GetSubGraphPerPerson.py

## Get and plot sub-graph from Theses main graph generated by GenerateGenealogyGraph
## 
## Vincent Crocher - 2022 - 2024

import pandas as pd
import networkx as nx
import pickle
import mpu
from rapidfuzz import fuzz, process

from wordcloud import WordCloud

## Extract a given field from pd
def get_this(indf, field, wherethisone, isthis):
	tmp=indf[indf[wherethisone]==isthis];
	return tmp.at[tmp.index[0], field];

## Find closest name and associated id
def find_closest(search):
	options=difflib.get_close_matches(search, search_l, n=5, cutoff=0.4)
	f=options[0]
	pid=[k for k,v in mapping.items() if v == f]
	return pid[0]

## Find closest possible names and associated ids
def find_closest_suggestions(search):
	options=process.extract(search, search_l, limit=5)
	pids=[]
	suggs=[]
	for o in options:
		pid=[k for k,v in mapping.items() if v == o[0]]
		pids.append(pid[0])
		suggs.append(o[0])
	return pids, suggs

def words_cloud(sub_g):
	words=[]
	exclude_l = ['-', 'à', 'le', 'la', 'les', 'des', 'un', 'une', 'de', 'du', 'par', 'pour', "d", "l", "et", "ou", "en", "sa", "son", "ses", "leur"
				'contribution', 'contributions', 'étude', 'études']
	text=""
	for n in sub_g:
		titre=people.loc[people['ID'] == n, 'TitreThese'].values[0]
		titre=titre.replace("'", ' ')
		titre=titre.lower()
		text=text+' '+titre
	wordcloud = WordCloud(stopwords = exclude_l,collocations=True,background_color="white").generate(text)
	print(wordcloud.to_svg())
	image = wordcloud.to_image()
	image.show()

## Subgrah around given node
def get_subgraph(start_node, mapping):
	Gd=nx.bfs_tree(G, start_node) #Get nodes downwards only
	nx.set_node_attributes(Gd, 'etud', name='class')
	Gu=nx.bfs_tree(G, start_node, reverse=True) #Get nodes upwards only
	nx.set_node_attributes(Gu, 'dir', name='class')
	G2=nx.compose(Gd,Gu.reverse()) #Merge both
	for gu in Gu: #For each upward, get downwards nodes
		G3=nx.bfs_tree(G, gu)
		G3=nx.compose(G2,G3) #Merge
	nx.set_node_attributes(G3, {start_node: 'auteur'}, name='class')
	words_cloud(G3)
	G3=nx.relabel_nodes(G3, mapping, copy=False)
	G3=nx.relabel_nodes(G3, lambda nom: nom.replace('\n','\\n'), copy=False)
	return G3


## Local quick display
def draw_local(start_node, mapping):
	G2=get_subgraph(start_node, mapping)
	nx.draw(G2, pos=nx.nx_agraph.graphviz_layout(G2, prog='dot'), arrows=None, with_labels=True, node_size=0, font_size=16)


## Format for agraph
def agraph_format(G2):
	A = nx.nx_agraph.to_agraph(G2)
	A.layout('dot', args='-Nfontsize=10 -Nwidth=".2" -Nheight=".2" -Nmargin=0 -Gfontsize=8')
	return A

## To neat PNG
def draw_png(start_node, mapping):
	G2=get_subgraph(start_node, mapping)
	A = agraph_format(G2)
	A.draw('out.png')

## To neat SVG
def draw_svg(start_node, mapping):
	G2=get_subgraph(start_node, mapping)
	A = agraph_format(G2)
	A.draw('out.svg')

## To dot file
def draw_dot(start_node, mapping):
	G2=get_subgraph(start_node, mapping)
	nx.nx_agraph.write_dot(G2,'out.dot')


## Load data
import time
start = time.process_time()
with open('ThesesAssocGraph.gpickle', 'rb') as f:
    G = pickle.load(f)

print(time.process_time() - start)
start = time.process_time()
mapping=mpu.io.read('ThesesMapping.pickle')
print(time.process_time() - start)
start = time.process_time()
people=mpu.io.read('ThesesPeople.pickle')
print(time.process_time() - start)
start = time.process_time()
#clean mapping list from NaNs
search_l=list(mapping.values())
print(time.process_time() - start)
start = time.process_time()
#search_l = {i.split('\n')[0] for i in search_l if type(i)==str} #TODO: but separate mapping and search_l
search_l = {i for i in search_l if type(i)==str}
print(time.process_time() - start)



## Testing


Key='Pierre Bourdieu'

start_nodes, sug=find_closest_suggestions(Key)
print(sug)


#quick test by node id
#start_nodes = ['16726785X']
start_nodes = ['269265910']
#print(start_nodes)

# Use drawing method:
#draw_local(start_nodes[0], mapping)
draw_svg(start_nodes[0], mapping)


#draw_png(start_nodes[0], mapping)
#draw_dot(start_nodes[0], mapping)