5 Commits

4 changed files with 316 additions and 230 deletions

View File

@@ -28,7 +28,7 @@ CACHE_EXPIRY_MINS=10
LOG_HTTP_RESPONSES = False
FEED_REJECT_IF_OLDER_THAN_DAYS = 7
FEED_REJECT_IF_OLDER_THAN_DAYS = 60
class PathHelper:
pathChar="/"

View File

@@ -30,6 +30,8 @@ class NewsFeed:
return False
def getItemsInAmericasNewsRoomFeed(self,url):
response = None
try:
now=datetime.now()
cachePathFileName=PathHelper.makePathFileName(VIDEODB_AMERICAS_NEWSROOM_FILENAME,self.pathDb)
if self.isFeedCacheAvailable(cachePathFileName,CACHE_EXPIRY_MINS):
@@ -39,10 +41,9 @@ class NewsFeed:
sections=Sections()
videos = {}
httpNetRequest=HttpNetRequest()
response=httpNetRequest=httpNetRequest.getHttpNetRequest(url)
response=httpNetRequest.getHttpNetRequest(url)
status=response.status_code
searchIndex=0
response.close()
if status!=200:
return None
if LOG_HTTP_RESPONSES:
@@ -57,8 +58,13 @@ class NewsFeed:
videoList=sorted(videoList, key=lambda x:x.getFeedTime(),reverse=False)
self.writeFeedCache(cachePathFileName,videoList)
return (videoList)
finally:
if None!= response:
response.close()
def getItemsInOutnumberedFeed(self,url):
response = None
try:
now=datetime.now()
cachePathFileName=PathHelper.makePathFileName(VIDEODB_OUTNUMBERED_FILENAME,self.pathDb)
if self.isFeedCacheAvailable(cachePathFileName,CACHE_EXPIRY_MINS):
@@ -68,10 +74,9 @@ class NewsFeed:
sections=Sections()
videos = {}
httpNetRequest=HttpNetRequest()
response=httpNetRequest=httpNetRequest.getHttpNetRequest(url)
response=httpNetRequest.getHttpNetRequest(url)
status=response.status_code
searchIndex=0
response.close()
if status!=200:
return None
if LOG_HTTP_RESPONSES:
@@ -86,8 +91,13 @@ class NewsFeed:
videoList=sorted(videoList, key=lambda x:x.getFeedTime(),reverse=True)
self.writeFeedCache(cachePathFileName,videoList)
return (videoList)
finally:
if None!=response:
response.close()
def getItemsInFeed(self,url):
response = None
try:
now=datetime.now()
cachePathFileName=PathHelper.makePathFileName(VIDEODB_FILENAME,self.pathDb)
if self.isFeedCacheAvailable(cachePathFileName,CACHE_EXPIRY_MINS):
@@ -99,10 +109,9 @@ class NewsFeed:
videos = {}
httpNetRequest=HttpNetRequest()
self.writeLog(f"Loading videos from {url}")
response=httpNetRequest=httpNetRequest.getHttpNetRequest(url)
response=httpNetRequest.getHttpNetRequest(url)
status=response.status_code
searchIndex=0
response.close()
if status!=200:
return None
if LOG_HTTP_RESPONSES:
@@ -113,11 +122,13 @@ class NewsFeed:
if video is not None and not (video.description in videos):
videos[video.description]=video
video.setFeedTime(DateTimeHelper.applyRelativeTime(now,video.feedTimeOffset))
# videoList=list(videos.values())
videoList=self.filterFeedMaxDays(list(videos.values()),FEED_REJECT_IF_OLDER_THAN_DAYS)
videoList=sorted(videoList, key=lambda x:x.getFeedTime(),reverse=True)
self.writeFeedCache(cachePathFileName,videoList)
return (videoList)
finally:
if None!=response:
response.close()
def filterFeedMaxDays(self, videos, days):
now = datetime.now()
@@ -127,13 +138,15 @@ class NewsFeed:
if delta.days <= days:
message = f"INCL. days={delta.days},feed time={video.getFeedTime()} feed time offset (strPublication)=:'{video.feedTimeOffset}', description={video.description}"
self.writeLog(message)
filteredList.insert(0,video)
filteredList.append(video)
else:
message = f"EXCL. days={delta.days},feed time={video.getFeedTime()} feed time offset (strPublication)=:'{video.feedTimeOffset}', description={video.description}"
self.writeLog(message)
