"""DMS search bounded context — models. Entities -------- DocumentSearchIndex Denormalized, always-current search snapshot of a Document. Updated by ``index_document()`` whenever a Document or its current DocumentVersion changes. Designed to be mirrored to OpenSearch / Elasticsearch in production without changing the API contract. Key design decisions: * One row per live Document (OneToOneField). * ``search_text`` is the concatenated plain-text FTS blob built from title, code, document-type name, version label, and file name. * ``tags`` is a JSONField list so tag filtering works on both SQLite (dev/test) and PostgreSQL (prod). A PostgreSQL ArrayField can replace it later without changing the service layer. * ``extra_json`` is reserved for Phase-4 structured metadata indexing. * ``index_version`` is a monotonic integer for detecting stale replicas when integrating an external search backend. SavedSearch A user-owned, optionally team-shared persisted search query. ``query_params`` stores the serialized SearchParams dict validated by ``SearchParamsSerializer``. """ from __future__ import annotations from typing import ClassVar from django.conf import settings from django.db import models from django.utils.translation import gettext_lazy as _ from simorgh.core.db.scoped import ScopedSoftDeleteManager from simorgh.core.models import AuditedModel, SoftDeleteModel, TenantScopedModel, UUIDModel # --------------------------------------------------------------------------- # DocumentSearchIndex # --------------------------------------------------------------------------- class DocumentSearchIndex(UUIDModel, TenantScopedModel, AuditedModel): """Denormalized search read-model — one row per live Document. No SoftDeleteModel: index entries are replaced or hard-deleted by services; they are never soft-deleted independently of their parent Document. """ document = models.OneToOneField( "dms.Document", on_delete=models.CASCADE, related_name="search_index", verbose_name=_("document"), ) # --- Denormalized document fields ---------------------------------------- title = models.TextField(_("title"), default="", blank=True) code = models.CharField(_("code"), max_length=100, blank=True, default="") document_type_code = models.CharField( _("document type code"), max_length=80, blank=True, default="" ) document_type_name = models.CharField( _("document type name"), max_length=200, blank=True, default="" ) repository_public_id = models.UUIDField( _("repository public id"), null=True, blank=True, db_index=True ) folder_public_id = models.UUIDField( _("folder public id"), null=True, blank=True, db_index=True ) status = models.CharField( _("status"), max_length=20, blank=True, default="", db_index=True ) workflow_status = models.CharField( _("workflow status"), max_length=100, blank=True, default="" ) # --- File info mirrored from current DocumentVersion --------------------- file_name = models.CharField(_("file name"), max_length=500, blank=True, default="") content_type = models.CharField( _("content type"), max_length=200, blank=True, default="", db_index=True ) file_size = models.BigIntegerField(_("file size bytes"), null=True, blank=True) version_label = models.CharField( _("version label"), max_length=50, blank=True, default="" ) # --- Full-text blob ------------------------------------------------------- search_text = models.TextField( _("search text"), blank=True, default="", help_text=_( "Concatenated searchable plain text: title + code + document-type " "name + version label + file name. Populated by index_document()." ), ) # --- Structured extras --------------------------------------------------- tags = models.JSONField( _("tags"), default=list, help_text=_( "Lowercase string labels, e.g. [\"contract\", \"urgent\"]. " "Populated from document.extra[\"tags\"] or AI classification." ), ) extra_json = models.JSONField( _("extra JSON"), default=dict, blank=True, help_text=_( "Reserved for Phase-4 structured metadata indexing. " "Arbitrary key/value pairs added by the metadata engine." ), ) # --- Index bookkeeping --------------------------------------------------- indexed_at = models.DateTimeField(_("indexed at"), auto_now=True) index_version = models.PositiveIntegerField(_("index version"), default=1) class Meta: verbose_name = _("document search index") verbose_name_plural = _("document search indices") indexes = [ models.Index( fields=["tenant", "status"], name="dms_srch_tenant_status_idx", ), models.Index( fields=["tenant", "content_type"], name="dms_srch_tenant_ctype_idx", ), models.Index( fields=["tenant", "document_type_code"], name="dms_srch_tenant_dtype_idx", ), models.Index( fields=["tenant", "indexed_at"], name="dms_srch_tenant_idxat_idx", ), ] def __str__(self) -> str: # pragma: no cover return f"SearchIndex({self.title!r})" # --------------------------------------------------------------------------- # SavedSearch # --------------------------------------------------------------------------- class SavedSearch(UUIDModel, TenantScopedModel, AuditedModel, SoftDeleteModel): """A user-owned, optionally team-shared persisted search query. ``query_params`` mirrors the fields of ``SearchParams`` and is validated at write time by ``SearchParamsSerializer``. When ``is_shared=True`` the entry is readable by all active tenant users but only the owner (or a superuser) may modify or delete it. """ owner = models.ForeignKey( settings.AUTH_USER_MODEL, on_delete=models.CASCADE, related_name="dms_saved_searches", verbose_name=_("owner"), ) name = models.CharField(_("name"), max_length=200) description = models.TextField(_("description"), blank=True, default="") query_params = models.JSONField( _("query params"), default=dict, help_text=_( "Serialized SearchParams dict. Validated by SearchParamsSerializer " "at create/update time." ), ) is_shared = models.BooleanField( _("shared"), default=False, help_text=_( "When True this saved search is visible (read-only) to all " "active users in the same tenant." ), ) last_used_at = models.DateTimeField(_("last used at"), null=True, blank=True) use_count = models.PositiveIntegerField(_("use count"), default=0) objects: ClassVar[ScopedSoftDeleteManager] = ScopedSoftDeleteManager() class Meta: verbose_name = _("saved search") verbose_name_plural = _("saved searches") ordering = ("-created_at",) indexes = [ models.Index( fields=["tenant", "owner", "is_deleted"], name="dms_savedsearch_owner_idx", ), models.Index( fields=["tenant", "is_shared", "is_deleted"], name="dms_savedsearch_shared_idx", ), ] def __str__(self) -> str: # pragma: no cover return f"SavedSearch({self.name!r})"