def parse1(text):
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
nose_kelim = nosekelim.open_file()
fixed = nosekelim.parse(nose_kelim)
links_netanel = []
netanel = 0
rosh = []
chapters = re.split(ur'@00', text)
for chapter_num, chapter in enumerate(chapters):
print chapter_num, chapter[0:10]
if len(chapter)<=1:
pass
else:
perek = []
a = re.split(ur'@22([^@]*)', chapter)
for seif, cont in zip(a[1::2], a[2::2]):
si = []
korban =[]
print seif
if ur'[*]' in seif:
print "hello1"
if ur'[*]' in seif and (os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet))) and netanel <= len(fixed):
print "hello", seif, netanel, len(fixed)
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)):
commentator = "Korban Netanel"
if os.path.isfile('source/PilPula_Charifta_on_{}.txt'.format(masechet)):
commentator = "Pilpula Charifta"
korban.append(fixed[netanel])
roash = "Rosh on %s." % masechet +str(len(rosh)+2) + "." + str(len(perek)+1) + ".1"
netanelink = commentator + " on " + masechet +"."+ str(len(links_netanel)+1) + ".1"
print roash, netanelink
links.append(link(netanelink, roash))
netanel += 1
content = re.split('@66', cont)
seif = re.sub(ur'[^א-ת]',"", seif)
seif = hebrew.heb_string_to_int(seif.strip())
for num, co in enumerate(content):
a = re.findall('\[\*\]', co)
for b in a:
print b, seif
if (os.path.isfile('source/Korban_netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet))) and netanel < len(fixed):
if os.path.isfile('source/Korban_netanel_on_{}.txt'.format(masechet)):
commentator = "Korban Netanel "
if os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
commentator = "Pilpula Charifta "
korban.append(fixed[netanel])
roash = "Rosh on %s." % masechet + str(len(rosh)+2) + "." + str(len(perek)+1) + "." + str(num+1)
netanelink = commentator + "on " + masechet + "." + str(len(links_netanel)+1)+ "."+ str(len(korban))
print roash, netanelink
links.append(link(netanelink, roash))
netanel +=1
si.append(co)
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
links_netanel.append(korban)
perek.append(si)
rosh.append(perek)
def parse1(text):
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
nose_kelim = nosekelim.open_file()
fixed = nosekelim.parse(nose_kelim)
links_netanel = []
netanel = 0
rosh = []
chapters = re.split(ur'@00', text)
for chapter_num, chapter in enumerate(chapters):
print chapter_num, chapter[0:10]
if len(chapter)<=1:
pass
else:
perek = []
a = re.split(ur'@22([^@]*)', chapter)
for seif, cont in zip(a[1::2], a[2::2]):
si = []
korban =[]
print seif
if ur'[*]' in seif:
print "hello1"
if ur'[*]' in seif and (os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet))) and netanel <= len(fixed):
print "hello", seif, netanel, len(fixed)
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)):
commentator = "Korban Netanel"
if os.path.isfile('source/PilPula_Charifta_on_{}.txt'.format(masechet)):
commentator = "Pilpula Charifta"
korban.append(fixed[netanel])
roash = "Rosh on %s." % masechet +str(len(rosh)+2) + "." + str(len(perek)+1) + ".1"
netanelink = commentator + " on " + masechet +"."+ str(len(links_netanel)+1) + ".1"
print roash, netanelink
links.append(link(netanelink, roash))
netanel += 1
content = re.split('@66', cont)
seif = re.sub(ur'[^א-ת]',"", seif)
seif = hebrew.heb_string_to_int(seif.strip())
for num, co in enumerate(content):
a = re.findall('\[\*\]', co)
for b in a:
print b, seif
if (os.path.isfile('source/Korban_netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet))) and netanel < len(fixed):
if os.path.isfile('source/Korban_netanel_on_{}.txt'.format(masechet)):
commentator = "Korban Netanel "
if os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
commentator = "Pilpula Charifta "
korban.append(fixed[netanel])
roash = "Rosh on %s." % masechet + str(len(rosh)+2) + "." + str(len(perek)+1) + "." + str(num+1)
netanelink = commentator + "on " + masechet + "." + str(len(links_netanel)+1)+ "."+ str(len(korban))
print roash, netanelink
links.append(link(netanelink, roash))
netanel +=1
si.append(co)
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
links_netanel.append(korban)
perek.append(si)
rosh.append(perek)
def parse(text):
if os.path.isfile("../source/Korban_Netanel_on_{}.txt".format(masechet)) or os.path.isfile(
"../source/Pilpula_Charifta_on_{}.txt".format(masechet)
):
# print "has korban netanel 2"
nose_kelim = nosekelim.open_file()
fixed = nosekelim.parse(nose_kelim)
links_netanel = []
netanel = 0
