def doInBackground(self):
#-----------------------------------------------------------------------
# Is it possible for an exception to occur?
#-----------------------------------------------------------------------
try:
#-------------------------------------------------------------------
# Inform the user of what is occurring, and try to retrieve the data
#-------------------------------------------------------------------
print 'connecting...'
doc = Jsoup.connect(self.url).get()
#-------------------------------------------------------------------
# Use Jsoup methods to locate the HTML links & associated text
#-------------------------------------------------------------------
print 'processing...'
#-------------------------------------------------------------------
# Traverse the HTML, looking for <H3> tags and their tables...
#-------------------------------------------------------------------
visitor = FormattingVisitor(self.url)
walker = NodeTraversor(visitor)
walker.traverse(doc)
# Tabs = visitor.Tabs
# print 'Tabs:', Tabs.getTabCount()
# print visitor.toString()
self.frame.add(visitor.Tabs)
self.frame.validate()
except:
Type, value = sys.exc_info()[:2]
Type, value = str(Type), str(value)
print '\nError:', Type
print 'value:', value
def doInBackground(self):
#-----------------------------------------------------------------------
# Is it possible for an exception to be thrown
#-----------------------------------------------------------------------
try:
#-------------------------------------------------------------------
# Inform the user of what is occurring, and try to retrieve the data
#-------------------------------------------------------------------
self.msg.setText('connecting...')
self.doc = Jsoup.connect(self.url).get()
#-------------------------------------------------------------------
# Use Jsoup methods to locate the HTML links & associated text
#-------------------------------------------------------------------
self.msg.setText('processing...')
#-------------------------------------------------------------------
# Warning: Don't update the visible model within the loop.
#-------------------------------------------------------------------
model = DefaultListModel()
for link in self.doc.getElementsByTag('a'):
name = link.text()
href = link.attr('abs:href')
self.docLinks[name] = href
model.addElement(name)
#-------------------------------------------------------------------
# Replace the visible model with the one containing the real data
#-------------------------------------------------------------------
self.List.setModel(model)
self.msg.setText('Make a selection')
except:
Type, value = sys.exc_info()[:2]
Type, value = str(Type), str(value)
print '\nError:', Type
print 'value:', value
self.msg.setText(value)
def getPS3News(self):
self.Doc = Jsoup.connect(self.PS3Url.decode('utf-8')).get()
news = self.Doc.select('table[class=tbl tbl_list_comm]')
news = news.select('tr[class= ]').get(0)
newsUrl = "http://bbs2.ruliweb.daum.net/gaia/do/ruliweb/default/ps/93/" + news.select('a').get(1).attr('href')
newsTitle = news.select('a').get(1).text().encode('utf-8')
self.PS3News = "\002[" + newsTitle.decode('utf-8') + "]\002 " + newsUrl.decode('utf-8')
def highlight_files(self, fnames):
from pygments import highlight
from org.jsoup import Jsoup
st = """
<style type="text/css">
@page {
size: 10in 11in;
position: relative;
margin: 0.2in 0.2in;
display: inline-table;
}
</style>
"""
for filename in fnames:
self.changeUpdate.post(AsyncEvent(self,
ACTION.FILEREAD, "Reading file: %s - 1/5" % filename))
filestring = self.file_to_string(filename)
self.changeUpdate.post(AsyncEvent(self,
ACTION.CONVERTTOHTML, "Hiliting file: %s - 2/5" % filename))
html = highlight(filestring, self.lexer, self.formatter)
self.changeUpdate.post(AsyncEvent(self,
ACTION.CLEANHTML, "Cleaning the HTML 3/5"))
doc = Jsoup.parseBodyFragment(html)
doc.head().html(st).after(doc.getElementsByTag('div').first())
yield doc
def getTodayD(self, group = ""):
url = "http://m.thisisgame.com/pad/"
Doc = Element
Doc = Jsoup.connect(url).get()
_param = 4;
requestStr = ""
if group.upper() == "A" or group.upper() == "A조".decode('utf-8'):
_param = 0
requestStr = "\002[A조 게릴라 일정]\002 ".decode('utf-8')
elif group.upper() == "B" or group.upper() == "B조".decode('utf-8'):
_param = 1
requestStr = "\002[B조 게릴라 일정]\002 ".decode('utf-8')
elif group.upper() == "C" or group.upper() == "C조".decode('utf-8'):
_param = 2
requestStr = "\002[C조 게릴라 일정]\002 ".decode('utf-8')
elif group.upper() == "D" or group.upper() == "D조".decode('utf-8'):
_param = 3
requestStr = "\002[D조 게릴라 일정]\002 ".decode('utf-8')
elif group.upper() == "E" or group.upper() == "E조".decode('utf-8'):
_param = 4
requestStr = "\002[E조 게릴라 일정]\002 ".decode('utf-8')
else:
_param = 4
requestStr = "\002[E조 게릴라 일정]\002 ".decode('utf-8')
requestStr = requestStr + Doc.select("div[class=content guerilla-time]").select("tr")[1].select("td")[_param].text()
self.requestAdd(requestStr)
def getMonsterInfo(self, monsterName):
url = self.getMonsterInfoUrl(monsterName)
Doc = Element
Doc = Jsoup.connect(url).get()
requestStr = "\002["
requestStr = requestStr + Doc.select("dl[class=monstercode]").select("dd").text() + " / "
requestStr = requestStr + Doc.select("dl[class=monsterclass1]").select("dd").text().replace("타입 ".decode('utf-8'), "타입, ".decode('utf-8')) + " / "
requestStr = requestStr + Doc.select("dl[class=monstergrade]").select("dd").text() + " "
requestStr = requestStr + Doc.select("dl[class=monstername]").select("dd").text() + "]\002 "
requestStr = requestStr + Doc.select("dl[class=maxlevel]").select("dd").text() + ", "
requestStr = requestStr + Doc.select("dl[class=teamcost]").select("dd").text() + ", "
requestStr = requestStr + Doc.select("dl[class=maxexp]").select("dd").text() + " | "
requestStr = requestStr + "능력치 [".decode('utf-8')
requestStr = requestStr + "체력 :".decode('utf-8')+ Doc.select("dl[class=hp]")[1].select("dd").text() + " "
requestStr = requestStr + "공격력 :".decode('utf-8')+ Doc.select("dl[class=attack]")[1].select("dd").text() + " "
requestStr = requestStr + "회복 :".decode('utf-8')+ Doc.select("dl[class=heal]")[1].select("dd").text() + "] "
requestStr2 = "\002[".decode('utf-8') + Doc.select("dl[class=leader grouped]").select("dl[class=skillname]").select("dd").text() + "]\002 "
requestStr2 = requestStr2 + Doc.select("dl[class=leader grouped]").select("dl[class=description]").select("dd").text() + " / "
requestStr2 = requestStr2 + Doc.select("dl[class=leader grouped]").select("dl[class=skillextra]").select("dd").text() + " "
requestStr2 = requestStr2 + "\002[".decode('utf-8') + Doc.select("dl[class=normal grouped").select("dl[class=skillname]").select("dd").text() + "]\002 "
requestStr2 = requestStr2 + Doc.select("dl[class=normal grouped]").select("dl[class=turn]").select("dd").text() + " / "
requestStr2 = requestStr2 + Doc.select("dl[class=normal grouped]").select("dl[class=description]").select("dd").text()
if 0 != len(Doc.select("dl[class=awakeskills]").select("span[class=name]")):
requestStr3 = "\002[각성스킬]\002 ".decode('utf-8')
for _i in range(len(Doc.select("dl[class=awakeskills]").select("span[class=name]"))):
requestStr3 = requestStr3 + Doc.select("dl[class=awakeskills]").select("span[class=name]")[_i].text() + " "
requestStr3 = requestStr3.strip().replace(" ", ", ")
self.requestAdd(requestStr)
self.requestAdd(requestStr2)
if 0 != len(Doc.select("dl[class=awakeskills]").select("span[class=name]")):
self.requestAdd(requestStr3)
def Migration(self):
url_ko = "http://ma.inven.co.kr/dataninfo/card/"
Doc = Element
Doc = Jsoup.connect(url_ko).get()
Doc = Doc.select("tbody[id=filteredList]")
Doc = Doc.select("tr")
self.stat.executeUpdate("DELETE FROM " + self.TABLENAME + ";")
for i in range(0, Doc.size()):
prep = self.db.prepareStatement("INSERT INTO " + self.TABLENAME + " VALUES (?, ?, ?, ?, ?, ?);")
prep.setString(1, Doc[i].select("a").attr("href")[Doc[i].select("a").attr("href").find("=")+1:]) #ID
prep.setString(2, "http://ma.inven.co.kr/dataninfo/card/" + Doc[i].select("a").attr("href")) #LINK
prep.setString(3, Doc[i].select("a").text()) #NAME
prep.setString(4, Doc[i].select("img").attr("src")) #FACE
rarity = ''
for j in range(0, Doc[i].select("img").size()):
if Doc[i].select("img")[j].attr("src").find("card_grade_star.png") != -1:
rarity = rarity + "☆".decode('utf-8')
if j == 5:
rarity = "★".decode('utf-8')
prep.setString(5, rarity) #레어리티
prep.setString(6, Doc[i].select("td[class=cost]").text()) #COST
prep.addBatch()
self.db.setAutoCommit(False);
prep.executeBatch();
self.db.setAutoCommit(True);
def doInBackground( self ) :