return filteredList
def getUSItemsInFeed(self,url):
response = None
try:
now=datetime.now()
cachePathFileName=PathHelper.makePathFileName(VIDEODB_US_FILENAME,self.pathDb)
if self.isFeedCacheAvailable(cachePathFileName,CACHE_EXPIRY_MINS):
@@ -146,7 +159,6 @@ class NewsFeed:
response=httpNetRequest.getHttpNetRequest(url)
status=response.status_code
searchIndex=0
response.close()
if status!=200:
return None
if LOG_HTTP_RESPONSES:
@@ -156,9 +168,9 @@ class NewsFeed:
videoId, searchIndex = sections.getVideoIdInSection(response.text,"article",searchIndex)
if videoId is None:
continue
url='https://video.foxnews.com/v/'+videoId
videoUrl='https://video.foxnews.com/v/'+videoId
httpNetRequest=HttpNetRequest()
innerResponse=httpNetRequest.getHttpNetRequest(url)
innerResponse=httpNetRequest.getHttpNetRequest(videoUrl)
status=innerResponse.status_code
innerResponse.close()
if status!=200:
@@ -171,8 +183,13 @@ class NewsFeed:
videoList=sorted(videoList, key=lambda x:x.getFeedTime(),reverse=True)
self.writeFeedCache(cachePathFileName,videoList)
return (videoList)
finally:
if None!=response:
response.close()
def getExclusiveItemsInFeed(self,url):
response = None
try:
now=datetime.now()
cachePathFileName=PathHelper.makePathFileName(VIDEODB_EXCLUSIVE_FILENAME,self.pathDb)
if self.isFeedCacheAvailable(cachePathFileName,CACHE_EXPIRY_MINS):
@@ -185,19 +202,18 @@ class NewsFeed:
response=httpNetRequest.getHttpNetRequest(url)
status=response.status_code
searchIndex=0
response.close()
if status!=200:
return None
if LOG_HTTP_RESPONSES:
self.writeLog(url)
self.writeLog(response.Text)
self.writeLog(response.text)
while -1!= searchIndex:
videoId, searchIndex = sections.getVideoIdInSection(response.text,"article",searchIndex)
if videoId is None:
continue
url='https://video.foxnews.com/v/'+videoId
videoUrl='https://video.foxnews.com/v/'+videoId
httpNetRequest=HttpNetRequest()
innerResponse=httpNetRequest.getHttpNetRequest(url)
innerResponse=httpNetRequest.getHttpNetRequest(videoUrl)
status=innerResponse.status_code
innerResponse.close()
if status!=200:
@@ -210,6 +226,9 @@ class NewsFeed:
videoList=sorted(videoList, key=lambda x:x.getFeedTime(),reverse=True)
self.writeFeedCache(cachePathFileName,videoList)
return (videoList)
finally:
if None!=response:
response.close()
def getItemsInArchiveFeed(self,url,archiveDbFileName):
cachePathFileName=PathHelper.makePathFileName(archiveDbFileName,self.pathDb)
@@ -221,11 +240,11 @@ class NewsFeed:
def readFeedCache(self,pathFileName):
try:
videos=[]
# 'with' will automatically close the stream
with open(pathFileName,"r",encoding='utf-8') as inputStream:
for line in inputStream:
video=Video.fromString(line)
videos.append(video)
inputStream.close()
return(videos)
except:
self.writeLog(traceback.format_exc())
@@ -236,7 +255,7 @@ class NewsFeed:
with open(pathFileName,"w",encoding='utf-8') as outputStream:
for video in videos:
outputStream.write(video.toString()+"\n")
outputStream.close()
# 'with' will automatically close the stream
return(videos)
except:
self.writeLog(traceback.format_exc())
@@ -246,24 +265,25 @@ class NewsFeed:
try:
self.writeLog('Inspecting cache file {pathFileName}'.format(pathFileName=pathFileName))
if not os.path.isfile(pathFileName):
return(False)
return False
modifiedTime = os.path.getmtime(pathFileName)
convertTime = time.localtime(modifiedTime)
formatTime = time.strftime('%d%m%Y %H:%M:%S', convertTime)
fileDateTime=DateTimeHelper.strptime(formatTime,'%d%m%Y %H:%M:%S')
