nordabiz/database/migrations/004_seo_metrics.sql

-- ============================================================
-- NordaBiz - Migration 004: SEO Quality Assessment Metrics
-- ============================================================
-- Created: 2026-01-08
-- Description:
--   - Extends company_website_analysis with PageSpeed Insights scores
--   - Adds on-page SEO detail columns
--   - Adds technical SEO fields
--   - Adds audit metadata for tracking
--
-- Usage:
--   PostgreSQL: psql -h localhost -U nordabiz_app -d nordabiz -f 004_seo_metrics.sql
--   SQLite: sqlite3 nordabiz_local.db < 004_seo_metrics.sql
-- ============================================================

-- ============================================================
-- 1. PAGESPEED INSIGHTS SCORES (0-100)
-- ============================================================

ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS pagespeed_seo_score INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS pagespeed_performance_score INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS pagespeed_accessibility_score INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS pagespeed_best_practices_score INTEGER;

-- PageSpeed Audit Details (JSONB for PostgreSQL, TEXT for SQLite fallback)
-- Stores detailed audit results from PageSpeed Insights API
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS pagespeed_audits JSONB;

COMMENT ON COLUMN company_website_analysis.pagespeed_seo_score IS 'Google PageSpeed SEO score 0-100';
COMMENT ON COLUMN company_website_analysis.pagespeed_performance_score IS 'Google PageSpeed Performance score 0-100';
COMMENT ON COLUMN company_website_analysis.pagespeed_accessibility_score IS 'Google PageSpeed Accessibility score 0-100';
COMMENT ON COLUMN company_website_analysis.pagespeed_best_practices_score IS 'Google PageSpeed Best Practices score 0-100';
COMMENT ON COLUMN company_website_analysis.pagespeed_audits IS 'Full PageSpeed audit results as JSON';

-- ============================================================
-- 2. ON-PAGE SEO DETAILS
-- ============================================================

-- Note: seo_title and seo_description already exist in migration 002
-- These are additional detailed on-page SEO metrics

ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS meta_title VARCHAR(500);
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS meta_description TEXT;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS meta_keywords TEXT;

-- Heading structure
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS h1_count INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS h2_count INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS h3_count INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS h1_text VARCHAR(500);

-- Image analysis
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS total_images INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS images_without_alt INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS images_with_alt INTEGER;

-- Link analysis
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS internal_links_count INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS external_links_count INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS broken_links_count INTEGER;

-- Structured data (Schema.org, JSON-LD, Microdata)
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS has_structured_data BOOLEAN DEFAULT FALSE;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS structured_data_types TEXT[];
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS structured_data_json JSONB;

COMMENT ON COLUMN company_website_analysis.h1_count IS 'Number of H1 tags on homepage (should be 1)';
COMMENT ON COLUMN company_website_analysis.h2_count IS 'Number of H2 tags on homepage';
COMMENT ON COLUMN company_website_analysis.h3_count IS 'Number of H3 tags on homepage';
COMMENT ON COLUMN company_website_analysis.h1_text IS 'Text content of first H1 tag';
COMMENT ON COLUMN company_website_analysis.images_without_alt IS 'Images missing alt attribute - accessibility issue';
COMMENT ON COLUMN company_website_analysis.has_structured_data IS 'Whether page contains JSON-LD, Microdata, or RDFa';
COMMENT ON COLUMN company_website_analysis.structured_data_types IS 'Schema.org types found: Organization, LocalBusiness, etc.';

-- ============================================================
-- 3. TECHNICAL SEO
-- ============================================================

-- Canonical URL handling
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS has_canonical BOOLEAN DEFAULT FALSE;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS canonical_url VARCHAR(500);

-- Indexability
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS is_indexable BOOLEAN DEFAULT TRUE;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS noindex_reason VARCHAR(200);

-- Mobile & Core Web Vitals
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS is_mobile_friendly BOOLEAN;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS viewport_configured BOOLEAN;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS largest_contentful_paint_ms INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS first_input_delay_ms INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS cumulative_layout_shift NUMERIC(5, 3);

-- Open Graph & Social Meta
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS has_og_tags BOOLEAN DEFAULT FALSE;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS og_title VARCHAR(500);
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS og_description TEXT;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS og_image VARCHAR(500);
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS has_twitter_cards BOOLEAN DEFAULT FALSE;

