tab-cli 0.1.9__tar.gz → 0.2.1__tar.gz

{tab_cli-0.1.9 → tab_cli-0.2.1}/CHANGELOG.md

@@ -1,3 +1,8 @@
+- 0.2.0:
+  - Added read-only SQLite/DuckDB input support via `{url}#{table_name}`, including remote SQLite/DuckDB files that are materialized locally before querying with Polars.
+  - Added `tab-cli[all]` to install all optional integrations.
+  - Renamed the `tab view` CLI flag from `--max-cell-len` to `--max-cell-length` to match the `max_cell_length` config key.
+  - Optimized cloud partition-glob expansion for fsspec-backed object stores so common patterns like `date=2026-01-*/*` avoid very slow `glob(...)` calls.
 - 0.1.9:
   - Fixed `tab view` to render cell contents verbatim instead of interpreting bracketed text like `[red]...[/red]` as Rich markup.
 - 0.1.8:
@@ -7,7 +12,7 @@
   - Fixed S3 Polars `storage_options` to avoid nested `client_kwargs` values that could break native reads.
   - Added `default_num_view_rows` config so the default `tab view` preview size can be customized.
   - Added `log_level` config so the CLI log level can default from `~/.config/tab/config.json` when `--log-level` is omitted.
-  - Added `max_cell_length` config so `tab view` can default to truncating long cell values without passing `--max-cell-len` every time.
+  - Added `max_cell_length` config so `tab view` can default to truncating long cell values without passing `--max-cell-length` every time.
   - Added `num_remote_workers` config to parallelize remote per-partition summary row counting.
   - Validated config file value types instead of silently accepting invalid JSON types.
   - Fixed the developer `Makefile` targets to point at `src/tab_cli`.
@@ -28,7 +33,7 @@
   - Automatic PyArrow fallback for Parquet files that fail to read with Polars' native reader.
 - 0.1.3:
   - Separate `tab view` from `tab cat`: `tab view` does not convert formats, `tab cat` does.
-  - Added `--max-cell-len` option to `tab view` to truncate long cell contents.
+  - Added `--max-cell-length` option to `tab view` to truncate long cell contents.
 - 0.1.2:
   - Bugfix on reading directories.
 - 0.1.1:

tab_cli-0.2.1/FSSPEC_GLOB_NOTES.md

@@ -0,0 +1,226 @@
+# Notes On Object-Store Glob Performance In `fsspec`
+
+This note is intended as background for a future upstream issue or PR against `fsspec` and/or object-store backends such as `s3fs`, `adlfs`, and `gcsfs`.
+
+It reflects two things:
+
+- current code inspection
+- practical measurements against a real S3 dataset
+
+## Motivation
+
+We observed severe slowdown on this pattern:
+
+```text
+s3://aws-public-blockchain/v1.0/btc/blocks/date=2026-01-*/*
+```
+
+This should match one month of partition directories, where each partition contains a single Parquet file.
+
+In practice, raw cloud globbing remained unacceptably slow, while a manual segmented expansion strategy completed in a few seconds.
+
+## Important Clarification
+
+An earlier simplified characterization of cloud globbing as just:
+
+```text
+find(root) + regex filter
+```
+
+is too coarse for modern async/cloud implementations.
+
+Current cloud backends may already do some server-side pruning by passing a fixed prefix down into backend-specific listing logic.
+
+However, that optimization is still not sufficient for common partition-style patterns like:
+
+```text
+.../date=2026-01-*/*
+.../year=2026/month=01/day=*/*
+.../chain=btc/date=*/part-*.parquet
+```
+
+The practical performance problem still exists.
+
+## What Object Stores Actually Provide
+
+Object stores like S3, GCS, and Azure Blob do not provide a true server-side glob primitive.
+
+They typically provide:
+
+- list by prefix
+- get object metadata
+- fetch object
+
+So wildcard matching is always a derived client-side operation.
+
+The main optimization opportunity is therefore not "make glob server-side", but rather:
+
+- minimize how much gets listed
+- keep listing aligned with prefix-oriented object-store APIs
+
+## What We Observed In Practice
+
+Using the public S3 dataset above:
+
+- backend authentication was not the bottleneck
+- listing the parent `blocks/` prefix was reasonable
+- per-file metadata lookup for the final 31 files was negligible
+- raw glob expansion was the expensive step
+
+Measured behavior:
+
+- `ls blocks/`: about 1.0 to 1.2 seconds
+- filter matching `date=2026-01-*` partition directories locally: negligible
+- `ls` the 31 matched directories: about 2.4 seconds
+- total manual segmented expansion: about 3.6 seconds
+- raw `s3fs.glob(...)`: still too slow to finish within 120 seconds in our testing
+- raw async `_glob(...)`: also too slow to be useful for this pattern
+
+So even if current cloud paths already do some fixed-prefix pruning, the remaining strategy is still too expensive for this type of layout.
+
+## Why Partition Patterns Are Special
+
+These layouts are common in data lakes:
+
+- many sibling partition directories
+- each segment encodes a structured partition key
+- only one or two path levels after the wildcard segment
+
+For this case, the optimal traversal is not a generic recursive discovery step.
+
+The optimal traversal is:
+
+1. list the fixed parent prefix
+2. match only the next path segment against the wildcard pattern
+3. recurse only into matched children
+4. continue one segment at a time
+
+That is a much closer match to object-store capabilities.
+
+## Current State By Backend
+
+### `s3fs`
+
+`s3fs` has custom async `_find()` behavior and can use prefix-aware listing to narrow traversal.
+
+This is better than the most naive possible implementation.
+
+However, for the tested partition glob, it was still too slow in practice.
+
+The missing piece is full segmented traversal by path component, not just fixed-prefix narrowing.
+
+### `gcsfs`
+
+`gcsfs` also has backend-specific listing behavior and can push fixed-prefix narrowing into object listing.
+
+That improves things compared with a totally naive subtree walk.
+
+But it still does not amount to path-segment-aware traversal.
+
+### `adlfs`
+
+`adlfs` similarly narrows object listing by fixed prefix.
+
+Again, this is helpful, but still not the same as matching segment-by-segment and descending only through matched directories.
+
+## Remaining Gap
+
+The important gap is:
+
+- current pruning is mostly based on the fixed prefix before wildcarding
+- partition-style globs often still need a better strategy after that point
+
+In other words, "prefix pruning exists" does not mean "glob is now efficient enough".
+
+For the tested workload, it was not.
+
+## Candidate Optimization
+
+For object-store-style backends, use segmented traversal when the pattern is compatible.
+
+### Proposed Strategy
+
+1. Split the path into segments.
+2. Keep the longest fixed prefix before the first wildcard segment.
+3. List that prefix one level at a time using delimiter-based listing semantics.
+4. Apply wildcard matching only to the current segment name.
+5. Recurse only into matched directory segments.
+6. Fall back to the current generic strategy for complex patterns.
+
+### Good Candidates For Segmented Traversal
+
+- `*`
+- `?`
+- character classes like `[abc]` or `[0-9]`
+- partition-style patterns with a fixed bucket and a fixed leading prefix
+
+### Fallback Cases
+
+- `**`
+- brace expansion if supported externally
+- bucket-level wildcarding
+- backend-specific edge cases that do not map cleanly to hierarchical traversal
+
+## Why This Still Seems Worthwhile
+
+Even if some newer implementations already do partial pruning:
+
+- the observed wall-clock behavior is still poor
+- the remaining inefficiency appears algorithmic
+- the manual segmented approach performed much better on the motivating example
+
+So there is still a strong case for improving glob expansion for object-store partition patterns.
+
+## Where This Could Live
+
+Two plausible options:
+
+### Option 1: `fsspec` Core
+
+Improve generic glob traversal in `fsspec` when the filesystem exposes object-store-style hierarchical listing semantics.
+
+Pros:
+
+- shared behavior across backends
+- one place to reason about segmented traversal
+
+Cons:
+
+- may be too specialized for a generic base implementation
+- care is needed not to regress non-object-store filesystems
+
+### Option 2: Backend Overrides
+
+Implement optimized segmented glob traversal directly in:
+
+- `s3fs`
+- `gcsfs`
+- `adlfs`
+
+Pros:
+
+- easier to exploit backend-specific listing semantics
+- simpler to tune for object stores
+
+Cons:
+
+- duplicated logic
+- less consistency across backends
+
+## Recommended Framing For An Upstream Discussion
+
+Suggested framing:
+
+- object stores do not provide server-side glob
+- fixed-prefix pruning already exists in some newer cloud backends
+- but partition-style patterns are still too slow in practice
+- segmented traversal by path component would better match object-store APIs
+- common data-lake glob patterns would benefit significantly
+
+## Minimal Repro Pattern
+
+```text
+s3://aws-public-blockchain/v1.0/btc/blocks/date=2026-01-*/*
+```
+
+Equivalent layouts are common across S3, GCS, and Azure.

