Mapping — Concept2

Source: a Concept2 Logbook (log.concept2.com) season CSV export — one row per workout, covering RowErg, SkiErg, and BikeErg. Pillars: A + B. Mapper: mapConcept2 — mapConcept2(csv) → { records, warnings }, one Session per row plus an avg-HR Measurement per workout.

Structural correspondence

Concept2 CSV	OpenBody
Log ID	clientRecordId; Description → Session.name; Comments → notes
Type (RowErg/SkiErg/BikeErg)	disciplines (rowing / concept2:skierg / cycling) and exerciseRef (row.erg / ski.erg / opaque-only — see notes)
a fixed piece ("2000m row", "30:00 ski")	a distance- or time-scored WorkUnit; the piece’s other metric rides in extension.concept2 (§5.5 — a non-continuous unit can’t carry a metric that contradicts its scoring)
a fixed-interval workout ("8x500m/0:30r row")	a Block of per-interval WorkUnits, each with performance.rest — the structure is recovered from the PM5-generated Description
a “just row” (arbitrary end)	a continuous WorkUnit carrying time + distance + energy together
Stroke Rate/Cadence / Avg Watts	performance.intensity ({ dimension: "cadence" } / { dimension: "power" }) on single-piece workouts (§5.13)
Avg Heart Rate	a Pillar A heart_rate_mean Measurement spanning the workout, linked via measuredBy (§7.2)
pace, stroke count, drag factor, cal/hr, …	extension.concept2 (canonical-plus-residue)

Input (excerpt of examples/concept2/concept2-season-sample.csv)

"Log ID",Date,Description,"Work Time (Formatted)","Work Time (Seconds)",...,"Work Distance",...,"Stroke Rate/Cadence",...,"Avg Watts",...,"Avg Heart Rate","Drag Factor",...,Type,...,Comments,...
90000001,"2026-03-02 06:45:00","2000m row",7:45.3,465.3,,,2000,,28,217,1:56.3,221,1063,137,172,118,40,Hwt,RowErg,Yes,"2k test — season best attempt",...
90000002,"2026-03-04 07:10:00","8x500m/0:30r row",15:20.0,920.0,4:00.0,240,4000,184,30,460,1:55.0,228,1085,185,158,116,40,Hwt,RowErg,No,,...

Output (the 2k test — a fixed-distance piece)

{
  "id": "c2-90000001",
  "recordType": "Session",
  "subject": "subj-001",
  "clientRecordId": "90000001",
  "name": "2000m row",
  "notes": "2k test — season best attempt",
  "disciplines": ["rowing"],
  "intent": "train",
  "startTime": "2026-03-02T06:45:00Z",
  "endTime": "2026-03-02T06:52:45Z",
  "provenance": { "method": "sensor", "sourceApp": "concept2", "device": { "manufacturer": "concept2", "model": "RowErg" } },
  "workUnits": [
    {
      "id": "c2-90000001-wu",
      "recordType": "WorkUnit",
      "exerciseRef": { "id": "row.erg", "opaque": "RowErg" },
      "scoring": "distance",
      "performance": {
        "distance": { "absolute": { "value": 2000, "unit": "m" } },
        "intensity": [
          { "dimension": "cadence", "unit": "/min", "value": { "absolute": { "value": 28 } } },
          { "dimension": "power", "unit": "W", "value": { "absolute": { "value": 221 } } }
        ]
      }
    }
  ],
  "links": [{ "type": "measuredBy", "ref": "c2-90000001-hr" }],
  "extension": { "concept2": { "pace": "1:56.3", "strokeCount": 217, "calHour": 1063, "dragFactor": 118, "workTimeSeconds": 465.3, "totalCal": 137 } }
}

The interval row ("8x500m/0:30r row") expands to a Block of eight WorkUnits:

{
  "id": "c2-90000002-blk",
  "recordType": "Block",
  "name": "8x500m/0:30r row",
  "children": [
    {
      "id": "c2-90000002-int1",
      "recordType": "WorkUnit",
      "exerciseRef": { "id": "row.erg", "opaque": "RowErg" },
      "scoring": "distance",
      "performance": {
        "distance": { "absolute": { "value": 500, "unit": "m" } },
        "rest": { "absolute": { "value": 30, "unit": "s" } }
      }
    }
  ]
}

Piece scoring is inferred, honestly

"2000m row" whose Work Distance is exactly 2000 ⇒ a fixed-distance piece (scoring: "distance"); "30:00 ski" whose Work Time is exactly 1800.0 ⇒ fixed-time; anything else ends at an arbitrary point ⇒ continuous, which per §5.5 MAY carry time + distance + energy together. The season CSV is one row per workout with no per-interval detail rows — fixed-interval structure is recoverable because the PM5 generates the Description deterministically; variable-interval workouts ("v2000m/3:00r...3") only disclose the first interval, so they degrade to a single continuous WorkUnit with the rest totals preserved as residue.

Notes

• Machine identity: RowErg ⇒ registry row.erg, SkiErg ⇒ ski.erg (both verified in the registry); BikeErg has no canonical registry id (the registry’s air-bike is a fan bike) so it stays opaque-only. The raw Type string always rides in exerciseRef.opaque.
• SkiErg’s discipline is concept2:skierg — there is no canonical skierg discipline token, and skiing would misstate an indoor erg (§5.9 namespaced fallback).
• For interval workouts the CSV’s stroke rate/watts are whole-workout averages — asserting them on any single interval would fabricate per-interval data, so they land in extension.concept2 instead.
• Timestamps are wall-clock with no offset — pass utcOffset to stamp them (default "Z").