mirror of
https://gitea.psi.ch/APOG/acsm-fairifier.git
synced 2026-01-24 04:06:27 +01:00
302 lines
12 KiB
Python
302 lines
12 KiB
Python
import os
|
|
import sys
|
|
import inspect
|
|
try:
|
|
thisFilePath = os.path.abspath(__file__)
|
|
print(thisFilePath)
|
|
except NameError:
|
|
print("[Notice] The __file__ attribute is unavailable in this environment (e.g., Jupyter or IDLE).")
|
|
print("When using a terminal, make sure the working directory is set to the script's location to prevent path issues (for the DIMA submodule)")
|
|
#print("Otherwise, path to submodule DIMA may not be resolved properly.")
|
|
thisFilePath = os.getcwd() # Use current directory or specify a default
|
|
|
|
|
|
projectPath = os.path.normpath(os.path.join(thisFilePath, "..", "..")) # Move up to project root
|
|
|
|
if projectPath not in sys.path:
|
|
sys.path.insert(0,projectPath)
|
|
|
|
import yaml
|
|
from collections import OrderedDict
|
|
|
|
def generate_command(base_command='python', inputs=None, outputs=None, parameters=None):
|
|
inputs = inputs or []
|
|
outputs = outputs or []
|
|
placeholders = [
|
|
f"${name}"
|
|
for name, value in inputs + parameters + outputs
|
|
if isinstance(value, dict) and not value.get('implicit', False)
|
|
]
|
|
return f"{base_command} {' '.join(placeholders)}"
|
|
|
|
class RenkuWorkflowBuilder:
|
|
"""
|
|
Build and manage a Renku workflow definition (YAML-based).
|
|
Steps can be added, merged, serialized, and reloaded.
|
|
"""
|
|
|
|
def __init__(self, name):
|
|
"""
|
|
Initialize a workflow builder.
|
|
|
|
Args:
|
|
name (str): Workflow name.
|
|
"""
|
|
self.name = name
|
|
self.directory_path = os.path.join(projectPath, 'workflows').replace(os.sep, '/')
|
|
self.steps = OrderedDict()
|
|
|
|
@staticmethod
|
|
def _hash_content(step_def):
|
|
"""
|
|
Compute a stable hash of a step definition (for collision detection).
|
|
"""
|
|
import json, hashlib
|
|
content_str = json.dumps(step_def, sort_keys=True, default=str)
|
|
return hashlib.md5(content_str.encode()).hexdigest()
|
|
|
|
@staticmethod
|
|
def _normalize_paths(items : list):
|
|
"""
|
|
Normalize file paths inside a list of (key, value) pairs.
|
|
"""
|
|
if not items:
|
|
return items
|
|
normalized = []
|
|
for key, value in items:
|
|
if isinstance(value, dict) and 'path' in value:
|
|
value['path'] = value['path'].replace(os.sep, '/')
|
|
normalized.append({key: value})
|
|
return normalized
|
|
|
|
def create_workflow_file(self):
|
|
"""
|
|
Create (or update) a workflow YAML file on disk.
|
|
"""
|
|
self.save_to_file(self.directory_path)
|
|
filepath = os.path.join(self.directory_path, f'{self.name}.yaml').replace(os.sep, '/')
|
|
if os.path.exists(filepath):
|
|
print(f'Workflow file created at : {filepath}')
|
|
# TODO: add else-case handling (currently silent if file not created)
|
|
|
|
def add_step(self, step_name, base_command,
|
|
inputs : list = [],
|
|
outputs : list = [],
|
|
parameters : list = []):
|
|
"""
|
|
Add a step to the workflow and persist it to file.
|
|
Normalizes input/output/parameter paths and avoids duplicates.
|
|
"""
|
|
command = generate_command(base_command, inputs, outputs, parameters)
|
|
step = {'command': command}
|
|
|
|
step['inputs'] = self._normalize_paths(inputs)
|
|
step['outputs'] = self._normalize_paths(outputs)
|
|
step['parameters'] = self._normalize_paths(parameters)
|
|
|
|
# Deduplicate or version step
|
|
if step_name not in self.steps:
|
|
self.steps[step_name] = step
|
|
elif self.steps[step_name] != step:
|
|
content_hash = self._hash_content(step)
|
|
hashed_name = f"{step_name}_{content_hash[:8]}"
|
|
print(f"[Added] Step '{step_name}' as '{hashed_name}'")
|
|
self.steps[hashed_name] = step
|
|
|
|
self.save_to_file(self.directory_path)
|
|
|
|
def run_and_add_step(self, step, *args, **kwargs):
|
|
"""
|
|
Run a step function, collect its provenance, and add it to the workflow.
