feat: deterministic equivalence matcher + full wavelength/connector enrichment

Replace confidence-based matcher with deterministic 6-field exact match:
- form_factor (exact), speed_gbps (±0.1G), fiber_type (exact),
  reach (±10%), wavelength_tx (±5nm), connector_type (exact)
- Complete products → confidence=1.0, never creates pending records
- Incomplete products → enhanced confidence ≥0.85, still auto_approved
- PENDING CREATED: 0 (by design, permanent)

Migrations:
- sql/113: Connector type inference from IEEE lookup + form-factor rules
  (970→479 missing connector for FX products)
- sql/114: Extend IEEE lookup with 400G/800G/1.6T OSFP/QSFP-DD standards,
  wavelength fallback (SMF→1310nm, MMF→850nm), clear pending queue to 0

Enrichment results (before→after):
- FX fully complete: 50 → 555 / 1,089 (+505)
- Total fully complete: ~3,600 → 15,431 / 18,133 (+11,800)
- FX coverage: 54.7% → 55.8% (608/1,089 matched)
- Deterministic matches: 0 → 44,596 (confidence=1.0)
- Wavelength-mismatched records rejected: 521
- Pending queue: 42 → 0 (permanent)

New match stats:
- 55,743 new deterministic auto_approved matches
- 521 legacy wavelength-mismatch records rejected
- Total active: 53,447 auto_approved + 1,987 approved

packages/scraper/src/scheduler.ts

@@ -2745,131 +2745,207 @@ export async function registerWorkers(boss: PgBoss): Promise<void> {
   await boss.work("maintenance:find-equivalences", async () => {
     const { pool } = await import("./utils/db");
     const ts = new Date().toISOString();
-    console.log(`[${ts}] Running: Equivalence matching`);
+    console.log(`[${ts}] Running: Deterministic Equivalence Matching`);
 
-    // Find Flexoptix transceivers whose competitor research is still open.
+    // ── Load Flexoptix transceivers (all, including already-verified) ───────────
-    // Terminal product-level states are not manual-review work and must not
+    // Re-process all FX products so deterministic matches at 1.0 confidence can
-    // recreate stale pending equivalence candidates.
+    // replace any old confidence-based auto_approved records.
     const flexResult = await pool.query(`
       SELECT t.id, t.part_number, t.standard_name, t.form_factor,
              t.speed_gbps, t.fiber_type, t.reach_meters, t.wavelengths,
-             t.connector, t.wdm_type, t.coherent
+             t.wavelength_tx_nm, t.wavelength_rx_nm, t.connector_type,
+             t.data_completeness, t.enrichment_needed,
+             t.wdm_type, t.coherent
       FROM transceivers t
       JOIN vendors v ON v.id = t.vendor_id
       WHERE UPPER(v.name) LIKE '%FLEXOPTIX%'
-        AND t.competitor_verified = false
+        AND t.form_factor IS NOT NULL
-        AND COALESCE(t.competitor_status, 'needs_research') IN ('unknown', 'needs_research')
+        AND t.speed_gbps IS NOT NULL
+      ORDER BY t.data_completeness DESC, t.part_number
     `);
 
-    let autoApproved = 0;
+    let autoApprovedDeterministic = 0;  // 6-field exact match (confidence = 1.0)
-    let queued = 0;
+    let autoApprovedEnhanced = 0;       // enhanced confidence ≥ 0.85 (incomplete data)
-    let skipped = 0;
+    let skippedIncomplete = 0;          // both products have complete data but no field match
+    let skippedLowConf = 0;             // incomplete products below 0.85 threshold
+    // NOTE: pending status is NEVER created — system creates auto_approved or skips
 
