pldatacli

A command-line tool for quick data analysis using Polars, with lazy execution for efficiency. Supports CSV, Parquet, and NDJSON files.

---

Tech Stack

• Polars – fast DataFrame engine with lazy execution for efficient data processing
• Typer – modern CLI framework for building command-line interfaces
• Rich – beautiful terminal rendering for clean table output
• PyYAML – YAML pipeline file parsing

---

PyPi Repository

Check the Repository on PyPI - https://pypi.org/project/pldatacli/

---

Installation

• Option 1: With pipx (Requires pipx to be installed)

pipx install pldatacli

• Option 2: with uv package manager (Requires uv to be installed)

uv tool install pldatacli

• Option 3: with homebrew tap (Requires homebrew installed)

brew tap utkarshg1/pldatacli
brew install pldatacli

---

Supported File Formats

Format	Extensions	Read	Write
CSV	.csv	✓	✓
Parquet	.parquet	✓	✓
NDJSON	.ndjson, .jsonl	✓	✓

All commands that accept a file input (query, schema, run, sql, pivot) support all three formats. Output (--output) is determined by the file extension you provide.

---

Commands

Command	Description
query	Filter, aggregate, sort, and explore a single file
schema	Inspect columns, dtypes, and null counts
run	Execute a multi-step pipeline defined in a YAML file
sql	Run arbitrary SQL queries against one or more files
pivot	Create pivot tables with flexible aggregations
init	Generate a boilerplate YAML pipeline file

⚡ Tip: query, pivot, and run all support --dry-run to preview the pipeline steps without loading or processing any data.

---

Usage

query — Exploratory Analysis

pldatacli query FILE [OPTIONS]

Example file:

SampleSuperstore.csv

---

Filter rows

Supported operators: =, !=, >, <, >=, <=

Single filter:

pldatacli query SampleSuperstore.csv \
  --filter "State = Texas"

Multiple filters:

pldatacli query SampleSuperstore.csv \
  --filter "State = Texas" \
  --filter "Category = Furniture"

Numeric filter:

pldatacli query SampleSuperstore.csv \
  --filter "Sales > 500"

---

Truncate date column

Truncate a date column into a period-based group. Format: col:period

Supported periods: year, quarter, month, week, day

This creates a new derived column named <col>_<period> that can be used in --groupby.

By month:

pldatacli query SampleSuperstore.csv \
  --truncate "Order Date:month"

By quarter:

pldatacli query SampleSuperstore.csv \
  --truncate "Order Date:quarter"

By year:

pldatacli query SampleSuperstore.csv \
  --truncate "Order Date:year"

---

Group by columns

Single column:

pldatacli query SampleSuperstore.csv \
  --groupby Region

Multiple columns:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --groupby Category

Group by truncated date column:

pldatacli query SampleSuperstore.csv \
  --truncate "Order Date:month" \
  --groupby "Order Date_month"

---

Aggregations

Supported aggregation operations for query: sum, mean, max, min, std, count, distinct_count

Single aggregation:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --agg "Profit:sum"

Multiple aggregations on one column:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --agg "Profit:sum,mean"

Multiple columns with aggregations:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --groupby Category \
  --agg "Sales:sum,mean" \
  --agg "Profit:sum"

---

Sorting

Single sort:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --agg "Profit:sum" \
  --sort "Profit_sum:desc"

Multiple sorts:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --agg "Profit:sum" \
  --sort "Region:asc" \
  --sort "Profit_sum:desc"

---

Rounding results

Round float columns to 2 digits:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --agg "Profit:mean" \
  --round 2

Custom rounding:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --agg "Profit:mean" \
  --round 4

---

Limiting rows

Head:

pldatacli query SampleSuperstore.csv \
  --head 5

Tail:

pldatacli query SampleSuperstore.csv \
  --tail 10

---

Save output to file

Save results as CSV:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --agg "Profit:sum" \
  --output result.csv

Save results as Parquet:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --agg "Profit:sum" \
  --output result.parquet

Save results as NDJSON:

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --agg "Profit:sum" \
  --output result.ndjson

⚡ Tip: Results are always printed to the terminal and saved to file simultaneously.

