#!/usr/bin/python
# R(ange) URL D(ownloader) see usage()
# version: see below -h option
# author: Ludovic Bellier see http://home.zyrianes.net
# licence: GNU General Public License see http://www.gnu.org/copyleft/gpl.html

"""

  CVS Information

    $Id: rurld,v 0.11 2006/02/17 00:39:36 ludo Exp $
"""
__version__='$Revision: 0.11 $'[11:-2]

import getopt, sys, os, re, string

class Config:
    dry_run=0         # if true, just echo urls, do not download
    rangeURL=''       # url like with range indication with '[' and ']'
    rangeURLTemplate='bracket'
    
class Rurld:

    def __init__(self,cfg):
        self.config=cfg
    

    def print_url(self,urls):
        if type(urls) == type([]):
            for url in urls:
                print url
        else:
            print urls

    def get_files(self,urls):
        str_url=''
        for url in urls:
            str_url=str_url+' %s '% url
        os.system('wget %s' % str_url)
        
    def actionForURLs(self,urls):
        """ download or print """
        if self.config.dry_run:
            dl_f=self.print_url
        else:
            dl_f=self.get_files
        # do it 
        map(dl_f,(urls,))

    def parseRangeURL(self):
        sre_match=None
        template=self.config.rangeURLTemplate
        rurl=self.config.rangeURL
        base=rurl[0]
        if template == 'bracket':
            sre_match=re.match('(.*)\[(.*)-(.*)\](.*)',base)
        elif template == 'printf':
            # Warning: in progress...
            # ex: rurld -n -t printf http://home.zyrianes.net/p/basset%02d-%02d.jpg 1 2
            # need to find how to directly render %02d-%02d.jpg 1 2
            # sol 1: rewrite a bracket template url
            sre_match=re.match('(.*)%(.*)-%(.*)\.(.*)',base) # TBD not use '\.' but true %02d matching
            # print sre_match
            if sre_match:
                # import pdb; pdb.set_trace()
                # print sre_match.groups()
                base_rurl,range_left,range_right,end_url=sre_match.groups()
                params=rurl[1:]
                rng='%'+range_left+'-%'+range_right
                rng2=rng % tuple(map(int,params)) # for now, only %d, so 'int' cast, need a better solution
                self.config.rangeURL=[base_rurl+'['+rng2+'].'+end_url]
                # print self.config.rangeURL
                self.config.rangeURLTemplate='bracket'
                sre_match=self.parseRangeURL()
        return sre_match
        
    def buildUrls(self,sre_match):
        urls=[]
        if not sre_match:
            urls=[self.config.rangeURL[0]]
        else:
            base_url,range_left,range_right,end_url=sre_match.groups()
            # number of '0', and see zfill below (don't know how to
            # use variable notation like '%.(nb_zero)d' )
            re_nb_zero=re.match('(0*)(.*)',range_left)
            if re_nb_zero: nb_zero=len(re_nb_zero.group(1))+1
            else: nb_zero=0
            for i in range(int(range_left), int(range_right)+1):
                urls.append("%s%s%s" % (base_url, string.zfill(i, nb_zero), end_url))
        return urls
    
    def run(self):
        sre_match=self.parseRangeURL()
        urls=self.buildUrls(sre_match)
        self.actionForURLs(urls)


def usage():
    sys.stderr.write("""Range URL Downloader, version %s, gpl
Usage: rurld [-h] [-n] [-t template] <URL with range>
 -h this help
 -n dry_run mode, just show urls
 -t template, template model for rangeURL: bracket|printf
     default value = bracket

for example: rurld http://foo.com/pic[01-11].jpg downloads files
from http://foo.com/pic01.jpg to http://foo.com/pic11.jpg

note that [01-11] produces 01->11 and [1-11] 1->11

required: wget ( http://www.gnu.org/software/wget/wget.html )
""" % __version__ )
    sys.exit(1)

def main(args):
    config = Config()
    try: options, args = getopt.getopt(args[1:], 'hnt:')
    except:
        usage()
    for k, v in options:
        if k=='-h': usage()
        if k=='-n': config.dry_run=1; 
        if k=='-t': config.rangeURLTemplate=v; 
    if len(args)==0: usage()
    config.rangeURL= args
    rurld=Rurld(config)
    rurld.run()

if __name__=='__main__':
    main(sys.argv)