xsv（XSV 搜索）

xsv - CSV Command Line Toolkit Skill

You are a CSV data manipulation specialist using xsv, a fast command-line CSV toolkit written in Rust. This skill provides comprehensive guidance for processing, analyzing, and transforming CSV data efficiently.

Why xsv?

xsv is designed for high-performance CSV operations:

• Extremely fast: Rust-based, optimized for speed
• Low memory: Streaming operations when possible
• Rich features: 20+ commands for CSV manipulation
• Indexing: Create indexes for faster random access
• Composable: Unix philosophy - pipe commands together

Core Capabilities

1. Selection: select, headers, slice, sample
2. Searching: search, frequency
3. Analysis: stats, frequency, count
4. Transformation: fmt, fixlengths, flatten
5. Combination: join, cat
6. Sorting: sort
7. Display: table
8. Splitting: split
9. Performance: index

Quick Start

View CSV Data as Table

xsv table data.csv
xsv table data.csv | less -S

# Limit field width for readability
xsv table -c 20 data.csv

Count Rows

xsv count data.csv

View Headers

xsv headers data.csv

Select Columns

xsv select 1,3,5 data.csv
xsv select Name,Email,Age data.csv

Essential Commands

headers - View Column Names

# Show all column names with indices
xsv headers data.csv

# Output format:
# 1   Name
# 2   Email
# 3   Age

Use case: Understand CSV structure before operations

count - Count Records

# Count records (excluding header)
xsv count data.csv

# Count with index (much faster)
xsv count data.csv.idx

Performance: O(1) with index, O(n) without

select - Select Columns

# By index (1-based)
xsv select 1,3,5 data.csv

# By name
xsv select Name,Email,Age data.csv

# Column ranges
xsv select 1-4 data.csv
xsv select Name-Age data.csv

# From column to end
xsv select 3- data.csv

# Exclude columns
xsv select '!1-2' data.csv

# Reorder and duplicate
xsv select 3,1,2,1 data.csv

# Disambiguate duplicate column names
xsv select 'Name[0],Name[1],Name[2]' data.csv

# Quote names with special characters
xsv select '"Date - Opening","Date - Closing"' data.csv

Common options:

• -o, --output <file>: Write to file
• -n, --no-headers: Treat first row as data
• -d, --delimiter <char>: Input delimiter (default:,)

search - Filter Rows by Regex

# Basic search
xsv search "pattern" data.csv

# Case insensitive
xsv search -i "pattern" data.csv

# Search specific columns
xsv search -s Email "gmail.com" data.csv
xsv search -s 1,3,5 "pattern" data.csv

# Invert match (exclude matching rows)
xsv search -v "pattern" data.csv

# Save results
xsv search "active" data.csv -o active_users.csv

Use case: Filter CSV rows like grep

slice - Extract Row Ranges

# First 10 rows
xsv slice -l 10 data.csv

# Rows 100-200
xsv slice -s 100 -e 200 data.csv

# Start at row 50, take 20 rows
xsv slice -s 50 -l 20 data.csv

# Last 10 rows (requires index)
xsv count data.csv  # Get total
xsv slice -s -10 data.csv

# Single row
xsv slice -i 42 data.csv

Performance: Much faster with index

stats - Compute Statistics

# Basic stats (mean, min, max, stddev)
xsv stats data.csv

# All statistics (includes median, mode, cardinality)
xsv stats --everything data.csv

# Specific columns
xsv stats -s Age,Salary data.csv

# Include median (requires memory)
xsv stats --median data.csv

# Include mode
xsv stats --mode data.csv

# Include cardinality (unique count)
xsv stats --cardinality data.csv

# Parallel processing
xsv stats -j 4 data.csv

# Output as table
xsv stats data.csv | xsv table

Output fields: field, type, sum, min, max, mean, stddev, median, mode, cardinality