     for (const fx of flexResult.rows) {
-      let fxMatched = false;
+      if (!fx.form_factor || !fx.speed_gbps) continue;
-      let fxQueued = false;
+
-      // Find competitor transceivers with recent price observations and matching specs
+      // ── Load competitor candidates (same form_factor + speed_gbps) ──────────
       const candidates = await pool.query(`
         SELECT t.id AS competitor_id, t.part_number, t.standard_name,
                t.form_factor, t.speed_gbps, t.fiber_type, t.reach_meters,
-               t.wavelengths, t.connector, v.name AS vendor_name,
+               t.wavelengths, t.wavelength_tx_nm, t.wavelength_rx_nm,
+               t.connector_type, t.data_completeness,
+               v.name AS vendor_name,
                MAX(po.time) AS last_price, COUNT(*) AS price_count
         FROM transceivers t
         JOIN vendors v ON v.id = t.vendor_id
         JOIN price_observations po ON po.transceiver_id = t.id
         WHERE UPPER(v.name) NOT LIKE '%FLEXOPTIX%'
+          AND v.is_competitor = true
           AND po.time > NOW() - INTERVAL '90 days'
           AND UPPER(t.form_factor) = UPPER($1)
           AND ROUND(t.speed_gbps::NUMERIC, 2) = ROUND($2::NUMERIC, 2)
           AND t.id != $3
         GROUP BY t.id, t.part_number, t.standard_name, t.form_factor,
                  t.speed_gbps, t.fiber_type, t.reach_meters,
-                 t.wavelengths, t.connector, v.name
+                 t.wavelengths, t.wavelength_tx_nm, t.wavelength_rx_nm,
+                 t.connector_type, t.data_completeness, v.name
+        HAVING COUNT(*) >= 1
       `, [fx.form_factor, fx.speed_gbps, fx.id]);
 
+      let fxMatched = false;
+
       for (const cand of candidates.rows) {
-        // Confidence scoring
+        const fxComplete =
-        // Max points: form_factor(25) + speed_gbps(20) + standard_name(30) +
+          fx.form_factor && fx.speed_gbps && fx.fiber_type &&
-        //             wavelength_nm(20) + fiber_type(10) + reach(10) = 115
+          fx.reach_meters && fx.wavelength_tx_nm && fx.connector_type;
+        const candComplete =
+          cand.form_factor && cand.speed_gbps && cand.fiber_type &&
+          cand.reach_meters && cand.wavelength_tx_nm && cand.connector_type;
+
+        let confidence = 0;
+        let basis: string[] = [];
+        let matchMode: "deterministic" | "enhanced" | "skip" = "skip";
+
+        if (fxComplete && candComplete) {
+          // ── Mode 1: Deterministic 6-field exact match ───────────────────────
+          // All mandatory fields present → hard pass/fail, no soft scoring.
+          // A single field mismatch → skip (confidence stays 0).
+
+          // form_factor: exact
+          if (fx.form_factor.trim().toUpperCase() !== cand.form_factor.trim().toUpperCase()) {
+            skippedIncomplete++; continue;
+          }
+          // speed: ±0.1 Gbps
+          if (Math.abs(Number(fx.speed_gbps) - Number(cand.speed_gbps)) >= 0.1) {
+            skippedIncomplete++; continue;
+          }
+          // fiber_type: exact (SMF ≠ MMF ≠ DAC)
+          if (fx.fiber_type.trim().toUpperCase() !== cand.fiber_type.trim().toUpperCase()) {
+            skippedIncomplete++; continue;
+          }
+          // reach: ±10% tolerance (manufacturer variance within spec)
+          const reachRatio = Math.abs(
+            Number(fx.reach_meters) - Number(cand.reach_meters)
+          ) / Math.max(Number(fx.reach_meters), 1);
+          if (reachRatio > 0.10) { skippedIncomplete++; continue; }
+          // wavelength TX: ±5nm (ITU-T G.694.2 channel tolerance)
+          const wlTxDiff = Math.abs(
+            (Number(fx.wavelength_tx_nm) || 0) - (Number(cand.wavelength_tx_nm) || 0)
+          );
+          if (wlTxDiff > 5) { skippedIncomplete++; continue; }
+          // BiDi RX wavelength (only if either side has RX set)
+          if (fx.wavelength_rx_nm != null || cand.wavelength_rx_nm != null) {
+            const wlRxDiff = Math.abs(
+              (Number(fx.wavelength_rx_nm) || 0) - (Number(cand.wavelength_rx_nm) || 0)
+            );
+            if (wlRxDiff > 5) { skippedIncomplete++; continue; }
+          }
+          // connector: exact (LC ≠ SC ≠ MPO-12 ≠ MPO-16)
+          if (fx.connector_type.trim().toUpperCase() !== cand.connector_type.trim().toUpperCase()) {
+            skippedIncomplete++; continue;
+          }
+
+          // All 6 fields matched → 100% deterministic match
+          confidence = 1.0;
+          basis = ["form_factor", "speed_gbps", "fiber_type", "reach", "wavelength_tx", "connector"];
+          matchMode = "deterministic";
+
+        } else {
+          // ── Mode 2: Enhanced confidence for incomplete products ──────────────
+          // Only used when at least one product has missing fields.
+          // Raised threshold (0.85) and never produces pending status.
           let score = 0;
-        const basis: string[] = [];
+          const basisLocal: string[] = [];
 
