Update documentation and dependencies for Excel file handling

- Enhanced `CHANGELOG.md` to include notable libraries in the documentation.
- Updated `README.md` to reference the new libraries documentation.
- Added `fastavro`, `fastexcel`, and `XlsxWriter` to `requirements.txt` to support Excel and Avro file operations.
- Modified `data_loader.py` to improve Excel file reading functionality, ensuring files open correctly on the first sheet unless specified otherwise.

Author: Power-Howdy

CHANGELOG.md

@@ -13,7 +13,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
   live preview on the first 200 rows. Results open as a view like AI results (`/api/transform/*`).
 - Open-source project files: license, contributing guide, code of conduct, security policy, issue and pull
   request templates, and CI.
-- Documentation under `docs/`.
+- Documentation under `docs/`, including [a list of notable libraries](docs/libraries.md) linked from the README.
 - Sidebar badges: loaded datasets show icons for whether they are indexed for searching and optimized for
   browsing (`GET /api/datasets/features`).
 - **Optimize for browsing**: for large files whose deep pages load slowly (Parquet written as huge row groups,
@@ -24,6 +24,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 ### Fixed
 
 - `.gitignore` no longer excludes `package.json`, `package-lock.json` and `tsconfig*.json`.
+- Excel and Avro files could not be opened, and Excel export failed: `requirements.txt` now includes fastexcel,
+  XlsxWriter and fastavro, and Excel files open on the first sheet unless a sheet is chosen.
 - Column details showed an empty distribution for list, array and struct columns. List items are now counted
   (number lists get a histogram of their items), struct values are shown as JSON, and the counts are computed
   lazily.

README.md

@@ -112,6 +112,7 @@ Full instructions: [docs/getting-started.md](docs/getting-started.md).
 | [Architecture](docs/architecture.md) | Components and data flow |
 | [API reference](docs/api-reference.md) | REST endpoints |
 | [Development](docs/development.md) | Setup, tests, conventions |
+| [Libraries](docs/libraries.md) | Notable libraries and what they are used for |
 | [Troubleshooting](docs/troubleshooting.md) | Common problems |
 
 ## Tech stack
@@ -123,6 +124,8 @@ Full instructions: [docs/getting-started.md](docs/getting-started.md).
 | NumPy (search index) | zustand, TanStack Query/Table |
 | httpx (OpenAI-compatible LLM APIs) | Recharts, CodeMirror 6 |
 
