parquet-browser 1.0.0__tar.gz

parquet_browser-1.0.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 abhandari73
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

parquet_browser-1.0.0/PKG-INFO

@@ -0,0 +1,208 @@
+Metadata-Version: 2.4
+Name: parquet-browser
+Version: 1.0.0
+Summary: Browse, filter and inspect parquet files in your web browser - a local web app powered by DuckDB.
+Author: abhandari73
+License: MIT License
+        
+        Copyright (c) 2026 abhandari73
+        
+        Permission is hereby granted, free of charge, to any person obtaining a copy
+        of this software and associated documentation files (the "Software"), to deal
+        in the Software without restriction, including without limitation the rights
+        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+        copies of the Software, and to permit persons to whom the Software is
+        furnished to do so, subject to the following conditions:
+        
+        The above copyright notice and this permission notice shall be included in all
+        copies or substantial portions of the Software.
+        
+        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+        SOFTWARE.
+        
+Project-URL: Homepage, https://github.com/abhandari73/parquet-viewer
+Project-URL: Issues, https://github.com/abhandari73/parquet-viewer/issues
+Keywords: parquet,duckdb,viewer,data,browser,dataframe
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Classifier: Environment :: Web Environment
+Classifier: Intended Audience :: Science/Research
+Classifier: Intended Audience :: Developers
+Classifier: Topic :: Scientific/Engineering :: Information Analysis
+Classifier: Topic :: Database :: Front-Ends
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: duckdb>=1.1
+Dynamic: license-file
+
+# Parquet Viewer
+
+Look inside a folder full of parquet files from your web browser — no Jupyter, no IPython, no
+desktop app. Point it at a folder, click through the sub-folders, open a file (or a whole folder
+of files as one table) and sort, search, filter and profile it.
+
+```bash
+pip install parquet-browser
+pqv ~/data
+```
+
+Or try it without installing: `uvx parquet-browser ~/data`
+
+![Opening a folder, filtering and checking column statistics](https://raw.githubusercontent.com/abhandari73/parquet-viewer/main/docs/demo.gif)
+
+Parquet is the best format for storing tables — small, fast, typed — but you can't just double-click
+it. Looking inside usually means starting Python, loading pandas and printing a few rows, or exporting
+to CSV so Excel can open it. This tool gives you the Excel-style "open, sort, filter, look" directly on
+the parquet files.
+
+Everything runs on your own computer: a tiny Python web server (standard library only) serves
+the page, and [DuckDB](https://duckdb.org) reads the parquet files. Only the rows on screen are
+read, so multi-GB files open instantly. The only dependency is `duckdb`; the page uses no CDN, so
+it also works offline.
+
+## Features
+
+**Folder browser**
+- Breadcrumbs, a filter box, file sizes and counts; only parquet files and folders are listed.
+- **Open all files in a folder as one table** — for monthly/daily files or Hive-partitioned data
+  (`year=2024/month=01/…`). Every `*.parquet` file below the folder is read together, columns are
+  matched by name, `key=value` folder names become columns, and a `_file` column tells you which
+  file each row came from.
+- Recent files and folders.
+
+**Data**
+- Paged grid (50–1,000 rows per page) with the column types in the header.
+- **Sort** by clicking a header (asc → desc → off); Shift+click sorts by several columns.
+- **Search** across columns: text contains the term, numbers equal it, dates start with it (`2024-06`).
+- **Filters** per column: `= ≠ < ≤ > ≥`, between, in / not in, contains, starts / ends with, regex,
+  is null, is NaN. Filters show as chips you can edit, switch off or remove.
+- **SQL condition** for anything else, e.g. `price > 100 AND country IN ('JP','KR')` or
+  `abs(ret) > 0.2 AND date >= '2024-01-01'` (any DuckDB expression).
+- **Column picker** with search — handy for files with hundreds of columns.
+- **Row detail** panel with every field of a row, a field search and ↑/↓ navigation; ⊕ filters on a value.
+- **Column statistics**: nulls, NaN, distinct, min/max/mean/std/sum, percentiles, a histogram
+  (trimmed to the 0.5–99.5 % range so outliers don't flatten it), rows per day/month/year for dates,
+  most frequent values. Click a bar or a value to filter. Statistics follow the current filters.
+- **Export CSV** of the filtered rows and visible columns (the whole result, not just the page).
