|
Revision 27, 1.4 kB
(checked in by gak, 11 months ago)
|
|
small fixes
|
-
Property svn:keywords set to
Id Revision
|
| Line | |
|---|
| 1 | import os |
|---|
| 2 | import urllib |
|---|
| 3 | |
|---|
| 4 | from BeautifulSoup import BeautifulSoup as bs |
|---|
| 5 | |
|---|
| 6 | def download_page(definition): |
|---|
| 7 | os.system('wget -q --user-agent= -Oyo http://www.google.com.au/search?q=%s' \ |
|---|
| 8 | % urllib.quote(definition)) |
|---|
| 9 | return open('yo').read() |
|---|
| 10 | |
|---|
| 11 | def get_definition(definition): |
|---|
| 12 | data = download_page(definition) |
|---|
| 13 | soup = bs(data) |
|---|
| 14 | answer = soup.find('h2', {'class': 'r'}) |
|---|
| 15 | if not answer: |
|---|
| 16 | return None |
|---|
| 17 | try: |
|---|
| 18 | answer = answer.font.b |
|---|
| 19 | except AttributeError: |
|---|
| 20 | return None |
|---|
| 21 | answer = strip_ml_tags(str(answer)) |
|---|
| 22 | return answer |
|---|
| 23 | |
|---|
| 24 | def strip_ml_tags(in_text): |
|---|
| 25 | """Description: Removes all HTML/XML-like tags from the input text. |
|---|
| 26 | Inputs: s --> string of text |
|---|
| 27 | Outputs: text string without the tags |
|---|
| 28 | |
|---|
| 29 | # doctest unit testing framework |
|---|
| 30 | |
|---|
| 31 | >>> test_text = "Keep this Text <remove><me /> KEEP </remove> 123" |
|---|
| 32 | >>> strip_ml_tags(test_text) |
|---|
| 33 | 'Keep this Text KEEP 123' |
|---|
| 34 | """ |
|---|
| 35 | # convert in_text to a mutable object (e.g. list) |
|---|
| 36 | s_list = list(in_text) |
|---|
| 37 | i,j = 0,0 |
|---|
| 38 | |
|---|
| 39 | while i < len(s_list): |
|---|
| 40 | # iterate until a left-angle bracket is found |
|---|
| 41 | if s_list[i] == '<': |
|---|
| 42 | while s_list[i] != '>': |
|---|
| 43 | # pop everything from the the left-angle bracket until the |
|---|
| 44 | # right-angle bracket |
|---|
| 45 | s_list.pop(i) |
|---|
| 46 | |
|---|
| 47 | # pops the right-angle bracket, too |
|---|
| 48 | s_list.pop(i) |
|---|
| 49 | else: |
|---|
| 50 | i=i+1 |
|---|
| 51 | |
|---|
| 52 | # convert the list back into text |
|---|
| 53 | join_char='' |
|---|
| 54 | return join_char.join(s_list) |
|---|
| 55 | |
|---|