kw_tools/choozmo/gsc_tree2.py

import searchconsole
import dataset
import os
import networkx as nx
from pyvis.network import Network


db = dataset.connect('sqlite:///:memory:')
table=db['tmp']
rid=0

def checkig():
    global instl
    global table
    global pname
    global rid
    lst=[]
    cntdict={}
    codelist={}
    idx=0
    flag_break=False

    fname=os.path.abspath(__file__)
    elmts=fname.split(os.path.sep)
    path2=os.path.sep.join(elmts[0:-1])
    keysdir=path2+os.path.sep+'../keys'+os.path.sep

    account = searchconsole.authenticate(client_config='/Users/zooeytsai/Downloads/client_secret4.json')
    webproperty = account['https://bennis.com.tw/']
    print(webproperty)
    report = webproperty.query.range('today', days=-2).dimension('query').get()
    # report = webproperty.query.range('2022-05-20', '2022-05-23').dimension('page', 'query').get()
#    account = searchconsole.authenticate(client_config='C:\\gitlab\\kw_tools\\kw_tools\\hhh\\keys\\client_secret.json',credentials='C:\\gitlab\\kw_tools\\kw_tools\\hhh\\keys\\credentials.json')
    G = nx.Graph()

#    webproperty = account['https://ipromise.com.tw/']
#    webproperty = account['sc-domain:face8ook.org']
#     webproperty = account['sc-domain:hhh.com.tw']


    result=[]
    rdict={}
    total_idx=0

    for r in report.rows:
        if 'hhh.com.tw/designers/cases/' not in r[0]:
            continue
        if rdict.get(r[0]) is None:
            total_idx+=1
            rid=total_idx
            rdict[r[0]]=rid
        else:
            rid=rdict[r[0]]
        entry={'page':r[0],'query':r[1],'rid':rid}
        result.append(entry)

    print('list done')

    for r in result:
        table.insert(r)
    db.commit()
    print('db done')

#    cursor=db.query('select query as q,page as url,rid from tmp where rid in (select rid from (select rid,count(*) from tmp group by rid having count(*) > 2 and count(*) < 6) as tbl1) order by rid ')
    cursor=db.query('select query as q,page as url,rid from tmp order by rid ')

    riddict={}
    prev=''
    curnode=''
    cururl=''

    total_idx=0
    for c in cursor:
        G.add_edge(c['q'],c['rid'],weight=3,width=3,borderwidth=3)
    remove=[]
    G.remove_edges_from(nx.selfloop_edges(G))
    G2=G
    pyG = Network(height="600px", width="100%",bgcolor="#444444",font_color="white")

    pyG.from_nx(G2)
    pyG.show('news.html')


r=checkig()