Skip to main content
Version: next

DBC Guide

CanMessenger::DBC parses DBC text and maps signal values to raw CAN bytes.

Load a DBC File​

dbc = CanMessenger::DBC.load('example.dbc')

You can also parse inline content:

dbc = CanMessenger::DBC.new(File.read('example.dbc'))

Supported DBC Elements​

Current parser support is intentionally narrow and focused:

  • BO_ message definitions.
  • SG_ signal definitions in the common start|length@endian+/- (factor,offset) pattern.
  • Ignores standard metadata directives such as VERSION, NS_, BS_, BU_, CM_, VAL_, BA_, BA_DEF_, BA_DEF_DEF_, and BO_TX_BU_.
  • Accepts SIG_VALTYPE_ declarations for integer type 0. Types 1 (IEEE float) and 2 (IEEE double) are unsupported and raise ArgumentError instead of being interpreted as integer signals.

Malformed BO_/SG_ lines and unknown non-empty directives, including unsupported BA_* variants, now raise ArgumentError with line context. Signal byte order must be @0 or @1; scaling factors and offsets must be valid numbers, and the factor must be nonzero. Signed decimals and scientific notation are supported. If your DBC uses advanced constructs (for example multiplexing or uncommon syntax variants), validate with tests before relying on runtime behavior.

Encode by Message Name​

frame = dbc.encode_can('Example', Speed: 42)
# => { id: 256, data: [42, ...] }

Validation performed during encoding includes:

  • Unknown message name rejection.
  • Signed/unsigned range checks.
  • Signal bit-bound checks against message DLC.

Decode by CAN ID​

decoded = dbc.decode_can(frame[:id], frame[:data])
# => { name: 'Example', signals: { Speed: 42.0 } }

Returns nil if no message with matching CAN ID is found.

Integral factors and offsets preserve integer precision, including signed and unsigned 64-bit signals. This also applies to equivalent decimal or scientific notation such as (1.0,0e0). Decoded values with integral scaling are integers; fractional scaling produces floating-point values. Encoding rounds to the nearest raw integer, with halfway values rounded away from zero.

Big-endian and Little-endian Signals​

  • @1 is treated as little-endian.
  • @0 is treated as big-endian (Motorola/sawtooth indexing).

Current releases include corrected multi-byte big-endian bit mapping and stricter unsigned overflow checks.