When an output column of extract.ai already exists, it affects the output of the extract. For example, by running the test below:
df = wrangles.recipe.run(
"""
wrangles:
- create.column:
output: length
value: 123
- extract.ai:
model: gpt-4o-mini
api_key: ${OPENAI_API_KEY}
seed: 1
timeout: 60
retries: 2
output:
length:
type: string
description: >-
Any lengths found in the data
such as cm, m, ft, etc.
""",
dataframe=pd.DataFrame({
"data": [
"wrench 25mm",
"6m cable",
"screwdriver 3mm"
],
})
)
The output becomes:
| Index |
data |
length |
| 0 |
wrench 25mm |
123 |
| 1 |
6m cable |
6m |
| 2 |
screwdriver 3mm |
123 |
Instead of the expected:
| Index |
data |
length |
| 0 |
wrench 25mm |
25mm |
| 1 |
6m cable |
6m |
| 2 |
screwdriver 3mm |
3mm |
When an output column of extract.ai already exists, it affects the output of the extract. For example, by running the test below:
The output becomes:
Instead of the expected: