Develop Reference

Scrapy MySQL foreign key constraint failures

I'm currently working on writing a web scraper for a website called Mountain Project and have come across an issue while inserting items into a MySQL (MariaDB) database.
The basic flow of my crawler is this:
Get response from link extractor (I'm subclassing CrawlSpider)
Extract data from response and send extracted items down the item pipeline
Items get picked up by the SqlPipeline and inserted into the database
An important note about step 2 is that the crawler is sending multiple items down the pipeline. The first item will be the main resource (either a route or an area), and then any items after that will be other important data about that resource. For areas, those items will be data on different weather conditions, for routes those items will be the different grades assigned to those routes.
My area and route tables look like this:
CREATE TABLE `area` (
`area_id` INT(10) UNSIGNED NOT NULL,
`parent_id` INT(10) UNSIGNED NULL DEFAULT NULL,
`name` VARCHAR(200) NOT NULL DEFAULT '',
`latitude` FLOAT(12) NULL DEFAULT -1,
`longitude` FLOAT(12) NULL DEFAULT -1,
`elevation` INT(11) NULL DEFAULT '0',
`link` VARCHAR(300) NOT NULL,
PRIMARY KEY (`area_id`) USING BTREE
)
CREATE TABLE `route` (
`route_id` INT(10) UNSIGNED NOT NULL,
`parent_id` INT(10) UNSIGNED NOT NULL,
`name` VARCHAR(200) NOT NULL DEFAULT '' COLLATE 'utf8_general_ci',
`link` VARCHAR(300) NOT NULL COLLATE 'utf8_general_ci',
`rating` FLOAT(12) NULL DEFAULT '0',
`types` VARCHAR(50) NULL DEFAULT NULL COLLATE 'utf8_general_ci',
`pitches` INT(3) NULL DEFAULT '0',
`height` INT(5) NULL DEFAULT '0',
`length` VARCHAR(5) NULL DEFAULT NULL COLLATE 'utf8_general_ci',
PRIMARY KEY (`route_id`) USING BTREE,
INDEX `fk_parent_id` (`parent_id`) USING BTREE,
CONSTRAINT `fk_parent_id` FOREIGN KEY (`parent_id`) REFERENCES `mountainproject`.`area` (`area_id`) ON UPDATE CASCADE ON DELETE CASCADE
)
And here's an example of one of my condition tables:
CREATE TABLE `temp_avg` (
`month` INT(2) UNSIGNED NOT NULL,
`area_id` INT(10) UNSIGNED NOT NULL,
`avg_high` INT(3) NOT NULL,
`avg_low` INT(3) NOT NULL,
PRIMARY KEY (`month`, `area_id`) USING BTREE,
INDEX `fk_area_id` (`area_id`) USING BTREE,
CONSTRAINT `fk_area_id` FOREIGN KEY (`area_id`) REFERENCES `mountainproject`.`area` (`area_id`) ON UPDATE CASCADE ON DELETE CASCADE
)
Here's where things start to get troublesome. If I run my crawler and just extract areas, everything works fine. The area is inserted into the database, and all the conditions data gets inserted without a problem. However, when I try to extract areas and routes, I get foreign key constraint failures when trying to insert routes because the area that the route belongs to (parent_id) doesn't exist. Currently to work around this I've been running my crawler twice. Once to extract area data, and once to extract route data. If I do that, everything goes smoothly.
My best guess as to why this doesn't work currently is that the areas that are being inserted haven't been committed and so when I attempt to add a route that belongs to an uncommitted area, it can't find the parent area. This theory quickly falls apart though because I'm able to insert condition data in the same run that I insert the area that the data belongs to.
My insertion code looks like this:
def insert_item(self, table_name, item):
encoded_vals = [self.sql_encode(val) for val in item.values()]
sql = "INSERT INTO %s (%s) VALUES (%s)" % (
table_name,
", ".join(item.keys()),
", ".join(encoded_vals)
)
logging.debug(sql)
self.cursor.execute(sql)
# EDIT: As suggested by #tadman I have moved to using the built in sql value
# encoding. I'm leaving this here because it doesn't affect the issue
def sql_encode(self, value):
"""Encode provided value and return a valid SQL value
Arguments:
value {Any} -- Value to encode
Returns:
str -- SQL encode value as a str
"""
encoded_val = None
is_empty = False
if isinstance(value, str):
is_empty = len(value) == 0
encoded_val = "NULL" if is_empty or value is None else value
if isinstance(encoded_val, str) and encoded_val is not "NULL":
encode_val = encoded_val.replace("\"", "\\\"")
encoded_val = "\"%s\"" % encoded_val
return str(encoded_val)
The rest of the project lives in a GitHub repo if any more code/context is needed

