Using break for nested for loop

I’d go with the review of the second version of the code since it’s the same as the first one with the addition of the flag_keyword logic. If you want guidance on how to produce code matching requirements or how to debug your code, you’d be better of asking on Stack Overflow.

A little bit on logic, though: since you want to move to the next record when the innermost filter validates your flag_keyword value should be reset at each iteration of the outermost loop. Preferably as the first condition after for record in fetch_record. If your dataset makes it through the whole for loop without entering the last if, then your

if flag_keyword == 1:
    raise GetOutOfLoop

will raise NameError: name 'flag_keyword' is not defined.

You are using the flag to run some code only if you broke out of a loop. Note, however, that if you change the logic a bit and want to run some code only if you didn't broke the loop you can use a for ... else construct.

Now let’s go to the code:

• Your GetOutOfLoop exception's only purpose is to let you break out of the outermost loop only if you broke from the innermost one (i.e. your innermost filter validates). The same thing you are doing with your flag_keyword at a deeper nesting level. Get consistent: don't mix implementations for similar tasks. There is an other approach if you read on but I’d raise the exception in replacement of the innermost break to avoid all your 'break if I broke' logic because:

• You have too much nesting levels, it makes code harder to read and follow. Remember that:

  for value in data:
      if is_valid(value):
          do_stuff()
  

  can also be written:

  for value in data:
      if not is_valid(value):
          continue
      do_stuff()
  

  when do_stuff is composed of nested loops, it can save on indentation levels. However, you can get rid of these test by using generators:

  for value in (v for v in data if is_valid(v)):
      do_stuff()
  

• Even if the purpose of the for ... else loop is to run some code only if you didn't break out of the loop, you can use it to run some code only if you did break. The general approach is:

  for value in data:
      # some code
  else:
      do_stuff_if_no_break()
  

  But for your use case, since you are within loops each time, you can change it to:

  for value in data:
      # some code
  else:
      continue
  do_stuff_if_break()
  

• You check for various things in u'{} {}'.format(record['title'], record['description']).lower() but you are building this string a large number of times. It is best, performance-wise, to build it once as soon as you get the record dictionary. Speaking of dictionary, you can improve the syntax a bit:

  u'{title} {description}'.format(**record).lower()
  

• You use lower on each items of your intermediate iterables (self.keyword_material_info, self.keyword_bse_list and the one returned by self.get_ngrams()). Can't you build those iterables with already formatted strings?

The code would become:

for record in fetch_record:
    search = u'{title} {description}'.format(**record).lower()
    total += 1
    for rec in cname:
        c_ngram = self.get_ngrams(rec['company_name'])
        # assuming c_ngrams contains items already formatted with .lower()+' '
        for gram in (g for g in c_ngram if g in search):
            filter_company_level += 1
            # print "Matched based on company name : ", record['article_link']
            company_write.write(record['article_link']+' - '+rec['company_name']+' - '+rec['company_code']+'\n')
            # assuming keyword_material_info contains items already formatted with .lower()+' '
            for value in (v for v in self.keyword_material_info if v in search):
                filter_with_material_info += 1
                materialinfo_write.write(record['article_link']+' - '+rec['company_name']+' - '+rec['company_code']+' - '+value+'\n')
                # assuming keyword_bse_list contains items already formatted with .lower()+' '
                for item in (i for i in self.keyword_bse_list if i in search):
                    filter_with_keyword_info += 1
                    keyword_write.write(record['article_link']+' - '+rec['company_name']+' - '+rec['company_code']+" - "+value+' - '+item+'\n')
                    print record['article_link']
                    print value
                    print item
                    break
                else:
                    # skip the next break if we didn't found an <item> matching <search>
                    continue
                break
            else:
                continue
            break
        else:
            continue
        break

However I’m a little bit worried on the nesting of your loops. Since neither self.keyword_material_info nor self.keyword_bse_list seems to depend on gram... They might be modified by the call of self.get_ngrams() but I can't verify.

One last note, that I didn’t try to enforce while re-writting your code: you may want to improve your variable naming. item or value does not carry much information, for instance.