tab_cli-0.2.1/PKG-INFO

@@ -0,0 +1,169 @@
+Metadata-Version: 2.4
+Name: tab-cli
+Version: 0.2.1
+Summary: A CLI tool for tabular data
+Author-email: Tongfei Chen <tongfei@pm.me>
+License-File: LICENSE
+Requires-Python: >=3.10
+Requires-Dist: blobfile>=3.0
+Requires-Dist: fsspec>=2026.1.0
+Requires-Dist: jmespath>=1.0
+Requires-Dist: loguru>=0.7.3
+Requires-Dist: polars-fastavro>=0.5.1
+Requires-Dist: polars>=1.0
+Requires-Dist: pyarrow>=20.0
+Requires-Dist: rich>=14.0
+Requires-Dist: typer>=0.21
+Provides-Extra: all
+Requires-Dist: adbc-driver-sqlite>=1.0.0; extra == 'all'
+Requires-Dist: adlfs>=2025.1.0; extra == 'all'
+Requires-Dist: azure-identity>=1.10.0; extra == 'all'
+Requires-Dist: boto3>=1.28.0; extra == 'all'
+Requires-Dist: duckdb>=1.0.0; extra == 'all'
+Requires-Dist: gcsfs>=2025.1.0; extra == 'all'
+Requires-Dist: google-auth>=2.27.0; extra == 'all'
+Requires-Dist: s3fs>=2025.1.0; extra == 'all'
+Provides-Extra: az
+Requires-Dist: adlfs>=2026.1.0; extra == 'az'
+Requires-Dist: azure-identity>=1.10.0; extra == 'az'
+Provides-Extra: duckdb
+Requires-Dist: duckdb>=1.0.0; extra == 'duckdb'
+Provides-Extra: gs
+Requires-Dist: gcsfs>=2026.1.0; extra == 'gs'
+Requires-Dist: google-auth>=2.27.0; extra == 'gs'
+Provides-Extra: s3
+Requires-Dist: boto3>=1.28.0; extra == 's3'
+Requires-Dist: s3fs>=2026.1.0; extra == 's3'
+Provides-Extra: sqlite
+Requires-Dist: adbc-driver-sqlite>=1.0.0; extra == 'sqlite'
+Description-Content-Type: text/markdown
+
+# tab
+
+[![pypi](https://img.shields.io/pypi/v/tab-cli)](https://pypi.org/project/tab-cli/)
+
+A CLI tool for viewing, querying, and converting tabular data files.
+Reads CSV, TSV, JSON Lines, Parquet, Avro, SQLite tables, and DuckDB tables -- locally or from S3, GCS, and Azure Blob Storage.
+
+```sh
+pip install tab-cli
+```
+
+**Documentation**: [tongfei.me/tab](https://tongfei.me/tab)
+
+---
+
+## Quick look
+
+### View any tabular file
+
+```bash
+tab view data.csv
+```
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/ctongfei/tab/refs/heads/gh-pages/assets/test.svg" alt="tab view" width="680">
+</p>
+
+### Can be remote, with many partitions that is globbed together:
+
+```bash
+tab view 's3://aws-public-blockchain/v1.0/btc/blocks/date=2026-01-*/*'
+````
+### Query with SQL
+
+The table is always available as `t`:
+
+```bash
+tab view --sql 'SELECT * FROM t WHERE Metric_A_Value > 80' data.csv
+```
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/ctongfei/tab/refs/heads/gh-pages/assets/test-where.svg" alt="tab view --sql" width="680">
+</p>
+
+### Reshape rows with JMESPath
+
+```bash
+tab view --jp '{id: participant.id, city: profile.address.city}' data.parquet 
+```
+
+### Convert between formats
+
+```bash
+tab convert data.csv data.parquet
+tab convert data.parquet data.jsonl -o jsonl
+tab convert data.csv output_dir/ -o parquet -n 4   # partitioned
+```
+
+### Concatenate files
+
+```bash
+tab cat part1.csv part2.csv part3.csv -o jsonl > combined.jsonl
+```
+
+### Inspect schema and summary
+
+```bash
+tab schema data.parquet
+tab summary data.parquet
+```
+
+### Read from stdin
+
+```bash
+curl -s https://example.com/data.csv | tab view -i csv -
+```
+
+### Read from cloud storage
+
+```bash
+tab view s3://bucket/path/data.parquet
+tab view gs://bucket/path/data.csv
+tab view az://container/path/data.jsonl
+```
+
+Globbing is supported for local and cloud paths:
+
+```bash
+tab view 'data/date=*/*.parquet'
+tab view 's3://bucket/path/date=2026-01-*/*'
+```
+
+### Read a SQLite table
+
+Use `{url}#{table_name}` for SQLite inputs:
+
+```bash
+tab view data.db#users
+tab view s3://bucket/path/data.db#users
+```
+
+### Read a DuckDB table
+
+Use `{url}#{table_name}` for DuckDB inputs:
+
+```bash
+tab view data.duckdb#users
+tab view s3://bucket/path/data.duckdb#users
+```
+
+Install cloud extras as needed:
+
+```sh
+pip install 'tab-cli[s3]'    # AWS S3
+pip install 'tab-cli[gs]'    # Google Cloud Storage
+pip install 'tab-cli[az]'    # Azure Blob Storage
+pip install 'tab-cli[duckdb]'  # DuckDB input support
+pip install 'tab-cli[sqlite]'  # SQLite via Polars ADBC
+pip install 'tab-cli[all]'   # Install all optional integrations
+```
+
+## Supported formats
+ - csv
+ - tsv
+ - jsonl
+ - parquet
+ - avro
+ - duckdb (input only; use `{url}#{table_name}`)
+ - sqlite (input only; use `{url}#{table_name}`)

