Tachograph parser integration

1. Scope

FLEET Mira parses driver-card C files and vehicle-unit M files through one TachographImportService. The original compressed archive object remains the evidence source; parsed tables are replaceable derived data.

2. Import pipeline

flowchart TD
    A[Archive ID] --> B[Read and verify GZIP/hash]
    B --> C[Classify binary content]
    C --> D[Parse C or M structure]
    D --> E[Verify certificates and signed sections]
    E --> F{Result}
    F -->|Valid| G[Atomically replace derived rows]
    F -->|Invalid/unverifiable| H[Persist quarantine results]
    F -->|Technical failure| I[Keep previous valid details]
    G --> J[Links, deadlines, UI cache invalidation]
    H --> J

Public integration points:

TachographImportService(db_path, logger).import_archive_id(archive_id)
DriverCardFullParser.parse_bytes(data)
VehicleUnitParser.parse(path)
DriverCardSignatureVerificationService.verify(parsed, trust_store, reference_time)
SignatureVerificationService.verify_sections(parsed, trust_store)

3. Classification

Classification uses protocol/file identifiers in the binary content, never the filename alone. Result values are DRIVER_CARD, VEHICLE_UNIT and UNKNOWN. C/M filename prefixes are validated only for readout-date and plausibility functions.

4. Driver-card data

The C parser preserves block kind, application/file ID, generation, order, raw data bytes and following signature bytes. It maps identity, card validity, authorities, driving-licence number, activities, events, faults, vehicles and positions.

Important identity mapping:

Parser value	Business use
holder surname/first names	person conflict check and display
birth date	optional person value
card number/generation	card link and C/M matching
card issue/valid-from/expiry	card record and expiry deadline
driving-licence number/authority	person licence data
preferred language/country	card metadata

No guessed values are generated from unrelated byte ranges. Text fields strip encoding indicators/control bytes and normalise whitespace without changing meaningful characters.

5. Vehicle-unit data

Supported download groups are G1 TREP 01–05, G2 v1 21–25 and G2 v2 31–35. Derived tables include overview/identity, card sessions, slot activities, events, faults, overspeed, second-by-second speed, technical/calibration records and position/border/loading data.

Unknown bounded optional records are retained as warnings. Invalid length/protocol sequences produce parse failure. Driver/workshop names use the same normalisation rules as C text.

6. Signature verification

C files

• G1: EF.Card_Certificate, EF.CA_Certificate, RSA PKCS#1 and SHA-1 over exact data-block bytes.
• G2/G2V2: Card_Sign, MSCA_Card, optional link certificate and ECDSA over exact signed EFs.

M files

Every available signed VU download section is verified separately against the applicable ERCA/MSCA/VU chain. Trust and certificate validity use the best plausible signed file time, otherwise archive/download time with warning.

Overall priority: INVALID, then UNVERIFIABLE, then VALID. cryptography missing, signature/certificate missing or trust anchor missing produces UNVERIFIABLE, not VALID.

7. Parser and quarantine status

Status	Meaning
PARSED_VERIFIED	parse and all required verification valid
PARSED_VERIFIED_WITH_WARNINGS	valid evidence with non-fatal parser warnings
QUARANTINED_INVALID_SIGNATURE	cryptographic invalidity
QUARANTINED_UNVERIFIABLE	required proof cannot be completed
PARSE_ERROR	structural/technical parse failure

Quarantine files remain visible/exportable but cannot create trusted person/vehicle links, contribute to combined C/M analysis or compliance snapshots.

8. C/M comparison

The selected archive remains the base. Opposite-source intervals match by normalised card number, card generation and UTC overlap. A sweep-line comparison cuts only on interval boundaries. Valid overlapping data is shown as C and M available; one-sided coverage remains C-only or M-only. C data has priority only in the complete person-history view, never by modifying raw source rows.

9. Reprocessing and idempotency

Archive hash prevents duplicate originals. Reprocessing writes new derived rows only after a successful parse/verification transaction. A failed reparse preserves the last valid details. Import, deletion, restore or reparse invalidates affected UI comparison caches.

10. Extending the parser

When adding a record type:

1. document normative file/record identifiers and generation;
2. validate lengths before reading;
3. preserve source section/index and raw reference;
4. keep UTC values unchanged;
5. add fixture and malformed-length tests;
6. add signature impact tests;
7. update both language versions of this document and API schemas;
8. never infer a business identity from an ambiguous value.

The older technical module and table names containing ddd or DLTNG remain compatibility identifiers.