nordabiz/database/migrations/005_zopk_knowledge_base.sql

-- ============================================================
-- NordaBiz - Migration 005: ZOPK Knowledge Base
-- ============================================================
-- Created: 2026-01-11
-- Description:
--   Zielony Okręg Przemysłowy Kaszubia (ZOPK) - Knowledge Base
--   Tables for projects, stakeholders, news, resources, and company links
-- ============================================================

-- ============================================================
-- 1. ZOPK PROJECTS (sub-initiatives)
-- ============================================================

CREATE TABLE IF NOT EXISTS zopk_projects (
    id SERIAL PRIMARY KEY,
    slug VARCHAR(100) UNIQUE NOT NULL,
    name VARCHAR(255) NOT NULL,
    description TEXT,

    -- Project details
    project_type VARCHAR(50),  -- energy, infrastructure, technology, defense
    status VARCHAR(50) DEFAULT 'planned',  -- planned, in_progress, completed
    start_date DATE,
    end_date DATE,

    -- Location
    location VARCHAR(255),
    region VARCHAR(100),  -- Wejherowo, Rumia, Gdynia

    -- Key metrics
    estimated_investment NUMERIC(15, 2),  -- PLN
    estimated_jobs INTEGER,

    -- Visual
    icon VARCHAR(50),  -- CSS icon or emoji
    color VARCHAR(20),  -- HEX color

    -- Display
    sort_order INTEGER DEFAULT 0,
    is_featured BOOLEAN DEFAULT FALSE,
    is_active BOOLEAN DEFAULT TRUE,

    created_at TIMESTAMP DEFAULT NOW(),
    updated_at TIMESTAMP DEFAULT NOW()
);

CREATE INDEX IF NOT EXISTS idx_zopk_projects_slug ON zopk_projects(slug);
CREATE INDEX IF NOT EXISTS idx_zopk_projects_status ON zopk_projects(status);
CREATE INDEX IF NOT EXISTS idx_zopk_projects_type ON zopk_projects(project_type);

COMMENT ON TABLE zopk_projects IS 'Sub-projects within ZOPK initiative (offshore, nuclear, data centers, etc.)';

-- ============================================================
-- 2. ZOPK STAKEHOLDERS (people and organizations)
-- ============================================================

CREATE TABLE IF NOT EXISTS zopk_stakeholders (
    id SERIAL PRIMARY KEY,
    stakeholder_type VARCHAR(20) NOT NULL,  -- person, organization
    name VARCHAR(255) NOT NULL,

    -- Role and affiliation
    role VARCHAR(255),  -- Koordynator, Minister, Starosta
    organization VARCHAR(255),  -- MON, Starostwo Wejherowskie

    -- Contact (public info)
    email VARCHAR(255),
    phone VARCHAR(50),
    website VARCHAR(500),

    -- Social media
    linkedin_url VARCHAR(500),
    twitter_url VARCHAR(500),

    -- Visual
    photo_url VARCHAR(500),

    -- Description
    bio TEXT,

    -- Categorization
    category VARCHAR(50),  -- government, local_authority, business, academic
    importance INTEGER DEFAULT 0,

    is_active BOOLEAN DEFAULT TRUE,

    created_at TIMESTAMP DEFAULT NOW(),
    updated_at TIMESTAMP DEFAULT NOW()
);

CREATE INDEX IF NOT EXISTS idx_zopk_stakeholders_type ON zopk_stakeholders(stakeholder_type);
CREATE INDEX IF NOT EXISTS idx_zopk_stakeholders_category ON zopk_stakeholders(category);

COMMENT ON TABLE zopk_stakeholders IS 'Key people and organizations involved in ZOPK';

-- ============================================================
-- 3. STAKEHOLDER-PROJECT LINKS
-- ============================================================

CREATE TABLE IF NOT EXISTS zopk_stakeholder_projects (
    id SERIAL PRIMARY KEY,
    stakeholder_id INTEGER NOT NULL REFERENCES zopk_stakeholders(id) ON DELETE CASCADE,
    project_id INTEGER NOT NULL REFERENCES zopk_projects(id) ON DELETE CASCADE,
    role_in_project VARCHAR(255),

    created_at TIMESTAMP DEFAULT NOW(),

    CONSTRAINT uq_stakeholder_project UNIQUE (stakeholder_id, project_id)
);

