Mercurial > hg > config
annotate python/html2flux.py @ 292:fbc033540a34
resurrect this ole beast
| author | Jeff Hammel <jhammel@mozilla.com> |
|---|---|
| date | Sat, 11 May 2013 03:18:55 -0700 |
| parents | 069a739d88ad |
| children | 45b970048ae2 |
| rev | line source |
|---|---|
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
1 #!/usr/bin/env python |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
2 |
| 292 | 3 """ |
| 4 transform an HTML <dl> file into a fluxbox menu | |
| 5 if no file give, read from stdin | |
| 6 | |
| 7 <dl><a>submenu name</a> | |
| 8 <dt>program label</dt><dd>command</dd> | |
| 9 <dt>another program label</dt><dd>command2</dd> | |
| 10 </dl> | |
| 11 | |
| 12 x-form -> internal format: | |
| 13 | |
| 14 ('submenu name': [('program label', 'command'), | |
| 15 ('another program label', 'command2')]) | |
| 16 """ | |
| 17 | |
| 18 import optparse | |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
19 import sys |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
20 from lxml import etree |
| 292 | 21 from lsex import lsex # local import |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
22 |
| 292 | 23 # available executables |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
24 executables = set([i.rsplit('/', 1)[-1] for i in lsex() ]) |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
25 |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
26 def printmenu(dl, output, top=True): |
| 292 | 27 |
| 28 menu_items = [] | |
| 29 name = None # menu name | |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
30 for child in dl.iterchildren(): |
| 292 | 31 |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
32 if not top and child.tag == 'a': |
| 292 | 33 # TODO: better way of labeling this! |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
34 print >> output, '[submenu] (%s)' % child.text |
| 292 | 35 |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
36 if child.tag == 'dt': |
| 292 | 37 # item label |
| 38 label = ' '.join([i.strip() for i in child.itertext() if i.strip()]) | |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
39 if child.tag == 'dd': |
| 292 | 40 command = ' '.join([i.strip() for i in child.itertext() if i.strip()]) |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
41 executable = command.split()[0] |
| 292 | 42 if executable in executables or os.path.isabs(executable): |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
43 print >> output, '[exec] (%s) {%s}' % (label, command) |
| 292 | 44 |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
45 if child.tag == 'dl': |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
46 printmenu(child, output, top=False) |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
47 if not top: |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
48 print >> output, '[end]' |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
49 |
| 292 | 50 def main(args=sys.argv[1:]): |
| 51 | |
| 52 # parse command line option | |
| 53 usage = '%prog [options] [menu.html]' | |
| 54 parser = optparse.OptionParser(usage=usage, | |
| 55 description=__doc__) | |
| 56 parser.add_option('--collapse', dest='collapse', | |
| 57 action='store_true', default=False, | |
| 58 help="collapse menus with a single item to that item") | |
| 59 parser.add_option('-o', '--output', dest='output', | |
| 60 help="output file [Default: <stdout>]") | |
| 61 options, args = parser.parse_args(args) | |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
62 |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
63 # setup input, output |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
64 if args: |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
65 htmlfile = file(args[0]) |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
66 else: |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
67 htmlfile = sys.stdin |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
68 html = htmlfile.read() |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
69 fluxout = sys.stdout |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
70 |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
71 # get first element |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
72 dom = etree.fromstring(html) |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
73 dl = dom.find('.//dl') |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
74 |
| 292 | 75 # print to stdout |
|
45
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
76 printmenu(dl, fluxout) |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
77 |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
78 if __name__ == '__main__': |
|
069a739d88ad
get fluxbox menu from a webpage, i.e. http://k0s.org/programs.html
Jeff Hammel <k0scist@gmail.com>
parents:
diff
changeset
|
79 main() |
