#!/usr/bin/python # render-dctrl # Copyright (C) 2009 Stefano Zacchiroli # # This program is free software: you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation, either version 3 of the License, or # (at your option) any later version. # Requirements (Debian packages): python-debian python-markdown usage = """Usage: render-dctrl [OPTION ...] [FILE ...] Render a 822-like listing of Debian packages (AKA "Packages" file) to XHTML, rendering (long) descriptions as Markdown text. Render text coming from FILEs, if given, or from standard input otherwise. Typical usage is within a dctrl-tools pipeline, example: grep-available -s Package,Depends,Description ocaml | render-dctrl > foo.html Warning: beware of #525525 and thus avoid using "-s Description" alone.""" import re import string import sys from debian import deb822 from markdown import markdown from optparse import OptionParser options = None # global, for cmdline options css = """ body { font-family: sans-serif; } dt { font-weight: bold; } dd { margin-bottom: 5pt; } div.package { border: solid 1pt; margin-top: 10pt; padding-left: 2pt; padding-right: 2pt; } .raw { font-family: monospace; background: #ddd; padding-left: 2pt; padding-right: 2pt; } .shortdesc { text-decoration: underline; margin-bottom: 5pt; display: block; } .longdesc { background: #eee; } span.package { font-family: monospace; font-size: 110%; } .uid { float: right; font-size: x-small; padding-right: 10pt; } """ html_header = """ """ % css html_trailer = """ """ mdwn_list_line = re.compile(r'^(\s*)[\*\+\-]') # Markdown list item line # mdwn_head_line = re.compile(r'^(\s*)#') # Markdown header padding = re.compile(r'^(\s*)') def get_indent(s): m = padding.match(s) if m: return len(m.group(1)) else: return 0 def render_longdesc(lines): print '

' lines = map(lambda s: s[1:], lines) # strip 822 heading space curpara, paragraphs = [], [] inlist, listindent = False, 0 store_para = lambda: paragraphs.append(string.join(curpara, '\n') + '\n') add_indent = lambda n, s: string.expandtabs('\t', n) + s for l in lines: # recognize Markdown paragraphs if l.rstrip() == '.': # RULE 1: split paragraphs at Debian's "." store_para() curpara, inlist, listindent = [], False, 0 else: if inlist: # currently in a list if get_indent(l) <= listindent: # RULE 3: leave list on underflow store_para() curpara, inlinst, linstindent = [l], False, 0 else: # the list goes on ... curpara.append(l) else: # currently not in a list if mdwn_list_line.match(l): # new list start if curpara: # RULE 2: handle list item *not* at para start store_para() curpara, inlist, listindent = [l], True, get_indent(l) elif get_indent(l) >= 1: # RULE 4: hande non-list verbatim if curpara and get_indent(curpara[-1]) < 4: store_para() curpara = [] curpara.append(add_indent(3, l)) else: curpara.append(l) if curpara: store_para() for p in paragraphs: # render paragraphs print markdown(p) print '

' def render_field(field, val): field = field.lower() print '

%s

' % field print '

' % field if field == 'description': lines = val.split('\n') print '%s' % lines[0] render_longdesc(lines[1:]) elif field == 'package': print 'id' % val print '%s' % (val, val) elif field in []: # fields not to be typeset as "raw" print '%s' % (field, val) else: print '%s' % val print '

' def render_file(f): global options, html_header, html_trailer if options.print_header: print html_header for pkg in deb822.Packages.iter_paragraphs(f): print '

' print '

' for (field, val) in pkg.iteritems(): render_field(field, val) print '' print '

\n' if options.print_header: print html_trailer def main(): global options, usage parser = OptionParser(usage=usage) parser.add_option("-n", "--no-headers", action="store_false", dest="print_header", default=True, help="suppress printing of HTML header/trailer") (options, args) = parser.parse_args() if len(args): for fname in args: render_file(open(fname)) else: render_file(sys.stdin) if __name__ == '__main__': main()