flywithu/epg2xml
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

정지 시간 추가

Browse Source
This commit is contained in:
wonipapa 2016-11-10 18:26:57 +09:00 committed by GitHub
parent d83851b4fa
commit 431d23c7d9
1 changed files with 56 additions and 60 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

98
epg2xml.py
View File

@ -16,7 +16,7 @@ import argparse
reload(sys) reload(sys)
sys.setdefaultencoding('utf-8') sys.setdefaultencoding('utf-8')
__version__ = '1.0.7' __version__ = '1.0.8'
# Set My Configuration # Set My Configuration
default_icon_url = '' # TV channel icon url (ex : http://www.example.com/Channels) default_icon_url = '' # TV channel icon url (ex : http://www.example.com/Channels)
@ -97,6 +97,7 @@ def GetEPGFromEPG(ChannelInfo):
ChannelId = ChannelInfo[0] ChannelId = ChannelInfo[0]
ChannelName = ChannelInfo[1] ChannelName = ChannelInfo[1]
ServiceId = ChannelInfo[3] ServiceId = ChannelInfo[3]
epginfo = []
url = 'http://www.epg.co.kr/epg-cgi/extern/cnm_guide_type_v070530.cgi' url = 'http://www.epg.co.kr/epg-cgi/extern/cnm_guide_type_v070530.cgi'
contenturl = 'http://www.epg.co.kr/epg-cgi/guide_schedule_content.cgi' contenturl = 'http://www.epg.co.kr/epg-cgi/guide_schedule_content.cgi'
for k in range(period): for k in range(period):
@ -113,7 +114,6 @@ def GetEPGFromEPG(ChannelInfo):
for i in range(1,4): for i in range(1,4):
thisday = day thisday = day
pid = ''
row = table[i].find_all('td', {'colspan':'2'}) row = table[i].find_all('td', {'colspan':'2'})
for j, cell in enumerate(row): for j, cell in enumerate(row):
hour = int(cell.text.strip().strip('시')) hour = int(cell.text.strip().strip('시'))
@ -124,50 +124,37 @@ def GetEPGFromEPG(ChannelInfo):
hour = 'AM ' + str(hour) hour = 'AM ' + str(hour)
thisday = day + datetime.timedelta(days=1) thisday = day + datetime.timedelta(days=1)
for celldata in cell.parent.find_all('tr'): for celldata in cell.parent.find_all('tr'):
matches = re.match("<tr>.*\[(.*)\]<\/td>\s.*ViewContent\('(.*)'\)\">(.*?)\s*(&lt;(.*)&gt;)?\s*(\(재\))?\s*(\(([\d,]+)회\))?(<img.*)?<\/a> <\/td><\/tr>", str(celldata)) pattern = "<tr>.*\[(.*)\]<\/td>\s.*\">(.*?)\s*(&lt;(.*)&gt;)?\s*(\(재\))?\s*(\(([\d,]+)회\))?(<img.*?)?(<\/a>)?\s*<\/td><\/tr>"
matches = re.match(pattern, str(celldata))
if not (matches is None): if not (matches is None):
minute = matches.group(1) if matches.group(1) else '' minute = matches.group(1) if matches.group(1) else ''
startTime = str(thisday) + ' ' + hour + ':' + minute startTime = str(thisday) + ' ' + hour + ':' + minute
startTime = datetime.datetime.strptime(startTime, '%Y-%m-%d %p %I:%M') startTime = datetime.datetime.strptime(startTime, '%Y-%m-%d %p %I:%M')
startTime = startTime.strftime('%Y%m%d%H%M%S') startTime = startTime.strftime('%Y%m%d%H%M%S')
endTime = '' image = matches.group(8) if matches.group(8) else ''
pid = matches.group(2) if matches.group(2) else '' grade = re.match('.*schedule_([\d,]+)?.*',image)
programName = matches.group(3) if matches.group(3) else '' if not (grade is None): rating = int(grade.group(1))
subprogramName = matches.group(5) if matches.group(5) else '' else : rating = 0
#programName, startTime, rating, subprogramName, rebroadcast, episode
epginfo.append([matches.group(2), startTime, rating, matches.group(4), matches.group(5), matches.group(7)])
for epg1, epg2 in zip(epginfo, epginfo[1:]):
programName = epg1[0] if epg1[0] else ''
subprogramName = epg1[3] if epg1[3] else ''
startTime = epg1[1] if epg1[1] else ''
endTime = epg2[1] if epg2[1] else ''
desc = '' desc = ''
actors = '' actors = ''
producers = '' producers = ''