Mysql insert lock wait timeout exceeded - auto increment

I'm having an issue with my application causing MySQL table to be locked due to inserts which take a long time, after reviewing online articles, it seems like it's related to auto increment, info below -
Python that inserts data (row at a time unfortunately as I need the auto incremented id for reference in future inserts) -
for i, flightobj in stats[ucid]['flight'].items():
flight_fk = None
# Insert flights
try:
with mysqlconnection.cursor() as cursor:
sql = "insert into cb_flights(ucid,takeoff_time,end_time,end_event,side,kills,type,map_fk,era_fk) values(%s,%s,%s,%s,%s,%s,%s,%s,%s);"
cursor.execute(sql, (
ucid, flightobj['start_time'], flightobj['end_time'], flightobj['end_event'], flightobj['side'],
flightobj['killnum'], flightobj['type'], map_fk, era_fk))
mysqlconnection.commit()
if cursor.lastrowid:
flight_fk = cursor.lastrowid
else:
flight_fk = 0
except pymysql.err.ProgrammingError as e:
logging.exception("Error: {}".format(e))
except pymysql.err.IntegrityError as e:
logging.exception("Error: {}".format(e))
except TypeError as e:
logging.exception("Error: {}".format(e))
except:
logging.exception("Unexpected error:", sys.exc_info()[0])
The above runs every 2 minutes on the same data and is supposed to insert only non duplicates as the MySQL would deny duplicates due to the unique ucid_takeofftime index.
MYSQL info, cb_flights table -
`pk` int(11) NOT NULL AUTO_INCREMENT,
`ucid` varchar(50) NOT NULL,
`takeoff_time` datetime DEFAULT NULL,
`end_time` datetime DEFAULT NULL,
`end_event` varchar(45) DEFAULT NULL,
`side` varchar(45) DEFAULT NULL,
`kills` int(11) DEFAULT NULL,
`type` varchar(45) DEFAULT NULL,
`map_fk` int(11) DEFAULT NULL,
`era_fk` int(11) DEFAULT NULL,
`round_fk` int(11) DEFAULT NULL,
PRIMARY KEY (`pk`),
UNIQUE KEY `ucid_takeofftime` (`ucid`,`takeoff_time`),
KEY `ucid_idx` (`ucid`) /*!80000 INVISIBLE */,
KEY `end_event` (`end_event`) /*!80000 INVISIBLE */,
KEY `side` (`side`)
) ENGINE=InnoDB AUTO_INCREMENT=76023132 DEFAULT CHARSET=utf8;
Now inserts into the table from the Python code, can take sometimes over 60 seconds.
I beleive it might be related to the auto increment that is creating the lock on the table, if so, I'm looking for a workaround.
innodb info -
innodb_autoinc_lock_mode 2
innodb_lock_wait_timeout 50
buffer is used up to 70% more or less.
Appreciate any assistance with this, either from application side or MySQL side.
EDIT
Adding the create statement for the cb_kills table which is also used with inserts but without an issue as far as I can see, this is in response to the comment on the 1st answer.
CREATE TABLE `cb_kills` (
`pk` int(11) NOT NULL AUTO_INCREMENT,
`time` datetime DEFAULT NULL,
`killer_ucid` varchar(50) NOT NULL,
`killer_side` varchar(10) DEFAULT NULL,
`killer_unit` varchar(45) DEFAULT NULL,
`victim_ucid` varchar(50) DEFAULT NULL,
`victim_side` varchar(10) DEFAULT NULL,
`victim_unit` varchar(45) DEFAULT NULL,
`weapon` varchar(45) DEFAULT NULL,
`flight_fk` int(11) NOT NULL,
`kill_id` int(11) NOT NULL,
PRIMARY KEY (`pk`),
UNIQUE KEY `ucid_killid_flightfk_uniq` (`killer_ucid`,`flight_fk`,`kill_id`),
KEY `flight_kills_fk_idx` (`flight_fk`),
KEY `killer_ucid_fk_idx` (`killer_ucid`),
KEY `victim_ucid_fk_idx` (`victim_ucid`),
KEY `time_ucid_killid_uniq` (`time`,`killer_ucid`,`kill_id`),
CONSTRAINT `flight_kills_fk` FOREIGN KEY (`flight_fk`) REFERENCES `cb_flights` (`pk`)
) ENGINE=InnoDB AUTO_INCREMENT=52698582 DEFAULT CHARSET=utf8;

