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 | ||
11 | def close(self): | |
12 | self.bk.close() | |
13 | ||
f3ad0817 FT |
14 | def read(self, sz = None): |
15 | if sz is None: | |
16 | return self.bk.read() | |
17 | else: | |
18 | return self.bk.read(sz) | |
19 | ||
20 | class page(lib.page): | |
3683ab38 FT |
21 | def __init__(self, chapter, stack, n, url): |
22 | self.stack = stack | |
f3ad0817 FT |
23 | self.chapter = chapter |
24 | self.volume = self.chapter.volume | |
25 | self.manga = self.volume.manga | |
26 | self.n = n | |
27 | self.url = url | |
28 | self.ciurl = None | |
29 | ||
30 | def iurl(self): | |
31 | if self.ciurl is None: | |
32 | page = soup(htcache.fetch(self.url)) | |
33 | self.ciurl = page.find("div", id="viewer").find("img", id="image")["src"] | |
34 | return self.ciurl | |
35 | ||
36 | def open(self): | |
37 | return imgstream(self.iurl()) | |
38 | ||
39 | class chapter(lib.pagelist): | |
3683ab38 FT |
40 | def __init__(self, volume, stack, name, url): |
41 | self.stack = stack | |
f3ad0817 FT |
42 | self.volume = volume |
43 | self.manga = volume.manga | |
44 | self.name = name | |
45 | self.url = url | |
46 | self.cpag = None | |
47 | ||
48 | def __getitem__(self, i): | |
49 | return self.pages()[i] | |
50 | ||
51 | def __len__(self): | |
52 | return len(self.pages()) | |
53 | ||
54 | def pages(self): | |
55 | if self.cpag is None: | |
56 | pg = soup(htcache.fetch(self.url + "1.html")) | |
57 | l = pg.find("form", id="top_bar").find("div", attrs={"class": "l"}) | |
58 | if len(l.contents) != 3: | |
59 | raise Exception("parse error: weird page list for %r" % self) | |
60 | m = l.contents[2].strip() | |
61 | if m[:3] != u"of ": | |
62 | raise Exception("parse error: weird page list for %r" % self) | |
3683ab38 | 63 | 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 |
64 | return self.cpag |
65 | ||
66 | def __str__(self): | |
67 | return self.name | |
68 | ||
69 | def __repr__(self): | |
70 | return "<mangafox.chapter %r.%r.%r>" % (self.manga.name, self.volume.name, self.name) | |
71 | ||
72 | class volume(lib.pagelist): | |
3683ab38 FT |
73 | def __init__(self, manga, stack, name): |
74 | self.stack = stack | |
f3ad0817 FT |
75 | self.manga = manga |
76 | self.name = name | |
77 | self.ch = [] | |
78 | ||
79 | def __getitem__(self, i): | |
80 | return self.ch[i] | |
81 | ||
82 | def __len__(self): | |
83 | return len(self.ch) | |
84 | ||
85 | def __str__(self): | |
86 | return self.name | |
87 | ||
88 | def __repr__(self): | |
89 | return "<mangafox.volume %r.%r>" % (self.manga.name, self.name) | |
90 | ||
91 | def nextel(el): | |
92 | while True: | |
93 | el = el.nextSibling | |
94 | if isinstance(el, BeautifulSoup.Tag): | |
95 | return el | |
96 | ||
97 | class manga(lib.manga): | |
98 | def __init__(self, lib, name, url): | |
99 | self.lib = lib | |
100 | self.name = name | |
101 | self.url = url | |
102 | self.cvol = None | |
3683ab38 | 103 | self.stack = [] |
f3ad0817 FT |
104 | |
105 | def __getitem__(self, i): | |
106 | return self.vols()[i] | |
107 | ||
108 | def __len__(self): | |
109 | return len(self.vols()) | |
110 | ||
111 | def vols(self): | |
112 | if self.cvol is None: | |
113 | page = soup(htcache.fetch(self.url)) | |
