From 910597b3ff556453060c5a98a6b807102982174e Mon Sep 17 00:00:00 2001
From: Andrei Stoica <andrei.stoica.365@gmail.com>
Date: Sun, 24 Jul 2022 15:21:57 -0400
Subject: [PATCH] added more units and instuctions to list

---
 src/scrape.py | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/src/scrape.py b/src/scrape.py
index 23d75e9..d3dd0e0 100644
--- a/src/scrape.py
+++ b/src/scrape.py
@@ -1,4 +1,5 @@
 from ast import alias
+from dis import Instruction
 import db
 import re
 from sqlalchemy import select, desc, exists, not_, except_
@@ -13,9 +14,22 @@ from argparse import ArgumentParser
 def parse_ingredient(ingredient_text):
     units = ['teaspoon', 'tablespoon', 'gram', 'ounce', 'jar', 'cup', 'pinch',
              'container', 'slice', 'package', 'pound', 'can', 'dash', 'spear',
-             'bunch', 'quart', 'cube', 'envelope', 'square', 'sprig', 'bags',
+             'bunch', 'quart', 'cube', 'envelope', 'square', 'sprig', 'bag',
              'box', 'drop', 'fluid ounce', 'gallon', 'head', 'link', 'loaf',
-             'pint', 'pod', 'sheet', 'stalk', 'whole', 'bar']
+             'pint', 'pod', 'sheet', 'stalk', 'whole', 'bar', 'bottle', 'bulb',
+             'year', 'fillet', 'litter', 'packet', 'slices']
+    instructions = ['and', 'or', 'chopped', 'diced', 'brewed', 'chilled', 'chunky',
+                    'small', 'medium', 'large', 'couarse(:?ly)?', 'cracked',
+                    'crushed', 'ground', 'cooked', 'cubed', 'crumbled', 'cut',
+                    'cold', 'hot', 'warm', 'day', 'old', 'drained', 'canned',
+                    'dried', 'dry', 'fine(?:ly)', 'firmly', 'fresh', 'frozen',
+                    'grated', 'grilled', 'hard', 'hot', 'juliened?', 'leftover',
+                    'light', 'lite', 'mashed', 'melted', 'minced', 'packed',
+                    'peeled', 'pitted', 'sliced', 'prepared', 'refrigerated',
+                    'rehydrated', 'seedless', 'shaved', 'shredded', 'sifted',
+                    'sieved', 'shucked', 'slivered', 'thick', 'sliced', 'thin',
+                    'toasted', 'trimmed', 'unbaked', 'uncooked', 'unpeeled',
+                    'unopened', 'unseasoned']
     number_regex = '((?:[\d\\./\\u00BC-\\u00BE\\u2150-\\u215E]*\s?(?:\(.+\))?)*)'
     ingredient_regex = '([a-zA-Z \'\-]+)'
     supplement_regex = ',?(.*)'