cookbook/mom.py

   1 # Copyright (C) 2010 W. Trevor King <wking@drexel.edu>
   2 #
   3 # This file is part of Cookbook.
   4 #
   5 # Cookbook is free software: you can redistribute it and/or modify it
   6 # under the terms of the GNU General Public License as published by the
   7 # Free Software Foundation, either version 3 of the License, or (at your
   8 # option) any later version.
   9 #
  10 # Cookbook is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License
  16 # along with Cookbook.  If not, see <http://www.gnu.org/licenses/>.
  17
  18 """Hack parser for standardizing my Mom's cookbook text.
  19 """
  20
  21 from .cookbook import (
  22     Cookbook, Recipe, IngredientBlock, Ingredient, Amount, Directions)
  23
  24
  25 class MomParser (object):
  26     def parse(self, filename):
  27         c = Cookbook()
  28         consecutive_blanks = 100
  29         recipe_lines = None
  30         for line in file(filename, 'r'):
  31             line = line.strip().decode('utf-8')
  32             if line == '':
  33                 if recipe_lines != None and consecutive_blanks == 0:
  34                     recipe_lines.append('')
  35                 consecutive_blanks += 1
  36                 continue
  37             if consecutive_blanks >= 2:
  38                 if recipe_lines != None:
  39                     c.append(self._parse_recipe(recipe_lines))
  40                 recipe_lines = [line]
  41             else:
  42                 recipe_lines.append(line)
  43             consecutive_blanks = 0
  44         return c
  45
  46     def _parse_recipe(self, lines):
  47         name = lines.pop(0)
  48         yield_,author,source,url,lines = self._parse_yield_line(lines)
  49         ingredient_blocks,lines = self._parse_ingredient_blocks(lines)
  50         directions,lines = self._parse_directions(lines)
  51         assert len(lines) == 0, lines
  52         return Recipe(
  53             name=name,
  54             ingredient_blocks=ingredient_blocks,
  55             directions=directions,
  56             yield_=yield_,
  57             author=author,
  58             source=source,
  59             url=url)
  60
  61     def _parse_yield_line(self, lines):
  62         while len(lines) > 0 and lines[0] == '':
  63             lines.pop(0)
  64         fields = ['yield', 'from', 'source', 'url']
  65         yield_ = author = source = url = None
  66         matching_line = False
  67         for field in fields:
  68             if field in lines[0].lower():
  69                 matching_line = True
  70                 break
  71         if matching_line == True:
  72             bits = lines.pop(0).split('\t')
  73             for bit in bits:
  74                 for field in fields:
  75                     if bit.lower().startswith(field+':'):
  76                         value = bit[len(field+':'):].strip()
  77                         if field == 'yield':
  78                             yield_ = value.replace('Serving', 'serving')
  79                         elif field == 'from':
  80                             author = value
  81                         elif field == 'source':
  82                             source = value
  83                         elif field == 'url':
  84                             url = value
  85                         break
  86         return (yield_, author, source, url, lines)
  87
  88     def _parse_ingredient_blocks(self, lines):
  89         ingredient_blocks = []
  90         first_block = True
  91         while True:
  92             while len(lines) > 0 and lines[0] == '': # scroll past blanks
  93                 lines.pop(0)
  94             if (len(lines) == 0
  95                 or not (first_block == True
  96                         or lines[0].endswith(':'))):
  97                 break
  98             if lines[0].endswith(':'):
  99                 line = lines.pop(0)
 100                 name = line[:-1].strip()
 101             else:
 102                 name = None
 103             block = IngredientBlock(name)
 104             while len(lines) > 0 and lines[0] != '':
 105                 block.append(self._parse_ingredient_line(lines.pop(0)))
 106             ingredient_blocks.append(block)
 107             first_block = False
 108         return (ingredient_blocks, lines)
 109
 110     def _parse_ingredient_line(self, line):
 111         if line.lower().startswith('1 red'):
 112             line = '1 # red'+line[len('1 red'):]
 113         try:
 114             value,units,name = line.split(' ', 2)
 115         except ValueError:
 116             print line,
 117             raise
 118         if units == '#':
 119             units = None
 120         elif units == 'Large':
 121             units = 'large'
 122         elif units == 'Cloves':
 123             units = 'cloves'
 124         return Ingredient(name, Amount(value, units))
 125
 126     def _parse_directions(self, lines):
 127         directions = Directions()
 128         paragraph = []
 129         for line in lines:
 130             if line == '':
 131                 if len(paragraph) > 0:
 132                     directions.append('\n'.join(paragraph))
 133                     paragraph = []
 134             else:
 135                 paragraph.append(line)
 136         return (directions, [])