DataTableViewer

A tabular data viewer for Windows - built as a native plugin for Seer, the quick-look file preview tool.

DataTableViewer lets you preview CSV, TSV, and SQLite files in a fast read-only table view. Press Space on a supported file in Seer to inspect rows, search values, sort columns, and copy selected cells without opening a spreadsheet or database tool.

Features

• CSV and TSV paging: bounded-memory paging backed by disk offset indexing, handling large files (>64 MiB, 100k+ rows) without memory exhaustion or truncation
• SQLite preview: browse database tables, pick one, then page through large tables without loading the whole database
• Table paging: configurable page size, asynchronous row counts and index progress, and current-page-only filtering
• Interactive table view: sortable columns, movable/resizable headers, alternating rows, and TSV copy
• Type-aware sorting: numeric columns sort by numeric value instead of plain text
• Live filtering: search across the current table while keeping the UI responsive
• Status bar metrics: format, row count, column count, file size, load time, warnings, and paging progress
• Async parsing: background-thread parsing and cancellation keep Seer responsive while switching files

Screenshots

Supported Formats

• .csv
• .tsv
• .sqlite
• .sqlite3
• .db
• .db3
• .sl3

Paging, Sorting & Resource Bounds

• Paged CSV and TSV preview: Large files are indexed into a temporary binary disk index in the background without reading the entire file into RAM. The first page renders in < 500 ms while background indexing continues in 256 KiB slices. Single-page files (total <= pageSize) hide the pager bar.
• Sorting behavior: Full-file sorting is available for SQLite tables with a usable rowid and small in-memory tables. Full-file sorting is unsupported for paged CSV/TSV files; clicking a column header displays an explanation in the status bar ("Sorting is not supported for paged CSV/TSV files") without reordering rows or showing sort indicators.
• Resource limits: Decoded table cells are bounded to 32 MiB per page with dynamic cell clamping at valid UTF-8 code point boundaries. Clamped cells are refetched asynchronously during clipboard copy up to an aggregate 64 MiB payload budget.

Building

Requirements:

• Qt 6.8
• CMake 3.16+
• Visual Studio 2022 or newer with MSVC
• vcpkg, available through VCPKG_ROOT

Command-line configure (Ninja, RelWithDebInfo):

cmake -B build -G Ninja `
  -DCMAKE_BUILD_TYPE=RelWithDebInfo `
  -DCMAKE_PREFIX_PATH="$env:QTDIR" `
  -DCMAKE_TOOLCHAIN_FILE="$env:VCPKG_ROOT/scripts/buildsystems/vcpkg.cmake" `
  -DVCPKG_TARGET_TRIPLET=x64-windows-static-md
cmake --build build --target datatableviewer

The checked-in CMakePresets.json only defines a hidden Qt base preset (vcpkg toolchain, x64-windows-static-md triplet). The convenient RelWithDebInfo preset lives in the gitignored, machine-local CMakeUserPresets.json, typically generated by Qt Creator or Visual Studio — fresh clones should configure manually as above.

SQLite is consumed through the vcpkg manifest in vcpkg.json. The checked-in CMake preset uses the x64-windows-static-md triplet so SQLite is linked into datatableviewer.dll while keeping the MSVC runtime dynamic and compatible with Qt. This avoids shipping a separate sqlite3.dll with the plugin.

Recommended Visual Studio flow:

1. Clone this repository.
2. Open the repository folder in Visual Studio with File -> Open -> Folder.
3. Let Visual Studio configure the CMake project.
4. Set datatableviewer_test as the startup item.
5. Build or run datatableviewer_test.

The plugin build produces:

• datatableviewer.dll - the Seer plugin
• plugin.json - copied next to the DLL after build

Unit tests build by default (BUILD_TESTS=ON); build and run them with cmake --build build --target run_all_tests (wraps ctest). A large-CSV benchmark test is opt-in with -DBUILD_CSV_LARGE_TESTS=ON.

Use With Seer

Seer is a quick-look file preview tool for Windows: press Space on a file to preview it without opening a full application.

1. Install Seer.
2. In Seer, open Settings -> Plugins.
3. Install or place the DataTableViewer plugin files.
4. Ensure datatableviewer.dll and plugin.json are in the same plugin folder.
5. Press Space on a supported CSV, TSV, or SQLite file.

SQLite support is statically linked into datatableviewer.dll; no separate SQLite runtime DLL is required in the plugin folder.

Configuration & Shortcuts

Settings are stored in DataTableViewer.ini located in the plugin DLL directory. If the plugin directory is read-only, compiled defaults remain active.

Keyboard shortcuts can be customized in the [Shortcuts] section:

[Shortcuts]
find=Ctrl+F
copy=Ctrl+C
viewText=Ctrl+Alt+T
pageFirst=Ctrl+Home
pagePrev=Ctrl+PageUp
pageNext=Ctrl+PageDown
pageLast=Ctrl+End

The file is used by this plugin only, so keys carry no plugin prefix. Keys written by older builds as DataTableViewer.find are still read, and a plain key wins if both forms are present.

Action Reference

Key	Description	Default Shortcut
find	Focus and select search bar filter	Ctrl+F
copy	Copy selected cells to clipboard	Ctrl+C
viewText	Open current file in Seer Text viewer	Ctrl+Alt+T
pageFirst	Jump to first page	Ctrl+Home
pagePrev	Navigate to previous page	Ctrl+PageUp
pageNext	Navigate to next page	Ctrl+PageDown
pageLast	Jump to last page	Ctrl+End

• Storage & syntax: Values use portable key sequence format (e.g., Ctrl+F, Ctrl+Shift+C).
• Fallback behavior: Missing, empty, or unparsable keys fall back to their compiled defaults. If two actions share the same shortcut, a warning is logged once and both actions retain the assignment.
• Read timing: Configuration is synchronized on viewer initialization and applies on the next file preview. Live in-flight shortcut refresh is not performed.
• Read-only directories: When the plugin directory is read-only, default settings cannot be written to disk; compiled defaults remain active and a warning is logged once.

Development Notes

Operational project rules live in AGENTS.md. Use it for architecture, format registration, lifecycle, threading, and release-check expectations.

scripts/generate_test_data.py is a manual developer tool: it writes deterministic 100k-row CSV and SQLite datasets (default ./data/test_100k.csv and ./data/test_100k.sqlite; the data/ output directory is gitignored) for manual paging benchmarks. Run it by hand; the build and the automated test suite do not depend on it.

License

Apache-2.0 — see LICENSE and NOTICE.