# -*- coding: utf-8 -*-
"""
Created on Fri May 14 22:46:21 2021
@author: bruger
"""
import dash
from jupyter_dash import JupyterDash
from jupyter_dash.comms import _send_jupyter_config_comm_request
try:
from dash import Dash, callback, html, dcc, dash_table, Input, Output, State, MATCH, ALL
import dash_bootstrap_components as dbc
from dash_interactive_graphviz import DashInteractiveGraphviz
# from dash.exceptions import PreventUpdate
# try:
# from dash import dcc
# except:
# import dash_core_components as dcc
# import dash_bootstrap_components as dbc
# try:
# from dash import html
# except:
# import dash_html_components as html
# from dash.dependencies import Input, Output, State
# from dash_interactive_graphviz import DashInteractiveGraphviz
except:
print('No Dash')
import plotly.graph_objs as go
from plotly.graph_objs.scatter.marker import Line
import webbrowser
from threading import Timer
import time
import json
import networkx as nx
from dataclasses import dataclass,field
from pathlib import Path
import pandas as pd
from modelhelp import cutout
import logging
log = logging.getLogger('werkzeug')
log.setLevel(logging.ERROR)
# log.setLevel(logging.INFO)
# log.setLevel(logging.DEBUG)
initial_dot_source = """
digraph {
node[style="filled"]
a ->b->d
a->c->d
}
"""
sidebar_width, rest_width = "15%", "82%"
# the style arguments for the sidebar. We use position:fixed and a fixed width
SIDEBAR_STYLE = {
"position": "fixed",
"top": 0,
"left": 0,
"bottom": 0,
"width": sidebar_width,
"padding": "2rem 1rem",
"background-color": "#f8f9fa",
"overflow": "scroll",
}
# the styles for the main content position it to the right of the sidebar and
# add some padding.
CONTENT_STYLE_TOP = {
"background-color": "f8f9fa",
"width": "100%",
"margin-left": "0%",
}
CONTENT_STYLE_GRAPH = {
"background-color": "f8f9fa",
"width":"82%",
"margin-left": '0%',
}
CONTENT_STYLE_TAB = {
"background-color": "f8f9fa",
"margin-left": sidebar_width,
"width": "85%",
"height": "auto",
}
[docs]
def app_setup(jupyter=False):
print('apprun')
if jupyter:
# Timer(1,_send_jupyter_config_comm_request).start()
# JupyterDash.infer_jupyter_proxy_config()
app = JupyterDash(__name__,external_stylesheets=[dbc.themes.BOOTSTRAP])
else:
app = dash.Dash(__name__, external_stylesheets=[dbc.themes.BOOTSTRAP])
return app
[docs]
def app_run(app,jupyter=False,debug=False,port=5000,inline=True):
def open_browser(port=port):
webbrowser.open_new(f"http://localhost:{port}")
# print('hest')
if jupyter:
if inline:
xx = app.run_server(debug=debug,port=port,mode='inline')
else:
Timer(1, open_browser).start()
# print('ko')
xx =app.run_server(debug=debug,port=port,mode='external')
# print('gris')
# print(f'{xx=}')
else:
# Timer(1, open_browser).start()
xx = app.run_server(debug=debug,port=port,jupyter_mode="tab")
[docs]
def get_stack(df,v='Guess it',heading='Yes',pct=True,threshold=0.5,desdict = {}):
pv = cutout(df,threshold)
template = '%{y:.1f}% explained by:<br><b>%{meta}</b>:' if pct else '%{y:.2f} explained by:<br><b>%{meta}</b>:'
trace = [go.Bar(x=pv.columns.astype('str'), y=pv.loc[rowname,:], name=rowname,hovertemplate = template,
meta = desdict.get(rowname,'hest')) for rowname in pv.index]
out = { 'data': trace,
'layout':
go.Layout(title=f'{heading}', barmode='relative',legend_itemclick='toggleothers',
yaxis = {'tickformat': ',7.0','ticksuffix':'%' if pct else ''})
}
return out
[docs]
def get_no_stack(df,v='No attribution for exogenous variables ',desdict = {}):
out = {'layout':
go.Layout(title=f'{v}')
}
return out
[docs]
def get_line_old(pv,v='Guess it',heading='Yes'):
trace = [go.Line(x=pv.columns.astype('str'), y=pv.loc[rowname,:], name=rowname) for rowname in pv.index]
out = { 'data': trace,
'layout':
go.Layout(title=f'{heading}')
}
return out
[docs]
def get_line(pv,v='Guess it',heading='Yes',pct=True):
trace = [go.Scatter(x=pv.columns.astype('str'),
y=pv.loc[rowname,:],
name=rowname
) for rowname in pv.index]
out = { 'data': trace,
'layout':
go.Layout(title=f'{heading}')
}
return out
[docs]
def generate_table(dataframe, max_rows=10):
return html.Table([
html.Thead(
html.Tr([html.Th('var')]+[html.Th(col) for col in dataframe.columns])
),
html.Tbody([
html.Tr([html.Td(dataframe.index[i])]+[html.Td(f'{dataframe.iloc[i][col]:25.2f}') for col in dataframe.columns])
