60 lines
5.2 KiB
Markdown
60 lines
5.2 KiB
Markdown
# Repo Working Notes
|
|
|
|
## Test And Debugging Baseline
|
|
|
|
- Use Docker/Linux for installs, builds, and tests by default.
|
|
- Treat the CI environment as the source of truth:
|
|
- Ubuntu/Linux
|
|
- Node 24
|
|
- `npm ci`
|
|
- `npm test`
|
|
- Dev server with auto-rebuild: `npm start` (port 8080). Production build: `npm run build` (output in `build/prod/`). If the production build OOMs, set `NODE_OPTIONS=--max_old_space_size=2048`.
|
|
- Do not spend time fixing Windows-only runtime or dependency issues unless explicitly requested.
|
|
- Do not commit repo changes whose only purpose is to make local Windows execution work.
|
|
- If a failure appears only in the local Windows shell, do not treat it as a code regression until it reproduces in Docker/Linux.
|
|
- When Docker is unavailable, restore Docker availability first rather than switching to Windows-specific debugging.
|
|
|
|
## Session Start
|
|
|
|
- At the start of a session, sync with `origin/master` before doing substantive work.
|
|
- Preferred command: `git pull --rebase origin master`
|
|
- Only do this automatically when the worktree is clean.
|
|
- If there are local changes already present, do not pull/rebase blindly; inspect first and avoid overwriting user work.
|
|
|
|
## Code Style
|
|
|
|
Follow `CONTRIBUTING.md` coding conventions: 4-space indentation, CamelCase class identifiers, camelCase function/variable names, UNDERSCORE_UPPER_CASE constants, UTF-8 source encoding, UNIX line endings, all files end with a newline.
|
|
|
|
## Commit Scope
|
|
|
|
- Keep commits small and reviewable by default.
|
|
- Prefer one commit per individual recipe change when that is practical.
|
|
- Otherwise group a commit around one coherent class of change, not multiple unrelated fixes or refactors.
|
|
- Split work before committing when a reviewer would benefit from evaluating the pieces independently.
|
|
- Only keep changes together when separating them would make the behavior harder to understand, test, or revert.
|
|
- Prefer squash or amend for related consecutive changes — if a follow-up commit only fixes or extends the immediately preceding commit, squash them into one rather than leaving a trail of iterative noise in the log.
|
|
|
|
## Payment Operation Maintenance
|
|
|
|
After completing any substantive payment operation work, ask: *"Did I learn anything in this session that isn't captured in AGENTS.md?"* If yes, add it before committing.
|
|
|
|
When adding, renaming, or removing a payment operation:
|
|
|
|
1. **Update `PAYMENT_RECIPES.md`** — add the operation to the correct numbered section and, if it introduces a new chaining pattern, add a lettered chaining pattern entry. Remove or mark deprecated any operations that are replaced.
|
|
2. **Follow the naming convention** — all payment operation display names use Title Case. Acronyms (DUKPT, AES, EMV, MAC, PAN, TR-31, TR-34, KCV) stay upper-case. Brand names keep their canonical form (`payShield`). Pattern: `[Domain Prefix] [Verb] [Qualifier]` — the domain/protocol prefix comes first so operations sort and scan by topic in the UI list. Example: `EMV Verify MAC`, `DUKPT Derive TDES Key`, `PIN Block Parse`. When a vendor name is a sub-specifier of a PIN method, embed it after the PIN domain prefix: `PIN IBM 3624 Offset Generate`, `PIN IBM 3624 Verify`. See the Naming Convention section in `PAYMENT_RECIPES.md`.
|
|
3. **Only operations written for this fork belong in the Payments category** — do not add upstream CyberChef ops (AES Encrypt, HMAC, CMAC, Triple DES Encrypt, AES Key Wrap, etc.) even as convenience shortcuts. If an op wasn't authored here, it stays in its own upstream category only.
|
|
4. **Keep `this.name` and file name consistent** — the CyberChef UI shows `this.name`; the file name is the class name in PascalCase. Both should reflect the same intent.
|
|
5. **Do not rename `this.name` without updating `PAYMENT_RECIPES.md`** — stale names in the doc are confusing and break recipe search.
|
|
6. **Review and update `this.description`, `this.inlineHelp`, and `this.testDataSamples`** whenever changing a recipe — operation descriptions, inline help text, and sample args must stay consistent with the current arg list and behavior. A renamed arg, added arg, or changed default silently breaks the tooltip if the description still references the old shape.
|
|
7. **Regenerate the build config after any add, rename, or delete** — three files are gitignored and auto-generated; editing `this.name` or `Categories.json` alone is not enough:
|
|
- `src/core/operations/index.mjs` — full op list; built by `generateOpsIndex.mjs`
|
|
- `src/core/config/modules/Payment.mjs` — maps `this.name` → constructor for the Payment module chunk; built by `generateConfig.mjs`
|
|
- `src/core/config/OperationConfig.json` — op metadata for the UI
|
|
Run from the project root after any op change:
|
|
```
|
|
node src/core/config/scripts/generateOpsIndex.mjs && node src/core/config/scripts/generateConfig.mjs
|
|
```
|
|
Or `npx grunt dev` / `npx grunt prod`, which runs both steps automatically. CI runs them on every build. **Symptom of a stale registry:** `TypeError: f[e.module][e.name] is not a constructor` at runtime.
|
|
**Grunt alias:** if using grunt tasks directly, the correct task is `npx grunt exec:generateConfig`. `npx grunt exec:generateNodeIndex` is a *different* task — it only regenerates the Node API wrapper (`src/node/index.mjs`) and does NOT update `OperationConfig.json` or `modules/Payment.mjs`.
|
|
|