You can check if autocommit is set to 1, this forces to commit every row and disabling it makes it somewhat faster
Instead of committing every insert try to bulk insert.
For that you should check
https://dev.mysql.com/doc/refman/8.0/en/optimizing-innodb-bulk-data-loading.html
and do something like
data = [
('city 1', 'MAC', 'district 1', 16822),
('city 2', 'PSE', 'district 2', 15642),
('city 3', 'ZWE', 'district 3', 11642),
('city 4', 'USA', 'district 4', 14612),
('city 5', 'USA', 'district 5', 17672),
]
sql = "insert into city(name, countrycode, district, population)
VALUES(%s, %s, %s, %s)"
number_of_rows = cursor.executemany(sql, data)
db.commit()

I want to put in here some of the ways I worked on finding a solution to this problem. I'm not an expert in MySQL but I think these steps can help anyone looking to find out why he has lock wait timeouts.
So the troubleshooting steps I took are as follows -
1- Check if I can find in the MySQL slow log the relevant query that is locking my table. Usually it's possible to find queries that run a long time and also locks with the info below and the query right after it
# Time: 2020-01-28T17:31:48.634308Z
# User#Host: # localhost [::1] Id: 980397
# Query_time: 250.474040 Lock_time: 0.000000 Rows_sent: 10 Rows_examined: 195738
2- The above should give some clue on what's going on in the server and what might be waiting for a long time. Next I ran the following 3 queries to identify what is in use:
check process list on which process are running -
show full processlist;
check tables in use currently -
show open tables where in_use>0;
check running transactions -
SELECT * FROM `information_schema`.`innodb_trx` ORDER BY `trx_started`;
3- The above 2 steps should give enough information on which query is locking the tables. in my case here I had a SP that ran an insert into <different table> select from <my locked table>, while it was inserting to a totally different table, this query was locking my table due to the select operation that took a long time.
To work around it, I changed the SP to work with temporary tables and now although the query is still not completely optimized, there are no locks on my table.
Adding here how I run the SP on temporary tables for async aggregated updates.
CREATE DEFINER=`username`#`%` PROCEDURE `procedureName`()
BEGIN
drop temporary table if exists scheme.temp1;
drop temporary table if exists scheme.temp2;
drop temporary table if exists scheme.temp3;
create temporary table scheme.temp1 AS select * from scheme.live1;
create temporary table scheme.temp2 AS select * from scheme.live2;
create temporary table scheme.temp3 AS select * from scheme.live3;
create temporary table scheme.emptytemp (
`cName1` int(11) NOT NULL,
`cName2` varchar(45) NOT NULL,
`cName3` int(11) NOT NULL,
`cName4` datetime NOT NULL,
`cName5` datetime NOT NULL,
KEY `cName1` (`cName1`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci;
INSERT into scheme.emptytemp
select t1.x,t2.y,t3.z
from scheme.temp1 t1
JOIN scheme.temp2 t2
ON t1.x = t2.x
JOIN scheme.temp3 t3
ON t2.y = t3.y
truncate table scheme.liveTable;
INSERT into scheme.liveTable
select * from scheme.emptytemp;
END
Hope this helps anyone that encounters this issue