#-----------------------------------------------------------------------
# Is it possible for an exception to occur?
#-----------------------------------------------------------------------
try :
#-------------------------------------------------------------------
# Inform the user of what is occurring, and try to retrieve the data
#-------------------------------------------------------------------
self.showMessage( 'Connecting...' )
doc = Jsoup.connect( self.url ).get()
#-------------------------------------------------------------------
# Use Jsoup methods to locate the HTML links & associated text
#-------------------------------------------------------------------
self.showMessage( 'Processing...' )
#-------------------------------------------------------------------
# Traverse the HTML, looking for <H3> tags and their tables...
#-------------------------------------------------------------------
visitor = FormattingVisitor( self.url )
walker = NodeTraversor( visitor )
walker.traverse( doc )
# Tabs = visitor.Tabs
# print 'Tabs:', Tabs.getTabCount()
# print visitor.toString()
self.sPane.setRightComponent( visitor.Tabs )
except :
Type, value = sys.exc_info()[ :2 ]
Type, value = str( Type ), str( value )
self.showMessage( '\nheaderTask Error: %s\nvalue: %s' % ( Type, value ) )
def get_html(self, n=10): #Stackoverflow Parser 시작점....!!!!!!!!
elements = []
so_question_list = self._stackoverflow_links(n)
for so_question in so_question_list:
print "Treat link", so_question["title"]
#doc = Jsoup.connect(so_question["link"]).userAgent("Mozilla/5.0 (Windows; U; WindowsNT 5.1; en-US; rv1.8.1.6) Gecko/20070725 Firefox/2.0.0.6").get()
doc = Jsoup.connect(so_question["link"]).userAgent(
'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_11_3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/48.0.2564.109 Safari/537.36'
).get()
#TODO 여기부터..
if not self.template: #template을 None으로 초기화 했기 때문에 template이 None이면 초기를 doc으로 설정
self.template = doc
elements.append({
"e": doc.select(".accepted-answer .post-text"),
"title": so_question["title"]
})
if self.template:
table_checkbox = self._table()
self.snippets = [
table_checkbox %
(element["title"], i, element["e"].outerHtml())
for i, element in enumerate(elements[:n])
]
self._relative_to_absolute()
self.template.body().html("".join(self.snippets))
self._add_style()
return self.template.outerHtml()
return "%s Results. Your search returned no matches." % (
len(so_question_list))
def remove_code_block(s):
from org.jsoup import Jsoup
doc = Jsoup.parse(s)
for element in doc.select("code"):
element.remove()
return doc.text()
def getMonsterPic(self, monsterName):
url = self.getMonsterInfoUrl(monsterName)
Doc = Element
Doc = Jsoup.connect(url).get()
requestStr = "\002[" + Doc.select("dl[class=monstername]").select("dd").text() + "]\002 "
requestStr = requestStr + Doc.select("div[class=image]").select("img").attr("src")
self.requestAdd(requestStr)
return
def fetch_document(scratch_project_url, timeout, user_agent):
connection = Jsoup.connect(scratch_project_url)
connection.userAgent(user_agent)
connection.timeout(timeout)
connection.ignoreContentType(True)
# connection.header("content-type", " text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8")
# return _ResponseJsoupDocumentWrapper(connection.get().text())
return json.loads(connection.get().text())
def connect(self):
try:
# r = requests.get(self.url, timeout=20)
self.doc = Jsoup.connect(self.url).userAgent(
"Mozilla/5.0 (Windows NT 6.3; rv:36.0) Gecko/20100101 Firefox/36.0"
).referrer("https://www.google.com").timeout(0).get()
except Exception as e:
print "Download - Error: ", self.url, e
def so_text(s):
""" Removes code tag and its content from SO body as well as all html tags"""
from org.jsoup import Jsoup
s = unescape_html(s)
doc = Jsoup.parse(s)
for element in doc.select("code"):
element.remove()
return doc.text()
def getProfileUrl(self, username):
Doc = Element
url = "http://osu.ppy.sh/u/" + username
Doc = Jsoup.connect(url).get()
id = Doc.select("script")[4].html()
id = id[id.find("var userId = ")+13:id.find(";")]
if id != "":
self.getUserInfo(id, username)
else:
self.requestAdd("?")