|
|
"""
|
|
file_path = inspect.getfile(step)
|
|
step_name = os.path.splitext(os.path.basename(file_path))[0]
|
|
provenance = step(*args, **kwargs)
|
|
# TODO: validate provenance has 'inputs', 'outputs', 'parameters'
|
|
self.add_step(step_name, "python",
|
|
provenance["inputs"],
|
|
provenance["outputs"],
|
|
provenance["parameters"])
|
|
return provenance
|
|
|
|
@staticmethod
|
|
def parse_workflow(yaml_content: str):
|
|
"""
|
|
Parse YAML content and return a populated RenkuWorkflowBuilder.
|
|
"""
|
|
data = yaml.safe_load(yaml_content)
|
|
builder = RenkuWorkflowBuilder(data['name'])
|
|
|
|
for step_name, step_def in data['steps'].items():
|
|
builder.steps[step_name] = {
|
|
'command': step_def.get('command'),
|
|
'inputs': [{k: v} for item in step_def.get('inputs', []) for k, v in item.items()] if step_def.get('inputs') else [],
|
|
'outputs': [{k: v} for item in step_def.get('outputs', []) for k, v in item.items()] if step_def.get('outputs') else [],
|
|
'parameters': [{k: v} for item in step_def.get('parameters', []) for k, v in item.items()] if step_def.get('parameters') else []
|
|
}
|
|
return builder
|
|
|
|
@staticmethod
|
|
def from_file(filepath):
|
|
"""
|
|
Load a workflow from file, or return a new empty one if file does not exist.
|
|
"""
|
|
if not os.path.exists(filepath):
|
|
workflow_name = os.path.splitext(os.path.basename(filepath))[0]
|
|
return RenkuWorkflowBuilder(name=workflow_name)
|
|
with open(filepath, 'r') as f:
|
|
return RenkuWorkflowBuilder.parse_workflow(f.read())
|
|
|
|
def to_dict(self):
|
|
"""
|
|
Return workflow definition as a dict, normalizing paths in inputs/outputs.
|
|
"""
|
|
for step_name, step_value in self.steps.items():
|
|
for segment in ['inputs', 'outputs']:
|
|
for item in step_value.get(segment, []):
|
|
for _, value in item.items():
|
|
if not isinstance(value, dict):
|
|
raise ValueError(
|
|
f"Invalid input. Step {step_name} must have {segment} as dict or str type."
|
|
)
|
|
if 'path' in value:
|
|
value['path'] = value['path'].replace(os.sep, '/')
|
|
|
|
return {'name': self.name, 'steps': dict(self.steps)}
|
|
|
|
def to_yaml(self):
|
|
"""
|
|
Serialize workflow definition to YAML.
|
|
"""
|
|
return yaml.dump(self.to_dict(), sort_keys=False)
|
|
|
|
def append_from(self, other, force=False):
|
|
"""
|
|
Merge steps from another workflow into this one.
|
|
If step content differs, a hashed suffix is added.
|
|
"""
|
|
if other.name != self.name:
|
|
raise ValueError(f"Cannot merge workflows with different names: {self.name} != {other.name}")
|
|
curr_steps = self.steps.copy()
|
|
for step_name, step_def in other.steps.items():
|
|
if step_name not in curr_steps:
|
|
self.steps[step_name] = step_def
|
|
elif self.steps[step_name] != step_def:
|
|
content_hash = self._hash_content(step_def)
|
|
hashed_name = f"{step_name}_{content_hash[:8]}"
|
|
if hashed_name not in self.steps:
|
|
self.steps[hashed_name] = step_def
|
|
print(f"[Added] Step '{step_name}' → '{hashed_name}'")
|
|
|
|
def save_to_file(self, directory):
|
|
"""
|
|
Save workflow definition to a YAML file in the given directory.
|
|
Merges with existing file if present.
