NBxl2gefx.py

# coding: utf8

from openpyxl import Workbook, load_workbook
from openpyxl.compat import range
import networkx as nx

# Detta skript läser en NB-export i Excel och skriver en gexf-fil för vidare bearbetning i Gephi

wb = Workbook()

wb = load_workbook('bio2014.xlsx')

# använd första arket
ws = wb.worksheets[0]

G=nx.DiGraph()

# loop spreadsheet rows
for row in ws.iter_rows(row_offset=1):

    # skip empty authors etc
    if row[1].value and row[2].value and row[5].value:

        # author
        auth = row[1].value.strip().replace(" ;","")
        G.add_node(auth, ttype="author")

        # work
        #work = row[2].value.strip()
        #G.add_node(work, ttype="work")

        # publ
        publisher = row[5].value.strip().replace(" ;","")
        G.add_node(publisher, ttype="publisher")

        cats = []

        # categories
        for col in range(6, 33):
            #if row[col].value:
            if row[col].value and "[650]" in row[col].value and " sao" in row[col].value:
                label = row[col].value.strip().replace("[650] ","").replace(" sao","")
                G.add_node(label, ttype="cat")
                #G.add_edge(work, row[col].value.strip())
                G.add_edge(publisher, label)

        # add edges
        #G.add_edge(work, auth)
        #G.add_edge(work, publisher)
        G.add_edge(auth, publisher)

nx.write_gexf(G, "nbtest650.gexf")