fileDateTime = time.strptime(formatTime, '%d%m%Y %H:%M:%S')
currentTime = datetime.now()
timedelta=currentTime-fileDateTime
hours, hremainder = divmod(timedelta.seconds,3600)
minutes, mremainder = divmod(timedelta.seconds,60)
elapsed = currentTime - datetime(*(fileDateTime[0:6]))
totalSeconds = int(elapsed.total_seconds())
hours, remainder = divmod(totalSeconds, 3600)
minutes, _ = divmod(remainder, 60)
self.writeLog('file is = "{age}" hours old'.format(age=hours))
self.writeLog('file is = "{age}" minutes old'.format(age=minutes))
if hours > 1 or minutes > expireMinutes:
self.archiveFile(pathFileName)
return(False)
return (True)
return False
return True
except:
self.writeLog(traceback.format_exc());
return(False)
self.writeLog(traceback.format_exc())
return False
def archiveFile(self, pathFileName):
if not os.path.isfile(pathFileName):
@@ -317,25 +337,35 @@ class Sections:
if "tokenvod" in previewUrl:
return video, searchIndex
indexDescription=strContainingString.index("alt=\"")
# Handle video description
indexDescription=strContainingString.find("alt=\"")
if -1 == indexDescription:
return video, searchIndex
description=strContainingString[indexDescription:]
description=self.betweenString(description,'"','"')
description=self.removeHtml(description)
description=description.replace("- Fox News","")
if "vod.foxbusiness" in description:
return video, searchIndex
indexDuration=strContainingString.index("<div class=\"duration\">")
# Handle video duration
indexDuration=strContainingString.find("<div class=\"duration\">")
if -1 != indexDuration:
strDuration=strContainingString[indexDuration:]
strDuration=self.betweenString(strDuration,">","<")
description=description+" - "+strDuration
indexPublication=strContainingString.index("<div class=\"pub-date\">")
# Handle video publication
strPublication = ""
indexPublication=strContainingString.find("<div class=\"pub-date\">")
if -1 != indexPublication:
strPublication=strContainingString[indexPublication:]
strPublication=self.betweenString(strPublication,"<time>","</time>")
description=description+" ("+strPublication+")"
# Handle the icon
icon=None
indexIcon=strContainingString.index("srcset=")
indexIcon=strContainingString.find("srcset=")
if -1 != indexIcon:
icon=strContainingString[indexIcon:]
icon=self.betweenString(icon,"\"","\"")
@@ -422,10 +452,12 @@ class Sections:
for code in codes:
strItem=strItem.replace(code,"'")
strItem=strItem.replace("&amp;","&")
strItem=strItem.replace("&#x2018;","'")
strItem=strItem.replace("&#x2019;","'")
strItem=strItem.replace("&#x2018;","")
strItem=strItem.replace("&#x2019;","")
strItem=strItem.replace("&#x2014;","-")
strItem=strItem.replace("&#39;","'")
strItem=strItem.replace("???","'")
strItem=strItem.replace("&quot;","\"")
return strItem
def pad(str,filler,length):
@@ -447,6 +479,19 @@ def parseDuration(strDuration):
# DON'T LEAVE ANYTHING OPEN BELOW THIS LINE BECAUSE THIS FILE IS IMPORTED BY OTHER MODULES AND ANY CODE NOT IN A CLASS WILL BE RUN
# strdate = "January 1, 2026"
# if DateTimeHelper.canstrptimeex(strdate):
# theDate = DateTimeHelper.strptimeex(strdate)
# if(not isinstance(theDate,datetime)):
# raise Exception('Invalid type for parameter')
# feedTimeOffset = "January 13, 2025"
# currentTime = datetime.now()
# for i in range(1,100):
# relativeTime = DateTimeHelper.applyRelativeTime(currentTime,feedTimeOffset)
# print(relativeTime)
#print(FOX_NEWS_URL)
# pathFileName='/home/pi/.kodi/addons/plugin.video.fox.news/resources/lib/videodb.txt'

