tabula-cli 2.0.0

Tabular data manipulation, statistics and plotting on the command line

Tabula

Process tabular data on the command line, apply statistical tests and create plots. Tabula is a command-line tool for manipulating and analyzing tabular data. It allows you to perform complex data operations using a simple expression syntax directly in your terminal.

Installation

Install Tabula using pip:

pip install tabula-cli

The three main components of Tabula are:

• tabula: For data manipulation and analysis
• tabula-plot: For creating plots from tabular data
• tabula-stats: For running statistical tests

Tabula

Tabula provides a chain-based syntax for data manipulation operations. Methods can be chained together using dot notation: method1().method2().method3(). One can perform operations like selecting columns, filtering rows, transforming data, and aggregating results all on the command line.

Data Selection Methods

select(col1, col2, ...)

Select specific columns from the dataset.

# Select single column
tabula -p "select(name)" data.csv

# Select multiple columns
tabula -p "select(name, age, salary)" data.csv

Data Transformation Methods

upper(col)

Convert text in specified column to uppercase.

tabula -p "select(name).upper(name)" data.csv

lower(col)

Convert text in specified column to lowercase.

tabula -p "select(name).lower(name)" data.csv

strlen(col)

Calculate the length of strings in specified column.

tabula -p "select(name).strlen(name)" data.csv

round(col, decimals)

Round numeric values to specified decimal places.

tabula -p "select(salary).round(salary, 2)" data.csv

Filtering Methods

where(condition)

Filter rows based on conditions. Supports comparison operators and logical operators.

# Simple condition
tabula -p "where(age > 30)" data.csv

# Multiple conditions with AND
tabula -p "where(age > 25 & salary >= 50000)" data.csv

# Multiple conditions with OR
tabula -p "where(department == 'IT' | department == 'HR')" data.csv

# Complex conditions with parentheses
tabula -p "where((age > 30 & department == 'IT') | salary < 40000)" data.csv

Data Limiting Methods

head(n)

Return the first n rows (default: 5).

tabula -p "head(10)" data.csv

tail(n)

Return the last n rows (default: 5).

tabula -p "tail(3)" data.csv

Sorting Methods

sortby(col, descending=False)

Sort data by specified column.

# Ascending sort
tabula -p "sortby(age)" data.csv

# Descending sort
tabula -p "sortby(salary, True)" data.csv

Aggregation Methods (Terminal)

count()

Count the number of rows.

tabula -p "count()" data.csv
tabula -p "where(age > 30).count()" data.csv

min(col), max(col), sum(col)

Calculate minimum, maximum, or sum of a column.

tabula -p "min(age)" data.csv
tabula -p "max(salary)" data.csv
tabula -p "sum(salary)" data.csv

mean(col), median(col), mode(col)

Calculate statistical measures.

tabula -p "mean(salary)" data.csv
tabula -p "median(age)" data.csv

std(col), var(col)

Calculate standard deviation and variance.

tabula -p "std(salary)" data.csv
tabula -p "var(age)" data.csv

first(col), last(col)

Get first or last value from a column.

tabula -p "first(name)" data.csv
tabula -p "last(name)" data.csv

Unique Value Methods

uniq(col)

Get unique values from a column.

tabula -p "uniq(department)" data.csv

uniqc(col)

Count unique values (group by and count).

tabula -p "uniqc(department)" data.csv

String Methods

strjoin(col, separator)

Join all values in a column with a separator.

tabula -p "strjoin(name, ', ')" data.csv

Utility Methods

columns()

List all column names.

tabula -p "columns()" data.csv

Complete Example Workflow

# Sample data.csv:
# name,age,salary,department
# Alice,25,50000,HR
# Bob,30,60000,IT
# Charlie,35,70000,Finance
# David,40,80000,IT

# Complex analysis: Find IT employees over 30, show their names and salaries, sorted by salary
tabula -p "where(department == 'IT' & age > 30).select(name, salary).sortby(salary)" data.csv

# Output:
# name,salary
# Bob,60000
# David,80000

Method Chaining Rules

1. Terminal Methods: Methods like count(), sum(), min(), max() must be the last in the chain
2. Column Selection: Use select() before applying column-specific operations
3. Filtering: where() conditions support parentheses for complex logic
4. String Operations: Methods like upper(), lower(), strlen() work on text columns

Output Formats

Use the -o flag to specify output format:

• --outtype polars: Default table format
• --outtype csv: CSV format
• --outtype tsv: Tab-separated values

tabula -p "select(name, age)"

tabula-plot

tabula-plot allows you to create plots from tabular data directly in the terminal. It's an ideal companion for command-line data processing tools like awk and grep.