sqlite3.OperationalError: no such table: main.source

I am making an API that uses Python Flask and Sqlite3. Most of it works. Specifically:
all GET endpoints/SELECT queries work
two POST endpoints/INSERT INTO queries work
However, the remaining POSTs/INSERT INTOs do not work. They all have the same sqlite3.OperationalError with the message:
no such table: main.source
This is weird because none of the queries use a table called "source" or "main.source". I print the queries before I execute them, and I have tried copy/pasting the queries into the sqlite3 command prompt. The queries have no issues when I do that.
The other weird thing is that all the INSERT INTO queries call the same function to create the actual query (which in turn calls the function to run queries ... used by ALL queries most of which work). Only some of the INSERT INTOs cause this error.
Some potentially useful information:
An excerpt from createdb.sql
CREATE TABLE transactions (
id INTEGER PRIMARY KEY,
buyer INTEGER NOT NULL,
seller INTEGER NOT NULL,
amount INTEGER NOT NULL,
currency VARCHAR(6) NOT NULL,
fee INTEGER NOT NULL,
source INTEGER NOT NULL,
description TEXT NOT NULL,
status VARCHAR(40) NOT NULL,
created TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP
);
The insert into that Python prints/that throws an error in execute:
INSERT INTO transactions (status, fee, description, source, seller, currency, amount, buyer) VALUES ('initiated', '1', 'nada', '1', '2', 'USD', '1000', '1');
And some stuff from Sqlite's prompt:
sqlite> .tables
conversations sources users
messages transactions withdrawals
sqlite> SELECT id, description FROM transactions;
1|hella mulah
2|payback
3|woohoo
sqlite> INSERT INTO transactions (status, fee, description, source, seller, currency, amount, buyer) VALUES ('initiated', '1', 'nada', '1', '2', 'USD', '1000', '1');
sqlite>
sqlite> SELECT id, description FROM transactions;
1|hella mulah
2|payback
3|woohoo
4|nada
For references, here is a POST command that has no error despite using most of the same stuff:
INSERT INTO users (session, balance, name, firebaseToken) VALUES ('ABCDEFG', '0', 'Mr Miyagi', 'ABCDEFG');
There are a lot of similar questions on SO but here's why they are not duplicates:
flask/sqlalchemy - OperationalError: (sqlite3.OperationalError) no such table Yes, I created the tables before using them.
Why am I suddenly getting "OperationalError: no such table"? my Flask app is able to find the database no problem (and most of the queries work without a hitch). To be safe I made my DB in connect an absolute path. No effect.
Error: sqlite3.OperationalError: no such table: main.m I don't do any weird indexing + if I did this the copy pasting wouldn't work
Python Sqlite3 - Data is not saved permanently: I do call commit in #app.teardown_appcontext. I also tried calling commit after every query. No effect.
Other stuff I considered but ruled out:
There is a list of disallowed names, but these are not included http://www.sqlite.org/lang_keywords.html. transactions is close to transaction but not the same. source is not on the list.
I'm sure this will end up being some kind of silly messup but any ideas on where to look would be much appreciated. I also tried Googling this error but I didn't see anything helpful.
--- more code ---
This is database.py
import sqlite3
import flask
import backend
def dict_factory(cursor, row):
output = {}
