DataSemver

Your data changed. DataSemver tells you whether that is a patch, a minor or a breaking release.

DataSemver compares two versions of a CSV, JSON or Parquet dataset, classifies every difference it finds against a configurable rule set, and returns the semantic version bump plus a ready-to-commit changelog entry. It is a CLI first and a Python library second, and it needs no schema registry, no database and no service running.

Install

pip install datasemver              # library and CLI
pip install "datasemver[web]"       # adds the dashboard
pipx install datasemver             # standalone command

Python 3.10 or newer. The package ships typed, so py.typed annotations reach type checkers.

Use it

datasemver diff old.csv new.csv --current-version 1.4.2

╭───────────── DataSemver ──────────────╮
│ Suggested bump: MAJOR                 │
│ 0.0.0 -> 1.0.0                        │
│                                       │
│ old: old.csv (8 rows)                 │
│ new: new.csv (10 rows)                │
╰───────────────────────────────────────╯
                                    Columns
┏━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
┃ column      ┃ status    ┃ type old ┃ type new ┃ nulls         ┃ cardinality ┃
┡━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
│ country     │ added     │ -        │ string   │ - -> 0.0%     │ - -> 4      │
│ phone       │ modified  │ int64    │ string   │ 0.0% -> 0.0%  │ 8 -> 10     │
│ email       │ modified  │ string   │ string   │ 25.0% -> 0.0% │ 6 -> 10     │
│ legacy_code │ removed   │ string   │ -        │ 0.0% -> -     │ 8 -> -      │
└─────────────┴───────────┴──────────┴──────────┴───────────────┴─────────────┘
                                        Changes
┏━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ severity ┃ rule                      ┃ description                                   ┃
┡━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ MAJOR    │ column_removed            │ Column 'legacy_code' was removed              │
│ MAJOR    │ type_changed_incompatible │ Column 'phone' changed type from int64 to     │
│          │                           │ string                                        │
│ MINOR    │ row_count_increased       │ Row count grew from 8 to 10 (+25.00%)         │
│ MINOR    │ column_added              │ Column 'country' was added                    │
│ PATCH    │ nulls_fixed               │ Column 'email' nulls dropped from 25.0% to    │
│          │                           │ 0.0%                                          │
└──────────┴───────────────────────────┴───────────────────────────────────────────────┘

A removed column and an int64 that became a string make this a breaking release. Without --output, the changelog entry is printed at the end; with it, the entry is prepended to the file you name.

datasemver diff old.csv new.csv --output CHANGELOG.md
datasemver diff old.csv new.csv --json | jq -r '.bump'
datasemver rules examples/lenient_rules.yaml

What it looks at

• Schema — columns added, removed and renamed, dtype changes, nullability.
• Content — row counts, cardinality, mean and standard deviation of numeric columns, mode and category sets of categorical ones.
• Semantics — renames inferred from the similarity of both the column name and its values, so user_name becoming username is one rename rather than a removal plus an addition.

The bump is the strongest severity across every classified change. Changes no rule covers are reported as unclassified and never inflate it.

Formats

Detected by extension: .csv, .tsv, .json, .jsonl, .ndjson, .parquet, .pq.

The delimiter of a .csv is detected from its first lines — comma, semicolon, tab and pipe are recognised, and a character that only appears inside quoted values does not win — while .tsv always uses the tab. Set DATASEMVER_CSV_DELIMITER to skip detection and force one character, the tab written as \t.

Nested JSON objects and Parquet structs are flattened with a ., so {"user": {"name": "..."}} is profiled as user.name. Types are inferred for the text formats; Parquet carries its own schema and is trusted as it stands.

Rules

Every severity is a list of rules, evaluated major, then minor, then patch. The first rule that matches a change assigns its severity.

major:
  - column_removed
  - type_changed_incompatible
  - row_count_decrease_greater_than: 20

minor:
  - column_added
  - row_count_decreased

patch:
  - nulls_fixed
  - minor_stat_change

Pass it with --rules custom.yaml, and check how it was parsed with datasemver rules custom.yaml. Threshold rules pair with their plain counterpart in a lower severity, which then acts as the fallback. Unknown rule names and severities are errors, not silent no-ops.

Python API

from datasemver import analyze

report = analyze("old.csv", "new.csv", current_version="1.4.2")

print(report.bump)          # Severity.MAJOR
print(report.next_version)  # 2.0.0

for item in report.classified:
    print(item.severity, item.rule, item.change.description)

analyze_schemas() takes two already loaded profiles, so dataframes from anywhere can be compared without touching the filesystem:

import pandas as pd
from datasemver.core.analyzer import analyze_schemas
from datasemver.formats.loader import schema_from_frame

report = analyze_schemas(
    schema_from_frame(pd.read_sql(query, engine), "warehouse@yesterday"),
    schema_from_frame(pd.read_sql(query, engine), "warehouse@today"),
)

Also in the box

• A web dashboard — FastAPI backend, no-build frontend — under the web extra. It is a local tool with no authentication: keep it on the loopback interface.
• A GitHub Action that analyses the datasets a pull request touches and posts the suggested bump as a comment, rewritten on each push.
• Two ready-made rule profiles, strict and lenient, and a full catalogue of rules, metrics and thresholds.

Those, the source, the changelog and a Spanish edition of this page live in the project repository, linked from this page's sidebar.

Security

Reading a dataset parses it. CSV and JSON go through pandas and the standard library, which do not execute file content; Parquet goes through pyarrow, and the dependency floor is pyarrow>=23.0.1 because earlier versions carried a critical code-execution flaw (CVE-2023-47248) triggered by a malicious Parquet file. Do not lower that floor. Rule files are YAML loaded with yaml.safe_load and cannot execute code. The library and the CLI never open a socket, and write nothing unless you pass --output.

License

MIT.