antares_get_data/.github/copilot-instructions.md

Copilot Instructions

Build, test, and lint

• cargo build
• cargo run - requires a local .env file with ANTARES_USERCODE and ANTARES_PASSWORD
• cargo test
• cargo test <test_name> to run one test
• cargo test -- --list to discover test names; the repository currently has no defined tests
• cargo fmt --all -- --check
• cargo clippy --all-targets -- -D warnings

High-level architecture

This is a modular single-binary Rust utility that fetches Antares B2B product data and exports it to Excel. The codebase is organized into separate concerns:

Modules:

• src/main.rs - orchestrates the pipeline: credential loading, API calls, JSON deserialization, and delegating to tools
• src/api/client.rs - handles URL construction for the Antares B2B API
• src/tools/excel.rs - implements Excel export with selective field mapping and business rule validation
• src/tools/logger.rs - provides dual output (terminal + file) logging to log/{YYYY-MM-DD}.log
• src/template/antares.rs - serde models with #[serde(rename_all = "PascalCase")]; top-level is Antares alias for Vec<AntaresItem>; also contains AntaresLogin struct holding credentials and base URL

Pipeline:

1. Initialize logger (creates log/ dir if needed)
2. Load credentials from .env via dotenv
3. Build Antares API URL using src/api/client::make_url() (takes an AntaresLogin struct)
4. Fetch with 600-second timeout via blocking reqwest::blocking::Client (in src/tools/request.rs)
5. Deserialize into typed Antares structs; if schema mismatch, preserve response as JSON/text
6. Write antares.json with pretty formatting
7. Export filtered rows to out/antares_export.xlsx via src/tools/excel::export_to_excel()
8. Log success/failure at each step

Logging:

• All significant events logged to terminal and log/{YYYY-MM-DD}.log
• Specialized methods: log_api_success(), log_api_failure(), log_export_success(), log_export_failure()
• Directory created automatically if missing; logs appended daily by date

Key conventions

Credentials:

• Load via URL, ANTARES_USERCODE, ANTARES_PASSWORD, and OUT env vars from .env
• URL is the Antares B2B base URL; OUT is the output file path for the Excel export
• .env.example documents required variables
• No hardcoded credentials; missing vars cause graceful failure with instructions
• PASSWORD is wrapped in Zeroizing<String> (from the zeroize crate) to securely wipe it from memory after use

Artifacts:

• temp/antares.json - saved in temp/ after each successful API fetch; directory created auto if missing
• Excel export path is configured via the OUT env var; parent directory created auto if missing
• log/{YYYY-MM-DD}.log - daily logs; log/ directory created auto if missing
• .gitignore excludes: *.json, *.xlsx, .env, /out, /log

Excel export (src/tools/excel.rs):

• Not a generic dump; maps selected Antares fields to fixed column layout per business rules
• Rows skipped if cikkszam empty OR both name fields (cikk_megnevezes_rovid, cikk_megnevezes) empty
• BESZCIKKNEV prefers cikk_megnevezes_rovid, falls back to cikk_megnevezes
• EGYSEGAR prefers netto_kisker_ar; if 0.0, searches cikk_jellemzok for entry with jellemzo_nev == "Alap ár"
• Headers are bold-formatted

API integration:

• When extending the schema, add/modify structs in src/template/antares.rs first
• Avoid ad hoc JSON traversal in main.rs; keep deserialization flow centered on typed structs
• Preserve response on schema mismatch (save as JSON or text) unless deliberately changing error handling

Modularity:

• Extend by adding new files to src/tools/ (e.g., csv.rs, database.rs)
• Each tool should be self-contained and exposed via src/tools/mod.rs
• HTTP client logic stays in src/tools/request.rs; reuse make_request() from there for consistency

Blocking HTTP:

• Uses reqwest::blocking::Client with 600-second timeout (in src/tools/request.rs)
• No async runtime; synchronous flow is intentional
• Only change to async if architecture overhaul is deliberate