Basic Usage

# Plot from a CSV file
tabula-plot data.csv -p "plot:relplot,kind:scatter,x:col1,y:col2,hue:col3"

# Plot from stdin (pipe data)
cat data.csv | tabula-plot - -p "plot:relplot,kind:scatter,x:col1,y:col2"

# Save plot to a file
tabula-plot data.csv -p "plot:relplot,kind:scatter,x:col1,y:col2" -o plot.png

# Specify column names if they're not in the first row
tabula-plot data.csv -p "plot:relplot,kind:scatter,x:col1,y:col2" -c "col1,col2,col3"

# Use a different separator for CSV data
tabula-plot data.tsv -p "plot:relplot,kind:scatter,x:col1,y:col2" -s "\t"

Supported Plot Types

• relplot (default): Scatter and line plots
  • kinds: scatter, line
• catplot: Categorical plots
  • kinds: strip (default), swarm, box, violin, boxen, point, bar, count
• displot: Distribution plots
  • kinds: hist (default), kde, ecdf
• pairplot: Pairwise relationships in dataset

Script Format

The plotting script uses a simple key:value format:

plot:plot_type,kind:plot_kind,x:x_column,y:y_column,hue:color_column,...

For example:

plot:catplot,kind:violin,x:category,y:value,hue:group

tabula-stats

Statsmodels on the command line - a powerful CLI for running statistical tests directly in your terminal.

Features

• Simple CLI interface: Run statistical tests without writing Python code
• Multiple test types: OLS, Logistic Regression, t-tests, ANOVA
• Input flexibility: Works with CSV files or piped stdin data
• Integration with Unix tools: Pairs perfectly with awk, grep, sed, jq, etc.

Usage

# Basic syntax
tabula-stats [INPUT_FILE] -p "test:TYPE,PARAM1:VALUE1,PARAM2:VALUE2"

# Example: OLS regression on data.csv
tabula-stats data.csv -p "test:ols,dependent:y,independent:x+z"

# Read from stdin
cat data.csv | tabula-stats -p "test:ttest,sample1:group1,sample2:group2"

Supported Tests

Ordinary Least Squares (OLS) Regression

tabula-stats data.csv -p "test:ols,dependent:y,independent:x+z+w"

Required parameters:

• dependent: The dependent variable column
• independent: Formula for independent variables (e.g., x+z+w or x*z)

Logistic Regression

tabula-stats data.csv -p "test:logit,dependent:binary_outcome,independent:x+z"

Required parameters:

• dependent: The binary dependent variable column
• independent: Formula for independent variables

Generalized Linear Models (GLM)

tabula-stats data.csv -p "test:glm,dependent:y,independent:x+z,family:poisson,link:log"

Required parameters:

• dependent: The dependent variable column
• independent: Formula for independent variables
• family: Distribution family - one of:
  • gaussian: For continuous data (normal distribution)
  • binomial: For binary data (0/1)
  • poisson: For count data
  • gamma: For positive continuous data with variance proportional to square of mean
  • inverse_gaussian: For positive continuous data
  • neg_binomial: For overdispersed count data
  • tweedie: For compound Poisson-gamma distribution

Optional parameters:

• link: Link function - depends on the family, common options include:
  • identity: No transformation (default for Gaussian)
  • log: Log transformation (default for Poisson and Gamma)
  • logit: Logit transformation (default for Binomial)
  • probit: Probit transformation
  • cloglog: Complementary log-log transformation
  • inverse: Inverse transformation
  • power: Power transformation
• alpha: Alpha parameter for NegativeBinomial family (default 1.0)
• var_power: Variance power for Tweedie family (default 1.5)
• power: Power parameter for Power link function (default 1.0)

Examples:

# Poisson regression with log link
tabula-stats data.csv -p "test:glm,dependent:count,independent:x+z,family:poisson"

# Gamma regression with log link
tabula-stats data.csv -p "test:glm,dependent:duration,independent:x+z,family:gamma"

# Binomial regression with probit link
tabula-stats data.csv -p "test:glm,dependent:success,independent:x+z,family:binomial,link:probit"

Two-sample t-test

tabula-stats data.csv -p "test:ttest,sample1:group1,sample2:group2,alternative:two-sided"

Required parameters:

• sample1: First sample column name
• sample2: Second sample column name

Optional parameters:

• alternative: Test type ('two-sided', 'larger', or 'smaller')

ANOVA

tabula-stats data.csv -p "test:anova,formula:y ~ C(group)"

Examples

CSV with headers

$ cat data.csv
x,y,group
1,3.4,A
2,5.7,A
3,6.3,B
4,8.1,B

$ tabula-stats data.csv -p "test:ols,x:y,y:x"

CSV without headers

$ cat data_no_header.csv
1,3.4,A
2,5.7,A
3,6.3,B
4,8.1,B

$ tabula-stats data_no_header.csv -p "test:ols,x:y,indepenydent:x" 

Saving output to file

$ tabula-stats data.csv -p "test:ols,dependent:y,independent:x" > results.txt