CREATE INDEX IF NOT EXISTS idx_zopk_sp_stakeholder ON zopk_stakeholder_projects(stakeholder_id);
CREATE INDEX IF NOT EXISTS idx_zopk_sp_project ON zopk_stakeholder_projects(project_id);

COMMENT ON TABLE zopk_stakeholder_projects IS 'Link table: stakeholders to projects';

-- ============================================================
-- 4. ZOPK NEWS (with approval workflow)
-- ============================================================

CREATE TABLE IF NOT EXISTS zopk_news (
    id SERIAL PRIMARY KEY,

    -- Source information
    title VARCHAR(500) NOT NULL,
    description TEXT,
    url VARCHAR(1000) NOT NULL,
    source_name VARCHAR(200),  -- trojmiasto.pl, etc.
    source_domain VARCHAR(200),

    -- Article details
    published_at TIMESTAMP,
    author VARCHAR(255),
    image_url VARCHAR(1000),

    -- Categorization
    news_type VARCHAR(50),  -- news, announcement, interview, press_release
    project_id INTEGER REFERENCES zopk_projects(id),

    -- AI Analysis
    relevance_score NUMERIC(3, 2),  -- 0.00-1.00
    sentiment VARCHAR(20),  -- positive, neutral, negative
    ai_summary TEXT,
    keywords TEXT[],  -- Extracted keywords array

    -- Moderation workflow
    status VARCHAR(20) DEFAULT 'pending',  -- pending, approved, rejected
    moderated_by INTEGER REFERENCES users(id),
    moderated_at TIMESTAMP,
    rejection_reason TEXT,

    -- Source tracking
    source_type VARCHAR(50) DEFAULT 'manual',  -- manual, brave_search, rss
    fetch_job_id VARCHAR(100),

    -- Deduplication
    url_hash VARCHAR(64) UNIQUE,

    is_featured BOOLEAN DEFAULT FALSE,
    views_count INTEGER DEFAULT 0,

    created_at TIMESTAMP DEFAULT NOW(),
    updated_at TIMESTAMP DEFAULT NOW()
);

CREATE INDEX IF NOT EXISTS idx_zopk_news_status ON zopk_news(status);
CREATE INDEX IF NOT EXISTS idx_zopk_news_project ON zopk_news(project_id);
CREATE INDEX IF NOT EXISTS idx_zopk_news_published ON zopk_news(published_at DESC);
CREATE INDEX IF NOT EXISTS idx_zopk_news_type ON zopk_news(news_type);
CREATE INDEX IF NOT EXISTS idx_zopk_news_url_hash ON zopk_news(url_hash);

COMMENT ON TABLE zopk_news IS 'News articles about ZOPK with approval workflow';

-- ============================================================
-- 5. ZOPK RESOURCES (documents, links, media)
-- ============================================================

CREATE TABLE IF NOT EXISTS zopk_resources (
    id SERIAL PRIMARY KEY,

    title VARCHAR(255) NOT NULL,
    description TEXT,

    -- Resource type
    resource_type VARCHAR(50) NOT NULL,  -- link, document, image, video, map

    -- URL or file
    url VARCHAR(1000),
    file_path VARCHAR(500),
    file_size INTEGER,
    mime_type VARCHAR(100),

    -- Thumbnail
    thumbnail_url VARCHAR(1000),

    -- Categorization
    category VARCHAR(50),  -- official, media, research, presentation
    project_id INTEGER REFERENCES zopk_projects(id),

    -- Tags
    tags TEXT[],

    -- Source
    source_name VARCHAR(255),
    source_date DATE,

    -- Moderation
    status VARCHAR(20) DEFAULT 'approved',
    uploaded_by INTEGER REFERENCES users(id),

    is_featured BOOLEAN DEFAULT FALSE,
    sort_order INTEGER DEFAULT 0,

    created_at TIMESTAMP DEFAULT NOW(),
    updated_at TIMESTAMP DEFAULT NOW()
);