114 | vls = page.find("div", id="chapters").findAll("div", attrs={"class": "slide"}) | |
115 | self.cvol = [] | |
3683ab38 FT |
116 | for i, vn in enumerate(reversed(vls)): |
117 | vol = volume(self, [(self, i)], vn.find("h3", attrs={"class": "volume"}).contents[0].strip()) | |
118 | cls = nextel(vn) | |
f3ad0817 FT |
119 | if cls.name != u"ul" or cls["class"] != u"chlist": |
120 | raise Exception("parse error: weird volume list for %r" % self) | |
3683ab38 | 121 | for o, ch in enumerate(reversed(cls.findAll("li"))): |
f3ad0817 FT |
122 | n = ch.div.h3 or ch.div.h4 |
123 | name = n.a.string | |
124 | for span in ch("span"): | |
125 | try: | |
126 | if u" title " in (u" " + span["class"] + u" "): | |
127 | name += " " + span.string | |
128 | except KeyError: | |
129 | pass | |
130 | url = n.a["href"].encode("us-ascii") | |
131 | if url[-7:] != "/1.html": | |
132 | raise Exception("parse error: unexpected chapter URL for %r: %s" % (self, url)) | |
3683ab38 FT |
133 | vol.ch.append(chapter(vol, vol.stack + [(vol, o)], name, url[:-6])) |
134 | self.cvol.append(vol) | |
f3ad0817 FT |
135 | return self.cvol |
136 | ||
137 | def __str__(self): | |
138 | return self.name | |
139 | ||
140 | def __repr__(self): | |
141 | return "<mangafox.manga %r>" % self.name | |
142 | ||
143 | def libalphacmp(a, b): | |
144 | return cmp(a.upper(), b.upper()) | |
145 | ||
146 | class library(lib.library): | |
147 | def __init__(self): | |
148 | self.base = "http://www.mangafox.com/" | |
149 | ||
150 | def alphapage(self, pno): | |
151 | page = soup(htcache.fetch(self.base + ("directory/%i.htm?az" % pno))) | |
152 | ls = page.find("div", id="mangalist").find("ul", attrs={"class": "list"}).findAll("li") | |
153 | ret = [] | |
154 | for m in ls: | |
155 | t = m.find("div", attrs={"class": "manga_text"}).find("a", attrs={"class": "title"}) | |
156 | name = t.string | |
157 | url = t["href"].encode("us-ascii") | |
158 | ret.append(manga(self, name, url)) | |
159 | return ret | |
160 | ||
161 | def alphapages(self): | |
162 | page = soup(htcache.fetch(self.base + "directory/?az")) | |
163 | ls = page.find("div", id="mangalist").find("div", id="nav").find("ul").findAll("li") | |
164 | return int(ls[-2].find("a").string) | |
165 | ||
166 | def byname(self, prefix): | |
167 | if not isinstance(prefix, unicode): | |
168 | prefix = prefix.decode("utf8") | |
169 | l = 1 | |
170 | r = self.alphapages() | |
171 | while True: | |
172 | if l > r: | |
173 | return | |
174 | c = l + ((r + 1 - l) // 2) | |
175 | ls = self.alphapage(c) | |
176 | if libalphacmp(ls[0].name, prefix) > 0: | |
177 | r = c - 1 | |
178 | elif libalphacmp(ls[-1].name, prefix) < 0: | |
179 | l = c + 1 | |
180 | else: | |
181 | pno = c | |
182 | break | |
183 | i = 0 | |
184 | while i < len(ls): | |
185 | m = ls[i] | |
186 | if libalphacmp(m.name, prefix) >= 0: | |
187 | break | |
188 | i += 1 | |
189 | while True: | |
190 | while i < len(ls): | |
191 | m = ls[i] | |
192 | if not m.name[:len(prefix)].upper() == prefix.upper(): | |
193 | return | |
194 | yield m | |
195 | i += 1 | |
196 | pno += 1 | |
197 | ls = self.alphapage(pno) | |
198 | i = 0 | |
943a9376 FT |
199 | |
200 | def __iter__(self): | |
201 | raise NotImplementedError("mangafox iterator") |