cpython/Demo/scripts/newslist.py

367 lines
11 KiB
Python
Raw Normal View History

#! /usr/bin/env python
1994-05-27 10:32:41 -03:00
#######################################################################
# Newslist $Revision$
#
# Syntax:
# newslist [ -a ]
#
# This is a program to create a directory full of HTML pages
1994-05-27 10:32:41 -03:00
# which between them contain links to all the newsgroups available
# on your server.
#
# The -a option causes a complete list of all groups to be read from
1994-05-27 10:32:41 -03:00
# the server rather than just the ones which have appeared since last
# execution. This recreates the local list from scratch. Use this on
1994-05-27 10:33:17 -03:00
# the first invocation of the program, and from time to time thereafter.
# When new groups are first created they may appear on your server as
1994-05-27 10:33:17 -03:00
# empty groups. By default, empty groups are ignored by the -a option.
# However, these new groups will not be created again, and so will not
# appear in the server's list of 'new groups' at a later date. Hence it
# won't appear until you do a '-a' after some articles have appeared.
#
1994-05-27 10:33:17 -03:00
# I should really keep a list of ignored empty groups and re-check them
# for articles on every run, but I haven't got around to it yet.
1994-05-27 10:32:41 -03:00
#
# This assumes an NNTP news feed.
#
# Feel free to copy, distribute and modify this code for
# non-commercial use. If you make any useful modifications, let me
1994-05-27 10:32:41 -03:00
# know!
#
# (c) Quentin Stafford-Fraser 1994
# fraser@europarc.xerox.com qs101@cl.cam.ac.uk
# #
#######################################################################
import sys,nntplib, string, marshal, time, os, posix, string
#######################################################################
# Check these variables before running! #
# Top directory.
# Filenames which don't start with / are taken as being relative to this.
1994-05-27 10:33:17 -03:00
topdir='/anfs/qsbigdisc/web/html/newspage'
1994-05-27 10:32:41 -03:00
# The name of your NNTP host
# eg.
1994-05-27 10:33:17 -03:00
# newshost = 'nntp-serv.cl.cam.ac.uk'
# or use following to get the name from the NNTPSERVER environment
1994-05-27 10:32:41 -03:00
# variable:
1994-05-27 10:33:17 -03:00
# newshost = posix.environ['NNTPSERVER']
newshost = 'nntp-serv.cl.cam.ac.uk'
1994-05-27 10:32:41 -03:00
# The filename for a local cache of the newsgroup list
treefile = 'grouptree'
# The filename for descriptions of newsgroups
# I found a suitable one at ftp.uu.net in /uunet-info/newgroups.gz
# You can set this to '' if you don't wish to use one.
1994-05-27 10:33:17 -03:00
descfile = 'newsgroups'
1994-05-27 10:32:41 -03:00
# The directory in which HTML pages should be created
# eg.
# pagedir = '/usr/local/lib/html/newspage'
# pagedir = 'pages'
1994-05-27 10:32:41 -03:00
pagedir = topdir
# The html prefix which will refer to this directory
# eg.
# httppref = '/newspage/',
1994-05-27 10:33:17 -03:00
# or leave blank for relative links between pages: (Recommended)
# httppref = ''
1994-05-27 10:32:41 -03:00
httppref = ''
# The name of the 'root' news page in this directory.
1994-05-27 10:32:41 -03:00
# A .html suffix will be added.
1994-05-27 10:33:17 -03:00
rootpage = 'root'
