CODEWITHSUNDEEP
pythonsqlalchemyflask-sqlalchemycase-insensitive
Ganye
Ganye

Reputation: 2671

Case Insensitive Flask-SQLAlchemy Query

I'm using Flask-SQLAlchemy to query from a database of users; however, while

user = models.User.query.filter_by(username="ganye").first()

will return

<User u'ganye'>

doing

user = models.User.query.filter_by(username="GANYE").first()

returns

None

I'm wondering if there's a way to query the database in a case insensitive way, so that the second example will still return

<User u'ganye'>

Upvotes: 160

Views: 103253

Answers (8)

snakecharmerb
snakecharmerb

Reputation: 55600

If it fits your use case, you may consider setting a custom collation on the column, such that the column automatically handles comparisons in a case-insensitive manner.

It's worth noting:

  • the collation will apply to all queries on the column
  • the collation will apply to ORDER BY clauses as well
  • a collation may be specified directly in queries, rather than being defined on the column
    • this may incur performance penalties
  • collation definitions tend to be RDBMS / locale / language-specific - consult the relevant documentation
  • collation names may not be portable between different RDBMS
  • available collation attributes may vary by RDBMS

in other words, consult your RDBMS's docs before using this feature.

This example script shows how you might use collations for MySQL, Postgresql (note the special-casing) and Sqlite; each RDBMS returns all three possible results for the query value.

import sqlalchemy as sa
from sqlalchemy import orm

data = {
    'mysql': ('mysql:///test', 'utf8mb4_general_ci'),
    'postgresql': ('postgresql:///test', 'coll'),
    'sqlite': ('sqlite://', 'NOCASE'),
}


for dialect, (uri, collation) in data.items():
    Base = orm.declarative_base()

    class Test(Base):
        __tablename__ = 't16573095'

        id = sa.Column(sa.Integer, primary_key=True)
        name = sa.Column(sa.String(32, collation=collation))

    engine = sa.create_engine(uri, echo=False, future=True)
    Base.metadata.drop_all(engine)

    if dialect == 'postgresql':
        # Postgres collations are more complicated
        # - read the docs!
        with engine.begin() as conn:
            conn.execute(sa.text('DROP COLLATION IF EXISTS coll'))
            stmt = """CREATE COLLATION coll (provider='icu', locale='und-u-ks-level2', deterministic=false)"""
            conn.execute(sa.text(stmt))

    Base.metadata.create_all(engine)
    Session = orm.sessionmaker(engine, future=True)

    with Session.begin() as s:
        instances = [Test(name=name) for name in ['GANYE', 'ganye', 'gAnYe']]
        s.add_all(instances)

    with Session() as s:
        results = s.execute(sa.select(Test.name).where(Test.name == 'GaNyE')).scalars()
        print(f'{dialect:-<12}')
        for name in results:
            print(name)
        print('-' * 12)

To specify a collation in a query, use an attribute's collate method:

with Session() as s:
    query = sa.select(Test).where(Test.name.collate('coll') == 'GaNyE')
    results = s.execute(query)

Upvotes: 3

Princejr
Princejr

Reputation: 193

Best way is to use case insensitive contains function ie: .icontains(). Example:

.query.filter(Model.column.icontains("search_query"))

cause you are not sure if the user is going to provide the exact text you want to compare with, so using icontains will return a result(s) even if the user query does not contain the EXACT word or phrase he/she is searching for.

Upvotes: -1

Trimatix
Trimatix

Reputation: 11

For declarative users, the equivalent without using query is the ColumnOperators.ilike method:

class User(DeclarativeBase):
    username: str

req = select(User).where(User.username.ilike("GANYE"))
result = session.execute(req)

will render as lower(User.username) LIKE lower("GANYE")

This is still the case in SQLAlchemy 2.0, and is compatible with AsyncSession, for those using it.

Upvotes: 0

inauris
inauris

Reputation: 1

  • By one field 
    uname = "ganye"
user = models.User.query.filter(User.username.ilikef(f"%{uname}%").first()
  • By several fields 
    return db.session.query(Post).filter(
    Post.title.ilike(f"%{search_query}%")
    | Post.body.ilike(f"%{search_query}%")
)

Upvotes: -1

Никита Филонов
Никита Филонов

Reputation: 139

I think the best solution would be

func.lower(User.username).contains(username.lower())

It will produce sql like

WHERE (lower(user.username) LIKE '%%' || %s || '%%')

This means your db username will be converted to the lower case and your python value username also will be converted to lower case. So now you do not have to worry about the case sesitivity

If your username is Sara Smith It will work for ara, sar, Sara or event a S

Upvotes: 4

Mohammad Aarif
Mohammad Aarif

Reputation: 1705

you can do 

user = db.session.query(User).filter_by(func.lower(User.username)==func.lower("GANYE")).first()

Or you can use ilike function

 user = db.session.query(User).filter_by(User.username.ilike("%ganye%")).first()

Upvotes: 12

iChux
iChux

Reputation: 2376

Improving on @plaes's answer, this one will make the query shorter if you specify just the column(s) you need:

user = models.User.query.with_entities(models.User.username).\
filter(models.User.username.ilike("%ganye%")).all()

The above example is very useful in case one needs to use Flask's jsonify for AJAX purposes and then in your javascript access it using data.result:

from flask import jsonify
jsonify(result=user)

Upvotes: 21

plaes
plaes

Reputation: 32716

You can do it by using either the lower or upper functions in your filter:

from sqlalchemy import func
user = models.User.query.filter(func.lower(User.username) == func.lower("GaNyE")).first()

Another option is to do searching using ilike instead of like:

.query.filter(Model.column.ilike("ganye"))

Upvotes: 299

Related Questions