Commit | Line | Data |
---|---|---|
f3ad0817 FT |
1 | import urllib |
2 | import BeautifulSoup | |
3 | import lib, htcache | |
4 | soup = BeautifulSoup.BeautifulSoup | |
5 | ||
3bba3a7b | 6 | class imgstream(lib.imgstream): |
f3ad0817 FT |
7 | def __init__(self, url): |
8 | self.bk = urllib.urlopen(url) | |
9 | self.ctype = self.bk.info()["Content-Type"] | |
10 | ||
af730068 FT |
11 | def fileno(self): |
12 | return self.bk.fileno() | |
13 | ||
f3ad0817 FT |
14 | def close(self): |
15 | self.bk.close() | |
16 | ||
f3ad0817 FT |
17 | def read(self, sz = None): |
18 | if sz is None: | |
19 | return self.bk.read() | |
20 | else: | |
21 | return self.bk.read(sz) | |
22 | ||
23 | class page(lib.page): | |
3683ab38 FT |
24 | def __init__(self, chapter, stack, n, url): |
25 | self.stack = stack | |
f3ad0817 FT |
26 | self.chapter = chapter |
27 | self.volume = self.chapter.volume | |
28 | self.manga = self.volume.manga | |
29 | self.n = n | |
46b3b90e | 30 | self.id = str(n) |
f3ad0817 FT |
31 | self.url = url |
32 | self.ciurl = None | |
33 | ||
34 | def iurl(self): | |
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"] | |
38 | return self.ciurl | |
39 | ||
40 | def open(self): | |
41 | return imgstream(self.iurl()) | |
42 | ||
43 | class chapter(lib.pagelist): | |
46b3b90e | 44 | def __init__(self, volume, stack, id, name, url): |
3683ab38 | 45 | self.stack = stack |
f3ad0817 FT |
46 | self.volume = volume |
47 | self.manga = volume.manga | |
46b3b90e | 48 | self.id = id |
f3ad0817 FT |
49 | self.name = name |
50 | self.url = url | |
51 | self.cpag = None | |
52 | ||
53 | def __getitem__(self, i): | |
54 | return self.pages()[i] | |
55 | ||
56 | def __len__(self): | |
57 | return len(self.pages()) | |
58 | ||
59 | def pages(self): | |
60 | if self.cpag is None: | |
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() | |
66 | if m[:3] != u"of ": | |
67 | raise Exception("parse error: weird page list for %r" % self) | |
3683ab38 | 68 | self.cpag = [page(self, self.stack + [(self, n)], n + 1, self.url + ("%i.html" % (n + 1))) for n in xrange(int(m[3:]))] |
f3ad0817 FT |
69 | return self.cpag |
70 | ||
71 | def __str__(self): | |
72 | return self.name | |
73 | ||
74 | def __repr__(self): | |
75 | return "<mangafox.chapter %r.%r.%r>" % (self.manga.name, self.volume.name, self.name) | |
76 | ||
77 | class volume(lib.pagelist): | |
46b3b90e | 78 | def __init__(self, manga, stack, id, name): |
3683ab38 | 79 | self.stack = stack |
f3ad0817 | 80 | self.manga = manga |
46b3b90e | 81 | self.id = id |
f3ad0817 FT |
82 | self.name = name |
83 | self.ch = [] | |
84 | ||
85 | def __getitem__(self, i): | |
86 | return self.ch[i] | |
87 | ||
88 | def __len__(self): | |
89 | return len(self.ch) | |
90 | ||
91 | def __str__(self): | |
92 | return self.name | |
93 | ||
94 | def __repr__(self): | |
95 | return "<mangafox.volume %r.%r>" % (self.manga.name, self.name) | |
96 | ||
97 | def nextel(el): | |
98 | while True: | |
99 | el = el.nextSibling | |
100 | if isinstance(el, BeautifulSoup.Tag): | |
101 | return el | |
102 | ||
103 | class manga(lib.manga): | |
46b3b90e | 104 | def __init__(self, lib, id, name, url): |
f3ad0817 | 105 | self.lib = lib |
46b3b90e | 106 | self.id = id |
f3ad0817 FT |
107 | self.name = name |
108 | self.url = url | |
109 | self.cvol = None | |
3683ab38 | 110 | self.stack = [] |
f3ad0817 FT |
111 | |
112 | def __getitem__(self, i): | |
113 | return self.vols()[i] | |