tab_cli-0.1.9/PKG-INFO → tab_cli-0.2.1/README.md

@@ -1,36 +1,9 @@
-Metadata-Version: 2.4
-Name: tab-cli
-Version: 0.1.9
-Summary: A CLI tool for tabular data
-Author-email: Tongfei Chen <tongfei@pm.me>
-License-File: LICENSE
-Requires-Python: >=3.10
-Requires-Dist: blobfile>=3.0
-Requires-Dist: fsspec>=2025.1.0
-Requires-Dist: jmespath>=1.0
-Requires-Dist: loguru>=0.7.3
-Requires-Dist: polars-fastavro>=0.5.1
-Requires-Dist: polars>=1.0
-Requires-Dist: pyarrow>=15.0
-Requires-Dist: rich>=13.0
-Requires-Dist: typer>=0.21.1
-Provides-Extra: az
-Requires-Dist: adlfs>=2025.1.0; extra == 'az'
-Requires-Dist: azure-identity>=1.10.0; extra == 'az'
-Provides-Extra: gs
-Requires-Dist: gcsfs>=2025.1.0; extra == 'gs'
-Requires-Dist: google-auth>=2.27.0; extra == 'gs'
-Provides-Extra: s3
-Requires-Dist: boto3>=1.28.0; extra == 's3'
-Requires-Dist: s3fs>=2025.1.0; extra == 's3'
-Description-Content-Type: text/markdown
-
 # tab
 
 [![pypi](https://img.shields.io/pypi/v/tab-cli)](https://pypi.org/project/tab-cli/)
 
 A CLI tool for viewing, querying, and converting tabular data files.
-Reads CSV, TSV, JSON Lines, Parquet, and Avro -- locally or from S3, GCS, and Azure Blob Storage.
+Reads CSV, TSV, JSON Lines, Parquet, Avro, SQLite tables, and DuckDB tables -- locally or from S3, GCS, and Azure Blob Storage.
 
 ```sh
 pip install tab-cli
@@ -52,6 +25,11 @@ tab view data.csv
   <img src="https://raw.githubusercontent.com/ctongfei/tab/refs/heads/gh-pages/assets/test.svg" alt="tab view" width="680">
 </p>
 
+### Can be remote, with many partitions that is globbed together:
+
+```bash
+tab view 's3://aws-public-blockchain/v1.0/btc/blocks/date=2026-01-*/*'
+````
 ### Query with SQL
 
 The table is always available as `t`:
@@ -105,12 +83,40 @@ tab view gs://bucket/path/data.csv
 tab view az://container/path/data.jsonl
 ```
 
