units = {
    "15ml":["1큰술","1T","1Ts","3t"],
    "5ml":["1작은술","1t","1ts"],    
    "200ml":["1컵","1Cup","1C"],
    "250ml":["16T", "1C"],
    "180ml":["1종이컵"],
    "28.3g":["1oz"],
    "453g":["1파운드","lb"],
    "3780ml":["1갤런","gallon"],
    "2g":["1꼬집",'약간'],
    "4g":["조금"],
    "6g":["조금"],
    "10g":["적당량"],
    "50g":["1줌"], # 나물
    "100g":["1큰줌"], # 나물
    "13마리":["1줌"], # 멸치
    "26마리":["1큰줌"], # 멸치
    "100g":["1주먹"], # 여자 어른의 주먹크기
    "3cm":["1토막"],
    "1알":["1톨","1쪽"], # 마늘, 생강 등
    "600g":["1근"], # 고기
    "400g":["1근"], # 채소
    "200g":["1봉지"],
}

unit_tokens = []
for _ in [v  for k,v in units.items()]:
    unit_tokens += _
", ".join(sorted(set(unit_tokens)))

import json, re
import pandas as pd
df_safe   = pd.read_csv("data/food_recipe_safe.csv")
df_safe   = df_safe.loc[:, ['메뉴명','재료정보']]
print(df_safe.shape, df_safe.tail())

from collections import Counter
df_safe.재료정보 = [_.split('\n')  for _ in df_safe.재료정보.fillna('')]
df_safe_names  = {no:_ for no,_ in enumerate(df_safe.메뉴명)}
sorted(dict(Counter([len(_)  for no, _ in enumerate(df_safe["재료정보"])])).items(),\
       key=lambda x:x[0], reverse = False) # Value 기준 정렬

# 재료정보의 value 갯수
df_safe.insert(1, "Values", [len(_) for _ in df_safe["재료정보"]])

# 중간 구분가능 Key 갯수의 확인
result = []
result_key = []
result_recipe = []
for _ in range(len(df_safe)):
    # 레시피 구분데이터
    data = df_safe.loc[_]
    count, temp = 0, []
    for no,_ in enumerate(data["재료정보"]):
        # 재료정보 List 전처리를 위한 위치값 탐지기
        if "".join(re.findall("[가-힣]+", _)) == _.replace(' ',""):
            count += 1
            temp.append(no)
            
    # 유효한 Key 값이 발견되었으나
    if count != 0:
        # 0번 index 값이 없을때..
        if 0 not in temp:
            result_recipe.append([data['메뉴명']] + data['재료정보'])
            temp = [0] + [_+1  for _ in temp]
        else:
            result_recipe.append(data['재료정보'])
    else:
        result_recipe.append(data['재료정보'])
            
    result.append(count)
    result_key.append(temp)

df_safe.insert(2, "Keys", result)
df_safe.insert(3, "Key_index", result_key)
df_safe["재료정보"] = result_recipe
df_safe.head(2)

sorted(Counter([(df_safe.Values[_], df_safe.Keys[_])   
                for _ in range(len(df_safe))]), 
       key=lambda x :x[1], reverse=True)

Counter([(df_safe.Values[_], df_safe.Keys[_])   
                for _ in range(len(df_safe))])

for no, _ in enumerate(range(len(df_safe))):
    data = df_safe.loc[_]
    if data.Keys != 0:
        if 0 not in data["Key_index"]:
            print(no, data["Key_index"], data["Values"])

df_safe.head(2)

df_safe[df_safe.Values==12].index[0]
temp = df_safe.loc[df_safe[df_safe.Values==12].index[0]]
temp["재료정보"]

# 범위 index 를 추가하여 Key, Value 테스트
temp = df_safe.loc[456]
temp_index = temp.Key_index + [len(temp.재료정보)]
{temp["재료정보"][temp_index[_]]:temp["재료정보"][temp_index[_]+1:temp_index[_+1]]
 for _ in range(len(temp.Key_index))}

df_safe["재료정보"] = result
df_safe.head(2)

result = []
for item in df_safe["재료정보"]:
    temp = list(item.keys()) # Key 이름
    temp_dict = {_:len(item[_])   for _ in item}
    result.append(temp_dict)

df_safe.insert(4, "Key_Values", result)
df_safe = df_safe.loc[:, ["메뉴명","Key_Values","재료정보"]]
df_safe.insert(2, "Key_Value_nums", [list(_.values()) for _ in df_safe.Key_Values])
df_safe.head(2)

result = []
for _ in df_safe.Key_Value_nums:
    if list(set(_)) == [1]:
        result.append('pass')
    else: 
        result.append('work')

df_safe.Key_Value_nums = result
df_safe.head(2)

result = []
for _ in list(df_safe.index):
    if _ in [_[0] for _ in temp_list]:
        result.append('Yes')
    else: result.append('Pass')
        
df_safe = df_safe.loc[:, ["메뉴명", "재료정보"]]
df_safe.insert(1, "Mid_Key", result)
df_safe[df_safe.Mid_Key=='Yes'].tail(3)

import re
reg_token = "-.*:"
reg_token = "\[.*\]"
texts = "오이고추 160g [절임물] 해물육수 30g, 어간장 10g"
token = re.findall(reg_token,  texts)
token = "".join(token)
token

import re
reg_token = "-.*:"
texts = "오이고추 160g - 절임물 : 해물육수 30g, 어간장 10g"
token = re.findall(reg_token,  texts)
token = "".join(token)
token

texts.split(token)

import ast
import pandas as pd
# df_safe = pd.read_csv("data/food_recipe_safe_new_keys.csv")
df_safe = pd.read_csv('data/food_recipe_safe_new_keys.csv')
df_safe["재료정보"] = [ast.literal_eval(_)  for _ in df_safe["재료정보"]]
df_safe = df_safe.iloc[:,[0,-1]]
print(df_safe.shape)
df_safe.head(2)

result = []
for idx in list(df_safe.index):
    count, temp = 0, df_safe.loc[idx]
    temp_keys = list(temp['재료정보'].keys())
    for k in temp_keys:
        if temp["재료정보"][k].find(':') != -1: count += 1
        else: pass
    result.append(count)
df_safe.insert(1, 'Mid_Key', result)
df_safe.head(2)

idx_key = list(df_safe[df_safe.Mid_Key!=0].index)[-4]
print(df_safe[df_safe.Mid_Key!=0].head(3))
print(df_safe[df_safe.Mid_Key!=0].shape)
df_safe.loc[idx_key]["재료정보"]

df_safe[df_safe.Mid_Key==0].shape()

df_safe.loc[idx_num]["재료정보"]

# 0 ~ 235
num  = 0
temp = df_safe[df_safe.Mid_Key!=0]
idx_num = temp.iloc[num,:].name
print(list(df_safe.loc[idx_num]["재료정보"].keys()))
df_safe.loc[idx_num]["재료정보"]

# 수작업 수정한 내용
temp_dict = {
    '돼지머리수육맑은전골': '돼지머리 200g, 우동면 100g, 미나리 20g, 청고추 15g',
    '소스': '일본 된장 30g, 다진마늘 10g, 다진대파 10g, 청주 15g, 무 25g',
    '육수' : '해물육수 300g, 청양고추 20g, 맛간장 15g'}

print(temp_dict)
df_safe.loc[idx_num,"재료정보"] = str(temp_dict)
# df_safe.loc[idx_num,"재료정보"] = temp_dict
df_safe.to_csv('data/food_recipe_safe_new_keys.csv', index=None)
print("수정완료")

df_safe.iloc[idx_num,-1] 

str(temp_dict)

ast.literal_eval(str(temp_dict))