-- 030_ani_schema.sql
-- ANI declarații de avere și interese — flagship transparency feature.
--
-- Source: declaratii.integritate.eu (e-DAI 2022→), old-declaratii.integritate.eu
-- (archive 2008-2022). Public by Law 176/2010, GDPR-safe (no CNP stored).
--
-- ~1.3M PDF declarations of Romanian public officials. Cross-references
-- politicians × firms-they-own × procurement-contracts (firms.entities, seap.*).
--
-- See ANI-PLAN.md for full architecture, volume estimates, and rollout plan.
-- This file = Stage 0 (schema only, no data).

CREATE SCHEMA IF NOT EXISTS ani;
GRANT USAGE ON SCHEMA ani TO PUBLIC;


-- ── ani.officials ──────────────────────────────────────────────────────────
-- Distinct demnitar/funcționar public. Filled by Stage 4 (entity resolution),
-- not by the listing scraper. ani.declaratii.official_id is nullable until
-- dedup runs.
CREATE TABLE IF NOT EXISTS ani.officials (
  id                bigserial PRIMARY KEY,
  normalized_name   text NOT NULL,            -- lower(unaccent(name)) collapsed
  display_name      text NOT NULL,            -- "Popescu Ioan-Vasile"
  cnp_hash          char(64),                 -- SHA-256 if extractable (rare)
  first_seen_year   smallint,                 -- min(declaration year)
  last_seen_year    smallint,                 -- max(declaration year)
  slug              text UNIQUE,              -- "popescu-ioan-vasile" + suffix
  primary_function  text,                     -- most-frequent function
  primary_judet     text,                     -- most-frequent judet
  declaration_count integer DEFAULT 0,        -- materialized count for UI
  created_at        timestamptz DEFAULT now()
);

CREATE INDEX IF NOT EXISTS idx_officials_norm_name
  ON ani.officials (normalized_name);
CREATE INDEX IF NOT EXISTS idx_officials_norm_name_trgm
  ON ani.officials USING gin (normalized_name gin_trgm_ops);


-- ── ani.declaratii ─────────────────────────────────────────────────────────
-- One row per PDF declaration. Listing scraper fills the metadata; PDF
-- downloader fills pdf_path + pdf_sha256; parser fills parse_status.
CREATE TABLE IF NOT EXISTS ani.declaratii (
  id                  bigserial PRIMARY KEY,
  official_id         bigint REFERENCES ani.officials(id) ON DELETE SET NULL,
  -- raw fields straight from portal listing (pre-resolution)
  raw_official_name   text NOT NULL,
  raw_institution     text,
  raw_function        text,
  raw_localitate      text,
  raw_judet           text,
  -- declaration details
  year                smallint NOT NULL,
  declaration_type    text NOT NULL CHECK (declaration_type IN
                        ('avere','interese','avere+interese')),
  submission_kind     text CHECK (submission_kind IN
                        ('anuala','numire-functie','incetare-functie',
                         'rectificativa','periodica','altele') OR
                       submission_kind IS NULL),
  data_completare     date,
  -- source tracking (which portal, which ID)
  source_portal       text NOT NULL CHECK (source_portal IN
                        ('old','new','depozitar')),
  source_url          text NOT NULL,
  source_id           text,                   -- uniqueIdentifier (old) / _id (new)
  -- PDF storage
  pdf_path            text,                   -- relative to /opt/vreaudigital-data/ani
  pdf_sha256          char(64),
  pdf_size_bytes      integer,
  fetched_at          timestamptz,
  -- parser state
  parsed_at           timestamptz,
  parse_status        text DEFAULT 'pending' CHECK (parse_status IN
                        ('pending','ok','ocr_required','parse_failed',
                         'template_unknown','download_failed')),
  parse_template      text,                   -- '2008-2010' | '2011-2016' | '2017+' | 'edai'
  parse_error         text,
  inserted_at         timestamptz DEFAULT now()
);

-- one declaration per (portal, source_id) — primary dedup key
CREATE UNIQUE INDEX IF NOT EXISTS idx_declaratii_source
  ON ani.declaratii (source_portal, source_id) WHERE source_id IS NOT NULL;
-- content-hash dedup — same PDF re-uploaded under different IDs
CREATE UNIQUE INDEX IF NOT EXISTS idx_declaratii_sha
  ON ani.declaratii (pdf_sha256) WHERE pdf_sha256 IS NOT NULL;
CREATE INDEX IF NOT EXISTS idx_declaratii_official
  ON ani.declaratii (official_id, year DESC) WHERE official_id IS NOT NULL;
CREATE INDEX IF NOT EXISTS idx_declaratii_year
  ON ani.declaratii (year DESC, declaration_type);
CREATE INDEX IF NOT EXISTS idx_declaratii_pending
  ON ani.declaratii (parse_status) WHERE parse_status IN ('pending','ocr_required');
CREATE INDEX IF NOT EXISTS idx_declaratii_raw_name_trgm
  ON ani.declaratii USING gin (raw_official_name gin_trgm_ops);
CREATE INDEX IF NOT EXISTS idx_declaratii_raw_inst_trgm
  ON ani.declaratii USING gin (raw_institution gin_trgm_ops);


