From bdb1c5eb1e6796599b328f2ef16c1d7ef8088293 Mon Sep 17 00:00:00 2001
From: Alexander Pace <alexander.pace@ligo.org>
Date: Mon, 16 Sep 2024 18:15:23 +0000
Subject: [PATCH] Add `time-to-alert` to beta reports page
the reports page is accessible on production, but the link to it is exposed on playground and test.
---
config/urls.py | 2 +-
gracedb/events/reports.py | 189 ++++++++++++++++++++++---
gracedb/templates/gracedb/reports.html | 127 ++++++++++++-----
requirements.txt | 6 +-
4 files changed, 263 insertions(+), 61 deletions(-)
diff --git a/config/urls.py b/config/urls.py
index 142930a4f..ddfeae86e 100644
--- a/config/urls.py
+++ b/config/urls.py
@@ -43,7 +43,7 @@ urlpatterns = [
re_path(r'^other/$', TemplateView.as_view(template_name='other.html'),
name='other'),
re_path(r'^performance/$', events.views.performance, name="performance"),
- re_path(r'^reports/$', events.reports.histo, name="reports"),
+ re_path(r'^reports/$', events.reports.reports_page_context, name="reports"),
re_path(r'^latest/$', search.views.latest, name="latest"),
#(r'^reports/(?P<path>.+)$', 'django.views.static.serve',
# {'document_root': settings.LATENCY_REPORT_DEST_DIR}),
diff --git a/gracedb/events/reports.py b/gracedb/events/reports.py
index c249844f8..928c3c5f2 100644
--- a/gracedb/events/reports.py
+++ b/gracedb/events/reports.py
@@ -3,8 +3,10 @@ from django.http import HttpResponseForbidden
from django.template import RequestContext
from django.shortcuts import render
from django.conf import settings
+from django import forms
-from .models import Event, Group, Search, Pipeline
+from .models import Event, Group, Search, Pipeline, Label
+from superevents.models import Superevent, Labelling
from ligoauth.decorators import internal_user_required
from django.db.models import Q
@@ -16,43 +18,191 @@ from search.query.events import parseQuery
from django.db.models import Max, Min, Avg
from django.db.models.aggregates import StdDev
-from datetime import timedelta, datetime
+from datetime import datetime, timedelta
+from gpstime import gpstime
from django.utils import timezone
+from plotly.io import to_html
from plotly.offline import plot
+from tailslide import Median, Percentile
+
+import numpy as np
+import pandas as pd
+import plotly.express as px
import plotly.graph_objects as go
-plot_title = "Events Uploaded Since {0} UTC"
-plot_sub_title = "<br><sup> Online, Production Events in the Last Seven Days </sup>"
+YEAR_RANGE = list(range(datetime.now().year, 2013, -1))
+INITIAL_DAYS_BACK = 31
+
+EVENT_SEARCH_CHOICES = (
+ ('AllSky', 'AllSky'),
+ ('EarlyWarning', 'EarlyWarning'),
+ ('SSM', 'SSM'),
+ ('BBH', 'BBH'),
+ ('VTInjection', 'VTInjection'),
+)
+
+EVENT_SEARCH_CHOICES_INITAL = ['AllSky']
-plt_title = plot_title + plot_sub_title
+SUPEREVENT_CATEGORIES = (
+ (Superevent.SUPEREVENT_CATEGORY_PRODUCTION, 'Production'),
+ (Superevent.SUPEREVENT_CATEGORY_MDC, 'MDC'),
+)
-days_back = 7
+class lookback_days_form(forms.Form):
+ start_date = forms.DateField(label='Start Date',
+ widget=forms.SelectDateWidget(years=YEAR_RANGE),
+ )
+ end_date = forms.DateField(label='End Date',
+ widget=forms.SelectDateWidget(years=YEAR_RANGE),
+ initial = datetime.now()
+ )
+
+ searches_choice = forms.MultipleChoiceField(widget=forms.CheckboxSelectMultiple,
+ choices=EVENT_SEARCH_CHOICES,
+ required=False,
+ label='Event Searches',
+ initial=EVENT_SEARCH_CHOICES_INITAL)
+
+ superevent_type = forms.ChoiceField(widget=forms.RadioSelect,
+ choices=SUPEREVENT_CATEGORIES,
+ label='Superevent Category',
+ initial=Superevent.SUPEREVENT_CATEGORY_PRODUCTION)
+
@internal_user_required
-#@method_decorator(internal_user_required(raise_exception=True),
-# name='dispatch')
-def histo(request):
+def reports_page_context(request):
+
+ # start with a blank context dict:
+ context = {}
+
+ if request.method == 'POST':
+ form = lookback_days_form(request.POST)
+
+ if form.is_valid():
+ start_date = form.cleaned_data['start_date']
+ start_date = datetime.combine(start_date, datetime.min.time())
+ end_date = form.cleaned_data['end_date']
+ end_date = datetime.combine(end_date, datetime.min.time())