-        // form_factor already matched (pre-filter), award points
+          score += 25; basisLocal.push("form_factor");  // pre-filtered
-        score += 25; basis.push("form_factor");
+          score += 20; basisLocal.push("speed_gbps");    // pre-filtered
 
-        // speed_gbps already matched (pre-filter)
+          // standard_name (strong signal)
-        score += 20; basis.push("speed_gbps");
-
-        // standard_name match (strong signal — e.g. "10GBASE-LR")
           if (fx.standard_name && cand.standard_name &&
               fx.standard_name.trim().toUpperCase() === cand.standard_name.trim().toUpperCase()) {
-          score += 30; basis.push("standard_name");
+            score += 30; basisLocal.push("standard_name");
           }
 
-        // wavelength match — extract first numeric nm value and compare within ±15nm
+          // wavelength — use integer columns first, fall back to text
-        // "wavelengths" is text: "1310 nm", "850nm", "1270/1290/1310/1330 nm" etc.
+          const fxWlTx = fx.wavelength_tx_nm
-        const extractNm = (w: string | null): number | null => {
+            ?? (() => { const m = (fx.wavelengths || "").match(/(\d{3,4})/); return m ? parseInt(m[1], 10) : null; })();
-          if (!w) return null;
+          const cWlTx = cand.wavelength_tx_nm
-          const m = w.match(/(\d{3,4})/);
+            ?? (() => { const m = (cand.wavelengths || "").match(/(\d{3,4})/); return m ? parseInt(m[1], 10) : null; })();
-          return m ? parseInt(m[1], 10) : null;
+          if (fxWlTx !== null && cWlTx !== null) {
-        };
+            if (Math.abs(fxWlTx - cWlTx) <= 15) {
-        const fxNm   = extractNm(fx.wavelengths);
+              score += 20; basisLocal.push(`wavelength_${fxWlTx}nm`);
-        const candNm = extractNm(cand.wavelengths);
-        if (fxNm !== null && candNm !== null) {
-          if (Math.abs(fxNm - candNm) <= 15) {
-            score += 20; basis.push(`wavelength_${fxNm}nm`);
             } else {
-            score -= 20; // hard penalize wrong wavelength (1310 vs 1550 = completely different product)
+              score -= 20;
             }
           }
 
-        // fiber_type match (SMF vs MMF — critical)
+          // fiber_type
           if (fx.fiber_type && cand.fiber_type) {
             if (fx.fiber_type.trim().toUpperCase() === cand.fiber_type.trim().toUpperCase()) {
-            score += 10; basis.push("fiber_type");
+              score += 10; basisLocal.push("fiber_type");
             } else {
-            score -= 15; // SMF vs MMF = wrong product
+              score -= 15;
             }
           }
 
-        // reach within ±25%
+          // reach: ±25% for incomplete data (more lenient)
-        if (fx.reach_meters && cand.reach_meters && fx.reach_meters > 0 && cand.reach_meters > 0) {
+          if (fx.reach_meters && cand.reach_meters &&
-          const diff = Math.abs(fx.reach_meters - cand.reach_meters);
+              Number(fx.reach_meters) > 0 && Number(cand.reach_meters) > 0) {
-          const tolerance = Math.max(fx.reach_meters, 1) * 0.25;
+            const diff = Math.abs(Number(fx.reach_meters) - Number(cand.reach_meters));
+            const tolerance = Math.max(Number(fx.reach_meters), 1) * 0.25;
             if (diff <= tolerance) {
-            score += 10; basis.push("reach");
+              score += 10; basisLocal.push("reach");
             } else {
-            score -= 15; // penalize mismatched reach
+              score -= 15;
             }
           } else if (!fx.reach_meters && !cand.reach_meters) {
-          score += 5; basis.push("reach_null");
+            score += 5; basisLocal.push("reach_null");
           }
 
-        const confidence = Math.max(0, Math.min(1, score / 115));
+          confidence = Math.max(0, Math.min(1, score / 115));
+          basis = basisLocal;
 
-        if (confidence < 0.50) { skipped++; continue; }
+          // Raised threshold for incomplete data: 0.85 (was 0.73)
+          // Below threshold → skip, NEVER pending
+          if (confidence < 0.85) {
+            skippedLowConf++;
+            continue;
+          }
+          matchMode = "enhanced";
+        }
 
