1 #!/usr/bin/python

     2 

     3 # python voa-special-c5.py

     4 

     5 import re

     6 

     7 f = open('../gadict-voa-special-english-word-book.dict-c5', 'r')

     8 

     9 re_delim = re.compile(r'^_____\n')

    10 re_empty = re.compile(r'^\n')

    11 

    12 while True:

    13     s = f.readline()

    14     if re_delim.match(s):

    15         break

    16 state = 'delim'

    17 

    18 words = []

    19 word = None

    20 article = None

    21 

    22 while True:

    23     line = f.readline()

    24     if len(line) == 0:

    25         break

    26     if state == 'delim':

    27         if re_empty.match(line):

    28             state = 'skip_to_word'

    29     elif state == 'skip_to_word':

    30         if not re_empty.match(line):

    31             state = 'word'

    32             word = line.strip()

    33     elif state == 'word':

    34         if re_empty.match(line):

    35             state = 'skip_to_article'

    36         else:

    37             state = 'article'

    38             article += line

    39     elif state == 'skip_to_article':

    40         if not re_empty.match(line):

    41             state = 'article'

    42             article = line

    43     elif state == 'article':

    44         if re_delim.match(line):

    45             state = 'delim'

    46             words.append( (word, article.strip()) )

    47             word = None

    48             article = ''

    49         else:

    50             state = 'article'

    51             article += line

    52 

    53 f.close()

    54 

    55 for i in words:

    56     print("\\worddef{%s}{%s}" % (i[0], i[1]))