View File

@@ -40,25 +40,30 @@ class StringHelper:
if strBegin is None:
index = 0
else:
index = strItem.index(strBegin)
if -1==index:
return None
str=None
if strBegin is not None:
str=strItem[index+len(strBegin):]
try:
if strBegin.startswith("<") and strBegin.endswith(">"):
tag_name = strBegin[1:-1] # e.g. "time"
index = strItem.index("<" + tag_name)
index = strItem.index(">", index) + 1
else:
str=strItem
index = strItem.index(strBegin) + len(strBegin)
except ValueError:
return None
if index == -1:
return None
str = strItem[index:] if strBegin is not None else strItem
if strEnd is None:
return str
try:
index = str.index(strEnd)
if -1==index :
except ValueError:
return None
sb = ""
for strIndex in range(0, len(str) - 1):
if index == strIndex:
break
sb = sb + str[strIndex]
return (sb)
return sb
class HttpNetRequest:
def __init__(self):
@@ -117,66 +122,97 @@ class DateTimeHelper:
def getCurrentDateTime():
return datetime.now()
# January 1, 2026
@staticmethod
def strptime(theTime,theFormat):
try:
return datetime.strptime(theTime,theFormat)
except:
return datetime(*(time.strptime(theTime,theFormat)[0:6]))
def strptime(date_string):
month_map = {
'January': 1, 'February': 2, 'March': 3, 'April': 4,
'May': 5, 'June': 6, 'July': 7, 'August': 8,
'September': 9, 'October': 10, 'November': 11, 'December': 12
}
date_string = date_string.replace(',', '')
parts = date_string.split()
if len(parts) == 3:
month_str, day_str, year_str = parts
month = month_map.get(month_str)
day = int(day_str)
year = int(year_str)
if month is not None:
return datetime(year, month, day)
else:
raise ValueError("Invalid month name in date string")
else:
raise ValueError("Date string format is incorrect")
# January 1, 2026
@staticmethod
def canstrptime(theTime,theFormat):
try:
datetime.strptime(theTime,theFormat)
return True
except:
def canstrptime(date_string):
month_map = {
'January': 1, 'February': 2, 'March': 3, 'April': 4,
'May': 5, 'June': 6, 'July': 7, 'August': 8,
'September': 9, 'October': 10, 'November': 11, 'December': 12
}
date_string = date_string.replace(',', '')
parts = date_string.split()
if len(parts) != 3:
return False
month_str, day_str, year_str = parts
month = month_map.get(month_str)
if month is None:
return False
day = int(day_str)
year = int(year_str)
return True
# returns a datetime
@staticmethod
def applyRelativeTime(sometime,relativetime):
relativeTimeResult = sometime
if(not isinstance(sometime,datetime)):
raise Exception('Invalid type for parameter')
if(not isinstance(relativetime,str)):
raise Exception('Invalid type for parameter')
if DateTimeHelper.canstrptime(relativetime,'%B %d, %Y'):
sometime = DateTimeHelper.strptime(relativetime,'%B %d, %Y')
return sometime
if DateTimeHelper.canstrptime(relativetime):
relativeTimeResult = DateTimeHelper.strptime(relativetime)
return relativeTimeResult
if relativetime=='just now':
return sometime
return relativeTimeResult
if relativetime=='just in':
return sometime
return relativeTimeResult
relativetimesplit=relativetime.split()
if len(relativetimesplit)==2:
year=datetime.now().year
relativetimex=relativetime+', '+str(year)
relativeDate = DateTimeHelper.strptime(relativetimex, '%B %d, %Y')
relativeDate = DateTimeHelper.strptime(relativetimex)
if(relativeDate>datetime.now()):
year=datetime.now().year-1
relativetimex=relativetime+', '+str(year)
relativeDate=DateTimeHelper.strptime(relativetimex,'%B %d, %Y')
relativeDate=DateTimeHelper.strptime(relativetimex)
days=sometime-relativeDate
sometime=sometime-days
relativeTimeResult=sometime-days
elif relativetimesplit[1]=='hour' or relativetimesplit[1]=='hours':
hours=int(relativetimesplit[0])
sometime=sometime-timedelta(hours=hours)
relativeTimeResult=sometime-timedelta(hours=hours)
elif relativetimesplit[1]=='day' or relativetimesplit[1]=='days':
days=int(relativetimesplit[0])
sometime=sometime-timedelta(days=days)
relativeTimeResult=sometime-timedelta(days=days)
elif relativetimesplit[1]=='minute' or relativetimesplit[1]=='minutes':
minutes=int(relativetimesplit[0])
sometime=sometime-timedelta(minutes=minutes)
relativeTimeResult=sometime-timedelta(minutes=minutes)
elif len(relativetimesplit)==3: # '16 mins ago' '2 hours ago'
if relativetimesplit[1]=='mins':
minutes=int(relativetimesplit[0])
sometime=sometime-timedelta(minutes=minutes)
relativeTimeResult=sometime-timedelta(minutes=minutes)
elif relativetimesplit[1]=='hours':
hours=int(relativetimesplit[0])
sometime=sometime-timedelta(hours=hours)
relativeTimeResult=sometime-timedelta(hours=hours)
elif relativetimesplit[1]=='day' or relativetimesplit[1]=='days':
days=int(relativetimesplit[0])
sometime=sometime-timedelta(days=days)
return sometime
relativeTimeResult=sometime-timedelta(days=days)
return relativeTimeResult
class DateTime:
def __init__(self):

View File

@@ -68,7 +68,12 @@ class Video:
description=splits[0].strip()
url=splits[1].strip()
icon=splits[2].strip()
timestamp=DateTime(splits[3].strip())
datePart = splits[3].strip()
timestamp = DateTime.getCurrentTime()
try :
timestamp=DateTime(datePart)
except Exception as exception:
print(f"Encountered invalid date '{datePart}'")
return(Video(description,url,icon,timestamp))
@staticmethod