-- ── ani.bunuri ─────────────────────────────────────────────────────────────
-- Sections I (imobile) + II (mobile).  raw_row_text always preserved for
-- audit / debug.
CREATE TABLE IF NOT EXISTS ani.bunuri (
  id              bigserial PRIMARY KEY,
  declaration_id  bigint NOT NULL REFERENCES ani.declaratii(id) ON DELETE CASCADE,
  category        text NOT NULL CHECK (category IN
                    ('imobil-teren','imobil-cladire','mobil-vehicul',
                     'mobil-bijuterii','mobil-tablouri-arta','mobil-altele')),
  subcategory     text,                       -- "agricol"/"intravilan"/"apartament"
  localitate      text,
  judet           text,
  tara            text DEFAULT 'România',
  year_acquired   smallint,
  mode_acquired   text,                       -- "cumparare"/"mostenire"/"donatie"
  area_sqm        numeric,
  share_pct       numeric,                    -- 1.0 = full ownership
  co_owner        text,
  value_lei       numeric,
  value_currency  text DEFAULT 'RON',
  raw_row_text    text                        -- audit
);

CREATE INDEX IF NOT EXISTS idx_bunuri_decl ON ani.bunuri (declaration_id);
CREATE INDEX IF NOT EXISTS idx_bunuri_judet ON ani.bunuri (judet) WHERE judet IS NOT NULL;
CREATE INDEX IF NOT EXISTS idx_bunuri_category ON ani.bunuri (category);


-- ── ani.shareholdings ──────────────────────────────────────────────────────
-- Section IX (firme deținute / asociate).  THE flagship table — joins to
-- firms.entities via firm_cui (resolved in Stage 4) and to seap.announcements
-- via that CUI for "politician-with-firm-supplier-to-state" recipes.
CREATE TABLE IF NOT EXISTS ani.shareholdings (
  id                bigserial PRIMARY KEY,
  declaration_id    bigint NOT NULL REFERENCES ani.declaratii(id) ON DELETE CASCADE,
  firm_name_raw     text NOT NULL,            -- text from PDF
  firm_cui          text,                     -- resolved later
  firm_match_score  real,                     -- pg_trgm similarity
  firm_match_method text CHECK (firm_match_method IN
                      ('exact_name','trgm','manual','unmatched') OR
                     firm_match_method IS NULL),
  matched_at        timestamptz,
  role              text,                     -- "actionar"/"asociat"/"administrator"/"membru CA"
  share_pct         numeric,
  value_lei         numeric,
  category          text CHECK (category IN
                      ('societate','asociatie','fundatie','cooperativa',
                       'oNG','altele') OR category IS NULL),
  raw_row_text      text
);

CREATE INDEX IF NOT EXISTS idx_share_decl ON ani.shareholdings (declaration_id);
CREATE INDEX IF NOT EXISTS idx_share_cui
  ON ani.shareholdings (firm_cui) WHERE firm_cui IS NOT NULL;
CREATE INDEX IF NOT EXISTS idx_share_name_trgm
  ON ani.shareholdings USING gin (firm_name_raw gin_trgm_ops);
CREATE INDEX IF NOT EXISTS idx_share_unmatched
  ON ani.shareholdings (firm_match_method)
  WHERE firm_match_method IS NULL OR firm_match_method = 'unmatched';


-- ── ani.functii ────────────────────────────────────────────────────────────
-- Section VIII — funcții publice și private.  Joinable to seap.cui_authority
-- (when is_public + institution_cui matches an authority) and firms.entities
-- (when is_public = false).
CREATE TABLE IF NOT EXISTS ani.functii (
  id                bigserial PRIMARY KEY,
  declaration_id    bigint NOT NULL REFERENCES ani.declaratii(id) ON DELETE CASCADE,
  is_public         boolean,
  function_name     text NOT NULL,
  institution_name  text NOT NULL,
  institution_cui   text,                     -- resolved later
  start_year        smallint,
  end_year          smallint,                 -- NULL if active
  salary_lei        numeric,                  -- annual income from this function
  raw_row_text      text
);

CREATE INDEX IF NOT EXISTS idx_functii_decl ON ani.functii (declaration_id);
CREATE INDEX IF NOT EXISTS idx_functii_inst_cui
  ON ani.functii (institution_cui) WHERE institution_cui IS NOT NULL;
CREATE INDEX IF NOT EXISTS idx_functii_inst_name_trgm
  ON ani.functii USING gin (institution_name gin_trgm_ops);


-- ── ani.donatii ────────────────────────────────────────────────────────────
-- Section V (donații primite).
CREATE TABLE IF NOT EXISTS ani.donatii (
  id              bigserial PRIMARY KEY,
  declaration_id  bigint NOT NULL REFERENCES ani.declaratii(id) ON DELETE CASCADE,
  donor_name      text,
  donation_type   text CHECK (donation_type IN
                    ('bani','imobil','mobil','servicii','altele') OR
                   donation_type IS NULL),
  value_lei       numeric,
  currency        text DEFAULT 'RON',
  year_received   smallint,
  raw_row_text    text
);

CREATE INDEX IF NOT EXISTS idx_donatii_decl ON ani.donatii (declaration_id);


-- ── Comments ──────────────────────────────────────────────────────────────
COMMENT ON SCHEMA ani IS
  'ANI declarații de avere și interese. Sources: declaratii.integritate.eu + old-declaratii.integritate.eu. Public by Law 176/2010.';
COMMENT ON TABLE ani.declaratii IS
  'One row per PDF declaration. official_id resolved in Stage 4 dedup.';
COMMENT ON TABLE ani.shareholdings IS
  'Section IX firme deținute. THE flagship cross-reference: firm_cui joins to firms.entities → seap.announcements.';
COMMENT ON COLUMN ani.declaratii.pdf_path IS
  'Relative path under /opt/vreaudigital-data/ani/. Full path: /opt/vreaudigital-data/ani/$pdf_path';