csvsight 0.1.0

Eyeball a CSV in seconds: rows × columns, per-column type, null %, ranges, and top values. Zero dependencies, no pandas.

csvsight

Eyeball a CSV in seconds. Someone drops a CSV export on you and you just want to know: how many rows, what columns, which ones are mostly empty, what's the range of that amount field? csvsight prints exactly that. Zero dependencies — no pandas, no REPL.

csvsight data.csv

data.csv — 10,432 rows × 5 columns  (comma, utf-8)

#  column   type    nulls         unique  detail
─  ───────  ──────  ────────────  ──────  ─────────────────────────────────────────
1  id       int     0 (0.0%)      10,432  min 1 · max 10432 · mean 5216.5
2  email    string  12 (0.1%)     10,411  e.g. "[e-mail]" · len 9–48
3  amount   float   34 (0.3%)     2,015   min 0.01 · max 9999 · mean 42.3
4  status   string  0 (0.0%)      3       active (61%) · churned (28%) · trial (11%)
5  country  string  120 (1.1%)    47      US (40%) · GB (12%) · DE (7%)

Why

You don't need a DataFrame to answer "what's in this file?" — but the usual tools make you spin one up anyway:

• pandas means pip install pandas, a Python session, and remembering the API for .describe() / .isna().sum() / .nunique().
• csvkit is lovely but pulls in a handful of dependencies.
• Excel chokes on big files and isn't in your terminal.

csvsight is one command on a CSV. It auto-detects the delimiter, infers each column's type, counts 10+ spellings of "missing" (NULL, N/A, nan, -, none, empty, …), and shows ranges for numbers and value distributions for categorical columns.

Usage

csvsight data.csv              # profile a file
cat data.csv | csvsight        # or read from stdin
csvsight data.tsv              # delimiter auto-detected (, tab ; |)

Option
--delimiter <c>	force the field delimiter
--no-header	treat the first row as data (columns named col1, col2, …)
--top <n>	top N values for categorical columns (default 3)
--json	emit the analysis as JSON instead of the table

What it reports per column

• type — int / float / string (inferred from the non-null values)
• nulls — count and percentage, recognizing many "missing" spellings
• unique — distinct non-null values
• detail — numbers get min · max · mean; low-cardinality columns get their value distribution; free-text columns get an example and length range

Install

pip install csvsight        # Python >= 3.8
npx csvsight data.csv       # Node >= 18 (byte-for-byte port)

• PyPI: https://pypi.org/project/csvsight/
• npm: https://www.npmjs.com/package/csvsight
• GitHub: https://github.com/jjdoor/csvsight-py · csvsight (Node)

License

MIT