Relations and indexes - Pawabase

Relations

from pawabase_iac import BelongsTo, HasMany, String, Table

posts = Table(
    "posts",
    fields=[String("title", required=True), String("author_id", indexed=True)],
    relations=[
        BelongsTo("author", profiles, field="author_id"),
        HasMany("comments", comments, field="post_id"),
    ],
)

Class	The link is stored	`field` is
`BelongsTo(name, resource, field)`	On this table	The column here that holds the other’s id
`HasMany(name, resource, field)`	On the other table	The column there that holds this record’s id

Pass the other Table object (or its name). Your app then uses expand("author"); see Relations.

Indexes

from pawabase_iac import Desc, Index, Where

indexes = [
    Index("customer_id", Desc("created_at")),                                     # newest first per customer
    Index("customer_id", unique=True, where=[Where("status", "=", "cart")]),      # one open cart per customer
    Index("slug", where=[Where("deleted_at", "is_null")]),                        # only live rows
]

Argument	Meaning
`*columns`	Column names, in order. Wrap a column in `Desc(...)` for descending
`name`	Defaults to one derived from the columns. Lowercase letters, digits and underscores
`unique`	Enforce uniqueness across the combination
`where`	A partial index: up to four `Where(column, op, value)` conditions
`method`	`btree` (default); `hash` and `brin` on PostgreSQL
`concurrent`	Build without locking writes (PostgreSQL)
`note`	A comment for your team

Where operators: =, !=, <, <=, >, >=, is_null, not_null. A table can have up to 16 declared indexes of up to 8 columns. Single-column indexes are simpler as a field flag: String("email", unique=True) or String("author_id", indexed=True).

The server normalises these

The server returns index columns as [{"name": ...}] and omits method when it’s btree. plan allows for that, so you won’t see false differences. See Indexes for how to choose them.

Tables

The whole table.

Policies

Access.