+ days_back = (end_date - start_date).days
+ event_searches = form.cleaned_data['searches_choice']
+ superevent_category=form.cleaned_data['superevent_type']
+
+ else:
+ days_back = INITIAL_DAYS_BACK
+ end_date = datetime.now()
+ start_date = end_date - timedelta(days=days_back)
+ event_searches = EVENT_SEARCH_CHOICES_INITAL
+ superevent_category = Superevent.SUPEREVENT_CATEGORY_PRODUCTION
+ form = lookback_days_form(initial={'start_date': start_date})
+
+ # update context with form:
+ context.update({'form': form})
+
+ # update context with histogram div:
+ context.update({'hist': latency_histo(days_back, start_date,
+ end_date, event_searches)})
+
+ # update with the time-to-alert div:
+ context.update({'tta': time_to_alert(start_date, end_date, superevent_category)})
+
+ return render(request, 'gracedb/reports.html', context)
+
+
+def time_to_alert(start_date, end_date, superevent_category):
+ # Set up the query to get all production superevents within the date range
+ # that have the GCN_PRELIM_SENT label applied:
+
+ # Get the GCN_PRELIM_SENT label first so we only hit the db once:
+ gcn = Label.objects.get(name='GCN_PRELIM_SENT')
+
+ sevents = Superevent.objects.filter(category=superevent_category,
+ t_0__range=(gpstime.fromdatetime(start_date).gps(),
+ gpstime.fromdatetime(end_date).gps()),
+ labels=gcn,
+ ).order_by('superevent_id')
+
+ # Now get the GCN_PRELIM_SENT label objects:
+ labs = Labelling.objects.filter(label=gcn, superevent__in=sevents).order_by('superevent__superevent_id')
+
+ # Make lists of the relevant values. One has superevent_id and t_0, the
+ # other has the created time of the gcn label (which is a proxy for
+ # when the alert goes out). Everything should be ordered correctly, but
+ # we're going to add a verification step just in case.
+ sevents_values = list(sevents.values_list('superevent_id', 't_0', 'created'))
+ label_values = list(labs.values_list('superevent__superevent_id', 'created'))
+
+ # Construct a list where each item looks like:
+ # [superevent_id (string), created (datetime), time-to-alert (float)]
+ #results =
+ if sevents_values:
+ results = [[i[0][0], i[0][2], gpstime.fromdatetime(i[1][1]).gps() - float(i[0][1])] for i in zip(sevents_values, label_values) if i[0][0] == i[1][0]]
+ np_results = np.array(results)
+ else:
+ np_results = np.zeros((1,3))
+
+
+ # Try making a pandas dataframe:
+ pd_results = pd.DataFrame(np_results,
+ columns=['superevent_id', 'created', 'time_to_alert'],
+ )
+
+
+ # Make a scatter plot:
+ scatter_fig = px.scatter(pd_results,
+ x='created',
+ y='time_to_alert',
+ hover_data='superevent_id',
+ )
+ scatter_fig.update_traces(marker_size=10)
+
+ scatter_fig.update_layout(title={
+ 'text': 'Superevent Time-to-Alert',
+ 'xanchor': 'center',
+ 'x':0.5,},
+ xaxis_title="Date",
+ yaxis_title="Time-to-Alert (s)",
+ autosize=True,
+ paper_bgcolor='rgba(0, 0, 0, 0)',
+ margin={'l': 0, 'r': 0},
+ )
+
+
+ return {'tta_plot_div': to_html(scatter_fig,
+ include_plotlyjs=False,
+ full_html=False,
+ default_width='100%',),
+ 'tta_stats': {'med': pd_results['time_to_alert'].median(),
+ 'nfp': pd_results['time_to_alert'].quantile(0.95)},
+ }
+
+
+def latency_histo(days_back, start_date, end_date, event_searches):
+ plot_title = "Events Uploaded Between {start} and {end}".format(
+ start=start_date.strftime("%m/%d/%Y"),
+ end=end_date.strftime("%m/%d/%Y"),
+ )
+ plot_sub_title = f"<br><sup> Online, Production Events in {days_back} Days </sup>"
+ plt_title = plot_title + plot_sub_title
+
fig = go.Figure()
# Get the timedelta and get it in a queryable form:
t_now = timezone.now()
date_cutoff = t_now - timedelta(days=days_back)
-
- #all_events_latency = list(Event.objects.filter(created__gt=date_cutoff, reporting_latency__isnull=False).values_list('reporting_latency', flat=True))
-
# Zero out the list of pipeline statistics:
aggregated_stats =[]
# Loop over pipelines that are determined to be "Production" search pipelines and
# retrieve data, and populate histograms:
+
for pipeline in Pipeline.objects.filter(pipeline_type=Pipeline.PIPELINE_TYPE_SEARCH_PRODUCTION):
