Migrate from fitparse¶

Most fitparse scripts translate line-for-line. The mental shift: fitparse gives you messages to iterate; chiptime also gives you the interpreted workout (sessions, streams, totals) so most loops disappear.

fitparse is a long-standing, widely used library — this guide is mechanics, not a verdict. If your current setup works, it works.

API mapping¶

fitparse	chiptime
`FitFile("a.fit")`	`chiptime.parse("a.fit")`
`ff.get_messages("record")`	`(m for m in result.messages if m.name == "record")` — or skip straight to streams
`rec.get_value("heart_rate")`	`msg.get("heart_rate")`
`field.value`, `field.raw_value`	`fv.value`, `fv.raw` (`msg.fields["heart_rate"]`)
`field.units`	`fv.units` or `stream.units`
`FitFile(..., check_crc=False)`	default `lenient` mode (CRC triaged + reported, not fatal)
`try: ... except FitParseError:`	check `result.ok` / `result.errors` (nothing raises in lenient)
manual semicircle → degrees	`position_lat`/`position_long` streams already in degrees
loop-and-collect per field	`session.records.stream("power").values` — already columnar

Before / after¶

fitparse

import fitparse

ff = fitparse.FitFile("ride.fit", check_crc=False)
hr = []
for record in ff.get_messages("record"):
    v = record.get_value("heart_rate")
    if v is not None and v != 255:          # sentinel guard, by hand
        hr.append(v)
avg = sum(hr) / len(hr) if hr else None

chiptime

import chiptime

result = chiptime.parse("ride.fit")          # damaged files included
s = result.activity.sessions[0]
avg = s.derived.avg.get("heart_rate")        # sentinels were never values

What changes¶

Files that raise in fitparse may parse in chiptime's lenient mode — check result.ok and provenance[] instead of wrapping in try/except.
None always means absent in chiptime streams; zero is a real value.
Multi-part files and gzip-wrapped exports are unwrapped automatically.
If you want raise-on-problem semantics: chiptime.parse(data, mode="strict") raises coded exceptions on the first violation.

Streaming equivalence¶

If you used fitparse for constant-memory iteration, the direct analog is:

for msg in chiptime.iter_messages("big.fit"):
    if msg.name == "record":
        ...