Messenger_1Million_Stats/scripts/graph_invites.py

from million.model.message import Message
import million.parse.fb_exports as fb
import re
import math
import igraph as ig

DATA_PATH = './data/'

export = fb.parse_dirfiles(DATA_PATH)


def extraire_nom(entree: Message):
    if entree.content is None:
        return None

    regex = r"(.+) a(?:vez)? ajouté (.+)(?: et (.+))? au groupe."

    match = re.match(regex, entree.content)
    if match is not None:
        return (
            match.group(1) if match.group(1) != "Vous" else entree.sender_name, 
            *tuple(match.group(2).split(" et "))
        )
    else:
        return None


invites = []
for message in export.messages:
    noms = extraire_nom(message)
    if noms is None:
        continue
    
    invites.append((message, noms))


print(f"Total invites: {len(invites)}")

edges = []
for invite in invites:
    for invitee in invite[1]:
        edges.append((invite[0].sender_name, invitee))

vertices_as_names = list(set([edge[0] for edge in edges] + [edge[1] for edge in edges]))

# Create a graph
nb_vertices = len(vertices_as_names)
vertices_as_indexes = {name: i for i, name in enumerate(vertices_as_names)}
edges_as_indexes = [(vertices_as_indexes[edge[0]], vertices_as_indexes[edge[1]]) for edge in edges]

g = ig.Graph(nb_vertices, edges_as_indexes)
g.vs["name"] = vertices_as_names

# Plot the graph

visual_style = {
    "vertex_size": 2,
    "vertex_label": g.vs["name"],
    "vertex_label_size": 10,
    "vertex_color": "blue",
    "edge_width": 0.5,
    "edges_arrow_width": 20,
    "layout": g.layout("rt_circular"),
    "bbox": (2160, 1920),
    "margin": 100,
    "vertex_label_angle": math.pi * 1.5,
    "vertex_label_dist": 10,
}

ig.plot(g, "output/invite_graph.png", **visual_style)