CREATE INDEX IF NOT EXISTS idx_zopk_resources_type ON zopk_resources(resource_type);
CREATE INDEX IF NOT EXISTS idx_zopk_resources_project ON zopk_resources(project_id);
CREATE INDEX IF NOT EXISTS idx_zopk_resources_category ON zopk_resources(category);
CREATE INDEX IF NOT EXISTS idx_zopk_resources_status ON zopk_resources(status);

COMMENT ON TABLE zopk_resources IS 'Resources: documents, links, images, videos for ZOPK knowledge base';

-- ============================================================
-- 6. ZOPK COMPANY LINKS (Norda members)
-- ============================================================

CREATE TABLE IF NOT EXISTS zopk_company_links (
    id SERIAL PRIMARY KEY,

    company_id INTEGER NOT NULL REFERENCES companies(id) ON DELETE CASCADE,
    project_id INTEGER NOT NULL REFERENCES zopk_projects(id) ON DELETE CASCADE,

    -- Type of involvement
    link_type VARCHAR(50) NOT NULL,  -- potential_supplier, partner, investor, beneficiary

    -- Description
    collaboration_description TEXT,

    -- Scoring
    relevance_score INTEGER,  -- 1-100

    -- Status
    status VARCHAR(20) DEFAULT 'suggested',  -- suggested, confirmed, active, completed

    -- Admin notes
    admin_notes TEXT,
    created_by INTEGER REFERENCES users(id),

    created_at TIMESTAMP DEFAULT NOW(),
    updated_at TIMESTAMP DEFAULT NOW(),

    CONSTRAINT uq_company_project_link UNIQUE (company_id, project_id, link_type)
);

CREATE INDEX IF NOT EXISTS idx_zopk_cl_company ON zopk_company_links(company_id);
CREATE INDEX IF NOT EXISTS idx_zopk_cl_project ON zopk_company_links(project_id);
CREATE INDEX IF NOT EXISTS idx_zopk_cl_status ON zopk_company_links(status);

COMMENT ON TABLE zopk_company_links IS 'Links between ZOPK projects and Norda Biznes member companies';

-- ============================================================
-- 7. ZOPK NEWS FETCH JOBS (automation tracking)
-- ============================================================

CREATE TABLE IF NOT EXISTS zopk_news_fetch_jobs (
    id SERIAL PRIMARY KEY,
    job_id VARCHAR(100) UNIQUE NOT NULL,

    -- Config
    search_query VARCHAR(500),
    search_api VARCHAR(50),  -- brave, google, bing
    date_range_start DATE,
    date_range_end DATE,

    -- Results
    results_found INTEGER DEFAULT 0,
    results_new INTEGER DEFAULT 0,
    results_approved INTEGER DEFAULT 0,

    -- Status
    status VARCHAR(20) DEFAULT 'pending',  -- pending, running, completed, failed
    error_message TEXT,

    -- Timing
    started_at TIMESTAMP,
    completed_at TIMESTAMP,

    -- Trigger
    triggered_by VARCHAR(50),  -- cron, manual, admin
    triggered_by_user INTEGER REFERENCES users(id),

    created_at TIMESTAMP DEFAULT NOW()
);

CREATE INDEX IF NOT EXISTS idx_zopk_fetch_status ON zopk_news_fetch_jobs(status);
CREATE INDEX IF NOT EXISTS idx_zopk_fetch_created ON zopk_news_fetch_jobs(created_at DESC);

COMMENT ON TABLE zopk_news_fetch_jobs IS 'Tracking for automated ZOPK news fetch jobs';

-- ============================================================
-- 8. INITIAL DATA - PROJECTS
-- ============================================================