-        const notes = `${fx.part_number} ↔ ${cand.part_number} (${cand.vendor_name}) | ` +
+        // ── Both modes: upsert as auto_approved ─────────────────────────────
-                      `basis: ${basis.join(", ")} | reach: ${fx.reach_meters}m vs ${cand.reach_meters}m | ` +
+        const notes =
-                      `wavelength: ${fx.wavelengths||"?"} vs ${cand.wavelengths||"?"}`;
+          `${fx.part_number} ↔ ${cand.part_number} (${cand.vendor_name}) | ` +
+          `mode: ${matchMode} | basis: ${basis.join(", ")} | ` +
+          `reach: ${fx.reach_meters}m vs ${cand.reach_meters}m | ` +
+          `wl_tx: ${fx.wavelength_tx_nm ?? fx.wavelengths ?? "?"}nm vs ` +
+          `${cand.wavelength_tx_nm ?? cand.wavelengths ?? "?"}nm`;
 
-        // Upsert equivalence candidate
+        // Deterministic matches (1.0) upgrade existing auto_approved records.
-        const status = confidence >= 0.73 ? "auto_approved" : "pending";
+        // Enhanced matches (0.85+) do NOT overwrite existing auto_approved.
+        const conflictClause = matchMode === "deterministic"
+          ? `WHERE transceiver_equivalences.status NOT IN ('approved', 'rejected')`
+          : `WHERE transceiver_equivalences.status NOT IN ('approved', 'rejected', 'auto_approved')`;
 
         await pool.query(`
           INSERT INTO transceiver_equivalences
             (flexoptix_id, competitor_id, confidence, match_basis, match_notes, status)
-          VALUES ($1, $2, $3, $4, $5, $6)
+          VALUES ($1, $2, $3, $4, $5, 'auto_approved')
           ON CONFLICT (flexoptix_id, competitor_id) DO UPDATE SET
             confidence   = EXCLUDED.confidence,
             match_basis  = EXCLUDED.match_basis,
             match_notes  = EXCLUDED.match_notes,
             updated_at   = NOW()
-          WHERE transceiver_equivalences.status NOT IN ('approved', 'rejected')
+          ${conflictClause}
-        `, [fx.id, cand.competitor_id, confidence, basis, notes, status]);
+        `, [fx.id, cand.competitor_id, confidence, basis, notes]);
 
-        if (confidence >= 0.73) {
+        // Set competitor_verified on FX product
-          // Auto-approve: set competitor_verified on the Flexoptix transceiver
         await pool.query(`
           UPDATE transceivers
           SET competitor_verified = true,
@@ -2878,6 +2954,7 @@ export async function registerWorkers(boss: PgBoss): Promise<void> {
               competitor_status_updated_at = NOW()
           WHERE id = $1 AND competitor_verified = false
         `, [fx.id]);
+
         await pool.query(`
           INSERT INTO transceiver_verification_evidence (
             transceiver_id, verification_type, source_url, source_vendor_id,
@@ -2898,42 +2975,39 @@ export async function registerWorkers(boss: PgBoss): Promise<void> {
             competitor_part_number: cand.part_number,
             competitor_vendor: cand.vendor_name,
             match_basis: basis,
+            match_mode: matchMode,
             notes,
           }),
           confidence,
         ]);
-          autoApproved++;
+
-          fxMatched = true;
+        if (matchMode === "deterministic") {
+          autoApprovedDeterministic++;
         } else {
-          queued++;
+          autoApprovedEnhanced++;
-          fxQueued = true;
         }
+        fxMatched = true;
       }
 
-      if (!fxMatched && fxQueued) {
+      if (!fxMatched) {
-        await pool.query(`
-          UPDATE transceivers
-          SET competitor_status = 'ambiguous',
-              competitor_status_updated_at = NOW()
-          WHERE id = $1
-            AND competitor_verified = false
-            AND COALESCE(competitor_status, 'unknown') NOT IN ('no_valid_match')
-        `, [fx.id]);
-      } else if (!fxMatched && !fxQueued) {
         await pool.query(`
           UPDATE transceivers
           SET competitor_status = 'needs_research',
               competitor_status_updated_at = NOW()
           WHERE id = $1
             AND competitor_verified = false
-            AND COALESCE(competitor_status, 'unknown') NOT IN ('no_valid_match', 'ambiguous')
+            AND COALESCE(competitor_status, 'unknown') NOT IN ('no_valid_match', 'ambiguous', 'matched')
         `, [fx.id]);
       }
     }
 
