def parse_index(raw=None): # {{{
raw = raw or read(INDEX).decode('utf-8', 'replace')
dep_start = raw.index('>Deprecated/Renamed/Retired Plugins:<')
dpat = re.compile(r'''(?is)Donate\s*:\s*<a\s+href=['"](.+?)['"]''')
key_pat = re.compile(r'''(?is)(History|Uninstall)\s*:\s*([^<;]+)[<;]''')
seen = {}
for match in re.finditer(r'''(?is)<li.+?<a\s+href=['"](https://www.mobileread.com/forums/showthread.php\?[pt]=\d+).+?>(.+?)<(.+?)</li>''', raw):
deprecated = match.start() > dep_start
donate = uninstall = None
history = False
name, url, rest = u(match.group(2)), u(match.group(1)), match.group(3)
m = dpat.search(rest)
if m is not None:
donate = u(m.group(1))
for m in key_pat.finditer(rest):
k = m.group(1).lower()
if k == 'history' and m.group(2).strip().lower() in {'yes', 'true'}:
history = True
elif k == 'uninstall':
uninstall = tuple(x.strip() for x in m.group(2).strip().split(','))
thread_id = url_to_plugin_id(url, deprecated)
if thread_id in seen:
raise ValueError('thread_id for %s and %s is the same: %s' % (seen[thread_id], name, thread_id))
seen[thread_id] = name
entry = IndexEntry(name, url, donate, history, uninstall, deprecated, thread_id)
yield entry
def parse_plugin_zip_url(raw):
for m in re.finditer(
r'''(?is)<a\s+href=['"](attachment.php\?[^'"]+?)['"][^>]*>([^<>]+?\.zip)\s*<''',
raw):
url, name = u(m.group(1)), u(m.group(2).strip())
if name.lower().endswith('.zip'):
return MR_URL + url, name
return None, None