Preparing your workspaceloading the brief, studio & editor…

Loadingone moment…

Diff two snapshots of a table · Python scenario · D8LooP

Reading is open to everyone. Running code and saving drafts need an account so your work is yours and comes back on your next visit.

CODE WORKSPACE

Diff two snapshots of a table

A source system has no change feed, so the pipeline takes a full snapshot every night and works out what moved by comparing it against yesterday's. The source also stamps every row with an updated_at on every export, whether or not anything about the row changed, which makes a naive comparison report the entire table as modified. Write diff_snapshots(before, after, key, ignore_fields).

Function to write

diff_snapshots(before: list[dict], after: list[dict], key: str, ignore_fields: list[str]) -> dict

A dict with added and removed keys, changed rows with their field-level differences, and an unchanged count.

How to approach it

Index both sides by key, then walk the union and compare field by field.

Sample cases

+ 2 held back until you submit

a night of changes

One row unchanged apart from its timestamp, one changed in two fields, one added and one removed.

Input

Argument 1

id	name	tier	updated_at
c1	Ada	gold	2026-03-01T00:00:00Z
c2	Grace	silver	2026-03-01T00:00:00Z
c3	Alan	gold	2026-03-01T00:00:00Z

Argument 2

id	name	tier	updated_at
c1	Ada	gold	2026-03-02T00:00:00Z
c2	Grace Hopper	gold	2026-03-02T00:00:00Z
c4	Edsger	silver	2026-03-02T00:00:00Z

Argument 3

'id'

Argument 4

[
  'updated_at'
]

Returns

{
  'added': [
    'c4'
  ],
  'removed': [
    'c3'
  ],
  'changed': [
    {
      'key': 'c2',
      'fields': {
        'name': {
          'before': 'Grace',
          'after': 'Grace Hopper'
        },
        'tier': {
          'before': 'silver',
          'after': 'gold'
        }
      }
    }
  ],
  'unchanged': 1
}

nothing before

A first snapshot makes every row an addition rather than a change.

Input

Argument 1

[] (empty list)

Argument 2

id	name	tier	updated_at
c1	Ada	gold	2026-03-02T00:00:00Z
c2	Grace Hopper	gold	2026-03-02T00:00:00Z
c4	Edsger	silver	2026-03-02T00:00:00Z

Argument 3

'id'

Argument 4

[
  'updated_at'
]

Returns

{
  'added': [
    'c1',
    'c2',
    'c4'
  ],
  'removed': [],
  'changed': [],
  'unchanged': 0
}

ignoring nothing

Without the ignore list the volatile timestamp makes every surviving row look changed.

Input

Argument 1

id	name	tier	updated_at
c1	Ada	gold	2026-03-01T00:00:00Z
c2	Grace	silver	2026-03-01T00:00:00Z
c3	Alan	gold	2026-03-01T00:00:00Z

Argument 2

id	name	tier	updated_at
c1	Ada	gold	2026-03-02T00:00:00Z
c2	Grace Hopper	gold	2026-03-02T00:00:00Z
c4	Edsger	silver	2026-03-02T00:00:00Z

Argument 3

'id'

Argument 4

[] (empty list)

Returns

{
  'added': [
    'c4'
  ],
  'removed': [
    'c3'
  ],
  'changed': [
    {
      'key': 'c1',
      'fields': {
        'updated_at': {
          'before': '2026-03-01T00:00:00Z',
          'after': '2026-03-02T00:00:00Z'
        }
      }
    },
    {
      'key': 'c2',
      'fields': {
        'name': {
          'before': 'Grace',
          'after': 'Grace Hopper'
        },
        'tier': {
          'before': 'silver',
          'after': 'gold'
        },
        'updated_at': {
          'before': '2026-03-01T00:00:00Z',
          'after': '2026-03-02T00:00:00Z'
        }
      }
    }
  ],
  'unchanged': 0
}

Constraints

1A key present only in after is added; only in before is removed.
2A key on both sides is changed when any compared field differs, and unchanged otherwise. Count the unchanged rather than listing them.
3Compare every field that appears on either side, except the key itself and anything in ignore_fields.
4A field present on one side only has changed — from None or to None. An absent field and a null field are the same thing here, because the snapshot cannot tell them apart.
5A boolean never equals a number, even though Python says True == 1.
6For each changed row report the key and a map of field to before and after.
7Rows without a usable key cannot be compared and are skipped.
8Report added, removed and changed in ascending key order so two nights can be compared to each other.

Worked example

c1 is byte-for-byte the same except that updated_at moved from the first of March to the second. With updated_at ignored it is counted as unchanged; without it, the diff reports a change on every row in the table every night, and the downstream job that reloads changed rows reloads everything. c2 changed in two fields at once — the name became Grace Hopper and the tier became gold. Reporting the row as 'changed' with no detail leaves whoever is debugging to fetch both snapshots and compare by hand, which is the work this function was written to avoid. The field-level detail is the deliverable, not the row-level verdict.

What this tests

Diffing at the right granularity, and the judgement that some columns carry no information about change. The volatile-column problem is the single most common reason a snapshot diff is useless in practice.

PYTHONdiff_snapshots(before: list[dict], after: list[dict], key: str, ignore_fields: list[str]) -> dict

AI evaluation

The reviewer reads your query.

Submit for review to find out what your query gets right, what it gets wrong, and how it compares with the best working query for this exercise.

Desktop only

This scenario runs a full workspace — editor, canvas and results side by side. It needs a laptop or desktop to be usable. Open this page on a bigger screen to start building.

Read the library instead Back to modules