for i in range(min(len(dataframe), max_rows))
])
])
[docs]
@dataclass
class Dash_graph():
mmodel : any = None
pre_var : str =''
filter: float = 0
up : int = 1
down : int = 0
time_att : bool = False
attshow :bool = False
all : bool = False
dashport : int = 5001
debug : bool = False
jupyter : bool = True
show_trigger : bool = False
inline : bool =False
lag : bool = False
threshold : float =0.5
growthshow : bool = False
def __post_init__(self):
self.fokusvar = set()
# print('Still worlking on the layout of this')
self.firstsession = True
selected_var = self.pre_var if self.pre_var else sorted(self.mmodel.allvar.keys())[0]
self.outvar_state = selected_var
sidebar = html.Div(
[
html.H3("Current Variable"),
dcc.Markdown(
id='outvar_state',
children=f'{selected_var}'),
html.H3("Select a variable"),
dcc.Dropdown(id='var',value=selected_var,options=[
dict(label=v, value=v)
for v in sorted(self.mmodel.allvar.keys())],persistence = True,persistence_type='session'),
html.H3("Tree walking"),
dbc.Row([
html.H5("Up"),
dcc.Dropdown(id="up",value=self.up,options=[
dict(label=engine, value=engine)
for engine in list(range(10))],persistence = True,persistence_type='session'),
html.H5("Down"),
dcc.Dropdown(id="down",value=self.down,options=[
dict(label=engine, value=engine)
for engine in list(range(10))],persistence = True,persistence_type='session'),
]),
dbc.Tooltip('Looking at preceding variabels up to this level',
target = 'up'),
dbc.Tooltip('Looking at dependent variabels down to this level',
target = 'down'),
html.H3("Graph filter%"),
dcc.Dropdown(id="filter",value=self.filter,options=[
dict(label=t, value=t)
for t in list(range(0,100,10))],persistence = True,persistence_type='session'),
dbc.Tooltip('All branches contribution less are pruned from the graph',
target = 'filter'),
html.H3("Graph orientation"),
dcc.RadioItems(id='orient',
options=[
{'label': 'Vertical', 'value':'v'},
{'label': 'Horisontal', 'value': 'h'},
],
value='v',labelStyle={'display': 'block'},persistence = True,persistence_type='session' ),
html.H3("Node Display"),
dcc.RadioItems(id='node',
options=[
{'label': 'Name', 'value':'name'},
{'label': '+values', 'value':'all'},
{'label': '+Attribution', 'value': 'attshow'},
],persistence = True, persistence_type='session',
value='all' if self.all else ('attshow' if self.attshow else 'name') ,labelStyle={'display': 'block'}),
html.H3("Behavior when clicking on node"),
dcc.RadioItems(id='onclick',
options=[
{'label': 'Center', 'value':'c'},
{'label': 'Display', 'value': 'd'},
],
value='c',persistence = True,persistence_type='session'),
],
style=SIDEBAR_STYLE
)
# breakpoint()
outvar= selected_var
tab0 = html.Div([
dbc.Tabs(id="tabs", children=[
dbc.Tab(id='Graph',label='Graph',children= [DashInteractiveGraphviz(id="gv" , style=CONTENT_STYLE_GRAPH,
dot_source = self.mmodel.draw(selected_var,up=self.up,down=self.down,showatt=False,lag=self.lag,
debug=0,dot=True,HR=False,filter = self.filter,
all=False,attshow=False))],
style=CONTENT_STYLE_TOP , ),
dbc.Tab(id='Chart',label='Chart',
children = [dcc.Graph(id='chart',
figure=get_line(self.mmodel.value_dic[selected_var].iloc[:2,:],selected_var,f'The values for {selected_var}'))
, dcc.Graph(id='chart_dif',
figure=get_line(self.mmodel.value_dic[selected_var].iloc[[2],:],selected_var,f'The impact for {selected_var}'))
],
style=CONTENT_STYLE_TOP),
dbc.Tab(id='Attribution',label='Attribution',
children = [dcc.Graph(id='att_pct',
figure = (get_stack(self.mmodel.att_dic[outvar],outvar,f'Attribution of the impact - pct. for {outvar}',
threshold=self.threshold,desdict=self.mmodel.var_description)
if outvar in self.mmodel.endogene else html.H3("Graph orientation")))
, dcc.Graph(id='att_level',
figure = (get_stack(self.mmodel.att_dic_level[outvar],outvar,f'Attribution of the impact - level for {outvar}',
pct=False,threshold=self.threshold,desdict=self.mmodel.var_description) if outvar in self.mmodel.endogene else html.H3("Graph orientation")))
],
style=CONTENT_STYLE_TOP)
],persistence=True, persistence_type = 'session'),
],style = CONTENT_STYLE_TAB)
# tabbed = tab0
# tabbed = dbc.Container(tab0,id='tabbed',style={"height": "100vh"},fluid=True)
tabbed = html.Div(tab0,id='tabbed',style={"height": "100vh"})
self.app = app_setup(jupyter=self.jupyter)
