4 soup = BeautifulSoup.BeautifulSoup
6 class imgstream(lib.imgstream):
7 def __init__(self, url):
8 self.bk = urllib.urlopen(url)
9 self.ctype = self.bk.info()["Content-Type"]
12 return self.bk.fileno()
17 def read(self, sz = None):
21 return self.bk.read(sz)
24 def __init__(self, chapter, stack, n, url):
26 self.chapter = chapter
27 self.volume = self.chapter.volume
28 self.manga = self.volume.manga
35 if self.ciurl is None:
36 page = soup(htcache.fetch(self.url))
37 self.ciurl = page.find("div", id="viewer").find("img", id="image")["src"]
41 return imgstream(self.iurl())
43 class chapter(lib.pagelist):
44 def __init__(self, volume, stack, id, name, url):
47 self.manga = volume.manga
53 def __getitem__(self, i):
54 return self.pages()[i]
57 return len(self.pages())
61 pg = soup(htcache.fetch(self.url + "1.html"))
62 l = pg.find("form", id="top_bar").find("div", attrs={"class": "l"})
63 if len(l.contents) != 3:
64 raise Exception("parse error: weird page list for %r" % self)
65 m = l.contents[2].strip()
67 raise Exception("parse error: weird page list for %r" % self)
68 self.cpag = [page(self, self.stack + [(self, n)], n + 1, self.url + ("%i.html" % (n + 1))) for n in xrange(int(m[3:]))]
75 return "<mangafox.chapter %r.%r.%r>" % (self.manga.name, self.volume.name, self.name)
77 class volume(lib.pagelist):
78 def __init__(self, manga, stack, id, name):
85 def __getitem__(self, i):
95 return "<mangafox.volume %r.%r>" % (self.manga.name, self.name)
100 if isinstance(el, BeautifulSoup.Tag):
103 class manga(lib.manga):
104 def __init__(self, lib, id, name, url):
112 def __getitem__(self, i):
113 return self.vols()[i]
116 return len(self.vols())
119 if self.cvol is None:
120 page = soup(htcache.fetch(self.url))
121 vls = page.find("div", id="chapters").findAll("div", attrs={"class": "slide"})
123 for i, vn in enumerate(reversed(vls)):
124 name = vn.find("h3", attrs={"class": "volume"}).contents[0].strip()
125 vid = name.encode("utf8")
126 vol = volume(self, [(self, i)], vid, name)
128 if cls.name != u"ul" or cls["class"] != u"chlist":
129 raise Exception("parse error: weird volume list for %r" % self)
130 for o, ch in enumerate(reversed(cls.findAll("li"))):
131 n = ch.div.h3 or ch.div.h4
133 chid = name.encode("utf8")
134 for span in ch("span"):
136 if u" title " in (u" " + span["class"] + u" "):
137 name += " " + span.string
140 url = n.a["href"].encode("us-ascii")
141 if url[-7:] != "/1.html":
142 raise Exception("parse error: unexpected chapter URL for %r: %s" % (self, url))
143 vol.ch.append(chapter(vol, vol.stack + [(vol, o)], chid, name, url[:-6]))
144 self.cvol.append(vol)
151 return "<mangafox.manga %r>" % self.name
153 def libalphacmp(a, b):
154 return cmp(a.upper(), b.upper())
156 class library(lib.library):
158 self.base = "http://mangafox.me/"
160 def alphapage(self, pno):
161 page = soup(htcache.fetch(self.base + ("directory/%i.htm?az" % pno)))
162 ls = page.find("div", id="mangalist").find("ul", attrs={"class": "list"}).findAll("li")
164 ubase = self.base + "manga/"
166 t = m.find("div", attrs={"class": "manga_text"}).find("a", attrs={"class": "title"})
168 url = t["href"].encode("us-ascii")
169 if url[:len(ubase)] != ubase or url.find('/', len(ubase)) != (len(url) - 1):
170 raise Exception("parse error: unexpected manga URL for %r: %s" % (name, url))
171 ret.append(manga(self, url[len(ubase):-1], name, url))
174 def alphapages(self):
175 page = soup(htcache.fetch(self.base + "directory/?az"))
176 ls = page.find("div", id="mangalist").find("div", id="nav").find("ul").findAll("li")
177 return int(ls[-2].find("a").string)
179 def byname(self, prefix):
180 if not isinstance(prefix, unicode):
181 prefix = prefix.decode("utf8")
183 r = self.alphapages()
187 c = l + ((r + 1 - l) // 2)
188 ls = self.alphapage(c)
189 if libalphacmp(ls[0].name, prefix) > 0:
191 elif libalphacmp(ls[-1].name, prefix) < 0:
199 if libalphacmp(m.name, prefix) >= 0:
205 if not m.name[:len(prefix)].upper() == prefix.upper():
210 ls = self.alphapage(pno)
214 url = self.base + ("manga/%s/" % id)
215 page = soup(htcache.fetch(url))
216 if page.find("div", id="title") is None:
217 # Assume we got the search page
219 name = page.find("div", id="series_info").find("div", attrs={"class": "cover"}).img["alt"]
220 return manga(self, id, name, url)
223 raise NotImplementedError("mangafox iterator")