Error while loading pipelines from yaml files

[IronD7]

Describe the bug
I've tried loading a Pipeline from a yaml file, it fails with an Error. The same pipeline as a String can be deserialized successfully (see To Reproduce section)

Error message

File ~/anaconda3/envs/env/lib/python3.10/site-packages/haystack/core/pipeline/pipeline.py:151, in Pipeline.from_dict(cls, data, callbacks, **kwargs)
135 @classmethod
136 def from_dict(
137 cls: Type[T], data: Dict[str, Any], callbacks: Optional[DeserializationCallbacks] = None, **kwargs
138 ) -> T:
139 """
140 Deserializes the pipeline from a dictionary.
141
(...)
149 Deserialized component.
150 """
--> 151 metadata = data.get("metadata", {})
152 max_loops_allowed = data.get("max_loops_allowed", 100)
153 debug_path = Path(data.get("debug_path", ".haystack_debug/"))

Expected behavior
The pipeline is loaded.

Additional context

To Reproduce
The first call is fine, the second (the string was copied from the yml file) fails

test_pipeline = """
components:
  first_addition:
    init_parameters:
      add: 2
    type: haystack.testing.sample_components.add_value.AddFixedValue
  double:
    init_parameters: {}
    type: haystack.testing.sample_components.double.Double
connections:
- receiver: double.value
  sender: first_addition.result
max_loops_allowed: 100
metadata: {}
"""
pipeline = Pipeline.loads(test_pipeline)
pipeline = Pipeline.loads("test_pipeline_01.yml")

FAQ Check

• Have you had a look at our new FAQ page?

System:

• Haystack version (commit or version number): 2.1.0

[anakin87]

Hello!

When loading a Pipeline from YAML, you should use Pipeline.load (instead of Pipeline.loads).

The following code works properly:

from haystack import Pipeline

test_pipeline = """
components:
  first_addition:
    init_parameters:
      add: 2
    type: haystack.testing.sample_components.add_value.AddFixedValue
  double:
    init_parameters: {}
    type: haystack.testing.sample_components.double.Double
connections:
- receiver: double.value
  sender: first_addition.result
max_loops_allowed: 100
metadata: {}
"""

pipeline = Pipeline.loads(test_pipeline)
print(pipeline)


# Save pipeline to YAML file
with open("test_pipeline_01.yaml", "w") as f:
    f.write(test_pipeline)

# Load pipeline from YAML file
with open("test_pipeline_01.yaml", "r") as f:
    pipeline = Pipeline.load(f)
print(pipeline)

Can you close the issue if the problem is solved?

[IronD7]

That works, thanks a lot for the quick response!!