+- **Code**: the DuckDB SQL and a pandas snippet that reproduce the current view.
+- The address bar holds the whole view (file, filters, sort, page), so you can bookmark or share it.
+
+**Schema** — DuckDB and parquet types, compression, compressed/uncompressed size and nulls per column;
+file metadata (rows, row groups, writer) and key/value metadata such as the pandas or Arrow schema.
+
+**Summary** — min, max, approx. distinct, mean, std, quartiles and null % for all visible columns
+(DuckDB `SUMMARIZE`), on demand.
+
+Light and dark mode follow your system.
+
+![Folder of partitioned files opened as one table, dark mode](https://raw.githubusercontent.com/abhandari73/parquet-viewer/main/docs/screenshot-folder-dark.png)
+
+## Install
+
+Python 3.9 or newer, on macOS or Linux. Windows is not supported yet (the tests don't pass there).
+
+```bash
+pip install parquet-browser      # or: pipx install parquet-browser
+uvx parquet-browser ~/data       # run once without installing (needs uv)
+```
+
+Latest development version:
+
+```bash
+pip install git+https://github.com/abhandari73/parquet-viewer
+```
+
+or from a clone:
+
+```bash
+git clone https://github.com/abhandari73/parquet-viewer
+cd parquet-viewer
+pip install .            # or: pip install -e .   while developing
+```
+
+Without installing, `pip install duckdb` and run `PYTHONPATH=src python -m parquet_viewer ~/data`.
+
+## Usage
+
+```bash
+pqv ~/data                    # opens http://localhost:8770 in your browser
+pqv                           # browse the current folder
+pqv ~/data --port 9000 --no-browser
+```
+
+| Option | Meaning |
+|---|---|
+| `ROOT` | Folder to browse (default: current folder). Nothing above it can be opened. |
+| `--port N` | Port (default 8770). |
+| `--no-browser` | Don't open a browser tab. |
+| `--memory-limit 2GB` | Cap DuckDB's memory (useful on small machines). |
+| `--threads N` | DuckDB threads. |
+| `--show-hidden` | Also list hidden files and folders. |
+| `--quiet` | Don't log requests. |
+| `--host ADDR` | Listen address (default `127.0.0.1`). See Security before changing it. |
+| `--version` | Show versions. |
+
+`parquet-browser` and `parquet-viewer` are aliases for `pqv`, and `python -m parquet_viewer` works too.
+
+![Filtered table with column statistics](https://raw.githubusercontent.com/abhandari73/parquet-viewer/main/docs/screenshot-data.png)
+
+### Files on another machine
+
+Run the viewer on the remote machine and tunnel the port over SSH; the data never leaves that machine:
+
+```bash
+ssh -L 8770:localhost:8770 user@server 'pqv /path/to/data --no-browser --memory-limit 1GB'
+# then open http://localhost:8770 locally
+```
+
+## Security
+
+- Listens on `127.0.0.1` and only answers requests addressed to `localhost`, so other machines and
+  DNS-rebinding tricks can't reach it.
+- API calls need a custom header, so other web sites open in your browser can't call it.
+- Only files below `ROOT` can be opened. DuckDB itself is locked to that folder
+  (`allowed_directories`, `enable_external_access = false`, configuration locked), so the SQL box
+  can't read or write files elsewhere, and only a single `SELECT` statement is ever run.
+- With `--host 0.0.0.0` anyone who can reach the port can read every file under `ROOT`.
+
+## How it compares
+
+| | This tool | [DuckDB UI](https://duckdb.org/docs/stable/core_extensions/ui) | Drag-and-drop web viewers | Desktop / terminal viewers |
+|---|---|---|---|---|
+| Browse a folder tree | ✓ | – (write SQL) | – (one file at a time) | some |
+| Folder of files as one table | ✓ | via SQL | – | rarely |
+| Click-to-filter, column stats | ✓ | partly | partly | varies |
+| Files larger than RAM | ✓ (DuckDB streams) | ✓ | limited by browser memory | varies |
+| Works offline | ✓ | – (UI loads from ui.duckdb.org) | ✓ | ✓ |
+| Install | `pip install parquet-browser` | DuckDB CLI | none | app install |
+
+If you mainly write SQL, the DuckDB UI is great. This tool is for quickly looking around a folder of
+data files and slicing them without writing code.
+
+## Notes
+
+- Integers beyond ±2^53 (JavaScript's exact range) are shown as text so they stay exact.
+- Timestamps with a time zone are shown as text.
+- Floats in the grid are shortened to about 6 significant decimals; the row panel shows full precision.
+- Opening a folder with thousands of files works, but the first count or sort reads every file.
+
+## Development
+
+```bash
+pip install -e .
+python tests/test_parquet_viewer.py      # or: pytest
+```
+
+The server is `src/parquet_viewer/server.py`; the front end is plain HTML/CSS/JavaScript in
+`src/parquet_viewer/static/` with no build step.
+
+## License
+
+MIT

