~/.newsboat/urls
$ cat feeds/blog.stargrave.org_russian_feed.atom/url
http://blog.stargrave.org/russian/feed.atom
@end example

@item Download your feed(s) data

Downloading is implemented in @command{redo}'s
@file{default.download.do} file. Probably you want to change its default
@env{$PROXY} value. It uses @command{curl}, that is aware of
@code{If-Modified-Since} and @code{ETag} headers, compressed content
encodings and HTTP redirections.

You can invoke feed downloading like that:

@example
$ redo feeds/blog.stargrave.org_russian_feed.atom/feed.download
@end example

If you want to see verbose output, then set @env{FEEDER_CURL_VERBOSE=1}.

As a rule, you wish to run all feeds downloading in parallel. You can
use @file{feeds-download.zsh}, that just invokes @command{redo-ifchange}
with @option{-f} option (forceful rebuild). Why that? Because most
@command{redo} implementations (that forces target building) do not
parallelize specified targets build. But you can also use @command{parallel}:

@example
$ parallel "redo @{@}/feed.download" ::: feeds/*
@end example

@item Parse your feeds

Parsing (and Maildir filling) is implemented in @command{redo}'s
@file{default.parse.do} file. It calls @command{cmd/feed2mdir/feed2mdir}
utility, that read @file{feeds/FEED/feed} from stdin, takes Maildir
directory as its first argument and prints feed's title.

You can use @file{feeds-parse.zsh} helper or invoke @command{parallel}
as in example above, replacing @code{.download} with @code{.parse}.

@example
$ ./feeds-parse.zsh
@end example

@item Run Mutt

@example
$ ./feeds-browse.zsh
@end example

That will read all feeds titles and create @file{mutt.rc} sourceable
configuration file with predefined helpers and @code{mailboxes}
commands. Mutt will be started in mailboxes browser mode (I will skip
many entries):

@verbatim
  1   N [  1|101] 2021-02-17 20:41 Cryptology ePrint Archive/
  3     [  0|  8] 2021-12-02 19:28 Thoughts/
 32     [  0|  8] 2021-02-17 19:32 apenwarr/
101     [ 10| 50] 2021-02-14 13:40 Блог Stargrave на русском comments/
102     [  0| 51] 2021-02-17 19:37 Блог Stargrave на русском/
316     [  0| 44] 2021-02-17 19:33 Eaten By A Grue: Infocom, Text Adventures, and Interactive Fiction/
@end verbatim

ePrint has new entries since last downloading/parsing. Stargrave's blog
comments have nothing new, but still ten unread entries.

If we open "Eaten By A Grue" mailbox, then will see its entries:

@verbatim
   1     [2021-01-30 11:00] Zork Zero: The Revenge of Megaboz (0,8K)
   2     [2021-06-12 11:01] Journey: The Quest Begins (0,8K)
   3     [2021-04-28 11:00] Eaten By A Cruise (0,8K)
[...]
---Mutt: feeds/monsterfeet.com_grue.rss [Nachr:44 60K]---
@end verbatim

@item Press @code{q} to return to mailbox browser again

This is made for convenience, because you will often switch your
mailboxes (feeds), but @code{q} quits Mutt by default.

@item Press @code{A} to mark all messages read

And again this is made for convenience. It will mark both new
(@strong{N}) and old-but-unread (@strong{O}) messages as read. You will
see left tag-marks near each message to understand what was touched.

@item Index your messages

@example
$ ./feeds-index.sh
@end example

That will create @file{mu/} and @file{search/} directories and run
@command{mu index} indexing, that is safely can be done incrementally
after each download/parse cycle.

@item Search something

Press @code{<F8>} in Mutt's index and enter your mu/Xapian search query.
Let's search for articles mentioning
@url{https://en.wikipedia.org/wiki/Planetfall, Planetfall} during
2019-2021 period: @code{Planetfall date:2019..2021}. @command{mu} will
create symbolic links in @file{search/} subdirectory to the message. Press
@code{<F9>} to switch that mailbox:

@verbatim
   1 [2021-12-20 07:08] Missed Classic: Stationfall - When Food Dispensers Attack (The Adventurers Guild)
   2 [2021-11-20 04:52] Missed Classic 102: Stationfall - Introduction (1987) (The Adventurers Guild)
   3 [2021-11-19 17:54] Boffo Games (The Digital Antiquarian)
   4 [2021-10-30 23:05] Missed Classic 100: The Manhole (1988) (The Adventurers Guild)
   5 [2020-05-17 22:16] Round 04 Reveal (unWinnable State)
   6 [2020-05-16 22:29] Round 03 Reveal (unWinnable State)
   7 [2020-04-20 11:00] Planetfall (Eaten By A Grue: Infocom, Text Adventures, and Interactive Fiction)
   8 [2020-04-09 11:00] Beyond Zork (Eaten By A Grue: Infocom, Text Adventures, and Interactive Fiction)
-%-Mutt: =search [Nachr:8 215K]---
@end verbatim

Pay attention that there is different index format, lacking unnecessary
message flags display and adding name of the feed in parenthesis.

@item Cleanup excess number of messages

@example
$ ./feeds-clear.zsh
@end example

That will remove all messages in all feeds @file{cur/} directory that is
not first hundred of ones, ordered by @code{mtime}. Pay attention that
@file{new/} directory is not touched, so you won't loose completely new
and unread messages when you are on vacation and left @command{cron}-ed
workers. @command{cmd/feed2mdir/feed2mdir} command by default has
@option{-max-entries 100} option set.

@item If you want to clean download state

@example
$ redo feeds/FEED/feed.clean
@end example

@end table