torrents.ru.parse - master - hubbitus/shell.scripts

torrents.ru.parse
66 строк3 KB
Hubbitus
Readd torrents.ru.parse script - on migration history was rewritten to do not contain login/pass to the service. Now script parametrized to use env variables.
15 апр 2014, 19:43
15 апр 2014, 19:43a608ca5
#! /bin/sh

###
# Script to grab data from torrents.ru tracker
#
# @version 1.0
# @author Pahan-Hubbitus (Pavel Alexeev) <Pahan [at] Hubbitus [ dot. ] info>
# @copyright Copyright (c) 2008, Pahan-Hubbitus (Pavel Alexeev)
###

: ${USERNAME?"For function properly script require set environment vars USERNAME and PASSWORD set"}
: ${PASSWORD?"For function properly script require set environment vars USERNAME and PASSWORD set"}
: ${1?"НЕОБХОДИМО передать аргумент, указывающий на файл содержащий ссылку (rutracker.org.URL) или просто саму ссылку на тему"}

FILE_FIRST=info.first.htm
#BASE_URL='http://torrents.ru/forum/'
BASE_URL='http://rutracker.org/forum/'

###################################################################################
# DO NOT change anything after that in untill you are really known what you doing #
###################################################################################

LOGIN_URL='http://login.rutracker.org/forum/login.php'

#LOG='--trace-ascii LOG'
LOG=''

	if [ -e "$1" ]; then
		URL="$( cat $1 )"
	else
		URL="$1"
	fi

# Log in to the server.  This can be done only once.
curl -L $LOG "$LOGIN_URL" --cookie-jar cook.file --cookie cook.file \
	--data "login_username=${USERNAME}&login_password=${PASSWORD}&login=login" -o - | grep Hubbitus &>/dev/null || { echo 'Error login! Check login/password'; return 1; }

#1. Load first page-description
curl -L -C - -L $LOG --cookie cook.file "$URL" -o $FILE_FIRST

enconv $FILE_FIRST

#2. Load image-cover
#<div class="postImg-wrap" style="padding: 4px 4px 2px; float: right; clear: right;"><img src="http://sec.web-da.ru/1000021918.jpg" id="postImgAligned" class="postImg" alt="pic"></div>
IMG_URLS=( `fgrep '<var class="postImg' $FILE_FIRST | sed -r 's/.*title="([^"]+?)".*/\1/'` )

	for (( i=0; i<${#IMG_URLS[*]}; i++ )); do
	echo $i - ${IMG_URLS[$i]}
	wget --timeout=15 --tries=1 "${IMG_URLS[$i]}" -O info$i.jpg
	done

#3.1. Extract torrent-file URL
#<p><a href="download.php?id=88138" class="genmed"><b>Download</b></a></p>
TORRENT_URL=`fgrep 'class="dl-stub dl-link"' $FILE_FIRST | sed -r 's/.*href="([^"]+)".*/\1/'`
echo TORRENT_URL=$TORRENT_URL

#3.2. Filename do not extracted properly from HTTP-headers! Do it manualy
#Перестали писать имя торрент-файла :) Теперь он просто как download.php?id=523865.torrent значится :(
TORRENT_filename=$( basename "`pwd`" | tr ' ' '_' | translit )".torrent"
echo '$TORRENT_filename='$TORRENT_filename

#3 Now grab the torrent file itself
curl -L $LOG --cookie cook.file --referer "$URL" --cookie bb_dl=$( echo $TORRENT_URL | grep -oP '\d+$' ) -o "$TORRENT_filename" "$TORRENT_URL"

#4 Make symlink in ~/Torents/
ln -s "$(pwd)" ~/Torents/"${TORRENT_filename%%.torrent}"