---

Generate YAML pipeline from query

Use --generate-yaml (or -y) to export the current query as a reusable YAML pipeline file. The results are still printed to the terminal as normal — the YAML is generated alongside.

pldatacli query SampleSuperstore.csv \
  --groupby Region \
  --agg "Profit:sum" \
  --generate-yaml my_pipeline.yaml

With a full query:

pldatacli query SampleSuperstore.csv \
  --filter "Region=West" \
  --truncate "Order Date:month" \
  --groupby "Order Date_month" \
  --groupby Category \
  --agg "Profit:sum,mean" \
  --agg "Sales:sum" \
  --sort "Profit_sum:desc" \
  --head 10 \
  --round 2 \
  --output monthly_west.csv \
  --generate-yaml monthly_west.yaml

The generated YAML can then be run directly with pldatacli run monthly_west.yaml.

⚡ Tip: Use --generate-yaml to capture ad-hoc queries as reproducible pipeline files for version control or reuse.

---

Dry run

Preview the pipeline steps that would be executed without loading or processing any data:

pldatacli query SampleSuperstore.csv \
  --filter "Sales > 500" \
  --groupby Region \
  --agg "Sales:sum" \
  --sort "Sales_sum:desc" \
  --dry-run

Example output:

=== DRY RUN: query ===
  Input file           : SampleSuperstore.csv
  Step 1: filter       → Sales > 500
  Step 2: truncate     → (none)
  Step 3: groupby      → Region
  Step 4: agg          → Sales:sum
  Step 5: sort         → Sales_sum:desc
  Step 6: limit        → (none)
  Step 7: round        → (none)
  Output               : (display only)
  Generate YAML        : (none)
No data was loaded or processed.

⚡ Tip: Use --dry-run to verify your pipeline arguments are correct before running on large files.

---

Full query example

pldatacli query SampleSuperstore.csv \
  --filter "Region=West" \
  --truncate "Order Date:month" \
  --groupby "Order Date_month" \
  --groupby Category \
  --agg "Profit:sum,mean" \
  --agg "Sales:sum" \
  --sort "Profit_sum:desc" \
  --head 10 \
  --round 2 \
  --output monthly_west.csv

---

schema — Inspect File Structure

Get columns, dtypes, and null counts without processing the full dataset:

pldatacli schema SampleSuperstore.csv

Example output:

LazyFrame Schema
┏━━━━━━━━━━━━━━┳━━━━━━━━━┳━━━━━━━┓
┃ Column       ┃ Dtype   ┃ Nulls ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━╇━━━━━━━┩
│ Ship Mode    │ String  │     0 │
│ Segment      │ String  │     0 │
│ Country      │ String  │     0 │
│ City         │ String  │     0 │
│ State        │ String  │     0 │
│ Postal Code  │ Int64   │     0 │
│ Region       │ String  │     0 │
│ Category     │ String  │     0 │
│ Sub-Category │ String  │     0 │
│ Sales        │ Float64 │     0 │
│ Quantity     │ Int64   │     0 │
│ Discount     │ Float64 │     0 │
│ Profit       │ Float64 │     0 │
└──────────────┴─────────┴───────┘
Rows: 9994, Columns: 13

⚡ Tip: Use schema before running queries to quickly inspect columns, types, and missing values.

---

run — YAML Pipelines

Execute a reusable, multi-step analysis defined in a YAML file:

pldatacli run query.yaml

Option 1 — Aggregation pipeline (query.yaml):

file: Superstore.csv
filter:
  - "Region=West"
truncate: "Order Date:month"
groupby:
  - "Order Date_month"
  - Category
agg:
  - "Profit:sum,mean"
  - "Sales:sum"
sort:
  - "Profit_sum:desc"
head: 10
round: 2
output: monthly_west.csv

Option 2 — Pivot pipeline (pivot.yaml):

file: Superstore.csv
truncate: "Order Date:month"
pivot:
  column: "Region"
  index: "Order Date_month"
  values: "Profit"
  aggregate: "sum"
round: 2
output: pivot_result.csv

index accepts a single string or a list of strings for multi-index pivots.

