Trainheroic Migration Prep

3. Apply deterministic mapping rules.

• Use mapping-rules.md.
• Parse supported set shapes for v0:
  • reps x load (rep x pound, rep x kilogram)
  • reps x seconds (rep x second)
  • seconds-only (second x)
  • reps-only (rep x)
  • seconds x pounds (second x pound)
• Skip unsupported patterns (distance/percentage/empty templates) without failing the whole run.

4. Build canonical records.

• Create canonical exercises from ExerciseTitle.
• Group workouts by (WorkoutTitle, ScheduledDate, RescheduledDate).
• Keep source row order for order_index and set expansion order for set_index.

5. Emit import-ready files.

• Follow output-contract.md exactly.
• Output CSV files matching Train table columns.
• Include deterministic IDs and FK IDs so files can be imported directly into Supabase in order.
• Include unmapped_rows.csv and run-report.json.

6. Validate consistency before returning.

• Every workout_exercises.workout_id exists in workouts.csv.
• Every workout_exercises.exercise_id exists in exercises.csv.
• Every exercise_sets.workout_exercise_id exists in workout_exercises.csv.
• order_index and set_index are positive and gapless per parent.
• Report skipped row count and reasons.

Guardrails

• Preserve source values where possible; do not invent reps/load/seconds.
• Never coerce unsupported units (distance/percentage) into supported units.
• If parsing is ambiguous, skip row and record reason in unmapped_rows.csv.
• Prefer explicit null (empty CSV value) for unknown optional fields.
• Keep output deterministic across runs given same input and user_id.

Output

Return:

1. File list created.
2. Row counts per output file.
3. Skipped row summary (with reasons).
4. Any schema-risk warnings before import.

Use output-contract.md as the authoritative format.