[SomethingNew71]

Fixes #78

Problem

MHD Tuning CSV exports opened with zero channels — the app appeared to do nothing.

The file from the issue looks like this:

<BOM>#Encoding: UTF-8
#Ecu CALID:
#Ecu PRGID: 7616431
#VIN: ...
Time,Amb pressure (Bar),Accel Ped. Pos. (%),Boost (Bar),...,MHD 5.26 IJE0S ST2V7.bin
539.607,1.008,16.1,-0.479,...

Two things broke it:

1. Every text parser inspects the first line to fingerprint the format. That line is #Encoding: UTF-8, not the header row, so no parser matched and detection fell through to the Haltech default — which only accepts HH:MM:SS data rows and returned 0 channels, 0 points.
2. The leading UTF-8 BOM was left attached to the first line (only UTF-16 BOMs were handled centrally; MegaSquirt and Woolich each stripped the UTF-8 one themselves).

@andris-silis reported the same failure with OBDLink exports, which use the same # preamble convention.

Fix

Central preprocessing in UltraLogApp::parse_text_content (src/app.rs), so no parser has to repeat it:

• strip a leading UTF-8 BOM
• drop a leading block of #-prefixed comment lines via the new parsers::strip_leading_comment_lines, so the real CSV header row is the first line detection sees

That alone fixes the OBDLink class of failure. Only the leading contiguous block is removed — a # after the header row stays as data.

New MHD parser (src/parsers/mhd.rs). Without it, MHD logs would now be picked up by RomRaider, which treats Time as milliseconds and would compress a 29-second pull into 29 ms. The parser:

• detects on the MHD-specific #Ecu CALID / #Ecu PRGID markers, or on the trailing tune column (MHD 5.26 IJE0S ST2V7.bin)
• reads units from the header and normalizes MHD's degree notation: (*C) → °C, (*) → °, (*CRK) → °CRK, (-) → unitless. rfind on the parens keeps names like MAF Req (wgdc) (g/s) intact
• treats Time as seconds, normalized relative to the first sample
• carries the last known value over unparseable fields, matching the Haltech/ECUMaster contract, so columns stay aligned
• drops out-of-order rows so times stays sorted (computed-channel time-shift lookups binary-search it)
• captures #Encoding / #Ecu CALID / #Ecu PRGID / #VIN and the tune name into MhdMeta

MHD detection deliberately runs on the raw text, ahead of the chain, because its fingerprint lives in the comment block the strip removes. That invariant is documented in CLAUDE.md.

Version

2.11.0 → 2.12.0. 2.11.0 was never released (latest tag is v2.10.1), but this adds a user-facing ECU format, so it gets its own minor.

Verification

cargo run --bin test_parser against the exact file attached to the issue:

Detected: MHD Tuning format
ECU Type: MHD
Channels: 31
Data points: 228
Time range: 0.000 to 29.438 seconds

  1. Amb pressure [Bar]
  5. Coolant [°C]
 13. Gear
 15. Lambda bank 1 [AFR]

228 data rows out of 233 file lines (4 comment + 1 header) — nothing dropped. Spot-checked value alignment against the raw row.

• cargo test — all pass, including 6 new MHD integration tests and 13 new unit tests
• cargo fmt --all -- --check — clean
• cargo clippy --all-targets --all-features — no findings in any changed file
• cargo build --release — clean

Notes

• The example log at exampleLogs/mhd/mhd-n55-pull.csv is the reporter's file with the VIN redacted — it is real customer data and the VIN identifies their car.
• cargo clippy --all-targets --all-features -- -D warnings currently fails on pre-existing findings in src/analysis/derived.rs, src/expression/engine.rs, tests/parsers/romraider_tests.rs, and dead code in tests/common/mod.rs. These are on main already (newer clippy than CI's toolchain) and are out of scope here.
• The README documents formats individually and is already missing DynamicEFI, Locomotive, and Motorsport Electronics sections. I added MHD only rather than backfilling the others.

[Copilot]

Pull request overview

This pull request fixes issue #78 by ensuring text-based log format detection sees the real CSV header (not a leading UTF-8 BOM or #-prefixed metadata preamble) and by adding a dedicated MHD Tuning CSV parser so MHD logs aren’t misdetected as other Time-prefixed formats.

Changes:

• Add central text preprocessing in UltraLogApp::parse_text_content to strip UTF-8 BOM and drop leading # metadata blocks before running the existing text parser detection chain (with an MHD-specific raw-text exception).
• Add a new src/parsers/mhd.rs parser (with unit normalization and metadata capture) and integration/unit tests for detection and parsing behavior.
• Update docs/examples and bump the crate version to 2.12.0.

Reviewed changes

Copilot reviewed 12 out of 13 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
tests/parsers/mod.rs	Registers the new MHD parser integration test module.
tests/parsers/mhd_tests.rs	Adds integration tests for MHD detection/parsing and comment-block stripping behavior.
tests/common/mod.rs	Adds the MHD example log path constant used by tests.
src/parsers/types.rs	Wires Mhd into Channel, Meta, and EcuType.
src/parsers/mod.rs	Exposes the new MHD parser and adds strip_leading_comment_lines() helper + tests.
src/parsers/mhd.rs	Implements the MHD Tuning CSV parser and its unit/detection logic.
src/bin/test_parser.rs	Mirrors the app’s preprocessing and adds explicit MHD detection/parsing in the CLI utility.
src/app.rs	Adds preprocessing + MHD special-case dispatch in parse_text_content.
README.md	Documents MHD support, updates the tree listing, and adds an MHD trademark note.
exampleLogs/mhd/mhd-n55-pull.csv	Adds a VIN-redacted MHD sample log used for tests/verification.
CLAUDE.md	Updates repo documentation to include MHD and documents the text preprocessing invariant.
Cargo.toml	Bumps version to 2.12.0.
Cargo.lock	Updates locked ultralog version to 2.12.0.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.