frequency - Frequency Tables

# Top 10 values per column
xsv frequency data.csv

# Specific columns
xsv frequency -s Status,Category data.csv

# Top 20 values
xsv frequency -l 20 data.csv

# All values (no limit)
xsv frequency -l 0 data.csv

# Ascending order
xsv frequency --asc data.csv

# Exclude nulls
xsv frequency --no-nulls data.csv

# View as table
xsv frequency -s Status data.csv | xsv table

Output: field, value, count

Use case: Value distribution analysis

sort - Sort Records

# Sort by first column
xsv sort data.csv

# Sort by specific columns
xsv sort -s Age data.csv
xsv sort -s LastName,FirstName data.csv

# Numeric sort
xsv sort -s Age -N data.csv

# Reverse order
xsv sort -s Age -R data.csv

# Numeric + reverse
xsv sort -s Salary -N -R data.csv

# Save sorted
xsv sort -s Name data.csv -o sorted.csv

Note: Requires reading entire file into memory

join - Join Two CSV Files

# Inner join
xsv join ID users.csv ID orders.csv

# Left outer join
xsv join --left ID users.csv ID orders.csv

# Right outer join
xsv join --right ID users.csv ID orders.csv

# Full outer join
xsv join --full ID users.csv ID orders.csv

# Case-insensitive join
xsv join --no-case Email users.csv Email contacts.csv

# Join on multiple columns
xsv join 'ID,Date' file1.csv 'ID,Date' file2.csv

# Include nulls in join
xsv join --nulls ID file1.csv ID file2.csv

# Cross join (cartesian product)
xsv join --cross 1 file1.csv 1 file2.csv

# Save result
xsv join ID users.csv ID orders.csv -o joined.csv

Join types:

• Default: Inner join (intersection)
• --left: Left outer join
• --right: Right outer join
• --full: Full outer join
• --cross: Cartesian product (use with caution)

table - Format as Aligned Table

# Basic table
xsv table data.csv

# With pager
xsv table data.csv | less -S

# Minimum column width
xsv table -w 10 data.csv

# Padding between columns
xsv table -p 4 data.csv

# Limit field length
xsv table -c 20 data.csv

# Combine limits
xsv slice -l 50 data.csv | xsv table -c 30

Note: Requires buffering entire file into memory

sample - Random Sampling

# Sample 100 rows
xsv sample 100 data.csv

# Sample 10% of large file
xsv count data.csv  # e.g., 1000000
xsv sample 100000 data.csv

# Save sample
xsv sample 1000 large.csv -o sample.csv

# Sample then analyze
xsv sample 10000 huge.csv | xsv stats --everything

Performance: Uses indexing for samples <10% of total

fmt - Format Output

# Convert to TSV
xsv fmt -t '\t' data.csv -o data.tsv

# Convert to pipe-delimited
xsv fmt -t '|' data.csv

# Add CRLF line endings
xsv fmt --crlf data.csv -o windows.csv

# Quote all fields
xsv fmt --quote-always data.csv

# Custom quote character
xsv fmt --quote "'" data.csv

# Custom escape character
xsv fmt --escape '\\' data.csv

cat - Concatenate Files

# Concatenate by rows (vertically)
xsv cat rows file1.csv file2.csv file3.csv

# Concatenate by columns (horizontally)
xsv cat columns file1.csv file2.csv

# Pad with empty values if different lengths
xsv cat rows-columns file1.csv file2.csv

split - Split Into Multiple Files

# Split into files of 1000 rows each
xsv split -s 1000 output_dir data.csv

# Creates: output_dir/0.csv, output_dir/1.csv, etc.

flatten - Show One Field Per Line

# Flatten first record
xsv slice -i 0 data.csv | xsv flatten

# Output format:
# field,value
# Name,John Doe
# Email,john@example.com
# Age,30

fixlengths - Fix Inconsistent Row Lengths

# Ensure all rows have same number of fields
xsv fixlengths data.csv -o fixed.csv

# Pads short rows with empty fields
# Useful for malformed CSVs

index - Create Index for Performance

# Create index
xsv index data.csv

# Creates: data.csv.idx

# Now operations are faster:
xsv count data.csv      # O(1) instead of O(n)
xsv slice -i 1000 data.csv  # Direct access
xsv sample 100 data.csv     # Fast random access

When to index:

• Large files (>100MB)
• Multiple operations on same file
• Random access patterns (slice, sample)
• Statistics with parallel processing

Common Workflows

Workflow 1: Data Exploration

# 1. Understand structure
xsv headers data.csv

# 2. Count records
xsv count data.csv

# 3. View sample
xsv slice -l 10 data.csv | xsv table

# 4. Get statistics
xsv stats data.csv | xsv table

# 5. Check value distributions
xsv frequency -s Status data.csv | xsv table

Workflow 2: Data Filtering and Selection

# 1. Select relevant columns
xsv select Name,Email,Age,Status data.csv |

# 2. Filter active users
xsv search -s Status "active" |

# 3. Filter by age
xsv search -s Age "^[3-9][0-9]$" |

# 4. Save result
xsv -o active_users_30plus.csv

Workflow 3: Data Analysis Pipeline

# 1. Create index for performance
xsv index large_data.csv

# 2. Sample for quick analysis
xsv sample 10000 large_data.csv |

# 3. Select columns of interest
xsv select Revenue,Region,Product |

# 4. Get statistics
xsv stats --everything |

# 5. View as table
xsv table

Workflow 4: Data Joining

# 1. Join users with orders
xsv join UserID users.csv UserID orders.csv |

# 2. Select relevant columns
xsv select 'UserName,Email,OrderID,OrderDate,Amount' |

# 3. Sort by amount
xsv sort -s Amount -N -R |

# 4. Top 100 orders
xsv slice -l 100 |

# 5. Format and save
xsv table -o top_orders.txt

Workflow 5: Data Cleaning

# 1. Fix row lengths
xsv fixlengths messy.csv |

# 2. Select valid columns
xsv select 1-10 |

# 3. Remove rows with empty email
xsv search -s Email '.+' |

# 4. Sort and deduplicate (using uniq)
xsv sort -s Email |
uniq |

# 5. Save cleaned data
xsv -o cleaned.csv

Workflow 6: Data Transformation

# 1. Select and reorder columns
xsv select 'LastName,FirstName,Email,Phone' data.csv |

# 2. Convert to TSV
xsv fmt -t '\t' |

# 3. Save
xsv -o output.tsv

Workflow 7: Large File Processing

# 1. Create index first
xsv index huge_file.csv

# 2. Get quick count
xsv count huge_file.csv

# 3. Sample for analysis
xsv sample 50000 huge_file.csv |

# 4. Analyze sample
xsv stats --everything |

# 5. View results
xsv table

Performance Tips

1. Create Indexes for Large Files

# One-time cost, speeds up many operations
xsv index large.csv

Speeds up:

• count (O(1) instead of O(n))
• slice (direct access)
• sample (efficient random access)
• stats -j (parallel processing)

2. Use Streaming Operations

These don't require reading entire file into memory:

• select
• search
• slice (with index)
• headers
• count (with index)

3. Avoid Memory-Intensive Operations on Large Files

These require full file in memory:

• sort
• table
• stats --median
• stats --mode
• frequency

Solution: Use sample or slice first:

xsv sample 100000 huge.csv | xsv stats --everything

4. Parallel Processing

# Use multiple cores for stats
xsv stats -j 0 data.csv  # Auto-detect CPUs

# Specific job count
xsv stats -j 4 data.csv

Requires: Indexed file for best performance

5. Chain Commands Efficiently

# Good: streaming pipeline
xsv select Name,Age data.csv | xsv search -s Age "^[3-9]" | xsv table

# Less efficient: multiple file reads
xsv select Name,Age data.csv -o temp1.csv
xsv search -s Age "^[3-9]" temp1.csv -o temp2.csv
xsv table temp2.csv

Advanced Patterns

Pattern 1: Top N Analysis

# Top 10 customers by revenue
xsv sort -s Revenue -N -R customers.csv | xsv slice -l 10 | xsv table

Pattern 2: Conditional Aggregation

# Count by status
xsv frequency -s Status -l 0 data.csv | xsv table

# Average age by region (requires external tools)
xsv select Region,Age data.csv | xsv sort -s Region | ...

Pattern 3: Multi-Column Search

# Search across multiple columns
xsv select Name,Email,Phone data.csv | xsv search "pattern"

Pattern 4: Data Validation

# Find rows with missing email
xsv search -s Email -v '.+' data.csv

# Find duplicates (by email)
xsv select Email data.csv | xsv sort | uniq -d

Pattern 5: Data Comparison

# Find differences between two files
xsv select ID,Value file1.csv > temp1
xsv select ID,Value file2.csv > temp2
diff temp1 temp2

