def node_page(path: str):
"""Creates a node visualization page including title, action buttons, etc."""
node, found = pipelines.find_node(path.split('/'))
if not found and node:
return flask.redirect(views.node_url(node), 302)
elif not node:
flask.abort(404, f'Node "{path}" not found')
title = [node.__class__.__name__, ' ',
[[_.a(href=views.node_url(parent))[parent.id], ' / '] for parent in node.parents()[1:-1]],
node.id] if node.parent else 'Data Integration'
return response.Response(
title=title,
action_buttons=action_buttons(node) if config.allow_run_from_web_ui() else [],
html=[_.script['''
var nodePage = null;
document.addEventListener('DOMContentLoaded', function() {
nodePage = NodePage("''' + flask.url_for('data_integration.node_page', path='') + '''");
});'''],
dependency_graph.card(node),
run_time_chart.card(node),
node_content(node),
last_runs.card(node)],
js_files=['https://www.gstatic.com/charts/loader.js',
flask.url_for('data_integration.static', filename='node-page.js'),
flask.url_for('data_integration.static', filename='utils.js'),
flask.url_for('data_integration.static', filename='run-time-chart.js'),
flask.url_for('data_integration.static', filename='system-stats-chart.js'),
flask.url_for('data_integration.static', filename='timeline-chart.js'),
flask.url_for('data_integration.static', filename='kolorwheel.js')],
css_files=[flask.url_for('data_integration.static', filename='common.css'),
flask.url_for('data_integration.static', filename='node-page.css'),
flask.url_for('data_integration.static', filename='timeline-chart.css')])
def pipeline_children_table(path: str):
"""Creates a table that documents all child nodes of a table"""
pipeline, __ = pipelines.find_node(path.split('/'))
assert (isinstance(pipeline, pipelines.Pipeline))
node_durations_and_run_times = node_cost.node_durations_and_run_times(pipeline.path())
rows = []
for node in pipeline.nodes.values():
[avg_duration, avg_run_time] = node_durations_and_run_times.get(tuple(node.path()), ['', ''])
rows.append(
_.tr[_.td[_.a(href=views.node_url(node))[node.id.replace('_', '_<wbr>')]],
_.td[node.description],
_.td[views.format_labels(node)],
_.td[node_cost.format_duration(avg_duration)],
_.td(style='color:#bbb' if avg_duration == avg_run_time else '')[
node_cost.format_duration(avg_run_time)],
_.td[node_cost.format_duration(
node_cost.compute_cost(node, node_durations_and_run_times))],
_.td[(_.input(class_='pipeline-node-checkbox', type='checkbox',
value=node.id, name='ids[]', onchange='runButtons.update()')
if config.allow_run_from_web_ui() else '')]])
return \
str(_.script['var runButtons = new PipelineRunButtons();']) \
+ str(bootstrap.table(['ID', 'Description', '', 'Avg duration', 'Avg run time', 'Cost', ''], rows)) \
+ str(_.script['floatMaraTableHeaders();'])
def dependency_graph(nodes: {str: pipelines.Node},
current_node: pipelines.Node = None) -> str:
"""
Draws a list of pipeline nodes and the dependencies between them using graphviz
Args:
nodes: The nodes to render
current_node: If not null, then this node is highlighted
Returns:
An svg representation of the graph
"""
graph = graphviz.Digraph(graph_attr={'rankdir': 'TD', 'ranksep': '0.25', 'nodesep': '0.1'})
for node in nodes.values():
node_attributes = {'fontname': ' ', # use website default
'fontsize': '10.5px' # fontsize unfortunately must be set
}
if node != current_node:
node_attributes.update(
{'href': views.node_url(node), 'fontcolor': '#0275d8',
'tooltip': node.description, 'color': 'transparent'})
else:
node_attributes.update({'color': '#888888', 'style': 'dotted'})
if isinstance(node, pipelines.Pipeline):
node_attributes.update({'shape': 'rectangle', 'style': 'dotted', 'color': '#888888'})
elif isinstance(node, pipelines.ParallelTask):