1994-05-27 10:32:41 -03:00
# Set skipempty to 0 if you wish to see links to empty groups as well.
# Only affects the -a option.
1994-05-27 10:33:17 -03:00
skipempty = 1
# pagelinkicon can contain html to put an icon after links to
# further pages. This helps to make important links stand out.
# Set to '' if not wanted, or '...' is quite a good one.
pagelinkicon='... <img src="http://pelican.cl.cam.ac.uk/icons/page.xbm"> '
1994-05-27 10:32:41 -03:00
# ---------------------------------------------------------------------
# Less important personal preferences:
# Sublistsize controls the maximum number of items the will appear as
# an indented sub-list before the whole thing is moved onto a different
# page. The smaller this is, the more pages you will have, but the
1994-05-27 10:32:41 -03:00
# shorter each will be.
sublistsize = 4
# That should be all. #
#######################################################################
for dir in os.curdir, os.environ['HOME']:
1998-09-14 13:44:15 -03:00
rcfile = os.path.join(dir, '.newslistrc.py')
if os.path.exists(rcfile):
print rcfile
execfile(rcfile)
break
1994-05-27 10:32:41 -03:00
from nntplib import NNTP
from stat import *
rcsrev = '$Revision$'
rcsrev = string.join(filter(lambda s: '$' not in s, string.split(rcsrev)))
1994-05-27 10:32:41 -03:00
desc = {}
# Make (possibly) relative filenames into absolute ones
treefile = os.path.join(topdir,treefile)
descfile = os.path.join(topdir,descfile)
page = os.path.join(topdir,pagedir)
# First the bits for creating trees ---------------------------
# Addtotree creates/augments a tree from a list of group names
def addtotree(tree, groups):
print 'Updating tree...'
for i in groups:
1998-09-14 13:44:15 -03:00
parts = string.splitfields(i,'.')
makeleaf(tree, parts)
1994-05-27 10:32:41 -03:00
# Makeleaf makes a leaf and the branch leading to it if necessary
def makeleaf(tree,path):
j = path[0]
l = len(path)
1994-05-27 10:32:41 -03:00
if not tree.has_key(j):
tree[j] = {}
if l == 1:
tree[j]['.'] = '.'
if l > 1:
makeleaf(tree[j],path[1:])
1994-05-27 10:32:41 -03:00
# Then the bits for outputting trees as pages ----------------
1994-05-27 10:32:41 -03:00
# Createpage creates an HTML file named <root>.html containing links
# to those groups beginning with <root>.
def createpage(root, tree, p):
filename = os.path.join(pagedir,root+'.html')
if root == rootpage:
detail = ''
else:
detail = ' under ' + root
f = open(filename,'w')
# f.write('Content-Type: text/html\n')
f.write('<TITLE>Newsgroups available' + detail + '</TITLE>\n')
f.write('<H1>Newsgroups available' + detail +'</H1>\n')
f.write('<A HREF="'+httppref+rootpage+'.html">Back to top level</A><P>\n')
printtree(f,tree,0,p)
f.write('<I>This page automatically created by \'newslist\' v. '+rcsrev+'.')
f.write(time.ctime(time.time()) + '</I><P>')
f.close()
1994-05-27 10:32:41 -03:00
# Printtree prints the groups as a bulleted list. Groups with
# more than <sublistsize> subgroups will be put on a separate page.
# Other sets of subgroups are just indented.
def printtree(f, tree, indent, p):
global desc
l = len(tree)
if l > sublistsize and indent>0:
# Create a new page and a link to it
f.write('<LI><B><A HREF="'+httppref+p[1:]+'.html">')
f.write(p[1:]+'.*')
f.write('</A></B>'+pagelinkicon+'\n')
createpage(p[1:], tree, p)
return
kl = tree.keys()
if l > 1:
kl.sort()
if indent > 0:
# Create a sub-list
f.write('<LI>'+p[1:]+'\n<UL>')
else:
# Create a main list
f.write('<UL>')
indent = indent + 1
for i in kl:
if i == '.':
# Output a newsgroup
f.write('<LI><A HREF="news:' + p[1:] + '">'+ p[1:] + '</A> ')
if desc.has_key(p[1:]):
f.write(' <I>'+desc[p[1:]]+'</I>\n')
else:
f.write('\n')
else:
# Output a hierarchy
printtree(f,tree[i], indent, p+'.'+i)
if l > 1:
f.write('\n</UL>')
1994-05-27 10:32:41 -03:00
# Reading descriptions file ---------------------------------------
# This returns an array mapping group name to its description
1994-05-27 10:33:17 -03:00
def readdesc(descfile):
global desc
1994-05-27 10:32:41 -03:00
desc = {}
1994-05-27 10:32:41 -03:00
if descfile == '':
1998-09-14 13:44:15 -03:00
return
1994-05-27 10:32:41 -03:00
try:
d = open(descfile, 'r')
print 'Reading descriptions...'