+Globbing is supported for local and cloud paths:
+
+```bash
+tab view 'data/date=*/*.parquet'
+tab view 's3://bucket/path/date=2026-01-*/*'
+```
+
+### Read a SQLite table
+
+Use `{url}#{table_name}` for SQLite inputs:
+
+```bash
+tab view data.db#users
+tab view s3://bucket/path/data.db#users
+```
+
+### Read a DuckDB table
+
+Use `{url}#{table_name}` for DuckDB inputs:
+
+```bash
+tab view data.duckdb#users
+tab view s3://bucket/path/data.duckdb#users
+```
+
 Install cloud extras as needed:
 
 ```sh
 pip install 'tab-cli[s3]'    # AWS S3
 pip install 'tab-cli[gs]'    # Google Cloud Storage
 pip install 'tab-cli[az]'    # Azure Blob Storage
+pip install 'tab-cli[duckdb]'  # DuckDB input support
+pip install 'tab-cli[sqlite]'  # SQLite via Polars ADBC
+pip install 'tab-cli[all]'   # Install all optional integrations
 ```
 
 ## Supported formats
@@ -119,3 +125,5 @@ pip install 'tab-cli[az]'    # Azure Blob Storage
  - jsonl
  - parquet
  - avro
+ - duckdb (input only; use `{url}#{table_name}`)
+ - sqlite (input only; use `{url}#{table_name}`)

{tab_cli-0.1.9 → tab_cli-0.2.1}/docs/cli-ref.md

@@ -4,6 +4,8 @@
 
 View tabular data from a data file in a rich CLI format, or a directory of partitions of data files.
 
+Paths may also use glob patterns such as `data/date=*/*.parquet` or `s3://bucket/path/date=2026-01-*/*`.
+
 ```bash
 tab view $path [OPTIONS]
 ```
@@ -12,17 +14,19 @@ Options:
 
 | Option                  | Description                                                                                               |
 |-------------------------|-----------------------------------------------------------------------------------------------------------|
-| `-i` / `--input-format` | Input format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`). Auto-detected from extension if omitted.         |
+| `-i` / `--input-format` | Input format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`, `sqlite`, `duckdb`). Auto-detected from extension if omitted. SQLite and DuckDB input use `{url}#{table_name}`. |
 | `--sql`                 | SQL query to apply before displaying. The table is available as `t`. Mutually exclusive with `--jmespath`. |
 | `--jmespath` / `--jp`   | JMESPath expression to apply to each row as JSON. Object outputs become columns; non-object outputs go to a `value` column. The result shape must stay consistent across rows. |
 | `--limit`               | Maximum number of rows to display.                                                                        |
 | `--skip`                | Number of rows to skip from the beginning.                                                                |
-| `--max-cell-len`        | Truncate cell contents longer than this. If omitted, `max_cell_length` from config is used when set.    |
+| `--max-cell-length`    | Truncate cell contents longer than this. If omitted, `max_cell_length` from config is used when set.     |
 
 ## `tab schema`
 
 Display the schema of a tabular data file.
 
+Paths may also use glob patterns.
+
 ```bash
 tab schema $path [OPTIONS]
 ```
@@ -31,13 +35,15 @@ Options:
 
 | Option                  | Description                                                                                               |
 |-------------------------|-----------------------------------------------------------------------------------------------------------|
-| `-i` / `--input-format` | Input format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`). Auto-detected from extension if omitted.         |
+| `-i` / `--input-format` | Input format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`, `sqlite`, `duckdb`). Auto-detected from extension if omitted. SQLite and DuckDB input use `{url}#{table_name}`. |
 
 
 ## `tab summary`
 
 Display summary information about a tabular data file.
 
+Paths may also use glob patterns.
+
 ```bash
 tab summary $path [OPTIONS]
 ```
@@ -46,13 +52,15 @@ Options:
 
 | Option                  | Description                                                                                               |
 |-------------------------|-----------------------------------------------------------------------------------------------------------|
-| `-i` / `--input-format` | Input format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`). Auto-detected from extension if omitted.         |
+| `-i` / `--input-format` | Input format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`, `sqlite`, `duckdb`). Auto-detected from extension if omitted. SQLite and DuckDB input use `{url}#{table_name}`. |
 
 
 ## `tab convert`
 
 Convert tabular data from one format to another.
 