def typeCmdProc(self):
if self.getMessage("msg").find('http://') != -1 or self.getMessage("msg").find('https://') != -1:
try:
url = self.getMessage('msg')[self.getMessage('msg').find('http'):]
url = url.split(' ')[0]
Doc = Jsoup.connect(url).get()
title = '\002[ LINK TITLE ]\002 ' + Doc.select('title').text().replace('\n', ' ')
self.requestAdd( title );
except:
pass
def getCallNumber(self, number):
myUrl = self.Url + str(number)
self.Doc = Jsoup.connect(myUrl.decode('utf-8')).get()
myNumber = self.Doc.select("tr[class=bg1]").select("td").select("td")[1].select("a").text().split()[0]
numberUrl = self.Doc.select("tr[class=bg1]").select("td").select("td")[1].select("a").attr("href")[2:]
numberUrl = "http://www.thecall.co.kr/" + numberUrl
try:
self.requestAdd("\002["+ myNumber+ " / " + numberUrl + "]\002 " + self.Doc.select("tr[class=bg1]").select("td").select("td")[2].text())
except:
self.requestAdd("해당 번호를 검색할 수 없거나, 연결에 실패했습니다".decode('utf-8'))
def checkShortLink(self, keyword):
Doc = Element
checkUrl = self.BASEURL + "&action=url-stats&shorturl=" + urllib.unquote(keyword)
Doc = Jsoup.connect(checkUrl).get()
message = Doc.select("message").text()
shorturl = Doc.select("shorturl").text()
url = Doc.select("url").text()
title = Doc.select("title").text()
if message.find("Error: short URL not found") != -1:
self.requestAdd("해당 키워드로 정의된 숏링크를 수 없었어요.".decode('utf-8'))
elif message.find("success") != -1:
self.requestAdd("\002[ " + shorturl + " -> ".decode('utf-8') + url + " ]\002")
def getUserInfo(self, id):
Doc = Element
url = "http://www.noobmeter.com/player/kr/" + id
Doc = Jsoup.connect(url).get()
PR = ""
ER = ""
WN6 = ""
WN7 = ""
Battles = ""
WR = ""
AD = ""
AE = ""
ME = ""
ABC = ""
ABD = ""
AK = ""
BS = ""
KD = ""
PR = Doc.select("div[class=pageContent]").select("tbody").select("tr")[0].select("td")[1].text()
ER = Doc.select("div[class=pageContent]").select("tbody").select("tr")[1].select("td")[1].text()
WN6 = Doc.select("div[class=pageContent]").select("tbody").select("tr")[2].select("td")[1].text()
WN7 = Doc.select("div[class=pageContent]").select("tbody").select("tr")[3].select("td")[1].text()
Battles = Doc.select("div[class=pageContent]").select("tbody").select("tr")[4].select("td")[1].text()
WR = Doc.select("div[class=pageContent]").select("tbody").select("tr")[5].select("td")[1].text()
AD = Doc.select("div[class=pageContent]").select("tbody").select("tr")[6].select("td")[1].text()
AE = Doc.select("div[class=pageContent]").select("tbody").select("tr")[7].select("td")[1].text()
ME = Doc.select("div[class=pageContent]").select("tbody").select("tr")[8].select("td")[1].text()
ABC = Doc.select("div[class=pageContent]").select("tbody").select("tr")[9].select("td")[1].text()
ABD = Doc.select("div[class=pageContent]").select("tbody").select("tr")[10].select("td")[1].text()
AK = Doc.select("div[class=pageContent]").select("tbody").select("tr")[11].select("td")[1].text()
BS = Doc.select("div[class=pageContent]").select("tbody").select("tr")[12].select("td")[1].text()
KD = Doc.select("div[class=pageContent]").select("tbody").select("tr")[13].select("td")[1].text()
requestStr = "\002[" + id + "]\002"
requestStr = requestStr + " 퍼포먼스 레이팅: ".decode('utf-8') + PR
requestStr = requestStr + " | 이피신시 레이팅: ".decode('utf-8') + ER
requestStr = requestStr + " | WN6: ".decode('utf-8') + WN6
requestStr = requestStr + " | WN7: ".decode('utf-8') + WN7
requestStr = requestStr + " | 총배틀: ".decode('utf-8') + Battles
requestStr = requestStr + " | 승율: ".decode('utf-8') + WR
requestStr = requestStr + " | 평균데미지: ".decode('utf-8') + AD
requestStr = requestStr + " | 평균경험치: ".decode('utf-8') + AE
requestStr = requestStr + " | 최대경험치: ".decode('utf-8') + ME
requestStr = requestStr + " | 점거율: ".decode('utf-8') + ABC
requestStr = requestStr + " | 방어율: ".decode('utf-8') + ABD
requestStr = requestStr + " | 격파율: ".decode('utf-8') + AK
requestStr = requestStr + " | 생존율: ".decode('utf-8') + BS
requestStr = requestStr + " | 킬뎃: ".decode('utf-8') + KD
self.requestAdd(requestStr)
def doInBackground(self):
#-----------------------------------------------------------------------
# Is it possible for an exception to be thrown
#-----------------------------------------------------------------------
try:
#-------------------------------------------------------------------
# Inform the user of what is occurring, and try to retrieve the data
#-------------------------------------------------------------------
self.area.setText('connecting...')
doc = Jsoup.connect(self.url).get()
self.area.setText(str(doc.normalise()))
except:
Type, value = sys.exc_info()[:2]
Type, value = str(Type), str(value)
self.area.setText('\nError: %s\nValue: %s' % (Type, value))
def Migration(self):
url_ko = "http://web.ruliweb.daum.net/etcs/mia/mia_card.htm"
url_jp = "http://web.ruliweb.daum.net/etcs/mia_jp/mia_card.htm"
Doc = Element
Doc = Jsoup.connect(url_ko).get()
Doc = Doc.select("table[class=mia_tbl]")
Doc = Doc.select("div").select("a")
self.stat.executeUpdate("DELETE FROM " + self.TABLENAME + ";")
for i in range(0, Doc.size()):
if i % 2 == 1:
prep = self.db.prepareStatement("INSERT INTO " + self.TABLENAME + " VALUES (?, ?, ?, ?, ?);")
prep.setString(1, Doc[i].select("a").attr("href")[Doc[i].select("a").attr("href").find("=") + 1:])
prep.setString(2, "KO")
prep.setString(3, "http://web.ruliweb.daum.net/etcs/mia/" + Doc[i].select("a").attr("href"))
prep.setString(4, Doc[i].html().replace("<br />", "*").split("*")[0])
prep.setString(5, Doc[i].html().replace("<br />", "*").split("*")[0])
prep.addBatch()
self.db.setAutoCommit(False);
prep.executeBatch();
self.db.setAutoCommit(True);
Doc = Jsoup.connect(url_jp).get()
Doc = Doc.select("table[class=mia_tbl]")
Doc = Doc.select("div").select("a")
for i in range(0, Doc.size()):
if i % 2 == 1:
prep = self.db.prepareStatement("INSERT INTO " + self.TABLENAME + " VALUES (?, ?, ?, ?, ?);")
prep.setString(1, Doc[i].select("a").attr("href")[Doc[i].select("a").attr("href").find("=") + 1:])
prep.setString(2, "JP")
prep.setString(3, "http://web.ruliweb.daum.net/etcs/mia_jp/" + Doc[i].select("a").attr("href"))
prep.setString(4, Doc[i].html().replace("<br />", "*").split("*")[0])
prep.setString(5, Doc[i].html().replace("<br />", "*").split("*")[0])
prep.addBatch()
self.db.setAutoCommit(False);
prep.executeBatch();
self.db.setAutoCommit(True);
def getCheckFocus(self):
url = "http://m.thisisgame.com/pad/"
Doc = Element
Doc = Jsoup.connect(url).get()
requestStr = ""
requestStr = requestStr + Doc.select("div[class =content week-focus]").select("li")[0].html()
requestStr = requestStr + " / " + Doc.select("div[class =content week-focus]").select("li")[1].html()
requestStr = requestStr + " / " + Doc.select("div[class =content week-focus]").select("li")[2].html()
requestStr = requestStr + "\n"
requestStr = requestStr + Doc.select("div[class =content week-focus]").select("li")[3].html()
requestStr = requestStr + " / " + Doc.select("div[class =content week-focus]").select("li")[4].html()
requestStr = requestStr + " / " + Doc.select("div[class =content week-focus]").select("li")[5].html()
requestStr = re.sub("</?[a][a-z0-9]*[^<>]*>", "", requestStr).replace("<span>", "").replace("</span>", "").replace(" 오후 11시 59분".decode('utf-8'), "")
requestStr = requestStr.replace("<strong>", "\002").replace("</strong>", "\002").replace("<br />", " ")
self.requestAdd(requestStr)
def getMonsterSearchResult(self, monsterName):
monsterName = monsterName.replace(" ", "%20")
url = "http://www.thisisgame.com/pad/info/monster/list.php?sf=name&sw=" + monsterName
Doc = Element
Doc = Jsoup.connect(url).get()
Doc = Doc.select("div[class=pad-table2 pad-info-light]").select("tr")
requestStr = monsterName + ": "
if 1 != len(Doc):
for _i in range(len(Doc)):
if _i <= 1:
tmp = Doc[_i].select("td").text()
requestStr = requestStr + tmp[0:tmp.find(" [")]
else:
tmp = Doc[_i].select("td").text()
requestStr = requestStr + " / " + tmp[0:tmp.find(" [")]
if _i == 5:
break
self.requestAdd(requestStr)
def doInBackground(self):
#-----------------------------------------------------------------------
# Disable the text (input) field, if it exists
#-----------------------------------------------------------------------
try:
self.msg.setText('working...')