for idx, col in enumerate(cursor.description):
output[col[0]] = row[idx]
return output
def get_db():
if not hasattr(flask.g, 'sqlite_db'):
flask.g.sqlite_db = sqlite3.connect("/my/absolute/path/var/data.db"
)
flask.g.sqlite_db.row_factory = dict_factory
flask.g.sqlite_db.execute("PRAGMA foreign_keys = ON;")
return flask.g.sqlite_db
def query(query, args=(), islast=False):
print(query) # this is where the print from before is
cur = get_db().execute(query, args)
rowvector = cur.fetchall()
if islast:
cur.close()
return rowvector
#backend.app.teardown_appcontext
def close_db(error):
if hasattr(flask.g, 'sqlite_db'):
flask.g.sqlite_db.commit()
flask.g.sqlite_db.close()
This is selected sections from apiimpl.py
QUERY_INSERT = "INSERT INTO"
QUERY_SELECT = "SELECT"
QUERY_UPDATE = "UPDATE"
def queryhelper(*args, **kwargs):
sqltxt = None
selectstr = None
if kwargs["action"] == QUERY_INSERT:
sqltxt = "{} {} ({}) VALUES ({});".format(
QUERY_INSERT,
kwargs["table"],
", ".join(["{}".format(x) for x in kwargs["cols"]]),
", ".join(["'{}'".format(x) for x in kwargs["vals"]]),
)
# pretty sure this next bit is not relevant but here it is anyway
selectstr = "SELECT * FROM {} WHERE ROWID=(SELECT last_insert_rowid());".format(
kwargs["table"],
)
elif kwargs["action"] == QUERY_SELECT:
# not relevant
elif kwargs["action"] == QUERY_UPDATE:
# not relevant
else:
assert(kwargs["action"] in [QUERY_INSERT, QUERY_SELECT, QUERY_UPDATE,])
try:
rv = db.query(sqltxt) # this is where the error is thrown
if selectstr:
return db.query(selectstr)
else:
return rv
except sqlite3.OperationalError as e:
# this is where the error is caught
return api_error("SQL error (1): {}", str(e), code=500)
def append(tablename, args):
tabledata = TABLES().tablenamemap[tablename]
print("tablename: " + tablename) # "tablename: transactions"
# a bunch of error detection
rv = queryhelper(
action=QUERY_INSERT,
table=tablename,
cols=args.keys(),
vals=args.values(),
)
# not shown: potentially returning json.dumps(rv)
return rv
def transactions_post(req):
# a lot of stuff to turn req into validargs
# printed validargs: {'status': 'initiated', u'fee': u'1', u'description': u'nada', u'source': u'1', u'seller': u'2', u'currency': u'USD', u'amount': u'1000', u'buyer': u'1'}
return append("transactions", validargs)
#backend.app.route("/transactions", methods=["GET", "POST", "PUT"])
def transactions_route():
return {
"GET": transactions_get, # get list of transactions
"POST": transactions_post, # initiate a transaction
"PUT": transactions_put, # change transaction status
}[flask.request.method](flask.request)
P.S. the purpose of this question is not to discuss the implementation, but if you want to leave a comment that's ok with me.
--- in response to comment --
sqlite> SELECT * FROM sqlite_master WHERE type="table" AND name="transactions";
table|transactions|transactions|4|CREATE TABLE transactions (
id INTEGER PRIMARY KEY,
buyer INTEGER NOT NULL,
seller INTEGER NOT NULL,
amount INTEGER NOT NULL,
currency VARCHAR(6) NOT NULL,
fee INTEGER NOT NULL,
source INTEGER NOT NULL,
description TEXT NOT NULL,
status VARCHAR(40) NOT NULL,
created TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
FOREIGN KEY (buyer) REFERENCES users(id), -- do not want to delete on CASCADE
FOREIGN KEY (seller) REFERENCES users(id), -- do not want to delete on CASCADE
FOREIGN KEY (source) REFERENCES source(id) -- do not want to delete on CASCADE
)