# Generate the queryset for production (G) events, uploaded online, that have a
# valid value of reporting_latency. There has to be a way to combine these so you
# hit the db once, but for now its just once for a list of values, and once for
# aggregated values.
- pipeline_query = Event.objects.filter(graceid__contains='G',
+ pipeline_query = Event.objects.filter(search__name__in=event_searches,
offline=False,
created__gt=date_cutoff, reporting_latency__isnull=False,
pipeline=pipeline).order_by('reporting_latency')
@@ -69,8 +219,8 @@ def histo(request):
))
# Update the statistics list with aggregated values.
- aggregated_stats.append(pipeline_query.aggregate(avg=Avg('reporting_latency'),
- std=StdDev('reporting_latency')))
+ aggregated_stats.append(pipeline_query.aggregate(med=Median('reporting_latency'),
+ nfp=Percentile('reporting_latency', 0.95)))
aggregated_stats[-1].update({'name': pipeline.name})
if pipeline_query:
aggregated_stats[-1].update({'min': pipeline_query.first().reporting_latency,
@@ -92,7 +242,7 @@ def histo(request):
#fig.update_layout(barmode='stack',
fig.update_layout(barmode='overlay',
title={
- 'text': plt_title.format(date_cutoff.strftime("%m/%d/%Y, %H:%M:%S")),
+ 'text': plt_title,
'xanchor': 'center',
'x':0.5,},
xaxis_title="Reporting Latency (s)",
@@ -102,12 +252,7 @@ def histo(request):
fig.update_traces(opacity=0.75)
latency_plot_div = plot(fig, output_type='div')
-
-
- return render(request, 'gracedb/reports.html',
- context=
- {'latency_plot_div': latency_plot_div,
+ return {'latency_plot_div': latency_plot_div,
'date_cutoff': date_cutoff,
'pipeline_stats': aggregated_stats,}
- )
diff --git a/gracedb/templates/gracedb/reports.html b/gracedb/templates/gracedb/reports.html
index 21f1ab9e5..c6af6a5b7 100644
--- a/gracedb/templates/gracedb/reports.html
+++ b/gracedb/templates/gracedb/reports.html
@@ -6,44 +6,101 @@
{% block content %}
<center>
-This page is currently in <span style="color:red;font-weight: bold;">beta</span>.
-Use for cross-checking and validation purposes only.
+ This page is currently in <span style="color:red;font-weight: bold;">beta</span>.
+ Use for cross-checking and validation purposes only.
</center>
-{% autoescape off %}
-{{ latency_plot_div }}
-{% endautoescape %}
-
-<div class="row my-3 justify-content-md-center">
- <div class="col-md-8">
- <table class="table-hover table-condensed table-resp-gracedb shadow p-3 mb-5 rounded" style="text-align:center">
- <thead>
- <td> Pipeline </td>
- <td> Uploads </td>
- <td> Min. Latency (s) </td>
- <td> Max Latency (s) </td>
- <td> Avg. Latency (s) </td>
- <td> Std. Dev. (s) </td>
- </thead>
- {% for data in pipeline_stats %}
- <tr>
- <!--<td style="font-family: monospace;"> {{data.name}} </td> -->
- <td style="font-family: 'Courier New', Courier, monospace; font-size: 12px; white-space: pre;"> {{data.name}} </td>
- <td> {{data.count}} </td>
- <td> {% if data.min %} {{data.min|floatformat:3}}
- (<a href="{% url "view" data.min_gid %}">{{ data.min_gid }}</a>)
- {% endif %}
- </td>
- <td> {% if data.max %} {{data.max|floatformat:3}}
- (<a href="{% url "view" data.max_gid %}">{{ data.max_gid }}</a>)
- {% endif %}
- </td>
- <td> {{data.avg|floatformat:3}} </td>
- <td> {{data.std|floatformat:3}} </td>
- </tr>
- {% endfor %}
- </table>
+<center>
+ <form action="" method="post">
+ {% csrf_token %}
+ <table>
+ {{ form.as_table }}
+ </table>
+ <input type="submit" value="Submit" class="btn btn-primary">
+ </form>
+</center>
+
+<ul class="nav nav-tabs" id="perfTab" role="tablist">
+ <li class="nav-item" role="presentation">
