如何为agrepl传入字符串列表实现模糊匹配检测？

背景与需求 #

需使用base R自带的agrepl函数，检测数据框df的Ingredients列是否包含指定的食材字符串，兼容轻微拼写错误。受环境限制无法安装新包，因此优先使用agrepl。

待检测字符串列表 #

strings_to_check <- c("Molybdenum Salt",
                      "Mineral Salt \\(Molybdenum Sulfide)",
                      "Molybdenum Sulfide",
                      "Mineral Salt \\(444\\)",
                      "444")

当前遇到的问题 #

• 用grepl拼接正则表达式可正常检测：

ingredients_df <- df %>% 
  mutate(Molybdenum = grepl(paste(strings_to_check, collapse = "|"), Ingredients))

• 单个字符串传入agrepl时功能正常：

one_string_df <- ingredients_df %>% 
  mutate(One_String = agrepl("Molybdenum Sulfide", Ingredients, max.distance = 2, ignore.case = TRUE))

• 但将字符串列表拼接后传入agrepl，返回结果全为FALSE：

fuzzy_df <- ingredients_df %>% 
  mutate(Fuzzy_Molybdenum = agrepl(paste(strings_to_check, collapse = "|"), Ingredients, max.distance = 2, ignore.case = TRUE))

解决方案 #

agrepl不支持直接传入用|拼接的正则表达式进行模糊匹配，需对每个目标字符串单独执行模糊检测，再取所有结果的逻辑或（即只要匹配任意一个目标字符串就返回TRUE）。以下提供两种实现方式：

方法1：dplyr + purrr实现 #

library(dplyr)
library(purrr)

fuzzy_df <- ingredients_df %>%
  mutate(Fuzzy_Molybdenum = map_lgl(Ingredients, ~ any(map_lgl(strings_to_check, agrepl, x = ., max.distance = 2, ignore.case = TRUE))))

方法2：Base R原生实现（无需额外包） #

fuzzy_df <- ingredients_df
fuzzy_df$Fuzzy_Molybdenum <- rowSums(sapply(strings_to_check, function(pattern) {
  agrepl(pattern, fuzzy_df$Ingredients, max.distance = 2, ignore.case = TRUE)
})) > 0

验证结果 #

运行上述代码后，可得到符合预期的输出：

Product_Name                        Ingredients                 Issue Molybdenum Fuzzy_Molybdenum
   <chr>                               <chr>                       <chr> <lgl>      <lgl>           
 1 Cheesy Jalapeno Popcorn             Sugar | Croutons (10%) (Wh… Mino… FALSE      TRUE            
 2 Creamy Coconut Curry Soup           Premix [Salt | Mineral Sal… NA    TRUE       TRUE            
 3 Crunchy Cheddar Bites               Vegetable Oils (Palm | Can… NA    FALSE      FALSE
 4 Exotic Thai Basil Noodles           Natural Cheese Flavour [Ma… NA    TRUE       TRUE
 5 Golden Honey Wheat Bread            Sesame Seeds (3%) | Yeast … Lowe… FALSE      TRUE            
 6 Gourmet Truffle Macaroni & Cheese   Rice Flour | Thickener (14… NA    TRUE       TRUE            
 7 Heavenly Hazelnut Delight Ice Cream Acidity Regulator (339) | … Majo… FALSE      FALSE           
 8 Juicy Pineapple Burst Sorbet        Maltodextrin | Salt | Suga… NA    TRUE       TRUE            
 9 Maple Glazed Pecan Granola          Dried Vegetables (9%) (Pea… NA    TRUE       TRUE            
10 Mediterranean Herb Garden Hummus    Electrolytes 11.5% (Sodium… NA    FALSE      FALSE           
11 Roasted Garlic Parmesan Pretzels    Dextrose | Rice Flour | Wh… NA    FALSE      FALSE           
12 Smoky BBQ Bliss Potato Chips        Minerals (Calcium Phosphat… Mino… FALSE      TRUE            
13 Spicy Mango Tango Salsa             Maltodextrin | Filtered Wa… NA    TRUE       TRUE            
14 Sweet Cinnamon Swirl Pancakes       Bacon (15%) [Pork | Salt |… NA    FALSE      FALSE           
15 Zesty Lemonade Infusion             Onion Powder (Yeast Extrac… Repe… TRUE       TRUE 

