import string, re
redundant_chars='"\'.:;,-+<>()*~' # chars we need to strip from a word before we see if it matches, and from the searchwords to eliminate boolean mode chars
tr=string.maketrans(redundant_chars,' '*len(redundant_chars))
class Done(Exception):
pass
class Word(str):pass
class FoundWord(str):
def __str__(self):
return self.tags[0]+self+self.tags[1]
class Part:
def __init__(self,tags,trail):
self.chain=[]
self.limit=trail
self.trail=trail
self.has=False
self.tags=tags
def push(self,w):
self.chain.insert(0,Word(w))
if len(self.chain)>self.limit:
if self.has:
self.chain.reverse()
raise Done()
self.chain.pop()
def add(self,w):
self.chain.insert(0,FoundWord(w))
self.limit+=self.trail+1
self.has=True
def __str__(self):
return '...%s...' % ' '.join(map(str,self.chain))
def generateParts(context,text,sw,tags,trail,maxlines):
par=Part(tags,trail)
sw=sw.translate(tr).strip().split()
test=lambda w:w.translate(tr).strip() in sw
i=0
for aw in text:
if i==maxlines:
raise StopIteration
if test(aw):
par.add(aw)
else:
try:
par.push(aw)
except Done:
i+=1
yield par
par=Part(tags,trail)
def cutFound(context,txt,sw,tags,trail,maxlines):
# initialize class
FoundWord.tags=tags
# strip html tags (in case it is a web page - we show result without formatting)
r=re.compile('<script>.*?</script>',re.DOTALL|re.IGNORECASE)
r=re.compile('<head>.*?</head>',re.DOTALL|re.IGNORECASE)
txt=re.sub(r,'',txt)
r=re.compile('<([^>]+)>',re.DOTALL|re.IGNORECASE)
txt=re.sub(r,'',txt)
r=re.compile('\s+')
txt=re.sub(r,' ',txt)
text = ' '.join(txt.split('\n')).split(' ') # very rough tokenization
return [p for p in generateParts(context,text,sw,tags,trail,maxlines)]
if __name__=='__main__':
sw='pricing priority right acting proportion'
txt=' '.join([l.strip() for l in open('offer.txt').readlines()])
# configuration
tags=('<b>','</b>')
trail=5
maxlines=5
sw=sw.split()
for p in cutFound(txt,sw,tags,trail,maxlines):
print p
# vim: filetype=python syntax=python shiftwidth=2