INSERT INTO zopk_projects (slug, name, description, project_type, status, region, icon, color, sort_order, is_featured)
VALUES
    ('offshore-wind', 'Morska Energetyka Wiatrowa', 'Farmy wiatrowe na Baltyku - kluczowy element transformacji energetycznej regionu', 'energy', 'in_progress', 'Baltyk / Trojmiasto', 'wind', '#0891b2', 1, TRUE),
    ('nuclear-plant', 'Elektrownia Jadrowa', 'Budowa pierwszej polskiej elektrowni jadrowej w lokalizacji Lubiatowo-Kopalino', 'energy', 'planned', 'Choczewo', 'atom', '#7c3aed', 2, TRUE),
    ('data-centers', 'Centra Danych', 'Nowoczesne centra przetwarzania danych wykorzystujace czysta energie', 'technology', 'planned', 'Trojmiasto', 'server', '#2563eb', 3, FALSE),
    ('hydrogen-labs', 'Laboratoria Wodorowe', 'Badania i rozwoj technologii wodorowych dla przemyslu', 'technology', 'planned', 'Gdynia / Rumia', 'flask', '#059669', 4, FALSE),
    ('kongsberg', 'Inwestycja Kongsberg', 'Zaklad produkcyjny norweskiego koncernu zbrojeniowego Kongsberg', 'defense', 'in_progress', 'Rumia', 'shield', '#dc2626', 5, TRUE)
ON CONFLICT (slug) DO NOTHING;

-- ============================================================
-- 9. INITIAL DATA - STAKEHOLDERS
-- ============================================================

INSERT INTO zopk_stakeholders (stakeholder_type, name, role, organization, category, importance, is_active)
VALUES
    ('person', 'Maciej Samsonowicz', 'Koordynator projektu, Doradca MON', 'Ministerstwo Obrony Narodowej', 'government', 100, TRUE),
    ('person', 'Wladyslaw Kosiniak-Kamysz', 'Minister Obrony Narodowej', 'Ministerstwo Obrony Narodowej', 'government', 95, TRUE),
    ('person', 'Marcin Kaczmarek', 'Starosta Wejherowski', 'Starostwo Powiatowe w Wejherowie', 'local_authority', 80, TRUE),
    ('person', 'Michal Pasieczny', 'Burmistrz Rumi', 'Urzad Miasta Rumi', 'local_authority', 75, TRUE),
    ('organization', 'Ministerstwo Obrony Narodowej', 'Ministerstwo odpowiedzialne za obronnosc', NULL, 'government', 90, TRUE),
    ('organization', 'Kongsberg Defence & Aerospace', 'Norweski koncern zbrojeniowy', NULL, 'business', 85, TRUE),
    ('organization', 'Polskie Elektrownie Jadrowe', 'Spolka realizujaca program jadrowy', NULL, 'business', 85, TRUE)
ON CONFLICT DO NOTHING;

-- ============================================================
-- 10. PERMISSIONS
-- ============================================================

GRANT ALL ON TABLE zopk_projects TO nordabiz_app;
GRANT ALL ON TABLE zopk_stakeholders TO nordabiz_app;
GRANT ALL ON TABLE zopk_stakeholder_projects TO nordabiz_app;
GRANT ALL ON TABLE zopk_news TO nordabiz_app;
GRANT ALL ON TABLE zopk_resources TO nordabiz_app;
GRANT ALL ON TABLE zopk_company_links TO nordabiz_app;
GRANT ALL ON TABLE zopk_news_fetch_jobs TO nordabiz_app;

GRANT USAGE, SELECT ON SEQUENCE zopk_projects_id_seq TO nordabiz_app;
GRANT USAGE, SELECT ON SEQUENCE zopk_stakeholders_id_seq TO nordabiz_app;
GRANT USAGE, SELECT ON SEQUENCE zopk_stakeholder_projects_id_seq TO nordabiz_app;
GRANT USAGE, SELECT ON SEQUENCE zopk_news_id_seq TO nordabiz_app;
GRANT USAGE, SELECT ON SEQUENCE zopk_resources_id_seq TO nordabiz_app;
GRANT USAGE, SELECT ON SEQUENCE zopk_company_links_id_seq TO nordabiz_app;
GRANT USAGE, SELECT ON SEQUENCE zopk_news_fetch_jobs_id_seq TO nordabiz_app;

-- ============================================================
-- 11. ALTER TABLE - Multi-source cross-verification columns
-- ============================================================
-- These columns support automatic cross-verification from multiple sources

