ParVu - Parquet Viewer

Overview

A powerful desktop application for viewing and querying large Parquet, CSV, and JSON files. Built with PyQt6 and DuckDB for efficient data handling.

Features

File Support

• Parquet files (.parquet)
• CSV files (.csv)
• JSON files (.json)
• Lazy loading - Files are queried directly without loading into memory
• Efficient handling of huge files (8GB+) with pagination

SQL Querying

• Full SQL query support via DuckDB
• Syntax highlighting for SQL keywords
• Auto-completion for SQL keywords and column names
• Smart query validation

Expression Language (DSL)

• Dual-mode editor — switch between raw SQL and expression mode
• Column assignment — data[new_col] = old_col * 2 adds computed columns
• Built-in functions — ABS, UPPER, LEN, ROUND, REPLACE, DROP_DUPLICATES, DROP_NULL, etc.
• Auto-completion for functions, columns, table names, and methods in expression mode
• Function documentation popup — hover completer items or click function names to see signature, parameters, examples, and SQL mapping
• Seamless compilation to DuckDB SQL behind the scenes

Table Operations

• Pagination - Browse large datasets efficiently (configurable rows per page)
• Double-click editing - Edit cell values in the current view
• Applied steps - Every transform, sort, filter, and cell edit is tracked
• Save commits transforms - Ctrl+S writes applied transforms (sort, drop dup, drop null, replace, math) back to the file alongside cell edits; Applied Steps clears after save
• Undo - Revert the last applied step (SQL transform or cell edit)
• Drop duplicates - Remove duplicate rows via Operations menu or expression
• Drop null values - Drop rows where a column IS NULL, or matches a user-supplied sentinel like -1 / '' / N/A
• Replace values - Replace text/regex in columns via Operations menu or expression
• Column sorting - Click any column header to sort (toggles asc/desc); sort indicator shown in header
• Column statistics - Right-click a header → Column Statistics: rows, non-null, null %, distinct, min, max, mean, std
• Unique values filter - Excel-like dropdown showing unique column values
• Copy operations - Copy cell selection as TSV (Ctrl+C), CSV, or Markdown; copy column values as Python tuple
• Drag-and-drop - Drop files onto the window to open each as a new tab
• Large file warnings - Warns when calculating unique values on files >1M rows

Search

• Search dialog (Ctrl+F) — finds matches across every page, not just the visible one. Pick a specific column or search all. Match case / Whole cell / Regex options. Next/Prev step one match at a time and cache visited matches for instant re-navigation. Selecting a match paginates to the right page and selects the cell.

Export Options

• Export query results to:
  • CSV format
  • Parquet format
  • JSON format

User Interface

• Clean, modern PyQt6 interface
• Theme System - 3 built-in themes (Light, Excel, ParVu Black)
• Internationalization (i18n) - 3 languages: English, Russian, Azerbaijani
• Operations menu — Math, Join, Append, Drop Duplicates, Drop Null Values, and Replace Values
• Edit menu — Find / Find Next / Find Previous
• Keyboard shortcut cheatsheet — Ctrl+/ shows every shortcut grouped by File / Table / Find / Help
• Collapsible applied steps — Toggle the steps panel to save screen space
• Exit warning — Warns about unsaved transforms with Save & Close option
• Import/Export custom themes
• Customizable colors, fonts, and layouts
• Recent files history
• Keyboard shortcuts and auto-complete

Installation

For End Users

Download pre-built packages:

• Linux (Ubuntu/Debian): Download the .deb package

  sudo dpkg -i ParVu-<version>-amd64.deb
  parvu

• Windows: Download ParVu-<version>-setup.exe (installer with desktop shortcut + .parquet/.csv/.json file associations) or ParVu-<version>-portable.zip (extract anywhere, run parvu.exe)
• See RELEASES.md for installation guide and docs/BUILD_WINDOWS.md for Windows build details

For Developers

# Clone the repository
git clone https://github.com/AzizNadirov/ParVu.git
cd ParVu

# Install dependencies using uv (recommended)
uv sync

# Run from source
uv run python src/app.py

Build distributable packages:

# Linux
./build.sh

# Windows
.\build.ps1

See BUILDING.md for complete build guide.