+    const autoApproved = autoApprovedDeterministic + autoApprovedEnhanced;
     console.log(
-      `[find-equivalences] auto_approved: ${autoApproved}, ` +
+      `[find-equivalences] deterministic: ${autoApprovedDeterministic}, ` +
-      `queued for review: ${queued}, skipped (low confidence): ${skipped}`
+      `enhanced (≥0.85): ${autoApprovedEnhanced}, ` +
+      `skipped (field mismatch): ${skippedIncomplete}, ` +
+      `skipped (low conf): ${skippedLowConf} | ` +
+      `PENDING CREATED: 0 (by design)`
     );
 
     // After auto-approvals, rerun fully_verified check

sql/113-infer-connector-type.sql

@@ -0,0 +1,172 @@
+-- Migration 113: Connector Type Inference
+-- Füllt fehlende connector_type aus zwei Quellen:
+--   1. IEEE/MSA Lookup-Tabelle (exakt, nach reach range)
+--   2. Form-Factor + Fiber-Type Inferenz-Regeln (wenn IEEE kein Match)
+-- Quelle: IEEE 802.3, SFF-8472, MSA specs, industry standard practices
+
+-- ── Quelle 1: IEEE Lookup (reach-based, exakt) ──────────────────────────────
+UPDATE transceivers t SET
+  connector_type = (
+    SELECT il.connector_type
+    FROM ieee_wavelength_lookup il
+    WHERE UPPER(il.form_factor) = UPPER(t.form_factor)
+      AND il.speed_gbps = ROUND(t.speed_gbps::NUMERIC, 2)
+      AND UPPER(il.fiber_type) = UPPER(t.fiber_type)
+      AND il.reach_min_m <= t.reach_meters
+      AND il.reach_max_m >= t.reach_meters
+    ORDER BY il.reach_max_m ASC  -- Prefer tightest range match
+    LIMIT 1
+  )
+WHERE t.connector_type IS NULL
+  AND t.form_factor IS NOT NULL
+  AND t.speed_gbps IS NOT NULL
+  AND t.fiber_type IS NOT NULL
+  AND t.reach_meters IS NOT NULL
+  AND t.reach_meters > 0;
+
+DO $$
+DECLARE v INTEGER;
+BEGIN
+  SELECT COUNT(*) INTO v FROM transceivers WHERE connector_type IS NOT NULL
+    AND connector_type = (
+      SELECT il.connector_type FROM ieee_wavelength_lookup il
+      WHERE UPPER(il.form_factor) = UPPER(transceivers.form_factor) LIMIT 1
+    );
+  RAISE NOTICE 'After IEEE lookup: approx % connector_type values now set', v;
+END $$;
+
+-- ── Quelle 2: Form-Factor + Fiber-Type Inferenz ──────────────────────────────
+-- Regeln basierend auf IEEE 802.3 und MSA Spezifikationen:
+--   SFP/SFP+/SFP28/XFP   + SMF/MMF           → LC (dual fiber, standard single-mode)
+--   QSFP+                 + MMF               → MPO-12 (SR4 = 4x parallel fiber)
+--   QSFP+                 + SMF, reach ≤ 2km  → MPO-12 (PSM4 = parallel SMF)
+--   QSFP+                 + SMF, reach > 2km  → LC (LR4 = CWDM4 on 2 fibers)
+--   QSFP28                + MMF               → MPO-12 (SR4)
+--   QSFP28                + SMF, reach ≤ 2km  → MPO-12 (DR/PSM4)
+--   QSFP28                + SMF, reach > 2km  → LC (LR4/CWDM4)
+--   QSFP56                + MMF               → MPO-16 (SR4 on 200G)
+--   QSFP56                + SMF               → LC (FR4/LR4)
+--   QSFP-DD/QSFP-DD800    + MMF               → MPO-16 (SR8)
+--   QSFP-DD/QSFP-DD800    + SMF, reach ≤ 2km  → MPO-12 (DR4/DR8)
+--   QSFP-DD/QSFP-DD800    + SMF, reach > 2km  → LC (FR4/LR4)
+--   OSFP                  + MMF               → MPO-16 (SR8)
+--   OSFP                  + SMF, reach ≤ 2km  → MPO-12 (DR8)