+See [docs/libraries.md](docs/libraries.md) for every notable library, its version, what it does here and its license.
+
 ## Project structure
 
 ```

backend/app/services/data_loader.py

@@ -226,9 +226,7 @@ def _read_orc_lazy(self, path: str, **options) -> pl.LazyFrame:
 
     def _read_excel_lazy(self, path: str, format: DataFormat, **options) -> pl.LazyFrame:
         """Read Excel file."""
-        sheet_name = options.get("sheet_name", 0)
-        df = pl.read_excel(path, sheet_name=sheet_name, **options)
-        return df.lazy()
+        return self._read_excel_eager(path, format, **options).lazy()
 
     def load_eager(self, path: str, format: Optional[DataFormat] = None, **options) -> pl.DataFrame:
         """Load data eagerly as DataFrame."""
@@ -295,9 +293,15 @@ def _read_orc_eager(self, path: str, **options) -> pl.DataFrame:
         return pl.from_arrow(table)
 
     def _read_excel_eager(self, path: str, format: DataFormat, **options) -> pl.DataFrame:
-        """Read Excel file eagerly."""
-        sheet_name = options.get("sheet_name", 0)
-        return pl.read_excel(path, sheet_name=sheet_name, **options)
+        """Read Excel file eagerly (the first sheet unless sheet_name or sheet_id is given)."""
+        options = dict(options)
+        sheet = options.pop("sheet_name", None)
+        if isinstance(sheet, int):
+            options["sheet_id"] = sheet + 1
+        elif sheet:
+            options["sheet_name"] = sheet
+        options.setdefault("sheet_id", None if "sheet_name" in options else 1)
+        return pl.read_excel(path, **options)
 
     def get_schema(self, lf: pl.LazyFrame) -> List[ColumnSchema]:
         """Extract schema from LazyFrame."""

backend/requirements.txt

@@ -9,6 +9,8 @@ colorama==0.4.6
 cramjam==2.13.0
 et_xmlfile==2.0.0
 fastapi==0.142.4
+fastavro==1.13.1
+fastexcel==0.21.0
 fastparquet==2026.9.0
 fsspec==2026.9.0
 h11==0.16.0
@@ -55,5 +57,6 @@ typing_extensions==4.16.0
 tzdata==2026.5
 uvicorn==0.54.0
 xlrd==2.0.2
+XlsxWriter==3.2.9
 xxhash==4.0.1
 zstandard==0.25.0

backend/tests/test_formats.py

@@ -0,0 +1,26 @@
+"""Formats that need optional packages (fastexcel, XlsxWriter, fastavro) read and write with requirements.txt."""
+import fastavro
+import polars as pl
+import pytest
+
+from app.models.schemas import DataFormat
+from app.services.data_loader import DataLoader
+
+DF = pl.DataFrame({"text": ["a", "b"], "n": [1, 2]})
+
+
+@pytest.mark.parametrize("sheet", [{}, {"sheet_name": "Sheet1"}, {"sheet_name": 0}, {"sheet_id": 1}])
+def test_excel_round_trip(tmp_path, sheet):
+    path = str(tmp_path / "data.xlsx")
+    loader = DataLoader()
+    loader.write(DF, path, DataFormat.XLSX)
+    assert loader.load_lazy(path, **sheet).collect().equals(DF)
+
+
+def test_avro_read(tmp_path):
+    path = tmp_path / "data.avro"
+    schema = {"type": "record", "name": "r", "fields": [
+        {"name": "text", "type": "string"}, {"name": "n", "type": "long"}]}
+    with open(path, "wb") as f:
+        fastavro.writer(f, schema, DF.to_dicts())
+    assert DataLoader().load_lazy(str(path)).collect().equals(DF)

docs/README.md

@@ -25,6 +25,7 @@ Polars and a React frontend shows them in the browser.
 | [API reference](api-reference.md) | Every REST endpoint |
 | [Frontend guide](frontend.md) | Structure, state, component rules |
 | [Development](development.md) | Local setup, tests, code style, adding features |
+| [Libraries](libraries.md) | Notable libraries, versions, licenses and what each is used for |
 
 ## Project
 

docs/libraries.md

@@ -0,0 +1,58 @@
+# Libraries
+
+Data Expert is built on open-source libraries. This page lists the notable ones, what each does in the app, and
+where to learn more. Exact versions are pinned in [`backend/requirements.txt`](../backend/requirements.txt) and
+[`frontend/package.json`](../frontend/package.json) (resolved in `package-lock.json`).
+
+## Backend (Python 3.14)
+
+| Library | Version | Used for | License |
+|---------|---------|----------|---------|
+| [FastAPI](https://fastapi.tiangolo.com/) | 0.142 | REST API: routes, request validation, OpenAPI docs at `/docs` | MIT |