self.doc = Jsoup.connect(self.url).get()
self.msg.setText('ready')
except:
Type, value = sys.exc_info()[:2]
print 'Error:', str(type)
print 'value:', str(value)
self.msg.setText(str(value))
#-----------------------------------------------------------------------
# Was the specified URL retrieved?
#-----------------------------------------------------------------------
if self.doc:
self.cb.removeAllItems()
for link in self.doc.getElementsByTag('a'):
self.cb.addItem(str(link.text()))
def requestShortLink(self, geturl, keyword):
Doc = Element
if keyword == "":
geturl = self.BASEURL + "&action=shorturl&url=" + urllib.quote(geturl.encode('utf-8')).replace("&", "%26").replace("#", "%23")
else:
geturl = self.BASEURL + "&action=shorturl&url=" + urllib.quote(geturl.encode('utf-8')).replace("&", "%26").replace("#", "%23") + "&keyword=" + keyword.encode('utf-8')
Doc = Jsoup.connect(geturl).get()
status = Doc.select("status").text()
message = Doc.select("message").text()
shorturl = Doc.select("shorturl").text()
url = Doc.select("url")[1].text()
if(status == "success"):
return shorturl
elif(status == "fail"):
if message.find("already exists in database") != -1:
if shorturl == "":
return message
else:
return shorturl
else:
return message
def requestShortLink(self, geturl, keyword):
Doc = Element
if keyword == "":
geturl = self.BASEURL + "&action=shorturl&url=" + urllib.quote(geturl.encode('utf-8')).replace("&", "%26").replace("#", "%23")
else:
geturl = self.BASEURL + "&action=shorturl&url=" + urllib.quote(geturl.encode('utf-8')).replace("&", "%26").replace("#", "%23") + "&keyword=" + keyword.encode('utf-8')
Doc = Jsoup.connect(geturl).get()
status = Doc.select("status").text()
message = Doc.select("message").text()
shorturl = Doc.select("shorturl").text()
url = Doc.select("url")[1].text()
if(status == "success"):
self.requestAdd("숏링크 생성 완료! ".decode('utf-8') + "\002[ " + url + " -> " + shorturl + " ]\002")
elif(status == "fail"):
if message.find("already exists in database") != -1:
if shorturl == "":
self.requestAdd("실패! 정의되지 않은 오류. ".decode('utf-8') + " 메세지: ".decode('utf-8') + message)
else:
self.requestAdd("실패!".decode('utf-8') + " 이미 생성된 링크일 수 있습니다 ".decode('utf-8') + " \002[ ".decode('utf-8') + shorturl + " ]\002")
else:
self.requestAdd("실패! 정의되지 않은 오류. ".decode('utf-8') + " 메세지: ".decode('utf-8') + message + " " + " \002[ ".decode('utf-8') + shorturl + " ]\002")
def so_tokenizer(s, remove_html=True, as_str=True):
if remove_html:
from org.jsoup import Jsoup
s = unescape_html(s)
doc = Jsoup.parse(s)
s = doc.text()
tokens = tokenize(s)
tokens = set(tokens)
res = []
for token in tokens:
res.extend(camel_case_split(token))
res.append(token.lower())
res = [item for item in res if item not in java_stopwords]
res = set(res)
if as_str:
return " ".join(res)
else:
return res
def doInBackground(self):
#-----------------------------------------------------------------------
# Is it possible for an exception to occur?
#-----------------------------------------------------------------------
try:
#-------------------------------------------------------------------
# Inform the user of what is occurring, and try to retrieve the data
#-------------------------------------------------------------------
self.result.setText('connecting...')
doc = Jsoup.connect(self.url).get()
#-------------------------------------------------------------------
# Use Jsoup methods to locate the HTML links & associated text
#-------------------------------------------------------------------
self.result.setText('processing...')
#-------------------------------------------------------------------
# Traverse the HTML, looking for <H#...> tags
#-------------------------------------------------------------------
self.text = self.getPlainText(doc)
except:
Type, value = sys.exc_info()[:2]
Type, value = str(Type), str(value)
print '\nError:', Type
print 'value:', value
self.result.setText('Exception: %s' % value)
def doInBackground( self ) :
#-----------------------------------------------------------------------
# Is it possible for an exception to be thrown
#-----------------------------------------------------------------------
try :
#-------------------------------------------------------------------
# Inform the user of what is occurring, and try to retrieve the data
#-------------------------------------------------------------------
model = self.List.getModel()
model.set( 0, 'Connecting...' )
self.doc = Jsoup.connect( self.url ).get()
#-------------------------------------------------------------------
# Use Jsoup methods to locate the HTML links & associated text
#-------------------------------------------------------------------
model.set( 0, 'Processing...' )