⚡ Tip: Store your YAML query files in version control alongside your data pipelines for reproducible analysis.

---

Dry run

Parse the YAML and preview what steps would be executed without loading or processing any data:

pldatacli run analysis.yaml --dry-run

Example output for a query pipeline:

=== DRY RUN: run ===
  YAML file            : analysis.yaml
  Input file           : Superstore.csv
  Output               : monthly_west.csv
  Pipeline type        : query
  Step 1: filter       → Region=West
  Step 2: truncate     → Order Date:month
  Step 3: groupby      → Order Date_month, Category
  Step 4: agg          → Profit:sum,mean
  Step 5: agg          → Sales:sum
  Step 6: sort         → Profit_sum:desc
  Step 7: limit        → head 10
  Step 8: round        → 2 decimal places
No data was loaded or processed.

⚡ Tip: Use --dry-run to validate a YAML pipeline before running it on large files. Config errors (missing file key, malformed pivot block, etc.) are caught and reported immediately.

---

sql — Ad-hoc SQL Queries

Run arbitrary Polars SQL queries against one or more files.

pldatacli sql FILE [FILES...] [OPTIONS]

The first file is always registered as the table data. Additional files are registered using their filename stem (lowercased, sanitized).

---

Single file query

pldatacli sql Superstore.csv \
  -q "SELECT Category, SUM(Profit) AS total_profit FROM data GROUP BY Category ORDER BY total_profit DESC"

---

Multi-file JOIN

pldatacli sql orders.csv customers.csv \
  -q "SELECT * FROM data o JOIN customers c ON o.customer_id = c.id LIMIT 100"

---

Load SQL from a file

pldatacli sql sales.parquet \
  --sql-file complex_query.sql

⚡ Tip: Use --sql-file to keep complex queries in .sql files for readability and reuse. You cannot use --sql and --sql-file together.

---

Limit output rows

pldatacli sql Superstore.csv \
  -q "SELECT * FROM data" \
  --head 20

pldatacli sql Superstore.csv \
  -q "SELECT * FROM data" \
  --tail 10

---

Save SQL results to file

pldatacli sql Superstore.csv \
  -q "SELECT Region, SUM(Sales) AS total_sales FROM data GROUP BY Region" \
  --output result.parquet

---

Full SQL example

pldatacli sql orders.csv customers.csv \
  --sql-file analysis.sql \
  --head 50 \
  --output joined_results.csv

---

pivot — Pivot Tables

Create spreadsheet-style pivot tables with flexible row groupings, column expansion, and aggregations.

pldatacli pivot FILE [OPTIONS]

--on defines the column whose unique values become new columns, --index sets the row grouping, and --values is the column to aggregate.

---

Basic pivot

pldatacli pivot SampleSuperstore.csv \
  --index Region \
  --on Category \
  --values Sales \
  --aggregate sum

---

Multi-index pivot

Group rows by multiple columns:

pldatacli pivot SampleSuperstore.csv \
  --index Region \
  --index "Ship Mode" \
  --on Category \
  --values Sales \
  --aggregate sum

---

Supported aggregations

Aggregation	Description
sum	Total (default)
mean	Average
min	Minimum value
max	Maximum value
first	First occurrence
last	Last occurrence
median	Median value
count	Count of non-null values
len	Row count

pldatacli pivot SampleSuperstore.csv \
  --index Region \
  --on "Ship Mode" \
  --values OrderID \
  --aggregate count

---

Pivot with date truncation

Combine --truncate with pivot to analyse trends over time:

pldatacli pivot SampleSuperstore.csv \
  --truncate "Order Date:month" \
  --index "Order Date_month" \
  --on Category \
  --values Sales \
  --aggregate sum

---

Rounding pivot results

pldatacli pivot SampleSuperstore.csv \
  --index Region \
  --on Category \
  --values Profit \
  --aggregate mean \
  --round 2

---

Save pivot output to file

pldatacli pivot SampleSuperstore.csv \
  --index Region \
  --on Category \
  --values Sales \
  --aggregate sum \
  --output pivot.csv