parquet_browser-1.0.0/README.md

@@ -0,0 +1,164 @@
+# Parquet Viewer
+
+Look inside a folder full of parquet files from your web browser — no Jupyter, no IPython, no
+desktop app. Point it at a folder, click through the sub-folders, open a file (or a whole folder
+of files as one table) and sort, search, filter and profile it.
+
+```bash
+pip install parquet-browser
+pqv ~/data
+```
+
+Or try it without installing: `uvx parquet-browser ~/data`
+
+![Opening a folder, filtering and checking column statistics](https://raw.githubusercontent.com/abhandari73/parquet-viewer/main/docs/demo.gif)
+
+Parquet is the best format for storing tables — small, fast, typed — but you can't just double-click
+it. Looking inside usually means starting Python, loading pandas and printing a few rows, or exporting
+to CSV so Excel can open it. This tool gives you the Excel-style "open, sort, filter, look" directly on
+the parquet files.
+
+Everything runs on your own computer: a tiny Python web server (standard library only) serves
+the page, and [DuckDB](https://duckdb.org) reads the parquet files. Only the rows on screen are
+read, so multi-GB files open instantly. The only dependency is `duckdb`; the page uses no CDN, so
+it also works offline.
+
+## Features
+
+**Folder browser**
+- Breadcrumbs, a filter box, file sizes and counts; only parquet files and folders are listed.
+- **Open all files in a folder as one table** — for monthly/daily files or Hive-partitioned data
+  (`year=2024/month=01/…`). Every `*.parquet` file below the folder is read together, columns are
+  matched by name, `key=value` folder names become columns, and a `_file` column tells you which
+  file each row came from.
+- Recent files and folders.
+
+**Data**
+- Paged grid (50–1,000 rows per page) with the column types in the header.
+- **Sort** by clicking a header (asc → desc → off); Shift+click sorts by several columns.
+- **Search** across columns: text contains the term, numbers equal it, dates start with it (`2024-06`).
+- **Filters** per column: `= ≠ < ≤ > ≥`, between, in / not in, contains, starts / ends with, regex,
+  is null, is NaN. Filters show as chips you can edit, switch off or remove.
+- **SQL condition** for anything else, e.g. `price > 100 AND country IN ('JP','KR')` or
+  `abs(ret) > 0.2 AND date >= '2024-01-01'` (any DuckDB expression).
+- **Column picker** with search — handy for files with hundreds of columns.
+- **Row detail** panel with every field of a row, a field search and ↑/↓ navigation; ⊕ filters on a value.
+- **Column statistics**: nulls, NaN, distinct, min/max/mean/std/sum, percentiles, a histogram
+  (trimmed to the 0.5–99.5 % range so outliers don't flatten it), rows per day/month/year for dates,
+  most frequent values. Click a bar or a value to filter. Statistics follow the current filters.
+- **Export CSV** of the filtered rows and visible columns (the whole result, not just the page).
+- **Code**: the DuckDB SQL and a pandas snippet that reproduce the current view.
+- The address bar holds the whole view (file, filters, sort, page), so you can bookmark or share it.
+
+**Schema** — DuckDB and parquet types, compression, compressed/uncompressed size and nulls per column;
+file metadata (rows, row groups, writer) and key/value metadata such as the pandas or Arrow schema.
+
+**Summary** — min, max, approx. distinct, mean, std, quartiles and null % for all visible columns
+(DuckDB `SUMMARIZE`), on demand.
+
+Light and dark mode follow your system.
+
+![Folder of partitioned files opened as one table, dark mode](https://raw.githubusercontent.com/abhandari73/parquet-viewer/main/docs/screenshot-folder-dark.png)
+
+## Install
+
+Python 3.9 or newer, on macOS or Linux. Windows is not supported yet (the tests don't pass there).
+
+```bash