#-------------------------------------------------------------------
# Warning: Don't update the visible model within the loop.
# Note: the 'abs:href' notation causes jsoup to return an absolute
# (i.e., complete) URL.
#-------------------------------------------------------------------
self.model = DefaultListModel()
for link in self.doc.getElementsByTag( 'a' ) :
name = link.text()
href = link.attr( 'abs:href' )
self.docLinks[ name ] = href
self.model.addElement( name )
#-------------------------------------------------------------------
# Replace the visible model with the one containing the real data
#-------------------------------------------------------------------
self.List.setModel( self.model )
except :
Type, value = sys.exc_info()[ :2 ]
Type, value = str( Type ), str( value )
print '\nsoupTask Error:', Type
print 'value:', value
sys.exit()
def fetch_document(scratch_project_url, timeout, user_agent):
connection = Jsoup.connect(scratch_project_url)
connection.userAgent(user_agent)
connection.timeout(timeout)
return _ResponseJsoupDocumentWrapper(connection.get())
def getCardInfo(self, requestStr ,cardName):
stmt = self.db.createStatement();
cardName = cardName.replace("\017", "").strip()
rs = ''
cardUrl = ''
cardNumber = 0
if cardName.split()[0] == "일".decode('utf-8'):
cardName = self.countSplit(cardName, 1)
try:
rs = stmt.executeQuery("SELECT * FROM "+ self.TABLENAME + " WHERE " + self.COUNTRY + " = 'JP' AND " + self.CARDNAME + " LIKE '" + cardName +"';")
cardUrl = rs.getString(self.CARDURL)
cardNumber = rs.getString(self.CARDID)
except:
rs = stmt.executeQuery("SELECT * FROM "+ self.TABLENAME + " WHERE " + self.COUNTRY + " = 'JP' AND " + self.CARDNAME + " LIKE '%" + cardName +"%';")
cardUrl = rs.getString(self.CARDURL)
cardNumber = rs.getString(self.CARDID)
elif cardName.split()[0] == "한".decode('utf-8'):
cardName = self.countSplit(cardName, 1)
try:
rs = stmt.executeQuery("SELECT * FROM "+ self.TABLENAME + " WHERE " + self.COUNTRY + " = 'KO' AND " + self.CARDNAME + " LIKE '" + cardName +"';")
cardUrl = rs.getString(self.CARDURL)
cardNumber = rs.getString(self.CARDID)
except:
rs = stmt.executeQuery("SELECT * FROM "+ self.TABLENAME + " WHERE " + self.COUNTRY + " = 'KO' AND " + self.CARDNAME + " LIKE '%" + cardName +"%';")
cardUrl = rs.getString(self.CARDURL)
cardNumber = rs.getString(self.CARDID)
else:
try:
rs = stmt.executeQuery("SELECT * FROM "+ self.TABLENAME + " WHERE " + self.CARDNAME + " LIKE '" + cardName +"';")
cardUrl = rs.getString(self.CARDURL)
cardNumber = rs.getString(self.CARDID)
except:
rs = stmt.executeQuery("SELECT * FROM "+ self.TABLENAME + " WHERE " + self.CARDNAME + " LIKE '%" + cardName +"%';")
cardUrl = rs.getString(self.CARDURL)
cardNumber = rs.getString(self.CARDID)
Doc = Element
Doc = Jsoup.connect(cardUrl).get()
cardImg1 = Doc.select("table[id=content_block_1]").select("td")[0].select("a").attr("href")
cardImg2 = Doc.select("table[id=content_block_1]").select("td")[1].select("a").attr("href")
cardImg3 = Doc.select("table[id=content_block_1]").select("td")[2].select("a").attr("href")
cardImg4 = Doc.select("table[id=content_block_1]").select("td")[3].select("a").attr("href")
cardImg5 = Doc.select("table[id=content_block_1]").select("td")[4].select("a").attr("href")
cardImg6 = Doc.select("table[id=content_block_1]").select("td")[5].select("a").attr("href")
cardIllustrator = ""
try:
cardIllustrator = Doc.select("table[id=content_block_1]").select("td")[6].text().split(":")[1].strip()
except:
try:
cardIllustrator = Doc.select("table[id=content_block_1]").select("td")[12].text().split(":")[1].strip()
except:
pass
cardImg7 = ""
cardImg8 = ""
cardImg9 = ""
cardImg10 = ""
cardImg11 = ""
cardImg12 = ""
try:
cardImg7 = Doc.select("table[id=content_block_1]").select("td")[6].select("a").attr("href")
cardImg8 = Doc.select("table[id=content_block_1]").select("td")[7].select("a").attr("href")
cardImg9 = Doc.select("table[id=content_block_1]").select("td")[8].select("a").attr("href")
cardImg10 = Doc.select("table[id=content_block_1]").select("td")[9].select("a").attr("href")
cardImg11 = Doc.select("table[id=content_block_1]").select("td")[10].select("a").attr("href")
cardImg12 = Doc.select("table[id=content_block_1]").select("td")[11].select("a").attr("href")
except:
cardImg7 = cardImg8 = cardImg9 = cardImg10 = cardImg11 = "오리지날 일러를 찾을 수 없었어요".decode('utf-8')
cardName = Doc.select("table[class=mia_tbl]").select("tr")[0].select("td")[0].text()
cardCode = Doc.select("table[class=mia_tbl]").select("tr")[0].select("td")[1].text()
cardInfo = Doc.select("table[class=mia_tbl]").select("tr")[1].text()
cardSkillName = Doc.select("table[class=mia_tbl]").select("tr")[2].select("p")[0].text()
cardSkillInfo = Doc.select("table[class=mia_tbl]").select("tr")[2].select("p")[1].text()
cardRarelity = Doc.select("table[class=mia_tbl]").select("tr")[3].select("td")[0].text().replace("★".decode('utf-8'), "☆".decode('utf-8')).replace('☆☆☆☆☆'.decode('utf-8'), '★'.decode('utf-8'))
cardCost = Doc.select("table[class=mia_tbl]").select("tr")[3].select("td")[1].text()
cardFaction = Doc.select("table[class=mia_tbl]").select("tr")[4].select("td")[0].text()
cardType = Doc.select("table[class=mia_tbl]").select("tr")[4].select("td")[1].text()
cardGender = Doc.select("table[class=mia_tbl]").select("tr")[5].select("td")[0].text()
cardForce = Doc.select("table[class=mia_tbl]").select("tr")[5].select("td")[1].text()
cardGrowthRate = Doc.select("table[class=mia_tbl]").select("tr")[6].select("td")[0].text()
cardLimitBreakCount = Doc.select("table[class=mia_tbl]").select("tr")[6].select("td")[1].text()
cardGetInfo = Doc.select("table[class=mia_tbl]").select("tr")[7].select("td")[0].text()
cardMLV = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[1].html().replace("<br />", "*")
cardMLV = cardMLV[3:cardMLV.find("*")]
cardLLV = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[2].html().replace("<br />", "*")
cardLLV = cardLLV[3:cardLLV.find("*")]
cardHP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[9].text()
cardATK = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[10].text()
cardMHP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[11].text()
cardMATK = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[12].text()
cardFHP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[13].text()
cardFATK = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[14].text()
cardSum = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[18].text()
cardMSum = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[19].text()
cardFSum = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[20].text()
cardCP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[21].text()
cardMCP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[22].text()
cardFCP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[23].text()
cardHHP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[33].text()
cardHATK = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[34].text()