-- Language & International
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS html_lang VARCHAR(10);
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS has_hreflang BOOLEAN DEFAULT FALSE;

COMMENT ON COLUMN company_website_analysis.has_canonical IS 'Whether page has canonical URL defined';
COMMENT ON COLUMN company_website_analysis.is_indexable IS 'Whether page can be indexed (no noindex directive)';
COMMENT ON COLUMN company_website_analysis.noindex_reason IS 'Reason if page is not indexable: meta tag, robots.txt, etc.';
COMMENT ON COLUMN company_website_analysis.largest_contentful_paint_ms IS 'Core Web Vital: LCP in milliseconds';
COMMENT ON COLUMN company_website_analysis.first_input_delay_ms IS 'Core Web Vital: FID in milliseconds';
COMMENT ON COLUMN company_website_analysis.cumulative_layout_shift IS 'Core Web Vital: CLS score';

-- ============================================================
-- 4. SEO AUDIT METADATA
-- ============================================================

ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS seo_audit_version VARCHAR(20);
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS seo_audited_at TIMESTAMP;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS seo_audit_errors TEXT[];
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS seo_overall_score INTEGER;

-- Custom SEO health score (calculated from all metrics)
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS seo_health_score INTEGER;
ALTER TABLE company_website_analysis ADD COLUMN IF NOT EXISTS seo_issues JSONB;

COMMENT ON COLUMN company_website_analysis.seo_audit_version IS 'Version of SEO audit script used';
COMMENT ON COLUMN company_website_analysis.seo_audited_at IS 'Timestamp of last SEO audit';
COMMENT ON COLUMN company_website_analysis.seo_audit_errors IS 'Errors encountered during SEO audit';
COMMENT ON COLUMN company_website_analysis.seo_overall_score IS 'Calculated overall SEO score 0-100';
COMMENT ON COLUMN company_website_analysis.seo_health_score IS 'On-page SEO health score 0-100';
COMMENT ON COLUMN company_website_analysis.seo_issues IS 'List of SEO issues found with severity levels';

-- ============================================================
-- 5. INDEXES FOR SEO QUERIES
-- ============================================================

CREATE INDEX IF NOT EXISTS idx_website_seo_score ON company_website_analysis(pagespeed_seo_score);
CREATE INDEX IF NOT EXISTS idx_website_seo_audited ON company_website_analysis(seo_audited_at);
CREATE INDEX IF NOT EXISTS idx_website_seo_overall ON company_website_analysis(seo_overall_score);

-- ============================================================
-- 6. SEO DASHBOARD VIEW
-- ============================================================

CREATE OR REPLACE VIEW v_company_seo_overview AS
SELECT
    c.id,
    c.name,
    c.slug,
    c.website,
    cat.name as category_name,
    wa.pagespeed_seo_score,
    wa.pagespeed_performance_score,
    wa.pagespeed_accessibility_score,
    wa.pagespeed_best_practices_score,
    wa.seo_overall_score,
    wa.seo_health_score,
    wa.h1_count,
    wa.images_without_alt,
    wa.has_structured_data,
    wa.has_ssl,
    wa.has_sitemap,
    wa.has_robots_txt,
    wa.is_indexable,
    wa.is_mobile_friendly,
    wa.seo_audited_at,
    wa.analyzed_at
FROM companies c
LEFT JOIN company_website_analysis wa ON c.id = wa.company_id
LEFT JOIN categories cat ON c.category_id = cat.id
WHERE c.website IS NOT NULL AND c.website != ''
ORDER BY wa.seo_overall_score DESC NULLS LAST;

COMMENT ON VIEW v_company_seo_overview IS 'SEO metrics overview for admin dashboard';

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

-- Verify migration (PostgreSQL only)
DO $$
BEGIN
    RAISE NOTICE 'Migration 004 completed successfully!';
    RAISE NOTICE 'Added columns to company_website_analysis:';
    RAISE NOTICE '  - PageSpeed scores (seo, performance, accessibility, best_practices)';
    RAISE NOTICE '  - On-page SEO details (headings, images, links, structured data)';
    RAISE NOTICE '  - Technical SEO (canonical, indexability, Core Web Vitals)';
    RAISE NOTICE '  - SEO audit metadata (version, timestamp, scores)';
    RAISE NOTICE 'Created view:';
    RAISE NOTICE '  - v_company_seo_overview';
END $$;