+pip install parquet-browser      # or: pipx install parquet-browser
+uvx parquet-browser ~/data       # run once without installing (needs uv)
+```
+
+Latest development version:
+
+```bash
+pip install git+https://github.com/abhandari73/parquet-viewer
+```
+
+or from a clone:
+
+```bash
+git clone https://github.com/abhandari73/parquet-viewer
+cd parquet-viewer
+pip install .            # or: pip install -e .   while developing
+```
+
+Without installing, `pip install duckdb` and run `PYTHONPATH=src python -m parquet_viewer ~/data`.
+
+## Usage
+
+```bash
+pqv ~/data                    # opens http://localhost:8770 in your browser
+pqv                           # browse the current folder
+pqv ~/data --port 9000 --no-browser
+```
+
+| Option | Meaning |
+|---|---|
+| `ROOT` | Folder to browse (default: current folder). Nothing above it can be opened. |
+| `--port N` | Port (default 8770). |
+| `--no-browser` | Don't open a browser tab. |
+| `--memory-limit 2GB` | Cap DuckDB's memory (useful on small machines). |
+| `--threads N` | DuckDB threads. |
+| `--show-hidden` | Also list hidden files and folders. |
+| `--quiet` | Don't log requests. |
+| `--host ADDR` | Listen address (default `127.0.0.1`). See Security before changing it. |
+| `--version` | Show versions. |
+
+`parquet-browser` and `parquet-viewer` are aliases for `pqv`, and `python -m parquet_viewer` works too.
+
+![Filtered table with column statistics](https://raw.githubusercontent.com/abhandari73/parquet-viewer/main/docs/screenshot-data.png)
+
+### Files on another machine
+
+Run the viewer on the remote machine and tunnel the port over SSH; the data never leaves that machine:
+
+```bash
+ssh -L 8770:localhost:8770 user@server 'pqv /path/to/data --no-browser --memory-limit 1GB'
+# then open http://localhost:8770 locally
+```
+
+## Security
+
+- Listens on `127.0.0.1` and only answers requests addressed to `localhost`, so other machines and
+  DNS-rebinding tricks can't reach it.
+- API calls need a custom header, so other web sites open in your browser can't call it.
+- Only files below `ROOT` can be opened. DuckDB itself is locked to that folder
+  (`allowed_directories`, `enable_external_access = false`, configuration locked), so the SQL box
+  can't read or write files elsewhere, and only a single `SELECT` statement is ever run.
+- With `--host 0.0.0.0` anyone who can reach the port can read every file under `ROOT`.
+
+## How it compares
+
+| | This tool | [DuckDB UI](https://duckdb.org/docs/stable/core_extensions/ui) | Drag-and-drop web viewers | Desktop / terminal viewers |
+|---|---|---|---|---|
+| Browse a folder tree | ✓ | – (write SQL) | – (one file at a time) | some |
+| Folder of files as one table | ✓ | via SQL | – | rarely |
+| Click-to-filter, column stats | ✓ | partly | partly | varies |
+| Files larger than RAM | ✓ (DuckDB streams) | ✓ | limited by browser memory | varies |
+| Works offline | ✓ | – (UI loads from ui.duckdb.org) | ✓ | ✓ |
+| Install | `pip install parquet-browser` | DuckDB CLI | none | app install |
+
+If you mainly write SQL, the DuckDB UI is great. This tool is for quickly looking around a folder of
+data files and slicing them without writing code.
+
+## Notes
+
+- Integers beyond ±2^53 (JavaScript's exact range) are shown as text so they stay exact.
+- Timestamps with a time zone are shown as text.
+- Floats in the grid are shortened to about 6 significant decimals; the row panel shows full precision.
+- Opening a folder with thousands of files works, but the first count or sort reads every file.
+
+## Development
+
+```bash
+pip install -e .
+python tests/test_parquet_viewer.py      # or: pytest
+```
+
+The server is `src/parquet_viewer/server.py`; the front end is plain HTML/CSS/JavaScript in
+`src/parquet_viewer/static/` with no build step.
+
+## License
+
+MIT

