awss3tui 0.7

Textual TUI to navigate S3 across multiple AWS profiles

awss3tui

awss3tui is a Textual terminal UI (TUI) and CLI helper for working with Amazon S3 across multiple AWS profiles.

It provides:

• An interactive S3 browser (s3 with no subcommand)
• Lightweight wrappers for aws s3 ls, aws s3 cp, and aws s3 sync
• Automatic "best profile" detection for managing multiple roles to reduce profile/session friction

Install

From PyPI:

pip3 install awss3tui

From source (dev/local):

pip3 install -e .

This installs the s3 console script for the TUI and CLI helpers.

Quick Start

Launch the TUI:

s3

Open directly to a bucket/prefix:

s3 s3://my-bucket/path/
# or
s3 my-bucket/path

Force a reindex (bypass cache on startup):

s3 reindex

See more:

s3 --help

Requirements

• Python >=3.10
• AWS credentials/profiles configured (for example in ~/.aws/config)
• AWS CLI available in PATH for:
  • s3 login
  • automatic SSO login flows from the TUI
  • s3 generate-config
  • wrapper subcommands (ls, cp, sync)

Commands

Browser mode (default)

When the first argument is not a known subcommand, s3 treats it as TUI browse mode with optional startup path:

s3 [--profiles ...] [--profile ...] [--region ...] [path]

Notes:

• path may be s3://bucket/prefix or bucket/prefix.
• browse is accepted as an alias (s3 browse ...).

s3 reindex

Same as browse mode, but forces a fresh bucket/permission refresh:

s3 reindex [--profiles ...] [--profile ...] [--region ...] [path]

s3 ls

Wrapper around aws s3 ls:

s3 ls [path] [--dry-run] [any aws s3 ls options...]

Behavior:

• Adds --human-readable unless you already supplied it.
• Normalizes path to s3://... when needed.
• Unknown args are passed through to AWS CLI.
• --dry-run prints the resolved command and does not execute it.
• auto-selects AWS profiles for s3 ls, s3 cp, and s3 sync using its cached bucket-to-profile mapping, so you usually don’t need to pass --profile. If you do provide --profile, it is respected as an explicit override.

s3 cp

Wrapper around aws s3 cp:

s3 cp <source> <destination> [--dry-run] [any aws s3 cp options...]

Behavior:

• Source/destination are normalized:
  • Local-like paths (., .., absolute paths, ~/..., Windows drive paths) stay local.
  • Non-local paths are treated as S3 and normalized to s3://....
• --dry-run adds AWS CLI --dryrun unless already present, and still runs AWS CLI.
• Unknown args are passed through to AWS CLI.

s3 sync

Wrapper around aws s3 sync:

s3 sync <source> <destination> [--dry-run] [any aws s3 sync options...]

Behavior matches cp path normalization and dry-run handling (--dry-run maps to AWS CLI --dryrun).

s3 login

Runs AWS SSO login only for profiles that currently need reauthentication:

s3 login [--profiles dev,prod] [-p dev -p prod]

s3 generate-config

Generates/merges AWS SSO role profiles and prints resulting config to stdout:

s3 generate-config [--sso-session SESSION]

Details:

• Reads AWS config (defaults to ~/.aws/config, respects AWS_CONFIG_FILE).
• Reads SSO access tokens from ~/.aws/sso/cache.
• Discovers accounts/roles via AWS SSO APIs.
• Creates/updates profile blocks with sso_session, sso_account_id, sso_role_name, region=us-east-1, and output=json.
• Rewrites existing profile sso_session references to canonical session names when equivalent sessions exist.
• Exits with guidance if no valid cached SSO token exists for a session.
• Writes merged config to stdout (it does not directly edit files for you).

TUI Features

Core UI

• Left pane: bucket/prefix tree.
• Right pane:
  • file explorer table (prefixes/objects)
  • preview panel
• Draggable split handles for pane resizing.

Profile-aware bucket access

For each bucket, the app probes available profiles and picks the best effective profile:

• good: list + read object bytes
• no_download: list allowed but object read denied
• no_view: listing denied

You can manually switch profile per bucket from the profile button in the path bar.

SSO handling

• Startup preflight checks for expired/near-expired SSO tokens and can run aws sso login.
• Runtime operations retry once after SSO reauth when token-expired errors are detected.
• In-flight reauth per profile is deduplicated.

Preview and stats

• File preview loads first 4096 bytes, with m/More to fetch additional chunks.
• .gz files are previewed as decompressed text automatically.
• .sam, .bam, and .cram previews use samtools view -h behind the scenes and show a head snippet.
• If samtools is not available, those files fall back to the normal byte-based preview.
• Syntax highlighting is enabled automatically for Textual-supported languages (for example: JSON, YAML, TOML, XML, Markdown, Python, SQL).
• .csv and .tsv highlighting is enabled when tree-sitter-language-pack is installed (included as a project dependency in current versions).
• m/More is only available for normal byte-based previews and folder recursive scans.
• Folder preview shows immediate child stats.
• Press m on a folder preview to run recursive scan (up to 50,000 keys; reports partials when truncated).

Download flows

• Download selected file(s) to chosen path.
• Download entire prefix recursively.
• Multi-select in table:
  • click: single select
  • Ctrl/Cmd-click: toggle
  • Shift-click: range select

Favorites and filters

• Favorite a bucket with Ctrl+F.
• Toggle filters from the left pane:
  • hide no_view
  • hide no_download
  • hide empty buckets
  • show only favorites

Keybindings

Global/app:

• q: quit immediately
• Esc twice within 1 second: confirm quit
• r: refresh buckets
• Ctrl+L or /: focus path input
• Alt+Left / Alt+Right: back/forward navigation history
• Backspace: go to parent prefix
• Ctrl+F: toggle favorite bucket
• Space: preview selected row
• m: preview more / recursive scan

Widget behaviors:

• Tree:
  • Right: move focus to table
• Table:
  • Left: move focus to tree
  • Enter on bucket/prefix: open
  • Enter on object: download
  • double-click on bucket/prefix: open
• Path input:
  • Down: move focus to table
  • Enter: navigate to typed path

Local State and Cache

State is stored under:

• ${XDG_CONFIG_HOME:-~/.config}/awss/bucket-cache.json
• ${XDG_CONFIG_HOME:-~/.config}/awss/config.json

bucket-cache.json stores:

• resolved bucket -> profile mapping
• access classification
• empty-bucket flags
• save timestamp
• hash of ~/.aws/config + ~/.aws/credentials

Cache behavior:

• automatically reindexes every ~30 days
• invalidates immediately when AWS config/credentials hash changes
• startup may use cache first; reindex or r triggers live refresh
• if live refresh fails and cache exists, cached buckets can still be used

config.json stores:

• favorite buckets
• bucket filter toggles

Development

Run tests:

python -m unittest discover -s tests -p "test_*.py"

Current suite covers CLI dispatch/wrappers, service logic, and headless Textual behavior.