except (IOError):
print 'Failed to open description file ' + descfile
return
l = d.readline()
while l != '':
bits = string.split(l)
try:
grp = bits[0]
dsc = string.join(bits[1:])
if len(dsc)>1:
desc[grp] = dsc
except (IndexError):
pass
l = d.readline()
1994-05-27 10:32:41 -03:00
1994-05-27 10:33:17 -03:00
# Check that ouput directory exists, ------------------------------
# and offer to create it if not
1994-05-27 10:32:41 -03:00
1994-05-27 10:33:17 -03:00
def checkopdir(pagedir):
if not os.path.isdir(pagedir):
print 'Directory '+pagedir+' does not exist.'
print 'Shall I create it for you? (y/n)'
if sys.stdin.readline()[0] == 'y':
try:
os.mkdir(pagedir,0777)
except:
print 'Sorry - failed!'
sys.exit(1)
else:
print 'OK. Exiting.'
1998-09-14 13:44:15 -03:00
sys.exit(1)
1994-05-27 10:32:41 -03:00
1994-05-27 10:33:17 -03:00
# Read and write current local tree ----------------------------------
1994-05-27 10:32:41 -03:00
1994-05-27 10:33:17 -03:00
def readlocallist(treefile):
print 'Reading current local group list...'
tree = {}
try:
treetime = time.localtime(os.stat(treefile)[ST_MTIME])
except:
print '\n*** Failed to open local group cache '+treefile
print 'If this is the first time you have run newslist, then'
print 'use the -a option to create it.'
sys.exit(1)
treedate = '%02d%02d%02d' % (treetime[0] % 100 ,treetime[1], treetime[2])
try:
dump = open(treefile,'r')
tree = marshal.load(dump)
dump.close()
except (IOError):
print 'Cannot open local group list ' + treefile
return (tree, treedate)
1994-05-27 10:33:17 -03:00
def writelocallist(treefile, tree):
try:
dump = open(treefile,'w')
groups = marshal.dump(tree,dump)
dump.close()
print 'Saved list to '+treefile+'\n'
except:
print 'Sorry - failed to write to local group cache '+treefile
print 'Does it (or its directory) have the correct permissions?'
sys.exit(1)
1994-05-27 10:33:17 -03:00
# Return list of all groups on server -----------------------------
def getallgroups(server):
print 'Getting list of all groups...'
treedate='010101'
info = server.list()[1]
groups = []
print 'Processing...'
if skipempty:
print '\nIgnoring following empty groups:'
for i in info:
grpname = string.split(i[0])[0]
if skipempty and string.atoi(i[1]) < string.atoi(i[2]):
print grpname+' ',
else:
groups.append(grpname)
print '\n'
if skipempty:
print '(End of empty groups)'
return groups
1994-05-27 10:33:17 -03:00
# Return list of new groups on server -----------------------------
def getnewgroups(server, treedate):
print 'Getting list of new groups since start of '+treedate+'...',
info = server.newgroups(treedate,'000001')[1]
print 'got %d.' % len(info)
print 'Processing...',
groups = []
for i in info:
grpname = string.split(i)[0]
groups.append(grpname)
print 'Done'
return groups
1994-05-27 10:33:17 -03:00
# Now the main program --------------------------------------------
1994-05-27 10:32:41 -03:00
1994-05-27 10:33:17 -03:00
def main():
global desc
tree={}
# Check that the output directory exists
checkopdir(pagedir);
try:
print 'Connecting to '+newshost+'...'
if sys.version[0] == '0':
s = NNTP.init(newshost)
else:
s = NNTP(newshost)
connected = 1
except (nntplib.error_temp, nntplib.error_perm) as x:
print 'Error connecting to host:', x
print 'I\'ll try to use just the local list.'
connected = 0
# If -a is specified, read the full list of groups from server
if connected and len(sys.argv) > 1 and sys.argv[1] == '-a':
groups = getallgroups(s)
# Otherwise just read the local file and then add
# groups created since local file last modified.
else:
(tree, treedate) = readlocallist(treefile)
if connected:
groups = getnewgroups(s, treedate)
if connected:
addtotree(tree, groups)
writelocallist(treefile,tree)
# Read group descriptions
readdesc(descfile)
print 'Creating pages...'
createpage(rootpage, tree, '')
print 'Done'
1994-05-27 10:32:41 -03:00
if __name__ == "__main__":
main()
1994-05-27 10:32:41 -03:00
# That's all folks
######################################################################