+Input paths may also use glob patterns when the input format supports multi-file reads.
+
 ```bash
 tab convert $src $dst [OPTIONS]
 ```
@@ -61,7 +69,7 @@ Options:
 
 | Option                  | Description                                                                                             |
 |-------------------------|---------------------------------------------------------------------------------------------------------|
-| `-i` / `--input-format` | Input format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`). Auto-detected from extension if omitted.       |
+| `-i` / `--input-format` | Input format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`, `sqlite`, `duckdb`). Auto-detected from extension if omitted. SQLite and DuckDB input use `{url}#{table_name}`. |
 | `-o` / `--output-format` | Output format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`). If not specified, inherits from input format. |
 | `--sql`                 | SQL query to apply before writing. The table is available as `t`. Mutually exclusive with `--jmespath`. |
 | `--jmespath` / `--jp`   | JMESPath expression to apply to each row as JSON. Object outputs become columns; non-object outputs go to a `value` column. The result shape must stay consistent across rows. |
@@ -72,6 +80,8 @@ Options:
 
 Concatenate tabular data from multiple files.
 
+Input paths may also use glob patterns.
+
 ```bash
 tab cat $paths [OPTIONS]
 ```
@@ -80,7 +90,7 @@ Options:
 
 | Option                  | Description                                                                                               |
 |-------------------------|-----------------------------------------------------------------------------------------------------------|
-| `-i` / `--input-format` | Input format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`). Auto-detected from extension if omitted.         |
+| `-i` / `--input-format` | Input format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`, `sqlite`, `duckdb`). Auto-detected from extension if omitted. SQLite and DuckDB input use `{url}#{table_name}`. |
 | `-o` / `--output-format` | Output format (`parquet`, `csv`, `tsv`, `jsonl`, `avro`). If not specified, print Rich table in terminal. |
 | `--sql`                 | SQL query to apply after concatenation. The table is available as `t`. Mutually exclusive with `--jmespath`. |
 | `--jmespath` / `--jp`   | JMESPath expression to apply to each row as JSON. Object outputs become columns; non-object outputs go to a `value` column. The result shape must stay consistent across rows. |
@@ -90,5 +100,5 @@ Options:
 
 | Option                  | Description                                                                                                                  |
 |-------------------------|------------------------------------------------------------------------------------------------------------------------------|
-| `--az-url-authority-is-account` | Interpret az:// URL authority as storage account name instead of container name. See [azure.md](Azure) for more information. |
-| `--log-level`               | Log level from `{DEBUG, INFO, WARNING, ERROR, CRITICAL}`. If omitted, uses `log_level` from config.                         |
+| `--az-url-authority-is-account` | Interpret az:// URL authority as storage account name instead of container name. See [Cloud](cloud.md) for more information. |
+| `--log-level`               | Log level from `{DEBUG, INFO, WARNING, ERROR, CRITICAL}`. If omitted, uses `log_level` from config.                          |

{tab_cli-0.1.9 → tab_cli-0.2.1}/docs/cloud.md

@@ -9,6 +9,20 @@ tab view az://$container/$path
 tab view abfss://$container@$account.dfs.core.windows.net/$path
 ```
 
+Cloud paths may also use glob patterns:
+
+```bash
+tab view 's3://$bucket/path/date=*/*.parquet'
+tab view 'gs://$bucket/path/date=2026-01-*/*'
+tab summary 'az://$container/path/date=*/*.jsonl'
+```
+
+Install all optional integrations with:
+
+```bash
+pip install 'tab-cli[all]'
+```
+
 
 ## AWS S3
 

{tab_cli-0.1.9 → tab_cli-0.2.1}/docs/configuration.md

@@ -27,7 +27,7 @@ EOF
 | `az_url_authority_is_account` | `bool` | `false` | Interpret `az://` URL authority as storage account name instead of container name. |
 | `default_num_view_rows` | `int` | `20` | Default number of rows shown by `tab view` when `--limit` is omitted. |
 | `log_level` | `str` | `"INFO"` | Default CLI log level when `--log-level` is omitted. |
-| `max_cell_length` | `int \| null` | `null` | Default maximum cell length for `tab view`. The CLI `--max-cell-len` flag overrides it. |
+| `max_cell_length` | `int \| null` | `null` | Default maximum cell length for `tab view`. The CLI `--max-cell-length` flag overrides it. |
 | `num_remote_workers` | `int` | `8` | Maximum worker threads for remote per-partition summary work such as Parquet row counts. |
 | `sampling_size_for_schema_inference` | `int` | `32` | Number of rows sampled for schema inference (e.g. when using `--jp`). |