transceiver-db/sql/117-spec-equivalence-matcher.sql

-- Migration 117: Spec-Based Equivalence Matcher
-- Matches FX products with competitor products by technical specification
-- when no OPN-based equivalence already exists.
--
-- Match criteria (ALL must apply):
--   1. Same form_factor (exact)
--   2. Same speed_gbps (exact)
--   3. Same reach tier (SR/IR/LR/ER/ZR — based on reach_meters)
--   4. Same primary wavelength (within ±10nm, extracted from wavelengths field)
--      OR both have no wavelength data (broadband / non-WDM products)
--   5. Target must be a competitor vendor (is_competitor = true)
--   6. Max 30 competitor matches per FX product (too many = too generic)
--
-- Match quality:
--   confidence = 0.85   (high but below OPN-confirmed 1.0)
--   match_basis = '{spec}'
--   status      = 'auto_approved'
--
-- Rules:
--   - Skips pairs that already have ANY equivalence (approved, auto_approved, rejected)
--   - Skips FX products that already have an OPN-based equivalence
--     (OPN match is preferred; spec is only a fallback)
--   - Minimum reach_meters = 10 on both sides (avoids reach=0 garbage data)
--   - Reach tier comparison handles DAC/AOC (SR ≤ 300m)

-- ── Helper: extract primary wavelength in nm from text field ─────────────────
-- Handles: "1310nm", "850nm", "1310/1550nm", "1270nm-1610nm", NULL
CREATE OR REPLACE FUNCTION tip_extract_wavelength_nm(wl text)
RETURNS integer LANGUAGE sql IMMUTABLE PARALLEL SAFE AS $$
  SELECT (regexp_match(wl, '(\d{3,4})\s*nm'))[1]::integer
$$;

-- ── Helper: reach tier label ─────────────────────────────────────────────────
CREATE OR REPLACE FUNCTION tip_reach_tier(reach integer)
RETURNS text LANGUAGE sql IMMUTABLE PARALLEL SAFE AS $$
  SELECT CASE
    WHEN reach <=    300 THEN 'SR'   -- ≤300m  (SR, VSR, DAC, AOC)
    WHEN reach <=   2000 THEN 'IR'   -- ≤2km   (LX, LH intermediate)
    WHEN reach <=  10000 THEN 'LR'   -- ≤10km  (LR, LX, standard LH)
    WHEN reach <=  40000 THEN 'ER'   -- ≤40km  (ER, extended reach)
    ELSE                      'ZR'   -- >40km  (ZR, ZR+, coherent)
  END
$$;

-- ── Insert spec-based equivalences ──────────────────────────────────────────

INSERT INTO transceiver_equivalences (
  flexoptix_id,
  competitor_id,
  confidence,
  status,
  match_basis,
  match_notes,
  created_at,
  updated_at
)
SELECT DISTINCT
  fx.id                          AS flexoptix_id,
  comp.id                        AS competitor_id,
  0.85                           AS confidence,
  'auto_approved'                AS status,
  ARRAY['spec']                  AS match_basis,
  'Spec match: ' || fx.form_factor || ' ' || fx.speed_gbps || 'G ' ||
    tip_reach_tier(fx.reach_meters) ||
    CASE WHEN tip_extract_wavelength_nm(fx.wavelengths) IS NOT NULL
         THEN ' @' || tip_extract_wavelength_nm(fx.wavelengths) || 'nm'
         ELSE '' END             AS match_notes,
  NOW()                          AS created_at,
  NOW()                          AS updated_at
FROM transceivers fx
JOIN vendors vfx ON vfx.id = fx.vendor_id AND UPPER(vfx.name) LIKE '%FLEXOPTIX%'
JOIN transceivers comp
  ON comp.form_factor            = fx.form_factor
  AND comp.speed_gbps            = fx.speed_gbps
  AND comp.reach_meters          >= 10                 -- no garbage reach=0
  AND tip_reach_tier(comp.reach_meters) = tip_reach_tier(fx.reach_meters)
  -- Wavelength: both must match within ±10nm, OR both have no wavelength
  AND (
    (tip_extract_wavelength_nm(fx.wavelengths) IS NULL
     AND tip_extract_wavelength_nm(comp.wavelengths) IS NULL)
    OR
    ABS( COALESCE(tip_extract_wavelength_nm(comp.wavelengths), 0)
       - COALESCE(tip_extract_wavelength_nm(fx.wavelengths), 0) ) <= 10
  )
JOIN vendors vcomp ON vcomp.id = comp.vendor_id AND vcomp.is_competitor = true
WHERE fx.reach_meters             >= 10                -- no garbage reach=0 on FX side
  AND fx.speed_gbps               > 0
  -- FX product has no OPN-based equivalence at all (spec is fallback only)
  AND NOT EXISTS (
    SELECT 1 FROM transceiver_equivalences e
    WHERE e.flexoptix_id = fx.id
      AND 'opn' = ANY(e.match_basis)
  )
  -- Skip pairs that already have ANY equivalence
  AND NOT EXISTS (
    SELECT 1 FROM transceiver_equivalences e
    WHERE e.flexoptix_id = fx.id
      AND e.competitor_id = comp.id
  )
  -- Safety cap: skip FX product if it would match > 30 competitors
  -- (indicates too-generic spec — needs stricter criteria)
  AND (
    SELECT COUNT(DISTINCT c2.id)
    FROM transceivers c2
    JOIN vendors vc2 ON vc2.id = c2.vendor_id AND vc2.is_competitor = true
    WHERE c2.form_factor = fx.form_factor
      AND c2.speed_gbps  = fx.speed_gbps
      AND c2.reach_meters >= 10
      AND tip_reach_tier(c2.reach_meters) = tip_reach_tier(fx.reach_meters)
      AND (
        (tip_extract_wavelength_nm(fx.wavelengths) IS NULL
         AND tip_extract_wavelength_nm(c2.wavelengths) IS NULL)
        OR ABS( COALESCE(tip_extract_wavelength_nm(c2.wavelengths), 0)
              - COALESCE(tip_extract_wavelength_nm(fx.wavelengths), 0) ) <= 10
      )
  ) <= 30
ON CONFLICT DO NOTHING;

-- ── Statistics ───────────────────────────────────────────────────────────────
DO $$
DECLARE
  new_cnt      INTEGER;
  fx_covered   INTEGER;
  comp_covered INTEGER;
BEGIN
  SELECT COUNT(*) INTO new_cnt
  FROM transceiver_equivalences WHERE 'spec' = ANY(match_basis);

  SELECT COUNT(DISTINCT flexoptix_id) INTO fx_covered
  FROM transceiver_equivalences WHERE 'spec' = ANY(match_basis);

  SELECT COUNT(DISTINCT competitor_id) INTO comp_covered
  FROM transceiver_equivalences WHERE 'spec' = ANY(match_basis);

  RAISE NOTICE 'Migration 117 complete: Spec-Based Equivalence Matcher';
  RAISE NOTICE '  Spec equivalences total:     %', new_cnt;
  RAISE NOTICE '  FX products newly covered:   %', fx_covered;
  RAISE NOTICE '  Competitor products matched: %', comp_covered;
END $$;