SQLAlchemy Migration Guides
Resolve Engine.execute attribute errors without guessing transaction behavior.
Start here
Use engine.connect() or engine.begin(), then execute on the returned Connection. Do not blindly replace engine.execute with session.execute.
Stop before automation when: Choosing connect() versus begin(). Implicit autocommit replacements.
with engine.connect() as conn:
result = conn.execute(stmt)
with engine.begin() as conn:
conn.execute(user_table.insert(), {"name": "Ada"})
Repo fit check
Run the matching scanner locally first. If the report shows 10+ supported findings for this kind of cleanup, compare the paid workflow or use the $9.90 SQLAlchemy/Pydantic fit report add-on before buying the full pack.
Example scan summary
supported_findings: 38
manual_review_findings: 6
files_uploaded: 0
confidence: reviewableWhy this page exists
Engine.execute was removed, so the replacement depends on whether the code is reading, writing, or wrapping another library.
Direct answer
Use engine.connect() or engine.begin(), then execute on the returned Connection. Do not blindly replace engine.execute with session.execute.
Example
result = engine.execute(stmt)
engine.execute(user_table.insert(), {"name": "Ada"})
with engine.connect() as conn:
result = conn.execute(stmt)
with engine.begin() as conn:
conn.execute(user_table.insert(), {"name": "Ada"})
Decision path
Start with the free scan, inspect the report, and only move to the cleanup pack if the repo has repeated supported patterns that are still expensive to fix by hand.
Before checkout
Temporary sale price: $30 during Migration Sprint Sale; normally $299.99 per team. Stripe Checkout applies the 90% off discount automatically through May 27, 2026 at 11:59 PM Pacific.
Use the exact-problem guides as a triage layer, then decide whether the repo is inside the supported subset.
FAQ
Use engine.connect() or engine.begin(), then execute on the returned Connection. Do not blindly replace engine.execute with session.execute.
Automate only the supported subset: Detection of removed engine.execute calls. Separation of read and write examples. Fail-closed reporting when transaction semantics are unclear. Keep these cases in manual review: Choosing connect() versus begin(). Implicit autocommit replacements. Helper APIs that hide the engine object.
Use SQLAlchemy 1.4 to 2.0 Migration Cleanup Pack only when this pattern repeats across enough files that manual cleanup is still costly. Run the public scan or read the proof first. If the report shows ten or more supported findings, buy the pack or use the $9.90 SQLAlchemy/Pydantic fit report add-on; if unsupported findings dominate, keep the work manual.
Purchase fit
One matching page is not enough by itself. Run the local scan, then use the matching fit report add-on where listed or buy only when the same supported pattern appears often enough to matter.
Replace session.query(...).get(...), Session.query migration guide, Fix engine.execute(...) removal, Clean up joinedload_all(...), Fix select([..]) list syntax, Replace string join paths, Replace string loader options, Move sqlalchemy.ext.declarative imports to sqlalchemy.orm
View all SQLAlchemy 1.4 to 2.0 Migration Cleanup Pack guides