# app.layout = html.Div([sidebar,body2])
self.app.layout = dbc.Container([sidebar,tabbed],style={"height": "100vh","width":"100%"},fluid=True)
@self.app.callback(
[Output("gv", "dot_source"),
Output('chart','figure'), Output('chart_dif','figure'),
Output('att_pct','figure'), Output('att_level','figure'),
Output('outvar_state','children'),
Output('tabs','active_tab')],
[Input('var', "value"),
Input('gv', "selected_node"), Input('gv', "selected_edge"),
Input('up', "value"),Input('down', "value"),Input('filter', "value"),
Input('orient', "value"),
Input('onclick','value'),
Input('node','value')
]
, State('outvar_state','children')
)
def display_output( var,
selected_node,selected_edge,
up,down,filter,
orient,
onclick,
node,
outvar_state
):
# time.sleep(3)
ctx = dash.callback_context
# if ctx.triggered[0]['prop_id']== 'gv.selected_node' and ctx.triggered[0]['value']== None:
# raise PreventUpdate
# print(f'new trigger :{ctx.triggered=}')
# breakpoint()
if ctx.triggered:
trigger = ctx.triggered[0]['prop_id'].split('.')[0]
if self.show_trigger:
print('\nNew Trigger:')
ctx_msg = json.dumps({
'states': ctx.states,
'triggered': ctx.triggered,
'inputs': ctx.inputs
}, indent=2)
print(ctx_msg)
# print(f'{outvar=},{data_show=}')
print(f'{outvar_state=}')
print(f'{self.fokusvar=} When triggerd')
update_var = True
if trigger == 'var':
outvar=var
# elif trigger == 'gv' :
# print(f'{selected_node=}')
elif selected_node:
xvar= selected_node.split('(')[0]
if onclick == 'c':
outvar = xvar if trigger == 'gv' else self.outvar_state
elif onclick == 'd' :
if trigger == 'gv':
if xvar in self.fokusvar:
self.fokusvar = self.fokusvar-{xvar}
else:
self.fokusvar = self.fokusvar | {xvar}
outvar=xvar
else:
outvar=self.outvar_state
self.fokusvar = set() if trigger in {'onclick','var'} else self.fokusvar
if onclick == 'c' or trigger == 'var' : #or outvar not in self.mmodel.value_dic.keys() : # or trigger in ['up','down','orient','filter','node'] :
dot_out = self.mmodel.draw(outvar,up=up,down=down,filter=filter,showatt=False,debug=0,
lag=self.lag,dot=True,HR=orient=='h',last=0,all = False,
attshow=False)
elif onclick == 'd' :
# print(f'{outvar=}')
if self.show_trigger:
print(f'{self.fokusvar=} on display ')
dot_out = self.mmodel.draw(self.outvar_state,up=up,down=down,filter=filter,showatt=False,debug=0,
lag=self.lag,dot=True,HR=orient=='h',fokus2=self.fokusvar,all= node == 'all',
attshow= node =='attshow',growthshow= node=='all' or node =='attshow',)
chart_out = get_line(self.mmodel.get_values(outvar).iloc[:2,:],outvar,f'The values for {outvar}:{self.mmodel.var_description.get(outvar,"")}')
chart_dif_out = get_line(self.mmodel.get_values(outvar).iloc[[2],:],outvar,f'The impact for {outvar}:{self.mmodel.var_description.get(outvar,"")}')
att_pct_out = (get_stack(self.mmodel.get_att_pct(
outvar.split('(')[0], lag=False, start='', end=''),outvar,f'Attribution of the impact - pct. for {outvar}:{self.mmodel.var_description.get(outvar,"")}'
,threshold=self.threshold,desdict=self.mmodel.var_description)
if outvar in self.mmodel.endogene else html.H3(""))
att_level_out = (get_stack(self.mmodel.get_att_level(
outvar.split('(')[0], lag=False, start='', end=''),outvar,f'Attribution of the impact - level for {outvar}',
pct=False,threshold=self.threshold,desdict=self.mmodel.var_description) if outvar in self.mmodel.endogene else html.H3("") )
else:
return [dash.no_update, dash.no_update,dash.no_update,dash.no_update,dash.no_update, dash.no_update, dash.no_update]
if self.show_trigger:
print(f'{self.fokusvar=} before render')
self.outvar_state = outvar if onclick == 'c' or trigger == 'var' else self.outvar_state
outvar_state = self.outvar_state
return [dot_out, chart_out, chart_dif_out, att_pct_out, att_level_out, outvar_state ,dash.no_update,]
app_run(self.app,jupyter=self.jupyter,debug=self.debug,port=self.dashport,inline=self.inline)
#%test
if __name__ == "__main__":
#%% testing
if not 'baseline' in locals() or 0 :
from modelclass import model
madam,baseline = model.modelload('../Examples/ADAM/baseline.pcim',run=1,silent=0 )
# make a simpel experimet VAT
scenarie = baseline.copy()
scenarie.TG = scenarie.TG + 0.05
_ = madam(scenarie)
_ = Dash_graph(madam,'FY',debug = 0,all=1,filter=30,show_trigger=True,jupyter=True,up=1,dashport=5005)