+ <button class="nav-link active" id="latency-tab" data-toggle="tab" data-target="#latency" type="button" role="tab" aria-controls="latency" aria-selected="true">Event Latency</button>
+ </li>
+ <li class="nav-item" role="presentation">
+ <button class="nav-link" id="tta-tab" data-toggle="tab" data-target="#tta" type="button" role="tab" aria-controls="tta" aria-selected="false">Time-to-Alert</button>
+ </li>
+</ul>
+
+<div class="tab-content" id="myTabContent">
+ <div class="tab-pane fade show active" id="latency" role="tabpanel" aria-labelledby="latency-tab">
+
+
+ {% autoescape off %}
+ {{ hist.latency_plot_div }}
+ {% endautoescape %}
+
+ <div class="row my-3 justify-content-md-center">
+ <div class="col-md-8">
+ <table class="table-hover table-condensed table-resp-gracedb shadow p-3 mb-5 rounded" style="text-align:center">
+ <thead>
+ <td> Pipeline </td>
+ <td> # Uploads </td>
+ <td> Median Latency (s) </td>
+ <td> 95th Percentile (s) </td>
+ <td> Min. Latency (s) </td>
+ <td> Max Latency (s) </td>
+ </thead>
+ {% for data in hist.pipeline_stats %}
+ <tr>
+ <!--<td style="font-family: monospace;"> {{data.name}} </td> -->
+ <td style="font-family: 'Courier New', Courier, monospace; font-size: 12px; white-space: pre;"> {{data.name}} </td>
+ <td> {{data.count}} </td>
+ <td> {{data.med|floatformat:3}} </td>
+ <td> {{data.nfp|floatformat:3}} </td>
+ <td> {% if data.min %} {{data.min|floatformat:3}}
+ (<a href="{% url "view" data.min_gid %}">{{ data.min_gid }}</a>)
+ {% endif %}
+ </td>
+ <td> {% if data.max %} {{data.max|floatformat:3}}
+ (<a href="{% url "view" data.max_gid %}">{{ data.max_gid }}</a>)
+ {% endif %}
+ </td>
+ </tr>
+ {% endfor %}
+ </table>
+ </div>
</div>
+
+ </div>
+
+ <div class="tab-pane fade" id="tta" role="tabpanel" aria-labelledby="tta-tab">
+ <br>
+ <p width='90%'>
+ The below chart shows the time-to-alert superevents currently in GraceDB within
+ the specified time range. Time-to-alert (in seconds) is calculated by subtracting
+ the <b>time of the astrophysical phenomenon</b>
+ (<span class="text-monospace">t<sub>0</sub></span> of the superevent) from the
+ <b>time when the <span class="text-monospace">GCN_PRELIM_SENT</span> label was applied.</b>
+ The actual receipt time of the alert by observing partners likely differs from this
+ estimate. Identify the Superevent ID by hovering over the data point.
+ </p>
+ <p width='90%'>
+ A single data-point at zero indicates that no superevents were labelled
+ <span class="text-monospace">GCN_PRELIM_SENT</span> within the specified time window.
+ </p>
+ {% autoescape off %}
+ {{ tta.tta_plot_div }}
+ {% endautoescape %}
+
+ <table class="table-hover table-condensed table-resp-gracedb shadow p-3 mb-5 rounded" style="text-align:center; width:auto; margin-left: auto; margin-right: auto;">
+ <thead>
+ <td> Median Time-to-Alert (s) </td><td> 95th Percentile (s)</td>
+ </thead>
+ <tr>
+ <td>{{tta.tta_stats.med|floatformat:1}}</td><td>{{tta.tta_stats.nfp|floatformat:1}}</td>
+ </tr>
+
+ </table>
+
+ </div>
</div>
{% endblock %}
diff --git a/requirements.txt b/requirements.txt
index bffc4c1ba..7257b4f0e 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -38,8 +38,9 @@ matplotlib==3.6.2
mock==4.0.3
numpy==1.25.2
packaging==23.1
+pandas==2.2.2
phonenumbers==8.13.18
-plotly==5.15.0
+plotly==5.23.0
psycopg==3.1.12
PyJWT==2.8.0
python-ldap==3.4.4
@@ -61,5 +62,4 @@ pytest-django==4.5.2
pytz==2023.3
pyasn1==0.4.8
pyasn1-modules==0.2.8
-# pinning setuptools because of 2to3 errors from ConcurrentLogHandler:
-#setuptools==54.2.0
+tailslide==0.2.0
--
GitLab