kw_tools/choozmo/gsc_tree.py

import traceback
import dataset
import codecs
import sys
import pickle
import os
import searchconsole
import pandas as pd
import networkx as nx
#import pysftp
import codecs
import pyvis
import sys
import pickle
import os
import searchconsole
from pyvis.network import Network
import jieba


#db = dataset.connect('mysql://choozmo:pAssw0rd@127.0.0.1:3306/hhh?charset=utf8mb4')
#db = dataset.connect('mysql://choozmo:pAssw0rd@db.ptt.cx:3306/hhh?charset=utf8mb4')

#db.begin()
db = dataset.connect('sqlite:///:memory:')
table=db['tmp']
#table=db['gsc_page_query_year']
#pname='korea'
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='c:/keys/client_secret.json',credentials='c:/keys/credentials.json')
#    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']
#   webproperty = account['sc-domain:hhh.com.tw']

#    webproperty = account['https://www.damanwoo.com/']
    webproperty = account['https://innews.com.tw/']

#    report=webproperty.query.range('2021-03-01', '2021-06-17').dimension('page','query').get()
#    report=webproperty.query.range('2021-06-01', '2021-06-17').dimension('page','query').get()
#    report=webproperty.query.range('2020-06-01', '2021-06-22').dimension('page','query').filter('page', '/designers/cases/(491|31|293|278|31|24|594|356|307|491|33|385)', 'equals').get()
#    report=webproperty.query.range('2020-03-01', '2021-06-22').dimension('page','query').filter('page', '/designers/cases/'+pgnum, 'contains').get()
#    report=webproperty.query.range('2020-03-01', '2021-06-22').dimension('page','query').filter('page', '/designers/cases/'+pgnum, 'contains').get()
    report=webproperty.query.range('2022-01-01', '2022-04-16').dimension('page','query').get()

    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()