rosh = []
a = re.split(ur"@22([^@]*)", text)
for seif, cont in zip(a[1::2], a[2::2]):
si = []
korban = []
if ur"[*]" in seif and (
os.path.isfile(
"../source/Korban_Netanel_on_{}.txt".format(masechet)
or os.path.isfile("../source/PilPula_Charifta_on_{}.txt".format(masechet))
)
and netanel <= len(fixed)
):
if os.path.isfile("../source/Korban_Netanel_on_{}.txt".format(masechet)):
commentator = "Korban Netanel on "
if os.path.isfile("../source/PilPula_Charifta_on_{}.txt".format(masechet)):
commentator = "Pilpula Charifta on "
korban.append(fixed[netanel])
# print len(links_netanel)
roash = "Rosh on %s." % masechet + str(len(links_netanel) + 1) + ".1"
netanelink = commentator + masechet + "." + str(len(links_netanel) + 1) + ".1"
links.append(link(netanelink, roash))
netanel += 1
# print "netanel one seif", seif, netanel
# print fixed[netanel]
content = re.split("@66", cont)
seif = re.sub(ur"[^א-ת]", "", seif)
seif = hebrew.heb_string_to_int(seif.strip())
for num, co in enumerate(content):
if ur"[*]" in co:
# print co
a = re.findall("\[\*\](.{6})", co)
for b in a:
if (
os.path.isfile("../source/Korban_netanel_on_{}.txt".format(masechet))
or os.path.isfile("../source/Pilpula_Charifta_on_{}.txt".format(masechet))
) and netanel < len(fixed):
if os.path.isfile("../source/Korban_netanel_on_{}.txt".format(masechet)):
commentator = "Korban Netanel "
if os.path.isfile("../source/Pilpula_Charifta_on_{}.txt".format(masechet)):
commentator = "Pilpula Charifta "
korban.append(fixed[netanel])
roash = "Rosh on %s." % masechet + str(len(links_netanel) + 1) + "." + str(num + 1)
netanelink = (
commentator + "on " + masechet + "." + str(len(links_netanel) + 1) + "." + str(len(korban))
)
links.append(link(netanelink, roash))
netanel += 1
si.append(co)
def parse(text):
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
# print "has korban netanel 2"
nose_kelim = nosekelim.open_file()
fixed = nosekelim.parse(nose_kelim)
links_netanel = []
netanel = 0
rosh = []
a = re.split(ur'@22([^@]*)', text)
for seif, cont in zip(a[1::2], a[2::2]):
si = []
korban =[]
if ur'[*]' in seif and (os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet) or os.path.isfile('source/PilPula_Charifta_on_{}.txt'.format(masechet))) and netanel <= len(fixed)):
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)):
commentator = "Korban Netanel on "
if os.path.isfile('source/PilPula_Charifta_on_{}.txt'.format(masechet)):
commentator = "Pilpula Charifta on "
korban.append(fixed[netanel])
#print len(links_netanel)
roash = "Rosh on %s." % masechet + str(len(links_netanel)+1) + ".1"
netanelink = commentator + masechet +"."+ str(len(links_netanel)+1) + ".1"
links.append(link(netanelink, roash))
netanel += 1
#print "netanel one seif", seif, netanel
#print fixed[netanel]
content = re.split('@66', cont)
seif = re.sub(ur'[^א-ת]',"", seif)
seif = hebrew.heb_string_to_int(seif.strip())
for num, co in enumerate(content):
if ur'[*]' in co:
print co
a = re.findall('\[\*\](.{6})', co)
for b in a:
if (os.path.isfile('source/Korban_netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet))) and netanel < len(fixed):
if os.path.isfile('source/Korban_netanel_on_{}.txt'.format(masechet)):
commentator = "Korban Netanel "
if os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
commentator = "Pilpula Charifta "
korban.append(fixed[netanel])
roash = "Rosh on %s." % masechet + str(len(links_netanel)+1) + "." + str(num+1)
netanelink = commentator + "on " +masechet + "." + str(len(links_netanel)+1)+ "."+ str(len(korban))
links.append(link(netanelink, roash))
netanel +=1
si.append(co)
def parse1(text):
old_numeri = 0
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(
masechet)) or os.path.isfile(
'source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
nose_kelim = nosekelim.open_file()
fixed = nosekelim.parse(nose_kelim)
links_netanel = []
netanel = 0
rosh = []
chapters = re.split(ur'(?:@00|@99)([^@]*)', text)
for chapter_num, chapter in zip(chapters[1::2], chapters[2::2]):
mispar = chapter_num.strip().split(" ")[1]
if mispar.encode('utf-8') in misparim.keys():
mispar_numeri = misparim[mispar.encode('utf-8')]
print mispar_numeri
if mispar_numeri - old_numeri > 1:
for i in range(1, mispar_numeri - old_numeri):
rosh.append([])
#print "length of rosh", len(rosh)
old_numeri = mispar_numeri
print mispar
#if len(chapter)<=1:
# pass
#else:
perek = []
a = re.split(ur'@22([^@]*)', chapter)
for seif, cont in zip(a[1::2], a[2::2]):