It looks like you are referencing a table that doesn't exist based on your .tables command.
sqlite> .tables
conversations sources users
messages transactions withdrawals
And this create table statement.
CREATE TABLE transactions (
id INTEGER PRIMARY KEY,
buyer INTEGER NOT NULL,
seller INTEGER NOT NULL,
amount INTEGER NOT NULL,
currency VARCHAR(6) NOT NULL,
fee INTEGER NOT NULL,
source INTEGER NOT NULL,
description TEXT NOT NULL,
status VARCHAR(40) NOT NULL,
created TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
FOREIGN KEY (buyer) REFERENCES users(id), -- do not want to delete on CASCADE
FOREIGN KEY (seller) REFERENCES users(id), -- do not want to delete on CASCADE
FOREIGN KEY (source) REFERENCES source(id) -- do not want to delete on CASCADE
-- ^ there is no source table
)
If you change source(id) to sources(id) you should be good.

MySQL datetime column WHERE col IS NULL fails

I cannot get my very basic SQL query to work as it returns 0 values despite the fact that there are clearly nulls
query
SELECT
*
FROM
leads AS l
JOIN closes c ON l.id = c.lead_id
WHERE
c.close_date IS NULL
DDL
CREATE TABLE closes
(
id INT AUTO_INCREMENT
PRIMARY KEY,
lead_id INT NOT NULL,
close_date DATETIME NULL,
close_type VARCHAR(255) NULL,
primary_agent VARCHAR(255) NULL,
price FLOAT NULL,
gross_commission FLOAT NULL,
company_dollar FLOAT NULL,
address VARCHAR(255) NULL,
city VARCHAR(255) NULL,
state VARCHAR(10) NULL,
zip VARCHAR(10) NULL,
CONSTRAINT closes_ibfk_1
FOREIGN KEY (lead_id) REFERENCES leads (id)
)
ENGINE = InnoDB;
CREATE INDEX lead_id
ON closes (lead_id);
I should mention that I am inserting the data with a python web scraper and SQLAlchemy. If the data is not scraped it will be None on insert.
Here is a screenshot of datagrip showing a null value in the row
EDIT
Alright so I went ahead and ran the following on some of the entries in the table where the value was already <null>
UPDATE closes
SET close_date = NULL
WHERE
lead_id = <INTEGERVAL>
;
What is interesting now is that when running the original query I do actually return the 2 records that I ran the update query for (the expected outcome). This would lead me to believe that the issues is with how my SQLAlchemy model is mapping the values on insert.
models.py
class Close(db.Model, ItemMixin):
__tablename__ = 'closes'
id = db.Column(db.Integer, primary_key=True)
lead_id = db.Column(db.Integer, db.ForeignKey('leads.id'), nullable=False)
close_date = db.Column(db.DateTime)
close_type = db.Column(db.String(255))
primary_agent = db.Column(db.String(255))
price = db.Column(db.Float)
gross_commission = db.Column(db.Float)
company_dollar = db.Column(db.Float)
address = db.Column(db.String(255))
city = db.Column(db.String(255))
state = db.Column(db.String(10))
zip = db.Column(db.String(10))
def __init__(self, item):
self.build_from_item(item)
def build_from_item(self, item):
for k, v in item.items():
setattr(self, k, v)
But I am fairly confident the value is a python None in the event no value is scraped from the website. My understanding is that SQLAlchemy would map a None to NULL on insert and given that nullable=True is the default setting, which can seen on the generated DDL, I am still at a loss as to why it appears to be NULL when in reality it is not behaving that way.
EDIT 2
Only place where I think the issue would be happening is where my spider actually scrapes the data and assigns it to the Item which is shown below
closes.py
# item['close_date'] = None at this point
try:
item['close_date'] = arrow.get(item['close_date'], 'MMM D, YYYY').format('YYYY-MM-DD')
except ParserError as e:
# Maybe item['close_date'] = None here?
spider.logger.error(f'Parse error: {item["close_date"]} - {e}')
In the python code I've written this would appear to be the place where the issue would arise. But if arrow.get throws an exception the value of item['close_date'] should still be None. If that is not the case and even if it is it does not explain why it appears that the record value is NULL even thought it does not behave like it is.

I'm guessing that you're having an issue with the join, not the NULL value. The query below returns 1 result for me. More info about your data, the software used for querying (I tested with SQL Yog), and applicable versions might help.
EDIT
It could be that you're having issues with MySQL's 'zero date'.
https://dev.mysql.com/doc/refman/5.7/en/date-and-time-types.html
MySQL permits you to store a “zero” value of '0000-00-00' as a “dummy
date.” This is in some cases more convenient than using NULL values,
and uses less data and index space. To disallow '0000-00-00', enable
the NO_ZERO_DATE mode.
I've updated the SQL data below to include a zero date in the INSERT and SELECT's WHERE.
DROP TABLE IF EXISTS closes;
DROP TABLE IF EXISTS leads;
CREATE TABLE leads (
id INT(11) NOT NULL AUTO_INCREMENT,
PRIMARY KEY (id)
) ENGINE=INNODB AUTO_INCREMENT=7 DEFAULT CHARSET=utf8;
INSERT INTO leads(id) VALUES (1),(2),(3);
CREATE TABLE closes (
id INT(11) NOT NULL AUTO_INCREMENT,
lead_id INT(11) NOT NULL,
close_date DATETIME DEFAULT NULL,
close_type VARCHAR(255) DEFAULT NULL,
primary_agent VARCHAR(255) DEFAULT NULL,
price FLOAT DEFAULT NULL,
gross_commission FLOAT DEFAULT NULL,
company_dollar FLOAT DEFAULT NULL,
address VARCHAR(255) DEFAULT NULL,
city VARCHAR(255) DEFAULT NULL,
state VARCHAR(10) DEFAULT NULL,
zip VARCHAR(10) DEFAULT NULL,
PRIMARY KEY (id),
KEY lead_id (lead_id),
CONSTRAINT closes_ibfk_1 FOREIGN KEY (lead_id) REFERENCES leads (id)
) ENGINE=INNODB AUTO_INCREMENT=4 DEFAULT CHARSET=latin1;
INSERT INTO closes(id,lead_id,close_date,close_type,primary_agent,price,gross_commission,company_dollar,address,city,state,zip)
VALUES
(1,3,'0000-00-0000',NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL),
(2,1,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL),
(3,2,'2018-01-09 17:01:44',NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL);
SELECT
*
FROM
leads AS l
JOIN closes c ON l.id = c.lead_id
WHERE
c.close_date IS NULL OR c.close_date = '0000-00-00';