category = '' category = ''
rebroadcast = True if matches.group(6) else False rebroadcast = True if epg1[4] else False
episode = matches.group(8) if matches.group(8) else '' episode = epg1[5] if epg1[5] else ''
image = matches.group(9) if matches.group(9) else '' rating = int(epg1[2]) if epg1[2] else 0
grade = re.match('.*schedule_([\d,]+)?.*',image)
if not (grade is None):
rating = int(grade.group(1))
else :
rating = 0
if(i == 3 and len(row) - 1 == j and pid) :
params = {'pid':pid}
try:
response = requests.get(contenturl, params=params, headers=ua)
response.raise_for_status()
html_data = response.content
data = unicode(html_data, 'euc-kr', 'ignore').encode('utf-8', 'ignore')
strainer = SoupStrainer('table', {'border':'0', 'cellpadding':'3'})
soup = BeautifulSoup(data, 'lxml', parse_only=strainer, from_encoding='utf-8')
td = soup.select('tr:nth-of-type(3) > td:nth-of-type(3)')
endTime = td[0].text.split('~')[1].replace('시',':').replace('분','').replace(': ', ':').strip()
if(endTime.startswith('0')): endTime = endTime.replace('0:','12:')
endTime = str(thisday) + ' ' + 'AM ' + endTime
if(endTime.endswith(':')) : endTime = endTime + '00'
endTime = datetime.datetime.strptime(endTime, '%Y-%m-%d %p %I:%M')
endTime = endTime.strftime('%Y%m%d%H%M%S')
except requests.exceptions.HTTPError:
printError(ChannelName + HTTP_ERROR)
programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating} programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating}
writeProgram(programdata) writeProgram(programdata)
except requests.exceptions.HTTPError: except requests.exceptions.HTTPError:
printError(ChannelName + HTTP_ERROR) if(debug): printError(ChannelName + HTTP_ERROR)
else: pass
# Get EPG data from KT # Get EPG data from KT
def GetEPGFromKT(ChannelInfo): def GetEPGFromKT(ChannelInfo):
@ -215,9 +202,12 @@ def GetEPGFromKT(ChannelInfo):
if not(matches is None): rating = int(matches.group()) if not(matches is None): rating = int(matches.group())
programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating} programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating}
writeProgram(programdata) writeProgram(programdata)
else: printError(ChannelName + CONTENT_ERROR) else:
if(debug): printError(ChannelName + CONTENT_ERROR)
else: pass
except requests.exceptions.HTTPError: except requests.exceptions.HTTPError:
printError(ChannelName + HTTP_ERROR) if(debug): printError(ChannelName + HTTP_ERROR)
else: pass
# Get EPG data from LG # Get EPG data from LG
def GetEPGFromLG(ChannelInfo): def GetEPGFromLG(ChannelInfo):
@ -249,17 +239,12 @@ def GetEPGFromLG(ChannelInfo):
matches = re.match('^(.*?)(\(([\d,]+)회\))?$', epg1[0].decode('string_escape')) matches = re.match('^(.*?)(\(([\d,]+)회\))?$', epg1[0].decode('string_escape'))
if not (matches is None): if not (matches is None):
programName = matches.group(1) if matches.group(1) else '' programName = matches.group(1) if matches.group(1) else ''
episode = int(matches.group(3)) if matches.group(3) else '' episode = matches.group(3) if matches.group(3) else ''
startTime = datetime.datetime.strptime(epg1[1], '%Y-%m-%d %H:%M') startTime = datetime.datetime.strptime(epg1[1], '%Y-%m-%d %H:%M')
startTime = startTime.strftime('%Y%m%d%H%M%S') startTime = startTime.strftime('%Y%m%d%H%M%S')
endTime = datetime.datetime.strptime(epg2[1], '%Y-%m-%d %H:%M') endTime = datetime.datetime.strptime(epg2[1], '%Y-%m-%d %H:%M')
endTime = endTime.strftime('%Y%m%d%H%M%S') endTime = endTime.strftime('%Y%m%d%H%M%S')
category = epg1[2] category = epg1[2]
if verbose=='y' :
desc = ''
actors = epgdata[6]
producers = epgdata[7]
else:
desc = '' desc = ''
actors = '' actors = ''
producers = '' producers = ''
@ -269,9 +254,12 @@ def GetEPGFromLG(ChannelInfo):
if not(matches is None): rating = int(matches.group(1)) if not(matches is None): rating = int(matches.group(1))
programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating} programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating}
writeProgram(programdata) writeProgram(programdata)
else: printError(ChannelName + CONTENT_ERROR) else:
if(debug): printError(ChannelName + CONTENT_ERROR)
else: pass
except requests.exceptions.HTTPError: except requests.exceptions.HTTPError:
printError(ChannelName + HTTP_ERROR) if(debug): printError(ChannelName + HTTP_ERROR)
else: pass
# Get EPG data from SK # Get EPG data from SK
def GetEPGFromSK(ChannelInfo): def GetEPGFromSK(ChannelInfo):
@ -321,9 +309,11 @@ def GetEPGFromSK(ChannelInfo):
programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating} programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating}
writeProgram(programdata) writeProgram(programdata)
except ValueError: except ValueError:
printError(ChannelName + CONTENT_ERROR) if(debug): printError(ChannelName + CONTENT_ERROR)
else: pass
except requests.exceptions.HTTPError: except requests.exceptions.HTTPError:
printError(ChannelName + HTTP_ERROR) if(debug): printError(ChannelName + HTTP_ERROR)
else: pass
# Get EPG data from SKY # Get EPG data from SKY
def GetEPGFromSKY(ChannelInfo): def GetEPGFromSKY(ChannelInfo):
@ -341,7 +331,8 @@ def GetEPGFromSKY(ChannelInfo):
try: try:
data = json.loads(json_data, encoding='utf-8') data = json.loads(json_data, encoding='utf-8')
if (len(data['scheduleListIn']) == 0) : if (len(data['scheduleListIn']) == 0) :
printError(ChannelName + CONTENT_ERROR) if(debug): printError(ChannelName + CONTENT_ERROR)
else: pass
else : else :
programs = data['scheduleListIn'] programs = data['scheduleListIn']
for program in {v['starttime']:v for v in programs}.values(): for program in {v['starttime']:v for v in programs}.values():
@ -370,9 +361,11 @@ def GetEPGFromSKY(ChannelInfo):
programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating} programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating}
writeProgram(programdata) writeProgram(programdata)
except ValueError: except ValueError:
printError(ChannelName + CONTENT_ERROR) if(debug): printError(ChannelName + CONTENT_ERROR)
else: pass
except requests.exceptions.HTTPError: except requests.exceptions.HTTPError:
printError(ChannelName + HTTP_ERROR) if(debug): printError(ChannelName + HTTP_ERROR)
else: pass
# Get EPG data from Naver # Get EPG data from Naver
def GetEPGFromNaver(ChannelInfo): def GetEPGFromNaver(ChannelInfo):
@ -417,9 +410,11 @@ def GetEPGFromNaver(ChannelInfo):
programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating} programdata = {'channelId':ChannelId, 'startTime':startTime, 'endTime':endTime, 'programName':programName, 'subprogramName':subprogramName, 'desc':desc, 'actors':actors, 'producers':producers, 'category':category, 'episode':episode, 'rebroadcast':rebroadcast, 'rating':rating}
writeProgram(programdata) writeProgram(programdata)
except ValueError: except ValueError:
printError(ChannelName + CONTENT_ERROR) if(debug): printError(ChannelName + CONTENT_ERROR)
else: pass
except requests.exceptions.HTTPError: except requests.exceptions.HTTPError:
printError(ChannelName + HTTP_ERROR) if(debug): printError(ChannelName + HTTP_ERROR)
else: pass
# Write Program # Write Program
@ -536,3 +531,4 @@ elif args.socket:
sys.stdout = sockfile sys.stdout = sockfile
getEpg() getEpg()