-- Confidence score (1-5 based on source count)
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM information_schema.columns
                   WHERE table_name = 'zopk_news' AND column_name = 'confidence_score') THEN
        ALTER TABLE zopk_news ADD COLUMN confidence_score INTEGER DEFAULT 1;
    END IF;
END $$;

-- Number of sources that found this story
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM information_schema.columns
                   WHERE table_name = 'zopk_news' AND column_name = 'source_count') THEN
        ALTER TABLE zopk_news ADD COLUMN source_count INTEGER DEFAULT 1;
    END IF;
END $$;

-- List of sources (e.g., ['brave', 'google_news', 'rss_trojmiasto'])
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM information_schema.columns
                   WHERE table_name = 'zopk_news' AND column_name = 'sources_list') THEN
        ALTER TABLE zopk_news ADD COLUMN sources_list TEXT[];
    END IF;
END $$;

-- Title hash for fuzzy deduplication (normalized title)
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM information_schema.columns
                   WHERE table_name = 'zopk_news' AND column_name = 'title_hash') THEN
        ALTER TABLE zopk_news ADD COLUMN title_hash VARCHAR(64);
    END IF;
END $$;

-- Auto-verified flag (True if 3+ sources confirmed)
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM information_schema.columns
                   WHERE table_name = 'zopk_news' AND column_name = 'is_auto_verified') THEN
        ALTER TABLE zopk_news ADD COLUMN is_auto_verified BOOLEAN DEFAULT FALSE;
    END IF;
END $$;

-- Update status to include 'auto_approved' option
-- (no alter needed, just documentation that status can be: pending, approved, rejected, auto_approved)

-- Index for title_hash (fuzzy matching)
CREATE INDEX IF NOT EXISTS idx_zopk_news_title_hash ON zopk_news(title_hash);

-- Index for confidence score (filtering high-confidence news)
CREATE INDEX IF NOT EXISTS idx_zopk_news_confidence ON zopk_news(confidence_score);

-- ============================================================
-- 12. ALTER TABLE - AI Relevance Evaluation columns
-- ============================================================
-- These columns support AI-based relevance evaluation using Google Gemini

-- AI relevance flag (True = relevant to ZOPK, False = not relevant, NULL = not evaluated)
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM information_schema.columns
                   WHERE table_name = 'zopk_news' AND column_name = 'ai_relevant') THEN
        ALTER TABLE zopk_news ADD COLUMN ai_relevant BOOLEAN;
    END IF;
END $$;

-- AI evaluation reason/explanation
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM information_schema.columns
                   WHERE table_name = 'zopk_news' AND column_name = 'ai_evaluation_reason') THEN
        ALTER TABLE zopk_news ADD COLUMN ai_evaluation_reason TEXT;
    END IF;
END $$;

-- When AI evaluation was performed
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM information_schema.columns
                   WHERE table_name = 'zopk_news' AND column_name = 'ai_evaluated_at') THEN
        ALTER TABLE zopk_news ADD COLUMN ai_evaluated_at TIMESTAMP;
    END IF;
END $$;

-- Which AI model was used for evaluation
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM information_schema.columns
                   WHERE table_name = 'zopk_news' AND column_name = 'ai_model') THEN
        ALTER TABLE zopk_news ADD COLUMN ai_model VARCHAR(100);
    END IF;
END $$;

-- AI relevance score (1-5 stars: 1=weak, 5=perfect match)
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM information_schema.columns
                   WHERE table_name = 'zopk_news' AND column_name = 'ai_relevance_score') THEN
        ALTER TABLE zopk_news ADD COLUMN ai_relevance_score INTEGER CHECK (ai_relevance_score >= 1 AND ai_relevance_score <= 5);
    END IF;
END $$;

-- Index for AI relevance filtering
CREATE INDEX IF NOT EXISTS idx_zopk_news_ai_relevant ON zopk_news(ai_relevant);
CREATE INDEX IF NOT EXISTS idx_zopk_news_ai_score ON zopk_news(ai_relevance_score);

-- ============================================================
-- MIGRATION COMPLETE
-- ============================================================