cardHMHP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[35].text()
cardHMATK = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[36].text()
cardHFHP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[37].text()
cardHFATK = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[38].text()
cardHSum = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[42].text()
cardHMSum = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[43].text()
cardHFSum = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[44].text()
cardHCP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[45].text()
cardHMCP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[46].text()
cardHFCP = Doc.select("table[class=mia_tbl mia_a003 mia_normal]").select("td")[47].text()
requestMsg = ""
if requestStr == "!카드2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardRarelity + " " + cardName + "]\002 "
requestMsg = requestMsg + "코스트 : ".decode('utf-8') + cardCost + ", 각성레벨(한계레벨): ".decode('utf-8') + cardMLV + "(" + cardLLV + ") \002|\002 "
requestMsg = requestMsg + "체력 (M/F): ".decode('utf-8') + cardHP + "(" + cardMHP + "/" + cardFHP +"), "
requestMsg = requestMsg + "공격력 (M/F): ".decode('utf-8') + cardATK + "(" + cardMATK + "/" + cardFATK +"), "
requestMsg = requestMsg + "H+A(M/F): ".decode('utf-8') + cardSum + "(" + cardMSum + "/" + cardFSum +") \002|\002 "
requestMsg = requestMsg + "CP (M/F): ".decode('utf-8') + cardCP + "(" + cardMCP + "/" + cardFCP +") \002|\002 "
requestMsg = requestMsg + cardSkillName + " : " + cardSkillInfo
if requestStr == "!키라카드2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardRarelity + " " + cardName + "]\002 "
requestMsg = requestMsg + "코스트 : ".decode('utf-8') + cardCost + ", 각성레벨(한계레벨): ".decode('utf-8') + cardMLV + "(" + cardLLV + ") \002|\002 "
requestMsg = requestMsg + "체력 (M/F): ".decode('utf-8') + cardHHP + "(" + cardHMHP + "/" + cardHFHP +"), "
requestMsg = requestMsg + "공격력 (M/F): ".decode('utf-8') + cardHATK + "(" + cardHMATK + "/" + cardHFATK +"), "
requestMsg = requestMsg + "H+A(M/F): ".decode('utf-8') + cardHSum + "(" + cardHMSum + "/" + cardHFSum +") \002|\002 "
requestMsg = requestMsg + "CP (M/F): ".decode('utf-8') + cardHCP + "(" + cardHMCP + "/" + cardHFCP +") \002|\002 "
requestMsg = requestMsg + cardSkillName + " : " + cardSkillInfo
if requestStr == "!일러2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + " / " + cardIllustrator + "]\002 "
requestMsg = requestMsg + " " + cardImg1
if requestStr == "!각성일러2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + " / " + cardIllustrator + "]\002 "
requestMsg = requestMsg + " " + cardImg2
if requestStr == "!만렙일러2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + " / " + cardIllustrator + "]\002 "
requestMsg = requestMsg + " " + cardImg3
if requestStr == "!키라일러2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + " / " + cardIllustrator + "]\002 "
requestMsg = requestMsg + " " + cardImg4
if requestStr == "!키라각성일러2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + " / " + cardIllustrator + "]\002 "
requestMsg = requestMsg + " " + cardImg5
if requestStr == "!키라만렙일러2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + " / " + cardIllustrator + "]\002 "
requestMsg = requestMsg + " " + cardImg6
if requestStr == "!오리지날일러2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + " / " + cardIllustrator + "]\002 "
requestMsg = requestMsg + " " + cardImg7
if requestStr == "!오리지날각성일러2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + " / " + cardIllustrator + "]\002 "
requestMsg = requestMsg + " " + cardImg8
if requestStr == "!오리지날키라일러2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + " / " + cardIllustrator + "]\002 "
requestMsg = requestMsg + " " + cardImg10
if requestStr == "!오리지날키라각성일러2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + " / " + cardIllustrator + "]\002 "
requestMsg = requestMsg + " " + cardImg11
if requestStr == "!카드설명2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardName + "]\002 "
hangul = re.compile('''[^ ㄱ-ㅣ가-힣0-9a-zA-Z.,"']+'''.decode('utf-8'))
cardInfo = hangul.sub('', cardInfo)
if len(cardInfo) > 320:
cardInfo = cardInfo[:140] + "\n" + cardInfo[140:] + "...(생략)".decode('utf-8')
if len(cardInfo) > 150:
cardInfo = cardInfo[:140] + "\n" + cardInfo[140:]
requestMsg = requestMsg + " " + cardInfo
if requestStr == "!좀더카드정보2".decode('utf-8') or requestStr == "!mtcdjb".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardRarelity + " " + cardName + "]\002 "
requestMsg = requestMsg + "넘버: ".decode('utf-8') + cardNumber + ", "
requestMsg = requestMsg + "성별: ".decode('utf-8') + cardGender + ", "
requestMsg = requestMsg + "포스: ".decode('utf-8') + cardForce + ", "
requestMsg = requestMsg + "성장률: ".decode('utf-8') + cardGrowthRate + ", "
requestMsg = requestMsg + "한돌횟수: ".decode('utf-8') + cardLimitBreakCount + ", "
requestMsg = requestMsg + "입수방법: ".decode('utf-8') + cardGetInfo
if requestStr == "!좀덜카드정보2".decode('utf-8'):
requestMsg = "\002[" + cardFaction + " " + cardRarelity + " " + cardName + "]\002 "
requestMsg = requestMsg + "넘버: ".decode('utf-8') + cardNumber + ", "
requestMsg = requestMsg + "성별: ".decode('utf-8') + cardGender + ", "
requestMsg = requestMsg + "포스: ".decode('utf-8') + cardForce + ", "
requestMsg = requestMsg + "성장률: ".decode('utf-8') + cardGrowthRate + ", "
requestMsg = requestMsg + "한돌횟수: ".decode('utf-8') + cardLimitBreakCount + ", "
requestMsg = requestMsg + "입수방법: ".decode('utf-8') + cardGetInfo
requestMsg = requestMsg[:len(requestMsg)/2]
self.requestAdd(requestMsg)
def getUserInfo(self, id, username):
Doc = Element
url = "http://osu.ppy.sh/pages/include/profile-general.php?u=" + id
if self.requestStr == "!오스".decode('utf-8'):
url = url + "&m=0"
elif self.requestStr == "!태고".decode('utf-8'):
url = url + "&m=1"
elif self.requestStr == "!캐치어비트".decode('utf-8'):
url = url + "&m=2"
elif self.requestStr == "!오스!매니아".decode('utf-8'):
url = url + "&m=3"
Doc = Jsoup.connect(url).get()
rank = ""
score = ""
accuracy = ""
playcount = ""
totalscore = ""
level = ""
hits = ""
maxcombo = ""
if Doc.select("div[class=profileStatLine]")[0].text().find("Performance Rank") != -1:
rank = Doc.select("div[class=profileStatLine]")[0].text()
rank = rank[rank.find(":")+2:]
score = Doc.select("div[class=profileStatLine]")[1].text()
score = score[score.find(":")+2:]
accuracy = Doc.select("div[class=profileStatLine]")[2].text()
accuracy = accuracy[accuracy.find(":")+2:]
playcount = Doc.select("div[class=profileStatLine]")[3].text()
playcount = playcount[playcount.find(":")+2:]
totalscore = Doc.select("div[class=profileStatLine]")[4].text()
totalscore = totalscore[totalscore.find(":")+2:]
level = Doc.select("div[class=profileStatLine]")[5].text()
level = level[level.find(":")+2:]
hits = Doc.select("div[class=profileStatLine]")[6].text()
hits = hits[hits.find(":")+2:]
maxcombo = Doc.select("div[class=profileStatLine]")[7].text()
maxcombo = maxcombo[maxcombo.find(":")+2:]
else:
rank = "#-"
score = Doc.select("div[class=profileStatLine]")[0].text()
score = score[score.find(":")+2:]