parquet_browser-1.0.0/pyproject.toml

@@ -0,0 +1,42 @@
+[build-system]
+requires = ["setuptools>=64"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "parquet-browser"
+dynamic = ["version"]
+description = "Browse, filter and inspect parquet files in your web browser - a local web app powered by DuckDB."
+readme = "README.md"
+license = {file = "LICENSE"}
+authors = [{name = "abhandari73"}]
+requires-python = ">=3.9"
+dependencies = ["duckdb>=1.1"]
+keywords = ["parquet", "duckdb", "viewer", "data", "browser", "dataframe"]
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "License :: OSI Approved :: MIT License",
+    "Operating System :: OS Independent",
+    "Environment :: Web Environment",
+    "Intended Audience :: Science/Research",
+    "Intended Audience :: Developers",
+    "Topic :: Scientific/Engineering :: Information Analysis",
+    "Topic :: Database :: Front-Ends",
+]
+
+[project.urls]
+Homepage = "https://github.com/abhandari73/parquet-viewer"
+Issues = "https://github.com/abhandari73/parquet-viewer/issues"
+
+[project.scripts]
+pqv = "parquet_viewer.server:main"
+parquet-browser = "parquet_viewer.server:main"
+parquet-viewer = "parquet_viewer.server:main"
+
+[tool.setuptools.dynamic]
+version = {attr = "parquet_viewer.__version__"}
+
+[tool.setuptools.packages.find]
+where = ["src"]
+
+[tool.setuptools.package-data]
+parquet_viewer = ["static/*"]