114 | ||
115 | def __len__(self): | |
116 | return len(self.vols()) | |
117 | ||
118 | def vols(self): | |
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"}) | |
122 | self.cvol = [] | |
3683ab38 | 123 | for i, vn in enumerate(reversed(vls)): |
46b3b90e FT |
124 | name = vn.find("h3", attrs={"class": "volume"}).contents[0].strip() |
125 | vid = name.encode("utf8") | |
126 | vol = volume(self, [(self, i)], vid, name) | |
3683ab38 | 127 | cls = nextel(vn) |
f3ad0817 FT |
128 | if cls.name != u"ul" or cls["class"] != u"chlist": |
129 | raise Exception("parse error: weird volume list for %r" % self) | |
3683ab38 | 130 | for o, ch in enumerate(reversed(cls.findAll("li"))): |
f3ad0817 FT |
131 | n = ch.div.h3 or ch.div.h4 |
132 | name = n.a.string | |
46b3b90e | 133 | chid = name.encode("utf8") |
f3ad0817 FT |
134 | for span in ch("span"): |
135 | try: | |
136 | if u" title " in (u" " + span["class"] + u" "): | |
137 | name += " " + span.string | |
138 | except KeyError: | |
139 | pass | |
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)) | |
46b3b90e | 143 | vol.ch.append(chapter(vol, vol.stack + [(vol, o)], chid, name, url[:-6])) |
3683ab38 | 144 | self.cvol.append(vol) |
f3ad0817 FT |
145 | return self.cvol |
146 | ||
147 | def __str__(self): | |
148 | return self.name | |
149 | ||
150 | def __repr__(self): | |
151 | return "<mangafox.manga %r>" % self.name | |
152 | ||
153 | def libalphacmp(a, b): | |
154 | return cmp(a.upper(), b.upper()) | |
155 | ||
156 | class library(lib.library): | |
157 | def __init__(self): | |
6fab7b55 | 158 | self.base = "http://mangafox.me/" |
f3ad0817 FT |
159 | |
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") | |
163 | ret = [] | |
46b3b90e | 164 | ubase = self.base + "manga/" |
f3ad0817 FT |
165 | for m in ls: |
166 | t = m.find("div", attrs={"class": "manga_text"}).find("a", attrs={"class": "title"}) | |
167 | name = t.string | |
168 | url = t["href"].encode("us-ascii") | |
46b3b90e FT |
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)) | |
f3ad0817 FT |
172 | return ret |
173 | ||
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) | |
178 | ||
179 | def byname(self, prefix): | |
180 | if not isinstance(prefix, unicode): | |
181 | prefix = prefix.decode("utf8") | |
182 | l = 1 | |
183 | r = self.alphapages() | |
184 | while True: | |
185 | if l > r: | |
186 | return | |
187 | c = l + ((r + 1 - l) // 2) | |
188 | ls = self.alphapage(c) | |
189 | if libalphacmp(ls[0].name, prefix) > 0: | |
190 | r = c - 1 | |
191 | elif libalphacmp(ls[-1].name, prefix) < 0: | |
192 | l = c + 1 | |
193 | else: | |
194 | pno = c | |
195 | break | |
196 | i = 0 | |
197 | while i < len(ls): | |
198 | m = ls[i] | |
199 | if libalphacmp(m.name, prefix) >= 0: | |
200 | break | |
201 | i += 1 | |
202 | while True: | |
203 | while i < len(ls): | |
204 | m = ls[i] | |
205 | if not m.name[:len(prefix)].upper() == prefix.upper(): | |
206 | return | |
207 | yield m | |
208 | i += 1 | |
209 | pno += 1 | |
210 | ls = self.alphapage(pno) | |
211 | i = 0 | |
943a9376 | 212 | |
46b3b90e FT |
213 | def byid(self, id): |
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 | |
218 | raise KeyError(id) | |
219 | name = page.find("div", id="series_info").find("div", attrs={"class": "cover"}).img["alt"] | |
220 | return manga(self, id, name, url) | |
221 | ||
943a9376 FT |
222 | def __iter__(self): |
223 | raise NotImplementedError("mangafox iterator") |