cookbook/mom.py

   1 # Copyright
   2
   3 """Hack parser for standardizing my Mom's cookbook text.
   4 """
   5
   6 from .cookbook import (
   7     Cookbook, Recipe, IngredientBlock, Ingredient, Amount, Directions)
   8
   9
  10 class MomParser (object):
  11     def parse(self, filename):
  12         c = Cookbook()
  13         consecutive_blanks = 100
  14         recipe_lines = None
  15         for line in file(filename, 'r'):
  16             line = line.strip().decode('utf-8')
  17             if line == '':
  18                 if recipe_lines != None and consecutive_blanks == 0:
  19                     recipe_lines.append('')
  20                 consecutive_blanks += 1
  21                 continue
  22             if consecutive_blanks >= 2:
  23                 if recipe_lines != None:
  24                     c.append(self._parse_recipe(recipe_lines))
  25                 recipe_lines = [line]
  26             else:
  27                 recipe_lines.append(line)
  28             consecutive_blanks = 0
  29         return c
  30
  31     def _parse_recipe(self, lines):
  32         name = lines.pop(0)
  33         yield_,author,source,url,lines = self._parse_yield_line(lines)
  34         ingredient_blocks,lines = self._parse_ingredient_blocks(lines)
  35         directions,lines = self._parse_directions(lines)
  36         assert len(lines) == 0, lines
  37         return Recipe(
  38             name=name,
  39             ingredient_blocks=ingredient_blocks,
  40             directions=directions,
  41             yield_=yield_,
  42             author=author,
  43             source=source,
  44             url=url)
  45
  46     def _parse_yield_line(self, lines):
  47         while len(lines) > 0 and lines[0] == '':
  48             lines.pop(0)
  49         fields = ['yield', 'from', 'source', 'url']
  50         yield_ = author = source = url = None
  51         matching_line = False
  52         for field in fields:
  53             if field in lines[0].lower():
  54                 matching_line = True
  55                 break
  56         if matching_line == True:
  57             bits = lines.pop(0).split('\t')
  58             for bit in bits:
  59                 for field in fields:
  60                     if bit.lower().startswith(field+':'):
  61                         value = bit[len(field+':'):].strip()
  62                         if field == 'yield':
  63                             yield_ = value.replace('Serving', 'serving')
  64                         elif field == 'from':
  65                             author = value
  66                         elif field == 'source':
  67                             source = value
  68                         elif field == 'url':
  69                             url = value
  70                         break
  71         return (yield_, author, source, url, lines)
  72
  73     def _parse_ingredient_blocks(self, lines):
  74         ingredient_blocks = []
  75         first_block = True
  76         while True:
  77             while len(lines) > 0 and lines[0] == '': # scroll past blanks
  78                 lines.pop(0)
  79             if (len(lines) == 0
  80                 or not (first_block == True
  81                         or lines[0].endswith(':'))):
  82                 break
  83             if lines[0].endswith(':'):
  84                 line = lines.pop(0)
  85                 name = line[:-1].strip()
  86             else:
  87                 name = None
  88             block = IngredientBlock(name)
  89             while len(lines) > 0 and lines[0] != '':
  90                 block.append(self._parse_ingredient_line(lines.pop(0)))
  91             ingredient_blocks.append(block)
  92             first_block = False
  93         return (ingredient_blocks, lines)
  94
  95     def _parse_ingredient_line(self, line):
  96         if line.lower().startswith('1 red'):
  97             line = '1 # red'+line[len('1 red'):]
  98         try:
  99             value,units,name = line.split(' ', 2)
 100         except ValueError:
 101             print line,
 102             raise
 103         if units == '#':
 104             units = None
 105         elif units == 'Large':
 106             units = 'large'
 107         elif units == 'Cloves':
 108             units = 'cloves'
 109         return Ingredient(name, Amount(value, units))
 110
 111     def _parse_directions(self, lines):
 112         directions = Directions()
 113         paragraph = []
 114         for line in lines:
 115             if line == '':
 116                 if len(paragraph) > 0:
 117                     directions.append('\n'.join(paragraph))
 118                     paragraph = []
 119             else:
 120                 paragraph.append(line)
 121         return (directions, [])