49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184 | class ARGOPipelineWriter(BasePipelineWriter):
"""
Class for pipeline file writer. Corresponds to "ARGO" framework.
"""
@property
def docker_template_name(self) -> str:
return "argo/argo_dockerfile.jinja"
def _write_dag(self) -> None:
# Check if the given DAG flavor is a supported/valid one
try:
dag_flavor = ARGODagFlavor[
self.dag_config.get("dag_flavor", "StepPerSession")
]
except KeyError:
raise ValueError(f'"{dag_flavor}" is an invalid ARGO dag flavor.')
# Construct DAG text for the given flavor
full_code = self._write_operators(dag_flavor)
# Write out file
file = self.output_dir / f"{self.pipeline_name}_dag.py"
file.write_text(prettify(full_code))
logger.info(f"Generated DAG file: {file}")
def _write_operators(
self,
dag_flavor: ARGODagFlavor,
) -> str:
DAG_TEMPLATE = load_plugin_template("argo/argo_dag.jinja")
if dag_flavor == ARGODagFlavor.StepPerSession:
task_breakdown = DagTaskBreakdown.TaskPerSession
# Get task definitions based on dag_flavor
task_defs, task_graph = get_task_graph(
self.artifact_collection,
pipeline_name=self.pipeline_name,
task_breakdown=task_breakdown,
)
task_defs["setup"] = get_noop_setup_task_definition(self.pipeline_name)
task_defs["teardown"] = get_noop_teardown_task_definition(
self.pipeline_name
)
# insert in order to task_names so that setup runs first and teardown runs last
task_graph.insert_setup_task("setup")
task_graph.insert_teardown_task("teardown")
task_names = list(task_defs.keys())
task_defs = {tn: task_defs[tn] for tn in task_names}
rendered_task_defs = self.get_rendered_task_definitions(task_defs)
task_loading_blocks = self.get_task_loading_blocks(task_defs)
# Handle dependencies
task_dependencies = reversed(
[f"{task0} >> {task1}" for task0, task1 in task_graph.graph.edges]
)
# Get DAG parameters for an ARGO pipeline
input_parameters_dict: Dict[str, Any] = {}
for parameter_name, input_spec in super().get_pipeline_args().items():
input_parameters_dict[parameter_name] = input_spec.value
# set kube config to be user given path or expand default ~/.kube/config
if "kube_config" in self.dag_config:
kube_config = f'"{self.dag_config.get("kube_config")}"'
else:
kube_config = 'os.path.expanduser("~/.kube/config")'
full_code = DAG_TEMPLATE.render(
DAG_NAME=self.pipeline_name,
MODULE_NAME=self.pipeline_name + "_module",
NAMESPACE=self.dag_config.get("namespace", "argo"),
HOST=self.dag_config.get("host", "https://localhost:2746"),
VERIFY_SSL=self.dag_config.get("verify_ssl", "False"),
WORFLOW_NAME=self.dag_config.get(
"workflow_name", self.pipeline_name.replace("_", "-")
),
IMAGE=self.dag_config.get(
"image", f"{self.pipeline_name}:lineapy"
),
IMAGE_PULL_POLICY=self.dag_config.get(
"image_pull_policy", "Never"
),
SERVICE_ACCOUNT=self.dag_config.get("service_account", "argo"),
KUBE_CONFIG_PATH_CODE_BLOCK=kube_config,
TOKEN=self.dag_config.get("token", "None"),
dag_params=input_parameters_dict,
task_definitions=rendered_task_defs,
tasks=task_defs,
task_loading_blocks=task_loading_blocks,
task_dependencies=task_dependencies,
)
return full_code
def get_task_loading_blocks(
self,
task_defs: Dict[str, TaskDefinition],
) -> Dict[str, str]:
"""
Returns a dictionary to lookup previous task outputs.
The returned dictionary is used by the DAG to connect the right input files to
output files for inter task communication.
This method originates from:
https://github.com/argoproj-labs/hera-workflows/blob/4efddc85bfce62455db758f4be47e3acc0342b4f/examples/k8s_sa.py#L12
"""
task_loading_blocks: Dict[str, str] = {}
for _, task_def in task_defs.items():
for return_variable in task_def.return_vars:
task_loading_blocks[return_variable] = return_variable
return task_loading_blocks
def get_rendered_task_definitions(
self,
task_defs: Dict[str, TaskDefinition],
) -> List[str]:
"""
Returns rendered tasks for the pipeline tasks
"""
rendered_task_defs: List[str] = render_task_definitions(
task_defs,
self.pipeline_name,
task_serialization=TaskSerializer.TmpDirPickle,
include_imports_locally=True,
)
return rendered_task_defs
|