From: Fredrik Tolf Date: Fri, 20 Dec 2019 02:20:13 +0000 (+0100) Subject: ann.py: Various improvements. X-Git-Url: http://git.dolda2000.com/gitweb/?a=commitdiff_plain;h=307f4e9308c8c11069f48e3bd88fbedddf4f6de3;p=utils.git ann.py: Various improvements. --- diff --git a/ann.py b/ann.py index 2f3970a..ee3d472 100644 --- a/ann.py +++ b/ann.py @@ -63,6 +63,8 @@ def cstr(soup): for el in soup: ret += cstr(el) return ret + elif soup is None: + return None else: return soup.string @@ -101,17 +103,14 @@ class anime(object): def _main(self): return afind(self._page, "div", id="maincontent") - @cproperty - def _info(self): - ret = {} + def _info(self, nm): for t in afind(self._main, "div", id="content-zone")("div", "encyc-info-type"): - if t.strong: - ret[t.strong.text.lower().strip()[:-1]] = t.contents[t.contents.index(t.strong) + 1:] - return ret + if t.strong and t.strong.text.lower().strip()[:-1] == nm: + return t.contents[t.contents.index(t.strong) + 1:] @cproperty def rawname(self): - afind(self._main, "h1", id="page_header").text + return afind(self._main, "h1", id="page_header").text _nre = re.compile(r"^(.*\S)\s+\(([^\)]+)\)$") @cproperty def _sname(self): @@ -125,8 +124,26 @@ class anime(object): def type(self): return self._sname[1] @cproperty + def names(self): + ret = [] + for el in self._info("alternative title"): + if isinstance(el, bs4.Tag) and el.name == "div" and "tab" in el.get("class", []): + m = self._nre.search(el.text) + if m: + ret.append((m.groups()[0], m.groups()[1])) + else: + ret.append((el.text, None)) + if (self.name, None) in ret: + ret.remove((self.name, None)) + ret.insert(0, (self.name, None)) + return ret + + @cproperty def eps(self): - return int(cstr(self._info["number of episodes"])) + ret = cstr(self._info("number of episodes")) + if ret is None: + return ret + return int(ret) def __repr__(self): return "" % (self.name, self.id) @@ -134,6 +151,10 @@ class anime(object): def __str__(self): return self.name + @classmethod + def byid(cls, id): + return cls(id) + linkpat = re.compile("^/encyclopedia/anime\\.php\\?id=(\d+)$") def getlist(name): name = s(name, "^the\s+", "") @@ -156,7 +177,7 @@ def getlist(name): m = linkpat.match(link["href"]) if not m: raise incompatible() - found = anime(int(m.groups()[0])) + found = anime.byid(int(m.groups()[0])) found.rawname = mn ret.append(found) return ret