1 import urllib.request, urllib.parse, http.cookiejar, re, bs4, os, time
2 from . import profile, lib, htcache
3 soup = bs4.BeautifulSoup
4 soupify = lambda cont: soup(cont, "html.parser")
6 class pageerror(Exception):
7 def __init__(self, message, page):
8 super().__init__(message)
11 def iterlast(itr, default=None):
12 if default is not None:
20 def find1(el, *args, **kwargs):
21 ret = el.find(*args, **kwargs)
23 raise pageerror("could not find expected element", iterlast(el.parents, el))
26 def byclass(el, name, cl):
27 for ch in el.findAll(name):
28 if not isinstance(ch, bs4.Tag): continue
29 cll = ch.get("class", [])
37 if isinstance(el, bs4.Tag):
40 def fetchreader(lib, readerid, page):
41 pg = soupify(lib.sess.fetch(lib.base + "areader?" + urllib.parse.urlencode({"id": readerid,
43 "supress_webtoon": "t"}),
44 headers={"Referer": "http://bato.to/reader"}))
48 def __init__(self, chapter, stack, readerid, n):
50 self.lib = chapter.lib
51 self.chapter = chapter
54 self.name = "Page %s" % n
55 self.readerid = readerid
59 if self.ciurl is None:
60 page = fetchreader(self.lib, self.readerid, self.n)
61 img = find1(page, "img", id="comic_page")
62 self.ciurl = img["src"]
66 return lib.stdimgstream(self.iurl())
72 return "<batoto.page %r.%r.%r.%r>" % (self.chapter.manga.name, self.chapter.group.name, self.chapter.name, self.name)
74 class chapter(lib.pagelist):
75 def __init__(self, group, stack, id, name, readerid):
78 self.manga = group.manga
79 self.lib = self.manga.lib
82 self.readerid = readerid
85 def __getitem__(self, i):
86 return self.pages()[i]
89 return len(self.pages())
91 pnre = re.compile(r"page (\d+)")
94 pg = fetchreader(self.lib, self.readerid, 1)
96 for opt in find1(pg, "select", id="page_select").findAll("option"):
97 n = int(self.pnre.match(opt.string).group(1))
98 cpag.append(page(self, self.stack + [(self, len(cpag))], self.readerid, n))
106 return "<batoto.chapter %r.%r.%r>" % (self.manga.name, self.group.name, self.name)
108 class group(lib.pagelist):
109 def __init__(self, manga, stack, id, name):
116 def __getitem__(self, i):
126 return "<batoto.group %r.%r" % (self.manga.name, self.name)
128 class manga(lib.manga):
129 def __init__(self, lib, id, name, url):
139 def __getitem__(self, i):
143 return len(self.ch())
147 if page.find("div", id="register_notice"):
149 if not byclass(page, "table", "chapters_list"):
153 cure = re.compile(r"/reader#([a-z0-9]+)")
156 page = self.sess.lfetch(self.url, self.vfylogin)
157 cls = byclass(page, "table", "chapters_list")
158 if cls.tbody is not None:
160 scl = "lang_" + self.lib.lang
162 for ch in cls.childGenerator():
163 if isinstance(ch, bs4.Tag) and ch.name == "tr":
164 cll = ch.get("class", [])
165 if "row" in cll and scl in cll:
166 url = ch.td.a["href"]
167 m = self.cure.search(url)
168 if m is None: raise pageerror("Got weird chapter URL: %r" % url, page)
169 readerid = m.group(1)
171 gname = nextel(nextel(ch.td)).text.strip()
172 cch.append((readerid, name, gname))
175 for n, (readerid, name, gname) in enumerate(cch):
176 groups.setdefault(gname, [n, []])[1].append((readerid, name))
177 groups = sorted(groups.items(), key=lambda o: o[1][0])
179 for n, (gname, (_, gch)) in enumerate(groups):
180 ngrp = group(self, [(self, n)], gname, gname)
181 for m, (readerid, name) in enumerate(gch):
182 ngrp.ch.append(chapter(ngrp, ngrp.stack + [(ngrp, m)], readerid, name, readerid))
188 if self.cnames is None:
189 page = soupify(self.sess.fetch(self.url))
191 for tbl in page.findAll("table", attrs={"class": "ipb_table"}):
192 if tbl.tbody is not None: tbl = tbl.tbody
193 for tr in tbl.findAll("tr"):
194 if "Alt Names:" in tr.td.text:
196 if nls.name != "td" or nls.span is None:
197 raise pageerror("Weird altnames table in " + self.id, page)
198 cnames = [nm.text.strip() for nm in nls.findAll("span")]
200 if cnames is not None:
203 raise pageerror("Could not find altnames for " + self.id, page)
211 return "<batoto.manga %r>" % self.name
213 class credentials(object):
214 def __init__(self, username, password):
215 self.username = username
216 self.password = password
219 def fromfile(cls, path):
220 username, password = None, None
221 with open(path) as fp:
222 for words in profile.splitlines(fp):
223 if words[0] == "username":
225 elif words[0] == "password":
227 elif words[0] == "pass64":
229 password = binascii.a2b_base64(words[1]).decode("utf8")
230 if None in (username, password):
231 raise ValueError("Incomplete profile: " + path)
232 return cls(username, password)
236 path = os.path.join(profile.confdir, "batoto")
237 if os.path.exists(path):
238 return cls.fromfile(path)
241 class session(object):
242 def __init__(self, base, credentials):
244 self.creds = credentials
245 self.jar = http.cookiejar.CookieJar()
246 self.web = urllib.request.build_opener(urllib.request.HTTPCookieProcessor(self.jar))
249 rlre = re.compile(r"Welcome, (.*) ")
250 def dologin(self, pre=None):
252 if now - self.lastlogin < 60:
253 raise Exception("Too soon since last login attempt")
255 with self.web.open(self.base) as hs:
256 page = soupify(hs.read())
260 cur = page.find("a", id="user_link")
262 m = self.rlre.search(cur.text)
263 if not m or m.group(1) != self.creds.username:
265 nav = page.find("div", id="user_navigation")
267 for li in nav.findAll("li"):
268 if li.a and "Sign Out" in li.a.string:
269 outurl = li.a["href"]
271 raise pageerror("Could not find logout URL", page)
272 with self.wep.open(outurl) as hs:
274 with self.web.open(self.base) as hs:
275 page = soupify(hs.read())
280 form = page.find("form", id="login")
282 return self.dologin()
284 for el in form.findAll("input", type="hidden"):
285 values[el["name"]] = el["value"]
286 values["ips_username"] = self.creds.username
287 values["ips_password"] = self.creds.password
288 values["rememberMe"] = "1"
289 values["anonymous"] = "1"
290 req = urllib.request.Request(form["action"], urllib.parse.urlencode(values).encode("ascii"))
291 req.add_header("User-Agent", self.useragent)
292 with self.web.open(req) as hs:
293 page = soupify(hs.read())
294 for resp in page.findAll("p", attrs={"class": "message"}):
295 if resp.strong and "You are now signed in" in resp.strong.string:
298 raise pageerror("Could not log in", page)
302 return self.web.open(url)
304 useragent = "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.22 (KHTML, like Gecko) Chrome/25.0.1364.160 Safari/537.22"
305 def fetch(self, url, headers=None):
306 req = urllib.request.Request(url)
307 req.add_header("User-Agent", self.useragent)
308 if headers is not None:
309 for k, v in headers.items():
311 with self.open(req) as hs:
314 def lfetch(self, url, ck):
315 page = soupify(self.fetch(url))
317 self.dologin(pre=page)
318 page = soupify(self.fetch(url))
320 raise pageerror("Could not verify login status despite having logged in", page)
323 class library(lib.library):
324 def __init__(self, *, creds=None):
326 creds = credentials.default()
327 self.base = "http://bato.to/"
328 self.sess = session(self.base, creds)
329 self.lang = "English"
332 url = self.base + "comic/_/comics/" + id
333 page = soupify(self.sess.fetch(url))
334 title = page.find("h1", attrs={"class": "ipsType_pagetitle"})
337 return manga(self, id, title.string.strip(), url)
339 def _search(self, pars):
344 req = urllib.request.Request(self.base + "search?" + urllib.parse.urlencode(_pars))
345 req.add_header("User-Agent", session.useragent)
346 resp = urllib.request.urlopen(req)
348 page = soupify(resp.read())
351 rls = page.find("div", id="comic_search_results").table
352 if rls.tbody is not None:
355 for child in rls.findAll("tr"):
356 if child.th is not None: continue
357 if child.get("id", "")[:11] == "comic_rowo_": continue
358 if child.get("id") == "show_more_row":
361 link = child.td.strong.a
363 m = self.rure.search(url)
364 if m is None: raise Exception("Got weird manga URL: %r" % url)
366 name = link.text.strip()
367 yield manga(self, id, name, url)
372 rure = re.compile(r"/comic/_/([^/]*)$")
373 def search(self, expr):
374 return self._search({"name": expr, "name_cond": "c"})
376 def byname(self, prefix):
377 for res in self._search({"name": prefix, "name_cond": "s"}):
378 if res.name[:len(prefix)].lower() == prefix.lower():
381 for aname in res.altnames():
382 if aname[:len(prefix)].lower() == prefix.lower():
383 yield manga(self, res.id, aname, res.url)
387 print("eliding " + res.name)
388 print(res.altnames())