+| [Starlette](https://www.starlette.io/) | 1.7 | ASGI toolkit under FastAPI; streamed downloads | BSD-3-Clause |
+| [Uvicorn](https://www.uvicorn.org/) | 0.54 | ASGI server that runs the backend (`python -m app.main`) | BSD-3-Clause |
+| [sse-starlette](https://github.com/sysid/sse-starlette) | 3.5 | Server-sent events for streamed AI responses | BSD-3-Clause |
+| [Pydantic](https://docs.pydantic.dev/) v2 | 2.14 | Request and response models, AI plan validation | MIT |
+| [pydantic-settings](https://docs.pydantic.dev/latest/concepts/pydantic_settings/) | 2.15 | Settings from `config.yaml` and environment variables | MIT |
+| [PyYAML](https://pyyaml.org/) | 6.0 | Reads `config.yaml` | MIT |
+| [Polars](https://pola.rs/) | 2.0 | The data engine: lazy reading of every format, paging, search scans, edits, transforms, statistics, export | MIT |
+| [PyArrow](https://arrow.apache.org/docs/python/) | 25 | Parquet metadata (instant file open), streaming row readers, Arrow IPC browse copies, ORC | Apache-2.0 |
+| [NumPy](https://numpy.org/) | 2.5 | Search index fingerprints and vectorized statistics | BSD-3-Clause |
+| [HTTPX](https://www.python-httpx.org/) | 0.28 | Calls to OpenAI-compatible, Anthropic and Gemini model APIs | BSD-3-Clause |
+| [fastexcel](https://github.com/ToucanToco/fastexcel) | 0.21 | Reading Excel files (`.xlsx`, `.xls`) through Polars | MIT |
+| [XlsxWriter](https://xlsxwriter.readthedocs.io/) | 3.2 | Exporting to Excel through Polars | BSD-2-Clause |
+| [fastavro](https://fastavro.readthedocs.io/) | 1.13 | Reading Avro files | MIT |
+
+## Frontend (React + TypeScript)
+
+| Library | Version | Used for | License |
+|---------|---------|----------|---------|
+| [React](https://react.dev/) | 18 | UI components | MIT |
+| [TypeScript](https://www.typescriptlang.org/) | 5 | Static types for the whole frontend | Apache-2.0 |
+| [Vite](https://vite.dev/) | 8 | Dev server and production build | MIT |
+| [Tailwind CSS](https://tailwindcss.com/) | 3 | Styling | MIT |
+| [Radix UI](https://www.radix-ui.com/primitives) | 1–2 | Accessible primitives behind `src/components/ui/` (select, tabs, dropdown, checkbox, scroll area) | MIT |
+| [TanStack Table](https://tanstack.com/table) | 8 | The Browse data table: columns, sizing, row selection | MIT |
+| [TanStack Query](https://tanstack.com/query) | 5 | Server-state caching and request deduplication | MIT |
+| [zustand](https://zustand.docs.pmnd.rs/) | 4 | Global state stores (`src/stores/`) | MIT |
+| [axios](https://axios-http.com/) | 1 | HTTP client behind every `src/lib/*Api.ts` module | MIT |
+| [Recharts](https://recharts.org/) | 2.12 | Charts in Stats and column details | MIT |
+| [CodeMirror 6](https://codemirror.net/) via [@uiw/react-codemirror](https://uiwjs.github.io/react-codemirror/) | 6 / 4.25 | JSON row editor and code views | MIT |
+| [lucide-react](https://lucide.dev/) | 0.303 | Icons | ISC |
+| [react-hot-toast](https://react-hot-toast.com/) | 2 | Notifications | MIT |
+| [clsx](https://github.com/lukeed/clsx) + [tailwind-merge](https://github.com/dcastil/tailwind-merge) | 2 / 3 | Combining class names (`cn()` helper) | MIT |
+
+## Development and testing
+
+| Library | Used for | License |
+|---------|----------|---------|
+| [pytest](https://pytest.org/) and [pytest-asyncio](https://pytest-asyncio.readthedocs.io/) | Backend test suite (`cd backend && pytest`) | MIT / Apache-2.0 |
+| [mypy](https://mypy-lang.org/) | Optional static type checks for the backend | MIT |
+| [ESLint](https://eslint.org/) 9 with [typescript-eslint](https://typescript-eslint.io/), react-hooks and react-refresh plugins | Frontend linting (`npm run lint`) | MIT |
+| [PostCSS](https://postcss.org/) and [Autoprefixer](https://github.com/postcss/autoprefixer) | Tailwind build pipeline | MIT |
+
+## AI model servers
+
+Data Expert does not bundle a model. It talks to any OpenAI-compatible server (for example
+[llama.cpp](https://github.com/ggml-org/llama.cpp), [Ollama](https://ollama.com/), [LM Studio](https://lmstudio.ai/)
+or [vLLM](https://docs.vllm.ai/)) and to hosted providers. See [AI assistant](ai-assistant.md).