Pattern 6: Column Statistics

# Stats for specific column
xsv select Age data.csv | xsv stats | xsv table

# Multiple column stats
xsv select Age,Salary,Score data.csv | xsv stats --everything | xsv table

Common Options

Most commands support these options:

-h, --help              Display help
-o, --output <file>     Write to file instead of stdout
-n, --no-headers        First row is data, not headers
-d, --delimiter <char>  Input delimiter (default: ,)

Delimiter Support

Reading Different Formats

# TSV (tab-separated)
xsv select 1,3 -d '\t' data.tsv

# Pipe-delimited
xsv select Name,Age -d '|' data.txt

# Semicolon-delimited
xsv select 1-5 -d ';' data.csv

Converting Formats

# TSV to CSV
xsv fmt -d '\t' data.tsv -o data.csv

# CSV to TSV
xsv fmt -t '\t' data.csv -o data.tsv

# CSV to pipe-delimited
xsv fmt -t '|' data.csv -o data.txt

Error Handling

Common Issues

Issue: "CSV error: record has different length"

Solution: Use fixlengths

xsv fixlengths data.csv -o fixed.csv

Issue: "No such file or directory"

Solution: Check file path, use absolute paths if needed

Issue: Out of memory with large file

Solution: Use sampling or indexing

xsv index large.csv
xsv sample 10000 large.csv | xsv stats

Issue: Column name not found

Solution: Check headers first

xsv headers data.csv

Integration with Other Tools

With jq (for CSV→JSON)

# Convert CSV to JSON
xsv select Name,Age data.csv | xsv fmt -t ',' | \
  python -c 'import csv, json, sys; print(json.dumps([dict(r) for r in csv.DictReader(sys.stdin)]))'

With awk

# Add computed column
xsv select Price,Quantity data.csv | \
  awk -F, 'NR==1{print $0",Total"} NR>1{print $0","$1*$2}'

With sort/uniq

# Deduplicate by column
xsv select Email data.csv | sort | uniq

With grep

# Pre-filter before xsv
cat data.csv | grep "pattern" | xsv table

Quick Reference

# View structure
xsv headers data.csv
xsv count data.csv
xsv slice -l 5 data.csv | xsv table

# Select columns
xsv select 1,3,5 data.csv
xsv select Name,Email data.csv

# Filter rows
xsv search "pattern" data.csv
xsv search -s Email "gmail" data.csv

# Statistics
xsv stats data.csv
xsv frequency -s Status data.csv

# Sort
xsv sort -s Age -N data.csv

# Join
xsv join ID file1.csv ID file2.csv

# Format
xsv table data.csv
xsv fmt -t '\t' data.csv

# Sample
xsv sample 1000 data.csv

# Index (for performance)
xsv index large.csv

Comparison with xlsx

While xlsx handles Excel files, xsv is specialized for CSV:

When to use xsv:

• Working with CSV data
• Need maximum performance
• Large file processing
• Statistical analysis
• Data pipelines

When to use xlsx:

• Excel file format required
• Need formulas and formatting
• Multiple sheets
• Cell-level operations

Summary

Primary tool: xsv for fast CSV processing

Most common commands:

• xsv headers - Understand structure
• xsv select - Choose columns
• xsv search - Filter rows
• xsv stats - Analyze data
• xsv table - View formatted
• xsv join - Combine files
• xsv index - Speed up operations

Key advantages:

• Blazing fast (Rust-based)
• Composable (Unix pipes)
• Low memory (streaming)
• Rich analysis features
• Index support for large files

Best practices:

1. Index large files first
2. Use sampling for quick exploration
3. Chain commands with pipes
4. Check headers before operations
5. Use appropriate output formats