andrei-stoica
/
recipe-graph
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

fixed typos

This commit is contained in:
Andrei Stoica 2022-07-22 17:57:35 -04:00
parent fe8362b02c
commit 39e13a3454
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
src/scrape.py
View File

@ -13,7 +13,7 @@ from argparse import ArgumentParser
def parse_ingredient(ingredient_text): def parse_ingredient(ingredient_text):
units = ['teaspoon', 'tablespoon', 'gram', 'once', 'jar', 'cup', 'pinch', units = ['teaspoon', 'tablespoon', 'gram', 'once', 'jar', 'cup', 'pinch',
'container', 'slice', 'package', 'pound', 'can', 'dash', 'spear', 'container', 'slice', 'package', 'pound', 'can', 'dash', 'spear',
'bunch', 'quart', 'cube', 'envelope', 'squars', 'sprig'] 'bunch', 'quart', 'cube', 'envelope', 'square', 'sprig']
number_regex = '((?:[\d\\./\\u00BC-\\u00BE\\u2150-\\u215E]*\s?(?:\(.+\))?)*)' number_regex = '((?:[\d\\./\\u00BC-\\u00BE\\u2150-\\u215E]*\s?(?:\(.+\))?)*)'
ingredient_regex = '([a-zA-Z \'\-]+)' ingredient_regex = '([a-zA-Z \'\-]+)'
supplement_regex = ',?(.*)' supplement_regex = ',?(.*)'
@ -77,7 +77,7 @@ def parse_recipe(session, recipe, site):
name_div = name_candidates[0] name_div = name_candidates[0]
recipe.name = name_div.text recipe.name = name_div.text
logging.info(f"Adding Recipe {recipe}") logging.info(f"Adding Recipe {recipe.name} from {recipe_url}")
session.add(recipe) session.add(recipe)
session.flush() session.flush()
@ -156,7 +156,7 @@ with S.begin() as sess:
except KeyboardInterrupt as e: except KeyboardInterrupt as e:
savepoint.rollback() savepoint.rollback()
break break
except IntegrityError as e: except Exception as e:
savepoint.rollback() savepoint.rollback()
logging.error(e) logging.error(e)
continue continue