Data & ORM

Data in Semitexa PHP: a type-safe query builder, relations without N+1, filtering, pagination, repositories and schema sync instead of migrations.

• Domain-Level Models

  Semitexa separates persistence resources from business models. Resources map tables; domain models carry behavior and invariants.

• Repository Workflow

  The canonical Semitexa path: handlers depend on repository contracts, repositories return domain models, and persistence resources stay behind the boundary.

• Schema Sync, Not Migration Churn

  Semitexa creates SQL only when the real schema changed, blocks destructive drops by default, and logs the exact DDL plan as SQL and JSON.

• Query Builder

  Compose type-safe queries with a fluent API. No raw SQL, no magic strings.

• Filtering

  Mark a property #[Filterable] and the ORM handles the rest. No manual WHERE clauses.

• Pagination

  Offset and cursor pagination out of the box. Switch modes with a single query parameter.

• Relations

  Declare parent and child links on the resource itself, then read typed relations from the handler.

• Shared Table Extension

  Two modules can extend one table independently, and the ORM merges the schema without forcing either side to edit the other.

• N+1 Without Magic

  Semitexa avoids N+1 with resource slices that load the exact columns and relations each screen needs, instead of hiding queries behind implicit loading.

• ORM Working Guide

  Declaring a resource, querying it, relations and write policies, optimistic locking, transactions, wiring a store, and the gotchas.

• Cache

  Keep expensive results with CacheManagerInterface — per-worker memory or shared Redis, TTLs, single-flight remember(), namespaces, tags, and tenant isolation.

• File Storage and Uploads

  Keep files with StorageObjectStoreInterface on local disk or an S3-compatible store, receive multipart uploads as UploadedFile, and size the server for them.

• Search

  Declare a search index over a table, run free-text and filtered searches through SearchManagerInterface, and know what SQL LIKE matching, page-local ranking, tenant scoping and the LLM planner really do.