node_attributes.update({'shape': 'ellipse', 'style': 'dotted', 'color': '#888888'})
else:
node_attributes['shape'] = 'rectangle'
graph.node(name=node.id, label=node.id.replace('_', '\n'), _attributes=node_attributes)
for upstream in node.upstreams:
if upstream.id in nodes:
graph.edge(upstream.id, node.id, _attributes={'color': '#888888', 'arrowsize': '0.7'})
elif (not current_node) or node in current_node.upstreams:
graph.node(name=f'{upstream.id}_{node.id}',
_attributes={'style': 'invis', 'label': '', 'height': '0.1', 'fixedsize': 'true'})
graph.edge(f'{upstream.id}_{node.id}', node.id,
_attributes={'color': '#888888', 'arrowsize': '0.7',
'edgetooltip': upstream.id, 'style': 'dotted'})
for downstream in node.downstreams:
if downstream.id not in nodes and (not current_node or node in current_node.downstreams):
graph.node(name=f'{downstream.id}_{node.id}',
_attributes={'style': 'invis', 'label': '', 'height': '0.1', 'fixedsize': 'true'})
graph.edge(node.id, f'{downstream.id}_{node.id}',
_attributes={'color': '#888888', 'arrowsize': '0.7',
'edgetooltip': downstream.id, 'style': 'dotted'})
return graph.pipe('svg').decode('utf-8')
def run_page(path: str, with_upstreams: bool, ids: str):
if not config.allow_run_from_web_ui():
flask.abort(
403,
'Running piplelines from web ui is disabled for this instance')
# the pipeline to run
pipeline, found = pipelines.find_node(path.split('/'))
if not found:
flask.abort(404, f'Pipeline "{path}" not found')
assert (isinstance(pipeline, pipelines.Pipeline))
# a list of nodes to run selectively in the pipeline
nodes = []
for id in (ids.split('/') if ids else []):
node = pipeline.nodes.get(id)
if not node:
flask.abort(404, f'Node "{id}" not found in pipeline "{path}"')
else:
nodes.append(node)
stream_url = flask.url_for('data_integration.do_run',
path=path,
with_upstreams=with_upstreams,
ids=ids)
title = [
'Run ', 'with upstreams ' if with_upstreams else '', ' / '.join([
str(_.a(href=views.node_url(parent))[parent.id])
for parent in pipeline.parents()[1:]
])
]
if nodes:
title += [
' / [', ', '.join([
str(_.a(href=views.node_url(node))[node.id]) for node in nodes
]), ']'
]
return response.Response(
html=[
_.script['''
document.addEventListener('DOMContentLoaded', function() {
processRunEvents(''' + json.dumps(
flask.url_for('data_integration.node_page', path='')) + ', ' +
json.dumps(stream_url) + ', ' +
json.dumps(pipeline.path()) + ''');
});'''],
_.style[
'span.action-buttons > * {display:none}'], # hide reload button until run finishes
_.div(class_='row')
[_.div(class_='col-lg-7')[bootstrap.card(body=_.div(
id='main-output-area', class_='run-output')[''])],
_.div(class_='col-lg-5 scroll-container')[
bootstrap.
card(header_left='Timeline',
body=[
_.div(id='system-stats-chart', class_='google-chart'
)[' '],
_.div(id='timeline-chart')[' ']
]),
_.div(id='failed-tasks-container')[''],
_.div(id='running-tasks-container')[''],
_.div(id='succeeded-tasks-container')[''],
bootstrap.card(id='card-template',
header_left=' ',
header_right=' ',
body=[_.div(class_='run-output')['']])]]
],
js_files=[
'https://www.gstatic.com/charts/loader.js',
flask.url_for('data_integration.static',
filename='timeline-chart.js'),
flask.url_for('data_integration.static',
filename='system-stats-chart.js'),
flask.url_for('data_integration.static', filename='utils.js'),
flask.url_for('data_integration.static', filename='run-page.js')
],
css_files=[
flask.url_for('data_integration.static',
filename='timeline-chart.css'),
flask.url_for('data_integration.static', filename='run-page.css'),
flask.url_for('data_integration.static', filename='common.css')
],
action_buttons=[
response.ActionButton(
action='javascript:location.reload()',
label='Run again',
icon='play',
title='Run pipeline again with same parameters as before')
],
title=title,
)