python - Pandas 0.24 read_sql operational errors

Question

I just upgraded to Pandas 0.24.0 from 0.23.4 (Python 2.7.12), and many of my pd.read_sql queries are breaking. It looks like something related to MySQL, but it's strange that these errors only occur after updating my pandas version. Any ideas what's going on?

Here's my MySQL table:

CREATE TABLE `xlations_topic_update_status` (
  `run_ts` datetime DEFAULT NULL ON UPDATE CURRENT_TIMESTAMP
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;

Here's my query:

import pandas as pd
from sqlalchemy import create_engine
db_engine = create_engine('mysql+mysqldb://<><>/product_analytics', echo=False)
pd.read_sql('select max(run_ts) from product_analytics.xlations_topic_update_status', con = db_engine).values[0][0]

And here's the error:

OperationalError: (_mysql_exceptions.OperationalError) (1059, "Identifier name 'select max(run_ts) from product_analytics.xlations_topic_update_status;' is too long") [SQL: 'DESCRIBE `select max(run_ts) from product_analytics.xlations_topic_update_status;`']

I've also gotten this for other more complex queries, but won't post them here.

Answer 1

According to documentation the first argument is either a string (a table name) or SQLAlchemy Selectable (select or text object). In other words pd.read_sql() is delegating to pd.read_sql_table() and treating the entire query string as a table identifier.

Wrap your query string in a text() construct first:

stmt = text('select max(run_ts) from product_analytics.xlations_topic_update_status')
pd.read_sql(stmt, con = db_engine).values[0][0]

This way pd.read_sql() will delegate to pd.read_sql_query() instead. Another option is to call it directly.