accuracy = Doc.select("div[class=profileStatLine]")[1].text()
accuracy = accuracy[accuracy.find(":")+2:]
playcount = Doc.select("div[class=profileStatLine]")[2].text()
playcount = playcount[playcount.find(":")+2:]
totalscore = Doc.select("div[class=profileStatLine]")[3].text()
totalscore = totalscore[totalscore.find(":")+2:]
level = Doc.select("div[class=profileStatLine]")[4].text()
level = level[level.find(":")+2:]
hits = Doc.select("div[class=profileStatLine]")[5].text()
hits = hits[hits.find(":")+2:]
maxcombo = Doc.select("div[class=profileStatLine]")[6].text()
maxcombo = maxcombo[maxcombo.find(":")+2:]
if Doc.select("td")[1].text() != "":
#eventepic = Doc.select("td")[0].text() +": " + Doc.select("td")[1].select("div[class=event epic4]").text()
eventepic = Doc.select("td")[0].text() + " " + Doc.select("td")[1].text()
#Doc.select("td")[1].select("a")[0].text().decode('utf-8')
else:
eventepic = ""
requestStr = "\002[" + level + "레벨 ".decode('utf-8') + username + "]\002"
requestStr = requestStr + " 랭크: ".decode('utf-8') + rank + " / 정확도: ".decode('utf-8') + accuracy
requestStr = requestStr + " / 플레이카운트: ".decode('utf-8') + playcount + " / 토탈스코어: ".decode('utf-8') + totalscore
requestStr = requestStr + " / 총 히트 수: ".decode('utf-8') + hits + " / 최대콤보: ".decode('utf-8') + maxcombo + " | "
requestStr = requestStr + eventepic
self.requestAdd(requestStr)
def fetch_document(scratch_project_url, timeout, user_agent):
connection = Jsoup.connect(scratch_project_url)
connection.userAgent(user_agent)
connection.timeout(timeout)
return _ResponseJsoupDocumentWrapper(connection.get())
def remove_html_tags(s):
from org.jsoup import Jsoup
return Jsoup.parse(s).text()
def __init__(self, answer): self.answer = answer self.inline = [] self.block = [] self.doc = Jsoup.parse(answer)
import sys
import os
import test1
from org.jsoup import Jsoup
from com.pixshow.framework.utils import HttpUtility
url = "http://en.wikipedia.org/";
print test1.workDir()
html = HttpUtility.get(url);
doc = Jsoup.parse(html)
html = doc.select('#mp-itn b a').toString()
appContext.get('testService').save(html)
def getCardInfo(self, requestStr ,cardName):
stmt = self.db.createStatement();
lvBreak = 0
if requestStr == "!돌파".decode('utf-8') or requestStr == "!키라돌파".decode('utf-8'):
lvBreak = int(cardName.split(" ")[0])
cardName = cardName.split(" ")[1].replace("\017", "").strip()
else:
cardName = cardName.replace("\017", "").strip()
rs = ''
cardUrl = ''
cardNumber = 0
cardCost = ''
cardRarity = ''
try:
rs = stmt.executeQuery("SELECT * FROM "+ self.TABLENAME + " WHERE " + self.CARDNAME + " LIKE '" + cardName +"';")
cardUrl = rs.getString(self.CARDURL)
cardNumber = rs.getString(self.CARDID)
cardName = rs.getString(self.CARDNAME)
cardCost = rs.getString(self.CARDCOST)
cardRarity = rs.getString(self.CARDRARITY)
except:
rs = stmt.executeQuery("SELECT * FROM "+ self.TABLENAME + " WHERE " + self.CARDNAME + " LIKE '%" + cardName +"%';")
cardUrl = rs.getString(self.CARDURL)
cardNumber = rs.getString(self.CARDID)
cardName = rs.getString(self.CARDNAME)
cardCost = rs.getString(self.CARDCOST)
cardRarity = rs.getString(self.CARDRARITY)
Doc = Element
Doc = Jsoup.connect(cardUrl).get()
cardImg1 = Doc.select("div[id=cardG1A1]").select("img").attr("src")
cardImg2 = Doc.select("div[id=cardG1A2]").select("img").attr("src")
cardImg3 = Doc.select("div[id=cardG1A3]").select("img").attr("src")
cardImg4 = Doc.select("div[id=cardG1A4]").select("img").attr("src")
gender = Doc.select("div[class=cardInfo]").select("table").select("tr")[0].select("td")[0].text()
force = Doc.select("div[class=cardInfo]").select("table").select("tr")[0].select("td")[1].text()
growth = Doc.select("div[class=cardInfo]").select("table").select("tr")[1].select("td")[0].text()
illustrator = Doc.select("div[class=cardInfo]").select("table").select("tr")[1].select("td")[1].text()
faction = Doc.select("div[class=cardInfo]").select("table").select("tr")[2].select("td")[0].text()
if Doc.select("div[class=cardInfo]").select("table").select("tr")[2].select("td")[0].text() == "마법".decode('utf-8'):
faction = "마법의 파".decode('utf-8')
elif Doc.select("div[class=cardInfo]").select("table").select("tr")[2].select("td")[0].text() == "검술".decode('utf-8'):
faction = "검술의 성".decode('utf-8')
elif Doc.select("div[class=cardInfo]").select("table").select("tr")[2].select("td")[0].text() == "기교".decode('utf-8'):
faction = "기교의 장".decode('utf-8')
else:
pass
sellPrice = Doc.select("div[class=cardInfo]").select("table").select("tr")[2].select("td")[1].text()
skillInfo = Doc.select("div[class=foldedBox]")[1].select("div[class=content]").text()
cardStory = Doc.select("div[class=foldedBox]")[2].select("div[class=content]").text()
wakeupLV = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[0].select("td")[1].text()
lastLV = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[0].select("td")[2].text()
HP = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[1].select("td")[0].text()
wakeupHP = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[1].select("td")[1].text()
lastHP = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[1].select("td")[2].text()
ATK = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[2].select("td")[0].text()
wakeupATK = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[2].select("td")[1].text()
lastATK = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[2].select("td")[2].text()
CP = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[3].select("td")[0].text()
wakeupCP = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[3].select("td")[1].text()
lastCP = Doc.select("div[class=statTable]")[0].select("tbody").select("tr")[3].select("td")[2].text()
wakeupLVH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[0].select("td")[1].text()
lastLVH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[0].select("td")[2].text()
HPH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[1].select("td")[0].text()
wakeupHPH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[1].select("td")[1].text()
lastHPH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[1].select("td")[2].text()
ATKH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[2].select("td")[0].text()
wakeupATKH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[2].select("td")[1].text()
lastATKH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[2].select("td")[2].text()
CPH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[3].select("td")[0].text()
wakeupCPH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[3].select("td")[1].text()
lastCPH = Doc.select("div[class=statTable]")[1].select("tbody").select("tr")[3].select("td")[2].text()
requestMsg = ""
if requestStr == "!카드".decode('utf-8'):
requestMsg = "\002[" + faction + " " + cardRarity + " " + cardName + "]\002 "
requestMsg = requestMsg + "코스트 : ".decode('utf-8') + cardCost + ", 각성(한계)레벨 : ".decode('utf-8') + wakeupLV + "(" + lastLV + ")"
requestMsg = requestMsg + " \002|\002 "
requestMsg = requestMsg + "체력(M/L) : ".decode('utf-8') + HP + "(" + wakeupHP + "/" + lastHP + ")"
