feat: byte-driven BLAKE3 hashing progress bar, TTY detection & --quiet flag

[Itzzavdheshh]

📌 Related Issue

Fixes #46

---

📝 Description

🔹 What has been changed?

src/hasher.rs

• Global thread-safe QUIET AtomicBool added with set_quiet() and is_quiet() helpers — worker threads read it directly, no state propagation needed
• full_hash() updated to accept Option<&indicatif::ProgressBar> — bar advanced by exact bytes read per chunk:

if let Some(bar) = pb {
    bar.inc(read as u64);
}

• sparse_hash() signature left unchanged — small files (≤12KB) that fall back to full hashing internally are excluded from progress bar to prevent double-counting

src/main.rs

• Global -q / --quiet CLI flag added via Clap derive API — documented in custom GLOBAL FLAGS help template
• std::io::IsTerminal check on stdout — non-TTY (piped) automatically activates quiet mode; zero ANSI pollution in pipes
• byte_progress() helper builds bar with ProgressDrawTarget::stderr() + {bytes}/{total_bytes} ({bytes_per_sec}, {eta}) template
• scan_pipeline accumulates total bytes dynamically as duplicate candidates are identified — bar length set in bytes not file count
• All spinners/bars switch to ProgressBar::hidden() when quiet or non-TTY stdout detected
• Unused file-count progress helper function removed — zero dead code warnings

tests/integration_tests.rs

• test_quiet_suppresses_progress_output — verifies -q produces no stderr progress output while asserting command success
• test_nontty_stdout_suppresses_progress_output — verifies piped execution auto-suppresses all progress rendering

---

🔹 Why are these changes needed?

• The BLAKE3 full-hash pass — the most I/O-intensive part of any run — ran completely silently; on large directories the tool appeared frozen or hung
• Progress bars wrote raw ANSI \r sequences to stdout — piped commands like bdstorage dedupe /path | grep something produced garbled, unparseable output
• No way to suppress progress output in scripts or CI — --quiet flag was entirely absent

---

🛠️ Type of Change

• 🐛 Bug Fix
• ✨ New Feature
• 🎨 UI/UX Improvement
• 📖 Documentation Update
• ⚡ Performance Improvement
• 🔧 Refactoring
• 💥 Breaking Change

---

🧪 Testing

cargo fmt --check
cargo clippy --all-targets --all-features -- -D warnings
cargo test

✅ Tests Performed

• Tested locally
• Build runs successfully
• No console errors
• Responsive design checked
• Existing functionality verified
• cargo fmt --check — passes clean, zero violations
• cargo test — 17/17 tests pass including 2 new integration tests
• Large directory run — byte-driven progress bar with MB/s + ETA visible
• Bar disappears cleanly on completion — no leftover ANSI artifacts
• bdstorage dedupe /path | grep something — clean output, zero garbled \r sequences
• -q flag — all progress output suppressed, command exits successfully
• Double-counting verified absent — sparse_hash excluded from bar correctly

🌐 Browsers Tested

N/A — CLI tool

---

📷 Screenshots / Demo

1. Before — full-hash stage runs silently, tool appears hung

2. After — byte-driven progress bar with MB/s throughput and ETA

3. Bar disappears cleanly on completion

4. bdstorage dedupe /path | grep something — clean output, zero ANSI pollution

5. -q flag — zero progress output on stderr

6. cargo test — 17/17 tests passing including new integration tests

---

📋 Checklist

• I have read the project's CONTRIBUTING guidelines
• My code follows the project style guidelines
• I have performed a self-review of my code
• I have tested my changes locally
• I have added/updated documentation where necessary
• My changes do not introduce new warnings or errors
• This PR is linked to an existing issue
• Progress bar driven by bytes — shows real MB/s throughput and accurate ETA
• All progress output bound to ProgressDrawTarget::stderr() — stdout untouched
• Non-TTY stdout auto-activates quiet mode — pipes never polluted
• -q / --quiet flag suppresses all progress output globally
• sparse_hash excluded from bar — no double-counting on ≤12KB files
• QUIET stored as AtomicBool — thread-safe, no complex state propagation
• cargo fmt --check — pass
• cargo test — 17/17 pass
• Unused progress helper removed — zero dead code warnings
• Existing dedupe and scan behaviour completely unchanged

---

💬 Additional Notes

• Why AtomicBool for QUIET? Worker threads in the hashing pool need to check quiet state without locks or Arc passing — a global atomic is the idiomatic Rust pattern for this.
• Why bytes not file count? File count gives misleading ETAs when files vary wildly in size. Bytes processed reflects actual I/O work done and maps directly to MB/s throughput.
• Why stderr for progress? stdout must stay clean for piping. ProgressDrawTarget::stderr() is the standard indicatif pattern — progress never touches stdout.
• Only 3 files changed — surgical, zero regressions on existing scan and dedupe flows.

---

🙌 Contribution Note

Hi @Rakshat28 👋

• ✅ Silent hashing fixed — byte-driven bar with MB/s + ETA on the full BLAKE3 pass
• ✅ Pipe pollution fixed — TTY check auto-suppresses all ANSI on non-TTY stdout
• ✅ --quiet flag added — scripts and CI can fully silence progress output
• ✅ Double-counting prevented — sparse_hash correctly excluded from bar
• ✅ 17/17 tests pass — 2 new integration tests cover quiet mode and non-TTY suppression
• ✅ 3 files changed — zero regressions

Happy to address any feedback! 🚀

---

🏷️ Labels

#feature #cli #rust #progress-bar #indicatif #blake3 #tty #quiet-mode #integration-tests #NSoC26 #nsoc #GSSOC

---

Submitted as part of Open Source Contribution — NSoC (Nexus Spring of Code) & GSSoC

[Itzzavdheshh]

Hi @Rakshat28 👋

• ✅ Silent hashing fixed — byte-driven bar with MB/s + ETA on the full BLAKE3 pass
• ✅ Pipe pollution fixed — TTY check auto-suppresses all ANSI on non-TTY stdout
• ✅ --quiet flag added — scripts and CI can fully silence progress output
• ✅ Double-counting prevented — sparse_hash correctly excluded from bar
• ✅ 17/17 tests pass — 2 new integration tests cover quiet mode and non-TTY suppression
• ✅ 3 files changed — zero regressions

Happy to address any feedback! 🚀

[Rakshat28]

Please change README.md to reflect the addition of the --quiet flag @Itzzavdheshh

[Itzzavdheshh]

wow @Rakshat28 , thanku for your kind whatsapp message , i think we should not be working together .... i request you to close this PR unmerged and unassign me from your remaining issues ... its been great working with you , Thanku