parquet_browser-1.0.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

parquet_browser-1.0.0/src/parquet_browser.egg-info/PKG-INFO

@@ -0,0 +1,208 @@
+Metadata-Version: 2.4
+Name: parquet-browser
+Version: 1.0.0
+Summary: Browse, filter and inspect parquet files in your web browser - a local web app powered by DuckDB.
+Author: abhandari73
+License: MIT License
+        
+        Copyright (c) 2026 abhandari73
+        
+        Permission is hereby granted, free of charge, to any person obtaining a copy
+        of this software and associated documentation files (the "Software"), to deal
+        in the Software without restriction, including without limitation the rights
+        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+        copies of the Software, and to permit persons to whom the Software is
+        furnished to do so, subject to the following conditions:
+        
+        The above copyright notice and this permission notice shall be included in all
+        copies or substantial portions of the Software.
+        
+        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+        SOFTWARE.
+        
+Project-URL: Homepage, https://github.com/abhandari73/parquet-viewer
+Project-URL: Issues, https://github.com/abhandari73/parquet-viewer/issues
+Keywords: parquet,duckdb,viewer,data,browser,dataframe
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Classifier: Environment :: Web Environment
+Classifier: Intended Audience :: Science/Research
+Classifier: Intended Audience :: Developers
+Classifier: Topic :: Scientific/Engineering :: Information Analysis
+Classifier: Topic :: Database :: Front-Ends
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: duckdb>=1.1
+Dynamic: license-file
+
+# Parquet Viewer
+
+Look inside a folder full of parquet files from your web browser — no Jupyter, no IPython, no
+desktop app. Point it at a folder, click through the sub-folders, open a file (or a whole folder
+of files as one table) and sort, search, filter and profile it.
+
+```bash
+pip install parquet-browser
+pqv ~/data
+```
+
+Or try it without installing: `uvx parquet-browser ~/data`
+
+![Opening a folder, filtering and checking column statistics](https://raw.githubusercontent.com/abhandari73/parquet-viewer/main/docs/demo.gif)
+
+Parquet is the best format for storing tables — small, fast, typed — but you can't just double-click
+it. Looking inside usually means starting Python, loading pandas and printing a few rows, or exporting
+to CSV so Excel can open it. This tool gives you the Excel-style "open, sort, filter, look" directly on
+the parquet files.
+
+Everything runs on your own computer: a tiny Python web server (standard library only) serves
+the page, and [DuckDB](https://duckdb.org) reads the parquet files. Only the rows on screen are
+read, so multi-GB files open instantly. The only dependency is `duckdb`; the page uses no CDN, so
+it also works offline.
+
+## Features
+
+**Folder browser**
+- Breadcrumbs, a filter box, file sizes and counts; only parquet files and folders are listed.
+- **Open all files in a folder as one table** — for monthly/daily files or Hive-partitioned data
+  (`year=2024/month=01/…`). Every `*.parquet` file below the folder is read together, columns are
+  matched by name, `key=value` folder names become columns, and a `_file` column tells you which
+  file each row came from.
+- Recent files and folders.
+
+**Data**
+- Paged grid (50–1,000 rows per page) with the column types in the header.
+- **Sort** by clicking a header (asc → desc → off); Shift+click sorts by several columns.
+- **Search** across columns: text contains the term, numbers equal it, dates start with it (`2024-06`).
+- **Filters** per column: `= ≠ < ≤ > ≥`, between, in / not in, contains, starts / ends with, regex,
+  is null, is NaN. Filters show as chips you can edit, switch off or remove.
+- **SQL condition** for anything else, e.g. `price > 100 AND country IN ('JP','KR')` or
+  `abs(ret) > 0.2 AND date >= '2024-01-01'` (any DuckDB expression).
+- **Column picker** with search — handy for files with hundreds of columns.
+- **Row detail** panel with every field of a row, a field search and ↑/↓ navigation; ⊕ filters on a value.
+- **Column statistics**: nulls, NaN, distinct, min/max/mean/std/sum, percentiles, a histogram
+  (trimmed to the 0.5–99.5 % range so outliers don't flatten it), rows per day/month/year for dates,
+  most frequent values. Click a bar or a value to filter. Statistics follow the current filters.
+- **Export CSV** of the filtered rows and visible columns (the whole result, not just the page).
+- **Code**: the DuckDB SQL and a pandas snippet that reproduce the current view.
+- The address bar holds the whole view (file, filters, sort, page), so you can bookmark or share it.
+
+**Schema** — DuckDB and parquet types, compression, compressed/uncompressed size and nulls per column;
+file metadata (rows, row groups, writer) and key/value metadata such as the pandas or Arrow schema.
+
+**Summary** — min, max, approx. distinct, mean, std, quartiles and null % for all visible columns