requestMsg = requestMsg + ", "
requestMsg = requestMsg + "공격력(M/L) : ".decode('utf-8') + ATK + "(" + wakeupATK + "/" + lastATK + ")"
requestMsg = requestMsg + ", "
requestMsg = requestMsg + "CP(M/L) : ".decode('utf-8') + CP + "(" + wakeupCP + "/" + lastCP + ")"
requestMsg = requestMsg + " \002|\002 "
requestMsg = requestMsg + skillInfo
elif requestStr == "!키라카드".decode('utf-8') or requestStr == "!키라".decode('utf-8'):
requestMsg = "\002[" + faction + " " + cardRarity + " 키라 ".decode('utf-8') + cardName + "]\002 "
requestMsg = requestMsg + "코스트 : ".decode('utf-8') + cardCost + ", 각성(한계)레벨 : ".decode('utf-8') + wakeupLVH + "(" + lastLVH + ")"
requestMsg = requestMsg + " \002|\002 "
requestMsg = requestMsg + "체력(M/L) : ".decode('utf-8') + HPH + "(" + wakeupHPH + "/" + lastHPH + ")"
requestMsg = requestMsg + ", "
requestMsg = requestMsg + "공격력(M/L) : ".decode('utf-8') + ATKH + "(" + wakeupATKH + "/" + lastATKH + ")"
requestMsg = requestMsg + ", "
requestMsg = requestMsg + "CP(M/L) : ".decode('utf-8') + CPH + "(" + wakeupCPH + "/" + lastCPH + ")"
requestMsg = requestMsg + " \002|\002 "
requestMsg = requestMsg + skillInfo
elif requestStr == "!일러".decode('utf-8'):
requestMsg = "\002[" + faction + " " + cardName + " / " + illustrator +"]\002 "
requestMsg = requestMsg + cardImg1
elif requestStr == "!각성일러".decode('utf-8'):
requestMsg = "\002[" + faction + " 각성 ".decode('utf-8') + cardName + " / " + illustrator +"]\002 "
requestMsg = requestMsg + cardImg2
elif requestStr == "!키라일러".decode('utf-8'):
requestMsg = "\002[" + faction + " 키라 ".decode('utf-8') + cardName + " / " + illustrator +"]\002 "
requestMsg = requestMsg + cardImg3
elif requestStr == "!키라각성일러".decode('utf-8'):
requestMsg = "\002[" + faction + " 키라 각성".decode('utf-8') + cardName + " / " + illustrator +"]\002 "
requestMsg = requestMsg + cardImg4
elif requestStr == "!카드설명".decode('utf-8'):
requestMsg = "\002[" + faction + " " + cardName + " / " + illustrator +"]\002 "
if len(cardStory) > 320:
cardStory = cardStory[:140] + "\n" + cardStory[140:] + "...(생략)".decode('utf-8')
if len(cardStory) > 150:
cardStory = cardStory[:140] + "\n" + cardStory[140:]
requestMsg = requestMsg + " " + cardStory
elif requestStr == "!돌파".decode('utf-8'):
maxBreak = len(Doc.select("div[class=statTable]")[0].select("table").attr("data-wakeup").split("|")) - 2
if lvBreak >= maxBreak:
requestMsg = "\002[" + faction + " " + cardRarity + " " + str(maxBreak) + "회 풀 돌파 ".decode('utf-8') + cardName + "]\002 "
requestMsg = requestMsg + "레벨: ".decode('utf-8') + Doc.select("div[class=statTable]")[0].select("table").attr("data-wakeup").split("|")[maxBreak].split(",")[0] + ", "
HP = Doc.select("div[class=statTable]")[0].select("table").attr("data-wakeup").split("|")[maxBreak].split(",")[1]
requestMsg = requestMsg + "체력: ".decode('utf-8') + HP + ", "
ATK = Doc.select("div[class=statTable]")[0].select("table").attr("data-wakeup").split("|")[maxBreak].split(",")[2]
requestMsg = requestMsg + "공격력: ".decode('utf-8') + ATK + ", "
CP = (int(HP) + int(ATK)) / int(cardCost)
CP = str(CP)
requestMsg = requestMsg + "CP: ".decode('utf-8') + CP
else:
requestMsg = "\002[" + faction + " " + cardRarity + " " + str(lvBreak) + "회 돌파 ".decode('utf-8') + cardName + "]\002 "
requestMsg = requestMsg + "레벨: ".decode('utf-8') + Doc.select("div[class=statTable]")[0].select("table").attr("data-wakeup").split("|")[lvBreak].split(",")[0] + ", "
HP = Doc.select("div[class=statTable]")[0].select("table").attr("data-wakeup").split("|")[lvBreak].split(",")[1]
requestMsg = requestMsg + "체력: ".decode('utf-8') + HP + ", "
ATK = Doc.select("div[class=statTable]")[0].select("table").attr("data-wakeup").split("|")[lvBreak].split(",")[2]
requestMsg = requestMsg + "공격력: ".decode('utf-8') + ATK + ", "
CP = (int(HP) + int(ATK)) / int(cardCost)
CP = str(CP)
requestMsg = requestMsg + "CP: ".decode('utf-8') + CP
elif requestStr == "!키라돌파".decode('utf-8'):
maxBreak = len(Doc.select("div[class=statTable]")[1].select("table").attr("data-wakeup").split("|")) - 2
if lvBreak >= maxBreak:
requestMsg = "\002[" + faction + " " + cardRarity + " " + str(maxBreak) + "회 풀 돌파 키라 ".decode('utf-8') + cardName + "]\002 "
requestMsg = requestMsg + "레벨: ".decode('utf-8') + Doc.select("div[class=statTable]")[1].select("table").attr("data-wakeup").split("|")[maxBreak].split(",")[0] + ", "
HP = Doc.select("div[class=statTable]")[1].select("table").attr("data-wakeup").split("|")[maxBreak].split(",")[1]
requestMsg = requestMsg + "체력: ".decode('utf-8') + HP + ", "
ATK = Doc.select("div[class=statTable]")[1].select("table").attr("data-wakeup").split("|")[maxBreak].split(",")[2]
requestMsg = requestMsg + "공격력: ".decode('utf-8') + ATK + ", "
CP = (int(HP) + int(ATK)) / int(cardCost)
CP = str(CP)
requestMsg = requestMsg + "CP: ".decode('utf-8') + CP
else:
requestMsg = "\002[" + faction + " " + cardRarity + " " + str(lvBreak) + "회 돌파 키라 ".decode('utf-8') + cardName + "]\002 "
requestMsg = requestMsg + "레벨: ".decode('utf-8') + Doc.select("div[class=statTable]")[1].select("table").attr("data-wakeup").split("|")[lvBreak].split(",")[0] + ", "
HP = Doc.select("div[class=statTable]")[1].select("table").attr("data-wakeup").split("|")[lvBreak].split(",")[1]
requestMsg = requestMsg + "체력: ".decode('utf-8') + HP + ", "
ATK = Doc.select("div[class=statTable]")[1].select("table").attr("data-wakeup").split("|")[lvBreak].split(",")[2]
requestMsg = requestMsg + "공격력: ".decode('utf-8') + ATK + ", "
CP = (int(HP) + int(ATK)) / int(cardCost)
CP = str(CP)
requestMsg = requestMsg + "CP: ".decode('utf-8') + CP
self.requestAdd(requestMsg)
def getMonsterInfoUrl(self, monsterName):
monsterName = monsterName.replace(" ", "%20")
url = "http://www.thisisgame.com/pad/info/monster/list.php?sf=name&sw=" + monsterName
Doc = Element
Doc = Jsoup.connect(url).get()
return Doc.select("div[class=pad-table2 pad-info-light]").select("tr")[1].select("td")[0].select("a").attr("href")
def clean_question(html):
"""Removes code tag and its content. Subsequently, it removes html tags"""
doc = Jsoup.parse(html)
doc.select("code").empty()
return doc.text()
node["node"].replaceWith(new_div)
break
if len(argv) < 4:
infile = "/Users/mac/Downloads/im"
outfile = "/Users/mac/Downloads/dialogues.html"
textfile = "/Users/mac/Downloads/dialogues.txt"
else:
infile = argv[1]
outfile = argv[2]
textfile = argv[3]
with iopen(outfile, "w", encoding="utf-8", errors="ignore") as output:
input = File(infile)
soup = Jsoup.parse(input, "UTF-8", "")
# First, create a new document
new_doc = Jsoup.parse("<body></body>")
new_doc.updateMetaCharsetElement(True)
new_doc.charset(Charset.forName("UTF-8"))
new_body = new_doc.select("body").first()
for element in soup.select("*"):
if (element.tag().toString() == "ul" and element.className()
== "ui_clean_list im-mess-stack--mess _im_stack_messages") or (
element.tag().toString() == "div"
and element.className() == "im-mess-stack--pname"):
new_body.appendChild(element)
# Then remove empty tags from it and transform the labels
