PDF解析场景：对含条件表达式的函数使用Mapply的技术问询

Hey there! Let’s walk through how to use mapply with conditional expressions for your PDF data extraction workflow—sounds like you’ve got a solid setup with that CSV parameter map, so let’s make it work seamlessly with mapply.

Step 1: Prep Your CSV Data #

First, let’s get your CSV structured into a format mapply can easily work with. Let’s assume your CSV looks like this (with parameters as rows, hierarchical document types as columns, and identifiers in cells):

Parameter,type1.subtypeA,type2.subtypeB.s_subtypeX,type2.subtypeC
CustomerID,CUST-ID,Customer Number,Cust#
InvoiceDate,Inv Date,Invoice Dt,Inv. Date

Use this code to load and reshape the data:

library(tidyverse)

# Load your parameter CSV
param_df <- read_csv("your_parameter_file.csv")

# Extract core components
params <- param_df$Parameter
doc_types <- colnames(param_df)[-1]  # Grab all document type columns

# Create a full list of parameter + document type pairs
param_doc_pairs <- expand.grid(
  Parameter = params,
  DocType = doc_types,
  stringsAsFactors = FALSE
)

# Match each pair to its corresponding identifier from the CSV
param_doc_pairs$Identifier <- mapply(
  function(param, doc_type) param_df[param_df$Parameter == param, doc_type],
  param_doc_pairs$Parameter,
  param_doc_pairs$DocType
)

Step 2: Build Your Conditional Extraction Function #

Next, write a function that takes a parameter name, its identifier, and the hierarchical document type, then runs conditional logic based on the document’s structure. Replace the placeholder code with your actual PDF scanning/extraction logic (using tools like tesseract or pdftools):

extract_pdf_data <- function(param_name, identifier, doc_type) {
  # Split the document type into its hierarchical layers
  doc_hierarchy <- strsplit(doc_type, "\\.")[[1]]
  main_type <- doc_hierarchy[1]
  sub_type <- ifelse(length(doc_hierarchy) >= 2, doc_hierarchy[2], NA)

  # Clean conditional logic using case_when (easier to read than nested if-else)
  result <- case_when(
    main_type == "type1" ~ paste0("Extracted ", param_name, " (using '", identifier, "') from type1 document"),
    main_type == "type2" && sub_type == "subtypeB" ~ paste0("Extracted ", param_name, " (using '", identifier, "') from type2.subtypeB document"),
    main_type == "type2" && sub_type == "subtypeC" ~ paste0("Extracted ", param_name, " (using '", identifier, "') from type2.subtypeC document"),
    TRUE ~ paste0("Fallback extraction for ", param_name, " in ", doc_type, " (identifier: ", identifier, ")")
  )

  # Replace the above with your real PDF extraction code! Example:
  # pdf_path <- file.path("pdf_storage", doc_type, "target_document.pdf")
  # scanned_text <- tesseract::ocr(pdf_path)
  # result <- stringr::str_extract(scanned_text, identifier)

  return(result)
}

Step 3: Run with mapply #

Now use mapply to execute your extraction function across all parameter-document pairs. mapply will iterate over each element of your input vectors and pass them to the function in sync:

# Run the extraction across all pairs
extraction_results <- mapply(
  extract_pdf_data,
  param_name = param_doc_pairs$Parameter,
  identifier = param_doc_pairs$Identifier,
  doc_type = param_doc_pairs$DocType,
  SIMPLIFY = FALSE  # Keep results as a list (better for unstructured data)
)

# Combine pairs with results into a tidy data frame for easy analysis
final_results <- cbind(param_doc_pairs, ExtractedResult = unlist(extraction_results))

Key Tips #

• Handle Missing Identifiers: Add a check like if(is.na(identifier)) { return("No identifier found") } to skip or log cases where a parameter doesn’t have an identifier for a document type.
• Speed Up with Parallel Processing: If you’re processing hundreds of PDFs, use parallel::mclapply (Linux/macOS) or furrr::future_map (cross-platform) instead of mapply to run tasks in parallel.
• Deep Hierarchies: For document types with 3+ sublayers, extend the doc_hierarchy parsing (e.g., s_subtype <- doc_hierarchy[3]) and add more conditions to case_when.

内容的提问来源于stack exchange，提问作者jovianlynxdroid