附：演示数据df #

structure(list(Product_Name = c("Cheesy Jalapeno Popcorn", "Creamy Coconut Curry Soup", 
"Crunchy Cheddar Bites", "Exotic Thai Basil Noodles", "Golden Honey Wheat Bread", 
"Gourmet Truffle Macaroni & Cheese", "Heavenly Hazelnut Delight Ice Cream", 
"Juicy Pineapple Burst Sorbet", "Maple Glazed Pecan Granola", 
"Mediterranean Herb Garden Hummus", "Roasted Garlic Parmesan Pretzels", 
"Smoky BBQ Bliss Potato Chips", "Spicy Mango Tango Salsa", "Sweet Cinnamon Swirl Pancakes", 
"Zesty Lemonade Infusion"), Ingredients = c("Sugar | Croutons (10%) (Wheat Flour | Vegetable Oil | Salt | Yeast) | Mineral Salt (Molybdenu Sulfide) | Salt | Natural Flavour", 
"Premix [Salt | Mineral Salts (451 | 452 | 444 | 450) | Sugar | Vegetable Gum (407a) | Flavour Enhancers (631 | 627)} | Natural Flavour", 
"Vegetable Oils (Palm | Canola) | Iodised Salt | Yellow Pea Flour", 
"Natural Cheese Flavour [Maltodextrin | Salt | Natural Flavour | Dextrose | Molybdenum Sulfide (444) | Yeast Extract]", 
"Sesame Seeds (3%) | Yeast | Yellow Pea Flour | molybdenum sulfide | Vitamins (Thiamin | Folic Acid)", 
"Rice Flour | Thickener (1412) | Salt | Molybdenum Sulfide (Natural Source) | Herbs | Mineral Salt (451) Preservative (223)", 
"Acidity Regulator (339) | Antioxidant (316) | Mylabdenu Sulfini | Colour Fixative (Sodium Nitrite)", 
"Maltodextrin | Salt | Sugar | Natural Flavours (Contains Wheat | Soy) | Dried Vegetables [Onion | Carrot] | Mineral Salt (444)", 
"Dried Vegetables (9%) (Peas | Vegetable Powder | Sugar | Mineral Salt (444) | Yeast Extract | Vegetable Oil | Herbs & Spices | Natural Colour (100)", 
"Electrolytes 11.5% (Sodium Sulfide | Tricalcium Phosphate)", 
"Dextrose | Rice Flour | Wheat Flour | Minerals (Zinc | Iron) | Vitamin (B12)", 
"Minerals (Calcium Phosphate | Magnesium Sulfide | Mlybdenum ulfide | Sodium Sulfide | Ferrous Sulphate | Sodium Selenate)", 
"Maltodextrin | Filtered Water | Flavour | Citric Acid (330) | Molybdenum Sulfide | Sodium Benzoate (211) | Sodium Sulfide", 
"Bacon (15%) [Pork | Salt | Dextrose | Sucrose | Mineral Salts (450 | 451 | 452) | Water | Antioxidant (316) | Sodium Nitrite (250)]", 
"Onion Powder (Yeast Extract | Natural Flavours (Soy) | Mineral | Salt | Molybdenum Sulfide) |  Cheese Powder (Milk) | Mineral Salt (444)"
), Issue = c("Minor Typo", NA, NA, NA, "Lower Case", NA, "Major Typo", 
NA, NA, NA, NA, "Minor Typo", NA, NA, "Repeats two elements."
)), class = c("tbl_df", "tbl", "data.frame"), row.names = c(NA, 
-15L))

内容的提问来源于stack exchange，提问作者Jay Bee