This commit is contained in:
2024-07-26 12:46:36 +02:00
parent 42d11c1c8d
commit 41f356e5bb
2 changed files with 51 additions and 9 deletions

View File

@@ -3,7 +3,7 @@ import mmh3
import sys
#import requests
import httplib2
DEBUG = False
DEBUG = True
def log(*s):
if DEBUG:
@@ -67,13 +67,13 @@ def createnwview(file):
createFilterTable(file)
def createFilterTable(file):
with sqlite3.connect(file) as connection:
with sqlite3.connect(file,timeout=10) as connection:
cmd_create_filter_table = f"""CREATE TABLE IF NOT EXISTS filters(cmd TEXT);"""
cursor = connection.cursor()
cursor.execute(cmd_create_filter_table)
print("db connection:",connection.total_changes)
def addFineFilter(file,table,filterstr):
with sqlite3.connect(file) as connection:
with sqlite3.connect(file,timeout=10) as connection:
cmd_createFineFilter = f"""INSERT INTO {table}(cmd) VALUES(?);"""
cmd_checkIfExists = f"""SELECT * FROM {table} WHERE cmd = ?"""
cursor = connection.cursor()
@@ -81,7 +81,8 @@ def addFineFilter(file,table,filterstr):
cursor.execute(cmd_createFineFilter,(filterstr,))
def writedb(jobs):
with sqlite3.connect("../db/sqlite3.db") as connection:
with sqlite3.connect("../db/sqlite3.db",timeout=10) as connection:
connection.execute("pragma journal_mode=wal")
print("db connection", connection.total_changes)
cursor = connection.cursor()
# cursor.execute("CREATE TABLE jobs (title TEXT, location TEXT, company TEXT,link TEXT,hash INT)")
@@ -94,16 +95,19 @@ def writedb(jobs):
print("NEW_ENTRY")
cursor.execute("INSERT INTO jobs (star,tag,title,company,location,link,pubdate,hash) VALUES (?,?,?,?,?,?,?,?)",(job.starred,job.tag,job.title,job.company,job.location,job.link,job.date,hash1))
def isStillValid(file):
def isStillValid(file,skiprows):
rows = [0,0,0]
with sqlite3.connect(file) as connection:
with sqlite3.connect(file,timeout=10) as connection:
cmd_read_chunk = f"""SELECT link from jobs;"""
connection.execute("pragma journal_mode=wal")
cursor = connection.cursor()
cursor.execute(cmd_read_chunk)
#cursor.fetchmany(skiprows)#drop rows
while(len(rows)!=0):
isLink = True
rows = []
rows = cursor.fetchmany(256)
h = httplib2.Http()
for row in rows:
@@ -135,3 +139,5 @@ def isStillValid(file):
rm_itm = rm_cursor.execute(f"""DELETE from jobs WHERE link = ?;""",(row[0],))
print ("Deletion resultet in: ", rm_itm)
print("result of commit: ", connection.commit())
return 0