|
|
"""
|
|
os.makedirs(directory, exist_ok=True)
|
|
filepath = os.path.join(directory, f"{self.name}.yaml")
|
|
|
|
if os.path.exists(filepath):
|
|
existing = RenkuWorkflowBuilder.from_file(filepath)
|
|
if existing and existing.name == self.name:
|
|
existing.append_from(self)
|
|
with open(filepath, 'w') as f:
|
|
f.write(existing.to_yaml())
|
|
return
|
|
|
|
with open(filepath, 'w') as f:
|
|
f.write(self.to_yaml())
|
|
|
|
import os
|
|
import re
|
|
import yaml
|
|
from graphviz import Digraph
|
|
from IPython.display import Image
|
|
|
|
def plot_workflow_graph(yaml_file_path,
|
|
output_dir=".",
|
|
output_name="workflow_graph",
|
|
output_format="png",
|
|
dpi=300,
|
|
show_parameters=False):
|
|
|
|
def shorten_path(path, keep_start=1, keep_end=2):
|
|
"""
|
|
Shortens a long path by keeping a few elements from the start and end.
|
|
E.g. 'a/b/c/d/e/f.txt' -> 'a/.../e/f.txt'
|
|
"""
|
|
parts = path.strip('/').split('/')
|
|
if len(parts) <= (keep_start + keep_end):
|
|
return path
|
|
return '/'.join(parts[:keep_start]) + '/.../' + '/'.join(parts[-keep_end:])
|
|
|
|
def split_path_label(path):
|
|
parts = path.split('/')
|
|
if len(parts) >= 2:
|
|
return f"{'/'.join(parts[:-1])}/\n{parts[-1]}"
|
|
return path
|
|
|
|
# Load YAML workflow file
|
|
with open(yaml_file_path, 'r') as f:
|
|
workflow_full = yaml.safe_load(f)
|
|
|
|
dot = Digraph(format=output_format)
|
|
dot.attr(rankdir='LR') #'TB') # vertical layout
|
|
dot.node_attr.update(fontsize='48')
|
|
|
|
# Set DPI only if format supports it (like png)
|
|
if output_format.lower() == 'png':
|
|
dot.attr(dpi=str(dpi))
|
|
|
|
used_paths = set()
|
|
|
|
for step_name, step in workflow_full['steps'].items():
|
|
# Extract parameters
|
|
params = step.get("parameters", [])
|
|
|
|
# Extract parameters if enabled
|
|
param_lines = []
|
|
if show_parameters:
|
|
params = step.get("parameters", [])
|
|
for param in params:
|
|
for k, v in param.items():
|
|
val = v.get("value", "")
|
|
param_lines.append(f"{k} = {val}")
|
|
|
|
param_label = "\n".join(param_lines)
|
|
label = f"{param_label}\n{step_name}" if param_label else step_name
|
|
|
|
dot.node(step_name, label=label, shape="box", style="filled", fillcolor="lightblue")
|
|
|
|
for input_item in step.get('inputs', []):
|
|
for key, val in input_item.items():
|
|
if isinstance(val, dict) and 'path' in val:
|
|
path = shorten_path(val['path'])
|
|
label = split_path_label(path)
|
|
if path not in used_paths:
|
|
dot.node(path, label=label, tooltip=path, shape="ellipse", style="filled", fillcolor="lightgrey")
|
|
used_paths.add(path)
|
|
dot.edge(path, step_name)
|
|
|
|
for output_item in step.get('outputs', []):
|
|
for key, val in output_item.items():
|
|
if isinstance(val, dict) and 'path' in val:
|
|
path = shorten_path(val['path'])
|
|
label = split_path_label(path)
|
|
if path not in used_paths:
|
|
dot.node(path, label=label, tooltip=path, shape="ellipse", style="filled", fillcolor="lightgreen")
|
|
used_paths.add(path)
|
|
dot.edge(step_name, path)
|
|
|
|
os.makedirs(output_dir, exist_ok=True)
|
|
output_path = os.path.join(output_dir, output_name)
|
|
dot.render(output_path)
|
|
#dot.view()
|
|
|
|
# For SVG or PDF, you may want to return the file path or raw output instead of Image()
|
|
if output_format.lower() in ['png', 'jpg', 'jpeg', 'gif']:
|
|
return Image(output_path + f".{output_format}")
|
|
else:
|
|
print(f"Graph saved to: {output_path}.{output_format}")
|
|
return output_path + f".{output_format}"
|