MariaDB duplicates being inserted

I have the following Python code to check if a MariaDB record exists already, and then inserting. However, I am having duplicates being inserted. Is there something wrong with the code, or is there a better way to do it? I'm new to using Python-MariaDB.
import mysql.connector as mariadb
from hashlib import sha1
mariadb_connection = mariadb.connect(user='root', password='', database='tweets_db')
# The values below are retrieved from Twitter API using Tweepy
# For simplicity, I've provided some sample values
id = '1a23bas'
tweet = 'Clear skies'
longitude = -84.361549
latitude = 34.022003
created_at = '2017-09-27'
collected_at = '2017-09-27'
collection_type = 'stream'
lang = 'us-en'
place_name = 'Roswell'
country_code = 'USA'
cronjob_tag = 'None'
user_id = '23abask'
user_name = 'tsoukalos'
user_geoenabled = 0
user_lang = 'us-en'
user_location = 'Roswell'
user_timezone = 'American/Eastern'
user_verified = 1
tweet_hash = sha1(tweet).hexdigest()
cursor = mariadb_connection.cursor(buffered=True)
cursor.execute("SELECT Count(id) FROM tweets WHERE tweet_hash = %s", (tweet_hash,))
if cursor.fetchone()[0] == 0:
cursor.execute("INSERT INTO tweets(id,tweet,tweet_hash,longitude,latitude,created_at,collected_at,collection_type,lang,place_name,country_code,cronjob_tag,user_id,user_name,user_geoenabled,user_lang,user_location,user_timezone,user_verified) VALUES(%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s)", (id,tweet,tweet_hash,longitude,latitude,created_at,collected_at,collection_type,lang,place_name,country_code,cronjob_tag,user_id,user_name,user_geoenabled,user_lang,user_location,user_timezone,user_verified))
mariadb_connection.commit()
cursor.close()
else:
cursor.close()
return
Below is the code for the table.
CREATE TABLE tweets (
id VARCHAR(255) NOT NULL,
tweet VARCHAR(255) NOT NULL,
tweet_hash VARCHAR(255) DEFAULT NULL,
longitude FLOAT DEFAULT NULL,
latitude FLOAT DEFAULT NULL,
created_at DATETIME DEFAULT NULL,
collected_at DATETIME DEFAULT NULL,
collection_type enum('stream','search') DEFAULT NULL,
lang VARCHAR(10) DEFAULT NULL,
place_name VARCHAR(255) DEFAULT NULL,
country_code VARCHAR(5) DEFAULT NULL,
cronjob_tag VARCHAR(255) DEFAULT NULL,
user_id VARCHAR(255) DEFAULT NULL,
user_name VARCHAR(20) DEFAULT NULL,
user_geoenabled TINYINT(1) DEFAULT NULL,
user_lang VARCHAR(10) DEFAULT NULL,
user_location VARCHAR(255) DEFAULT NULL,
user_timezone VARCHAR(100) DEFAULT NULL,
user_verified TINYINT(1) DEFAULT NULL
);

Add unique constant to tweet_has filed.
alter table tweets modify tweet_hash varchar(255) UNIQUE ;

Every table should have a PRIMARY KEY. Is id supposed to be that? (The CREATE TABLE is not saying so.) A PK is, by definition, UNIQUE, so that would cause an error on inserting a duplicate.
Meanwhile:
Why have a tweet_hash? Simply index tweet.
Don't say 255 when there are specific limits smaller than that.
user_id and user_name should be in another "lookup" table, not both in this table.
Does user_verified belong with the user? Or with each tweet?
If you are expecting millions of tweets, this table needs to be made smaller and indexed -- else you will run into performance problems.