如何在R中不加载RDS文件获取其行列数？

Great question! You absolutely can retrieve the row and column counts of an RDS file without loading the entire dataset into memory—this is a total lifesaver for large files where loading everything would hog your RAM or crash your session. Below are a few reliable methods, depending on your needs:

Method 1: Base R (Quick Column Count, Limited Row Count Support) #

If you know your RDS file stores a data frame/tibble, you can grab the column count by reading just the structure (0 rows) of the dataset. This avoids loading any actual data:

# Get column count
col_count <- ncol(readRDS("your_file.rds", nrows = 0))

Note: Base R doesn't have a built-in way to get the row count without loading at least part of the data. For row counts, you'll want to use one of the package-based methods below.

Method 2: Arrow Package (Recommended for Most Cases) #

The arrow package provides efficient, memory-safe tools to work with RDS files (and many other data formats). It can read metadata directly without loading the full dataset:

1. First install the package if you haven't:

install.packages("arrow")

2. Use these commands to get your dimensions:

library(arrow)

# Read the RDS as an Arrow dataset (no full data loaded)
rds_data <- read_rds("your_file.rds", as_data_frame = FALSE)

# Extract row and column counts
row_count <- nrow(rds_data)
col_count <- ncol(rds_data)

# Print results
cat("Total rows:", row_count, "\nTotal columns:", col_count)

This works seamlessly with data frames, tibbles, and other tabular structures, and it's optimized for large files.

Method 3: rds Package (Lightweight RDS-Specific Tool) #

If you want a minimal, RDS-focused solution, the rds package lets you read the file's header metadata directly:

1. Install the package:

install.packages("rds")

2. Extract dimensions from the header:

library(rds)

# Read the RDS file header
rds_header <- rds_read_header("your_file.rds")

# Check if the object is a data frame, then get dimensions
if (rds_header$object_type == "data.frame") {
  dimensions <- rds_header$attributes$dim
  row_count <- dimensions[1]
  col_count <- dimensions[2]
  cat("Rows:", row_count, "\nColumns:", col_count)
}

This is perfect if you only need metadata and want to avoid heavier packages.

Important Note #

All these methods assume your RDS file contains a tabular object (like a data frame or tibble). If your RDS stores non-tabular data (e.g., a list, machine learning model, or custom object), the "row/column count" concept doesn't apply—you'll need to first check the object type via the header metadata before proceeding.

内容的提问来源于stack exchange，提问作者Joer