horrors2

toxhtml.py

  1#! /usr/bin/env python3
  2
  3import re
  4
  5block1_re = re.compile(r'{\\(?P<cmd>[\w*]+) (?P<txt>[^{}]+)}')
  6block2_re = re.compile(r'\\(?P<cmd>[\w*]+)(\[[^]]+\])?{(?P<txt>[^{}]*)}')
  7
  8
  9chapauth = None
 10chapimg = None
 11display = True
 12pfx = ''
 13
 14def texsub(m):
 15    global pfx, display
 16
 17    cmd = m.group('cmd')
 18    txt = m.group('txt')
 19    if cmd == 'em':
 20        return '<em>%s</em>' % txt
 21    elif cmd == 'bf':
 22        return '<strong>%s</strong>' % txt
 23    elif cmd == 'sf':
 24        return '<samp>%s</samp>' % txt
 25    elif cmd == 'sc':
 26        return '<span class="sc">%s</span>' % txt
 27    elif cmd == 'rm':
 28        return '<span class="rm">%s</span>' % txt
 29    elif cmd == 'url':
 30        return '<a href="%s">%s</a>' % (txt, txt)
 31    elif cmd == 'begin':
 32        if txt in ('center',):
 33            return
 34        elif txt in ('quotation', 'quote'):
 35            pfx = '> '
 36        elif txt == 'textblock':
 37            display = False
 38        else:
 39            print(cmd, txt)
 40            raise TypeError(cmd)
 41    elif cmd == 'end':
 42        if txt == 'textblock':
 43            display = True
 44        else:
 45            pfx = ''
 46        return ''
 47    elif cmd in ('include',
 48                 'chapter',
 49                 'chapimg',
 50                 'chapauth',
 51                 'illustration',
 52                 'scriptsize',
 53                 'section*',
 54                 'part'):
 55        return '#%s %s' % (cmd, txt)
 56    elif cmd in ('pagenumbering',
 57                 'includegraphics',
 58                 'newcommand',
 59                 'hbox'):
 60        return ''
 61    elif cmd in ('TeX',
 62                 'LaTeX'):
 63        return cmd
 64    else:
 65        print(cmd, txt)
 66        raise TypeError(cmd)
 67
 68decor_stack = []
 69decor_re = re.compile(r'({\\(?P<cmd>\w\w) ?|})')
 70
 71def decorsub(m):
 72    cmd = m.group('cmd')
 73    if not cmd:
 74        if not decor_stack:
 75            return m.group(0)
 76        cmd = decor_stack.pop()
 77        return '</%s>' % cmd
 78    else:
 79        if cmd == 'bf':
 80            cmd = 'strong'
 81        elif cmd == 'em':
 82            pass
 83        elif cmd == 'sc':
 84            decor_stack.append('span')
 85            return '<span class="sc">'
 86        else:
 87            raise TypeError(cmd)
 88        decor_stack.append(cmd)
 89        return '<%s>' % cmd
 90
 91def art(artist, url, title=None):
 92    alt = title or ("Artwork by %s" % artist)
 93    print('<div class="art">')
 94    print('<img src="%s" alt="%s" />' % (url, alt))
 95    if title:
 96        atxt = '<em>%s</em> by %s' % (title, artist)
 97    else:
 98        atxt = alt
 99    print('<p class="artist">%s</p>' % (atxt))
100    print('</div>')
101
102outbuf = ''
103
104def outline(l):
105    global chapimg, chapauth, outbuf
106
107    l = l.strip()
108    if not l:
109        print(outbuf)
110    elif l[0] == '%':
111        return
112    l = l.replace(r'\'e', 'é')
113    l = l.replace(r'\,c', 'ç')
114    l = l.replace(r'\'n', 'ń')
115    l = l.replace("''", '&rdquo;')
116    l = l.replace("``", '&ldquo;')
117    l = l.replace("'", '&rsquo;')
118    l = l.replace("`", '&lsquo;')
119    l = l.replace('---', '&mdash;')
120    l = l.replace('--', '&ndash;')
121    l = l.replace('\\\\', '<br />')
122    l = l.replace('\_', '_')
123    l = l.replace('\#', '#')
124    l = l.replace('\$', '$')
125    l = l.replace('\ ', ' ')
126    l = l.replace(r'\-', '')
127    l = l.replace(r'\~n', 'ñ')
128    l = l.replace(r'{\ldots}', '&hellip;')
129    l = l.replace(r'\ldots', '&hellip;')
130    l = l.replace(r'\copyright', '&copy;')
131    l = block1_re.sub(texsub, l)
132    l = block2_re.sub(texsub, l)
133    l = l.replace('{}', '')
134    l = decor_re.sub(decorsub, l)
135
136    if not l:
137        return
138    if l[0] == '#':
139        if l.startswith('#include'):
140            include(l[9:] + '.tex')
141        elif l.startswith('#chapimg'):
142            chapimg = l[9:-1].split('{')
143        elif l.startswith('#chapauth'):
144            chapauth = l[10:]
145        elif l.startswith('#chapter'):
146            print('<h1 class="chapter">%s</h1>' % l[9:])
147            if chapauth:
148                print('<h2 class="author">by %s</h2>' % chapauth)
149                chapauth = None
150            if chapimg:
151                art(chapimg[0], chapimg[1])
152                chapimg = None
153        elif l.startswith('#part'):
154            print('<h1 class="part">%s</h1>' % l[6:])
155        elif l.startswith('#illustration'):
156            artist, title, url = l[14:].split('{')
157            title = title[:-1]
158            url = url[:-1]
159            art(artist, url, title)
160        elif l.startswith('#section*'):
161            print('<h2 class="section">%s</h2>' % l[10:])
162        else:
163            print('<--! %s -->' % l)
164    elif l[0] == '\\':
165        what = l[1:5].lower()
166        if what in ('bigs', 'vfil'):
167            print('<br class="bigskip"/>')
168        elif what == 'newp':
169            print('<br class="pagebreak"/>')
170        elif what == 'noin':
171            print(l[10:])
172        elif what in ('hbox',
173                      'inde',
174                      'tabl',
175                      'appe',
176                      'page',
177                      'list'):
178            pass
179        elif l[1:9] == 'maketitl':
180            print('<h1>Horrors 2</h1>')
181            print('<h2>The Something Awful Forums</h2>')
182        else:
183            print('================= %r' % what)
184    elif display:
185        print('%s%s' % (pfx, l))
186
187def include(fn):
188    if fn == 'praise.tex':
189        return
190    f = open(fn)
191
192    for l in f:
193        outline(l)
194
195f = open('horrors2.ltx')
196
197# skip LaTeX crap
198for l in f:
199    if l.startswith('\\begin{document'):
200        break
201
202for l in f:
203    outline(l)