+(DuckDB `SUMMARIZE`), on demand.
+
+Light and dark mode follow your system.
+
+![Folder of partitioned files opened as one table, dark mode](https://raw.githubusercontent.com/abhandari73/parquet-viewer/main/docs/screenshot-folder-dark.png)
+
+## Install
+
+Python 3.9 or newer, on macOS or Linux. Windows is not supported yet (the tests don't pass there).
+
+```bash
+pip install parquet-browser      # or: pipx install parquet-browser
+uvx parquet-browser ~/data       # run once without installing (needs uv)
+```
+
+Latest development version:
+
+```bash
+pip install git+https://github.com/abhandari73/parquet-viewer
+```
+
+or from a clone:
+
+```bash
+git clone https://github.com/abhandari73/parquet-viewer
+cd parquet-viewer
+pip install .            # or: pip install -e .   while developing
+```
+
+Without installing, `pip install duckdb` and run `PYTHONPATH=src python -m parquet_viewer ~/data`.
+
+## Usage
+
+```bash
+pqv ~/data                    # opens http://localhost:8770 in your browser
+pqv                           # browse the current folder
+pqv ~/data --port 9000 --no-browser
+```
+
+| Option | Meaning |
+|---|---|
+| `ROOT` | Folder to browse (default: current folder). Nothing above it can be opened. |
+| `--port N` | Port (default 8770). |
+| `--no-browser` | Don't open a browser tab. |
+| `--memory-limit 2GB` | Cap DuckDB's memory (useful on small machines). |
+| `--threads N` | DuckDB threads. |
+| `--show-hidden` | Also list hidden files and folders. |
+| `--quiet` | Don't log requests. |
+| `--host ADDR` | Listen address (default `127.0.0.1`). See Security before changing it. |
+| `--version` | Show versions. |
+
+`parquet-browser` and `parquet-viewer` are aliases for `pqv`, and `python -m parquet_viewer` works too.
+
+![Filtered table with column statistics](https://raw.githubusercontent.com/abhandari73/parquet-viewer/main/docs/screenshot-data.png)
+
+### Files on another machine
+
+Run the viewer on the remote machine and tunnel the port over SSH; the data never leaves that machine:
+
+```bash
+ssh -L 8770:localhost:8770 user@server 'pqv /path/to/data --no-browser --memory-limit 1GB'
+# then open http://localhost:8770 locally
+```
+
+## Security
+
+- Listens on `127.0.0.1` and only answers requests addressed to `localhost`, so other machines and
+  DNS-rebinding tricks can't reach it.
+- API calls need a custom header, so other web sites open in your browser can't call it.
+- Only files below `ROOT` can be opened. DuckDB itself is locked to that folder
+  (`allowed_directories`, `enable_external_access = false`, configuration locked), so the SQL box
+  can't read or write files elsewhere, and only a single `SELECT` statement is ever run.
+- With `--host 0.0.0.0` anyone who can reach the port can read every file under `ROOT`.
+
+## How it compares
+
+| | This tool | [DuckDB UI](https://duckdb.org/docs/stable/core_extensions/ui) | Drag-and-drop web viewers | Desktop / terminal viewers |
+|---|---|---|---|---|
+| Browse a folder tree | ✓ | – (write SQL) | – (one file at a time) | some |
+| Folder of files as one table | ✓ | via SQL | – | rarely |
+| Click-to-filter, column stats | ✓ | partly | partly | varies |
+| Files larger than RAM | ✓ (DuckDB streams) | ✓ | limited by browser memory | varies |
+| Works offline | ✓ | – (UI loads from ui.duckdb.org) | ✓ | ✓ |
+| Install | `pip install parquet-browser` | DuckDB CLI | none | app install |
+
+If you mainly write SQL, the DuckDB UI is great. This tool is for quickly looking around a folder of
+data files and slicing them without writing code.
+
+## Notes
+
+- Integers beyond ±2^53 (JavaScript's exact range) are shown as text so they stay exact.
+- Timestamps with a time zone are shown as text.
+- Floats in the grid are shortened to about 6 significant decimals; the row panel shows full precision.
+- Opening a folder with thousands of files works, but the first count or sort reads every file.
+
+## Development
+
+```bash
+pip install -e .
+python tests/test_parquet_viewer.py      # or: pytest
+```
+
+The server is `src/parquet_viewer/server.py`; the front end is plain HTML/CSS/JavaScript in
+`src/parquet_viewer/static/` with no build step.
+
+## License
+
+MIT

parquet_browser-1.0.0/src/parquet_browser.egg-info/SOURCES.txt

@@ -0,0 +1,16 @@
+LICENSE
+README.md
+pyproject.toml
+src/parquet_browser.egg-info/PKG-INFO
+src/parquet_browser.egg-info/SOURCES.txt
+src/parquet_browser.egg-info/dependency_links.txt
+src/parquet_browser.egg-info/entry_points.txt
+src/parquet_browser.egg-info/requires.txt
+src/parquet_browser.egg-info/top_level.txt
+src/parquet_viewer/__init__.py
+src/parquet_viewer/__main__.py
+src/parquet_viewer/server.py
+src/parquet_viewer/static/app.css
+src/parquet_viewer/static/app.js
+src/parquet_viewer/static/index.html
+tests/test_parquet_viewer.py

parquet_browser-1.0.0/src/parquet_browser.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

parquet_browser-1.0.0/src/parquet_browser.egg-info/entry_points.txt

@@ -0,0 +1,4 @@
+[console_scripts]
+parquet-browser = parquet_viewer.server:main
+parquet-viewer = parquet_viewer.server:main
+pqv = parquet_viewer.server:main