si = []
korban = []
#print seif
if ur'[*]' in seif and (
os.path.isfile(
'source/Korban_Netanel_on_{}.txt'.format(masechet))
or os.path.isfile(
'source/Pilpula_Charifta_on_{}.txt'.format(masechet))
) and netanel < len(fixed):
# print "hello", seif, netanel, len(fixed)
if os.path.isfile(
'source/Korban_Netanel_on_{}.txt'.format(masechet)):
commentator = "Korban Netanel"
if os.path.isfile(
'source/PilPula_Charifta_on_{}.txt'.format(masechet)):
commentator = "Pilpula Charifta"
korban.append(fixed[netanel])
roash = "Rosh on %s." % masechet + str(
len(rosh) + 1) + "." + str(len(perek) + 1) + ".1"
netanelink = commentator + " on " + masechet + "." + str(
len(links_netanel) + 1) + ".1"
#print roash, netanelink
links.append(link(netanelink, roash))
netanel += 1
content = re.split('@66', cont)
seif = re.sub(ur'[^א-ת]', "", seif)
seif = hebrew.heb_string_to_int(seif.strip())
for num, co in enumerate(content):
a = re.findall('\[\*\]', co)
for b in a:
# print b, seif
if (os.path.isfile(
'source/Korban_netanel_on_{}.txt'.format(masechet))
or os.path.isfile(
'source/Pilpula_Charifta_on_{}.txt'.format(
masechet))) and netanel < len(fixed):
if os.path.isfile(
'source/Korban_netanel_on_{}.txt'.format(
masechet)):
commentator = "Korban Netanel "
if os.path.isfile(
'source/Pilpula_Charifta_on_{}.txt'.format(
masechet)):
commentator = "Pilpula Charifta "
korban.append(fixed[netanel])
roash = "Rosh on %s." % masechet + str(
len(rosh) + 1) + "." + str(len(perek) +
1) + "." + str(num + 1)
netanelink = commentator + "on " + masechet + "." + str(
len(links_netanel) + 1) + "." + str(len(korban))
#print roash, netanelink
links.append(link(netanelink, roash))
netanel += 1
si.append(co)
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(
masechet)) or os.path.isfile(
'source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
links_netanel.append(korban)
perek.append(si)
if len(perek) is not 0:
rosh.append(perek)
def parse1(text):
old_numeri = 0
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
nose_kelim = nosekelim.open_file()
fixed = nosekelim.parse(nose_kelim)
links_netanel = []
netanel = 0
rosh = []
chapters = re.split(ur'(?:@00|@99)([^@]*)', text)
for chapter_num, chapter in zip(chapters[1::2], chapters[2::2]):
mispar = chapter_num.strip().split(" ")[1]
if mispar.encode('utf-8') in misparim.keys():
mispar_numeri = misparim[mispar.encode('utf-8')]
print mispar_numeri
if mispar_numeri - old_numeri > 1:
for i in range(1,mispar_numeri-old_numeri):
rosh.append([])
#print "length of rosh", len(rosh)
old_numeri = mispar_numeri
print mispar
#if len(chapter)<=1:
# pass
#else:
perek = []
a = re.split(ur'@22([^@]*)', chapter)
for seif, cont in zip(a[1::2], a[2::2]):
si = []
korban =[]
#print seif
if ur'[*]' in seif and (os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet))) and netanel < len(fixed):
# print "hello", seif, netanel, len(fixed)
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)):
commentator = "Korban Netanel"
if os.path.isfile('source/PilPula_Charifta_on_{}.txt'.format(masechet)):
commentator = "Pilpula Charifta"
korban.append(fixed[netanel])
roash = "Rosh on %s." % masechet +str(len(rosh)+1) + "." + str(len(perek)+1) + ".1"
netanelink = commentator + " on " + masechet +"."+ str(len(links_netanel)+1) + ".1"
#print roash, netanelink
links.append(link(netanelink, roash))
netanel += 1
content = re.split('@66', cont)
seif = re.sub(ur'[^א-ת]',"", seif)
seif = hebrew.heb_string_to_int(seif.strip())
for num, co in enumerate(content):
a = re.findall('\[\*\]', co)
for b in a:
# print b, seif
if (os.path.isfile('source/Korban_netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet))) and netanel < len(fixed):
if os.path.isfile('source/Korban_netanel_on_{}.txt'.format(masechet)):
commentator = "Korban Netanel "
if os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
commentator = "Pilpula Charifta "
korban.append(fixed[netanel])
roash = "Rosh on %s." % masechet + str(len(rosh)+1) + "." + str(len(perek)+1) + "." + str(num+1)
netanelink = commentator + "on " + masechet + "." + str(len(links_netanel)+1)+ "."+ str(len(korban))
#print roash, netanelink
links.append(link(netanelink, roash))
netanel +=1
si.append(co)
if os.path.isfile('source/Korban_Netanel_on_{}.txt'.format(masechet)) or os.path.isfile('source/Pilpula_Charifta_on_{}.txt'.format(masechet)):
links_netanel.append(korban)
perek.append(si)
if len(perek) is not 0:
rosh.append(perek)