Save as Parquet:

pldatacli pivot sales.parquet \
  --index Region \
  --on "Ship Mode" \
  --values OrderID \
  --aggregate count \
  --output pivot.parquet

---

Generate YAML pipeline from pivot

Use --generate-yaml (or -y) to export the current pivot configuration as a reusable YAML pipeline file. The pivot results are still printed to the terminal as normal.

pldatacli pivot SampleSuperstore.csv \
  --index Region \
  --on Category \
  --values Sales \
  --aggregate sum \
  --generate-yaml my_pivot.yaml

With date truncation and output:

pldatacli pivot SampleSuperstore.csv \
  --truncate "Order Date:quarter" \
  --index "Order Date_quarter" \
  --index Region \
  --on Category \
  --values Revenue \
  --aggregate mean \
  --round 1 \
  --output quarterly_pivot.csv \
  --generate-yaml quarterly_pivot.yaml

The generated YAML can then be run directly with pldatacli run quarterly_pivot.yaml.

⚡ Tip: Use --generate-yaml to snapshot a pivot configuration into a YAML file for scheduled runs or sharing with teammates.

---

Dry run

Preview the pivot configuration that would be executed without loading or processing any data:

pldatacli pivot SampleSuperstore.csv \
  --index Region \
  --on Category \
  --values Sales \
  --aggregate sum \
  --dry-run

Example output:

=== DRY RUN: pivot ===
  Input file           : SampleSuperstore.csv
  Step 1: filter       → (none)
  Step 2: truncate     → (none)
  Step 3: pivot
    index              : Region
    on (column)        : Category
    values             : Sales
    aggregate          : sum
  Step 4: round        → (none)
  Output               : (display only)
  Generate YAML        : (none)
No data was loaded or processed.

⚡ Tip: Use --dry-run to confirm your pivot configuration before running it on large files.

---

Full pivot example

pldatacli pivot SampleSuperstore.csv \
  --truncate "Order Date:quarter" \
  --index "Order Date_quarter" \
  --index Region \
  --on Category \
  --values Revenue \
  --aggregate mean \
  --round 1 \
  --output quarterly_pivot.csv

---

init — Generate Boilerplate YAML

Generate a ready-to-edit YAML pipeline file with commented examples:

pldatacli init TEMPLATE [OPTIONS]

Accepts query or pivot as the template type. Will not overwrite an existing file — use --output to choose a different path.

---

Generate a query pipeline

pldatacli init query

Creates query.yaml:

file: data.csv
filter:
  - "Column=Value"
  - "Sales > 100"
truncate: "Order Date:month"   # periods: year, quarter, month, week, day
groupby:
  - "Order Date_month"
  - Category
agg:
  - "Profit:sum,mean"
  - "Sales:sum"
sort:
  - "Profit_sum:desc"
head: 10
round: 2
output: result.csv

---

Generate a pivot pipeline

pldatacli init pivot

Creates pivot.yaml:

file: data.csv
truncate: "Order Date:month"   # periods: year, quarter, month, week, day — optional
pivot:
  column: "Category"           # unique values of this column become new columns
  index:                       # row grouping — single string or list
    - "Order Date_month"
    - Region
  values: "Sales"              # column to aggregate
  aggregate: "sum"             # sum, mean, min, max, first, last, median, count, len
round: 2
output: pivot_result.csv

---

Custom output path

pldatacli init query --output my_analysis.yaml
pldatacli init pivot --output my_pivot.yaml

⚡ Tip: Run pldatacli init query or pldatacli init pivot at the start of a new analysis to get a fully commented template, then edit it and run with pldatacli run.

---

Version

Check the installed version:

pldatacli --version

or

pldatacli -v

---

Development

Clone the repo and install with dev dependencies:

git clone https://github.com/utkarshg1/pldatacli
cd pldatacli
uv sync --group dev

Run the test suite:

uv run pytest tests/ -v

The tests cover parsers, command logic, file loading (CSV/Parquet/NDJSON), and end-to-end YAML pipeline execution.