How to Use

Launch Application

# Using uv
uv run python src/app.py

# Or directly with file
uv run python src/app.py path/to/your/file.parquet

Basic Workflow

1. Load File - Click 'Browse' to select a Parquet, CSV, or JSON file
2. View Data - Table displays paginated results automatically
3. Run Queries - Write SQL in the editor and click 'Execute'
   • Auto-completion appears after typing 2+ characters
   • Press Tab or Enter to accept suggestions
4. Navigate - Use Previous/Next buttons to browse pages
5. Expression Mode - Switch the editor to expression mode and write:
   • data[discount] = price * 0.1 to add a computed column
   • DROP_DUPLICATES(data[id], data[name], 'first') to deduplicate
   • Press Execute to compile and run
6. Column Operations - Right-click column headers for:
   • Copy column name
   • Sort ascending/descending (or just click the header)
   • Copy values as tuple
   • Show unique values (with search and filter)
   • Column Statistics — rows, non-null, null %, distinct, min, max, mean, std
   • Drop Null Values — drop rows where this column is NULL (or matches a sentinel like -1)
7. Cell Operations - Right-click selected cells for:
   • Copy (TSV), Copy as CSV, Copy as Markdown, Copy with Headers
8. Search - Ctrl+F opens a dialog that searches every row across all pages
9. Operations Menu - Use Operations → Drop Duplicates / Drop Null Values / Math / Join / Append / Replace Values
10. Edit Cells - Double-click any cell to edit (tracked as an applied step)
11. Save - Ctrl+S commits cell edits and applied transforms back to the file; Applied Steps clears after save
12. Undo - Press the ↩ Undo button in the Applied Steps panel to revert changes
13. Export - File → Export Results to save query results
14. Change Theme - File → Change Theme to switch between Light, Excel, and ParVu Black themes
15. Drag-and-drop - Drop one or many files onto the window to open them as tabs
16. Shortcut cheatsheet - Press Ctrl+/ to see every keyboard shortcut

Expression Language Examples

-- Add a computed column
data[total] = price * quantity

-- Remove duplicates keeping the first occurrence
DROP_DUPLICATES(data[id], data[category], 'first')

-- Drop rows where a column is NULL (or matches a sentinel)
DROP_NULL(data[email])
DROP_NULL(data[score], -1)

-- String and math functions
data[full_name] = UPPER(first_name) || ' ' || UPPER(last_name)
data[abs_diff] = ABS(price - avg_price)

-- Use in expression mode; compiled to DuckDB SQL automatically

Example SQL Queries

-- View all data (default)
SELECT * FROM data

-- Filter rows
SELECT * FROM data WHERE age > 25

-- Aggregate data
SELECT category, COUNT(*) as count, AVG(price) as avg_price
FROM data
GROUP BY category
ORDER BY count DESC

-- Limit results
SELECT * FROM data LIMIT 100

-- Complex filtering
SELECT * FROM data
WHERE status = 'active' AND created_date > '2024-01-01'
ORDER BY created_date DESC

Configuration

Settings are stored in ~/.ParVu/settings/settings.json:

• default_data_var_name - Table name used in queries (default: "data")
• result_pagination_rows_per_page - Rows per page (default: 100)
• current_theme - Active theme name (default: "ParVu Light")
• sql_keywords - Keywords for syntax highlighting

Themes

ParVu includes 3 professionally designed themes:

• ParVu Light - Default light theme with blue/green accents
• Excel - Microsoft Excel-inspired green theme
• ParVu Black - Dark theme inspired by Visual Studio Code

Create custom themes or import community themes. See docs/THEMES.md for complete theme documentation.

Languages

ParVu supports multiple interface languages:

• 🇬🇧 English - Default
• 🇷🇺 Russian (Русский) - Full translation
• 🇦🇿 Azerbaijani (Azərbaycan) - Full translation

Change language in Settings → General tab. See docs/I18N.md for i18n documentation.

FAQ

Q: What SQL queries are supported? A: ParVu supports full DuckDB SQL syntax including SELECT, WHERE, GROUP BY, ORDER BY, aggregations, and window functions. See DuckDB SQL documentation for details.

Q: Can I edit the actual file? A: No, cell editing only modifies the current view. To save changes, export the results to a new file.

Q: How large files can it handle? A: ParVu uses lazy loading with DuckDB, querying files directly without materializing them in memory. It can efficiently handle huge files (8GB+) with millions of rows. Only the current page is loaded into memory at any time.

Q: What's the difference from the old version? A: This is a complete rewrite with:

• PyQt6 (modern UI framework)
• Lazy loading for huge files (8GB+) without memory issues
• Theme system with 3 built-in themes
• Expression language (DSL) with auto-completion
• Applied steps with undo for every transform
• Better SQL auto-completion
• Excel-like unique value filters
• Improved pagination performance
• More export formats (JSON added)
• Double-click cell editing

Q: Can I create my own themes? A: Yes! Export an existing theme as a template, edit the JSON file to customize colors and fonts, then import it back. See docs/THEMES.md for details.

Q: Can I add my own language translation? A: Yes! The i18n system is fully extensible. See docs/I18N.md for instructions on adding new languages.

Contact

• Telegram;

Project:

• GitHub;

It has bugs

Yes, it has. Please open an issue and lets solve this