Why is the API mad?
Top fixes when the scout throws shade. The dojo rule: don’t guess. Print one helpful thing, learn one fact, move forward.
Print debugging: your starter kit
print() debugging is just asking questions out loud. If you’re stuck, you need a fact. Prints are how
you get facts.
HTTP receipt (copy/paste into your project)
Print the URL and status every time. Print a tiny body slice only on failure.
def print_receipt(url, params, r, *, source='NETWORK'):
print('SOURCE:', source)
print('URL:', url)
print('PARAMS:', params)
print('STATUS:', r.status_code)
if r.status_code != 200:
print('BODY:', r.text[:200])Nested JSON: the “inspection ladder”
Don’t dive. Take stairs.
print(data.keys())
print(data['info'].keys())
print('participants:', len(data['info']['participants']))Small, high-signal prints
print(type(x))
print(len(x))
print(list(x.keys())[:10])
print(repr(x)[:200])The “everything is 403” moment
99% of the time: your dev API key expired (it’s a ~24h key). Regenerate it and rerun Step 02.
401 vs 403
- 401: missing/invalid auth (wrong header/key)
- 403: auth present but expired/not allowed
429 rate limit
- Fetch fewer matches.
- Add a small sleep + retry.
- Rely on disk cache for reruns.
Getting [] for match IDs
Usually: wrong routing value for match-v5, or overly strict filters. Print the exact host and URL you’re calling.
One-line self-check
print('HOST CHECK:', base_url)CSV looks weird
Don’t fix by vibes. Print the table facts.
Pandas sanity prints
print(df.shape)
print(df.dtypes)
print(df.head(5).to_string(index=False))
print('win unique:', sorted(df['win'].dropna().unique().tolist())[:10])
print('game_start min/max:', df['game_start'].min(), df['game_start'].max())Plot looks like spaghetti
Time plots need sorted time and real datetimes. Plotly is just the messenger. Your dataframe is the message.
Pre-plot checklist (copy/paste)
df = df.sort_values('game_start')
print(df[['game_start', 'win']].head(10).to_string(index=False))
print(df[['game_start', 'win']].tail(10).to_string(index=False))