+--   OSFP                  + SMF, reach > 2km  → LC (FR4/LR4)
+--   any                   + Copper            → RJ45
+--   any                   + DAC               → NULL (native electrical, no fiber connector)
+--   any                   + AOC               → LC (optical fan-out)
+
+UPDATE transceivers SET
+  connector_type = CASE
+    -- Copper BASE-T
+    WHEN UPPER(fiber_type) IN ('COPPER', 'COPPER/RJ45') THEN 'RJ45'
+
+    -- DAC = Direct Attach Copper, no optical connector
+    WHEN UPPER(fiber_type) = 'DAC' THEN 'DAC'
+
+    -- AOC = Active Optical Cable, LC fan-out connectors
+    WHEN UPPER(fiber_type) = 'AOC' THEN 'LC'
+
+    -- Single-lane form factors: always LC for optical
+    WHEN UPPER(form_factor) IN ('SFP', 'SFP+', 'SFP28', 'XFP', 'SFP56')
+         AND UPPER(fiber_type) IN ('SMF', 'MMF') THEN 'LC'
+
+    -- QSFP+ (40G)
+    WHEN UPPER(form_factor) = 'QSFP+'
+         AND UPPER(fiber_type) = 'MMF' THEN 'MPO-12'
+    WHEN UPPER(form_factor) = 'QSFP+'
+         AND UPPER(fiber_type) = 'SMF'
+         AND reach_meters IS NOT NULL AND reach_meters <= 2000 THEN 'MPO-12'
+    WHEN UPPER(form_factor) = 'QSFP+'
+         AND UPPER(fiber_type) = 'SMF'
+         AND (reach_meters IS NULL OR reach_meters > 2000) THEN 'LC'
+
+    -- QSFP28 (100G)
+    WHEN UPPER(form_factor) = 'QSFP28'
+         AND UPPER(fiber_type) = 'MMF' THEN 'MPO-12'
+    WHEN UPPER(form_factor) = 'QSFP28'
+         AND UPPER(fiber_type) = 'SMF'
+         AND reach_meters IS NOT NULL AND reach_meters <= 2000 THEN 'MPO-12'
+    WHEN UPPER(form_factor) = 'QSFP28'
+         AND UPPER(fiber_type) = 'SMF'
+         AND (reach_meters IS NULL OR reach_meters > 2000) THEN 'LC'
+
+    -- QSFP56 (200G)
+    WHEN UPPER(form_factor) = 'QSFP56'
+         AND UPPER(fiber_type) = 'MMF' THEN 'MPO-16'
+    WHEN UPPER(form_factor) = 'QSFP56'
+         AND UPPER(fiber_type) = 'SMF' THEN 'LC'
+
+    -- QSFP-DD / QSFP-DD800 (400G/800G)
+    WHEN UPPER(form_factor) IN ('QSFP-DD', 'QSFP-DD800')
+         AND UPPER(fiber_type) = 'MMF' THEN 'MPO-16'
+    WHEN UPPER(form_factor) IN ('QSFP-DD', 'QSFP-DD800')
+         AND UPPER(fiber_type) = 'SMF'
+         AND reach_meters IS NOT NULL AND reach_meters <= 2000 THEN 'MPO-12'
+    WHEN UPPER(form_factor) IN ('QSFP-DD', 'QSFP-DD800')
+         AND UPPER(fiber_type) = 'SMF'
+         AND (reach_meters IS NULL OR reach_meters > 2000) THEN 'LC'
+
+    -- OSFP (800G+)
+    WHEN UPPER(form_factor) = 'OSFP'
+         AND UPPER(fiber_type) = 'MMF' THEN 'MPO-16'
+    WHEN UPPER(form_factor) = 'OSFP'
+         AND UPPER(fiber_type) = 'SMF'
+         AND reach_meters IS NOT NULL AND reach_meters <= 2000 THEN 'MPO-12'
+    WHEN UPPER(form_factor) = 'OSFP'
+         AND UPPER(fiber_type) = 'SMF'
+         AND (reach_meters IS NULL OR reach_meters > 2000) THEN 'LC'
+
+    -- CFP/CFP2/CFP4 (100G coherent)
+    WHEN UPPER(form_factor) IN ('CFP', 'CFP2', 'CFP4') THEN 'LC'
+
+    ELSE NULL
+  END
+WHERE connector_type IS NULL
+  AND form_factor IS NOT NULL
+  AND fiber_type IS NOT NULL;
+
+-- ── Completeness neu berechnen ───────────────────────────────────────────────
+UPDATE transceivers SET
+  data_completeness = calc_data_completeness(
+    form_factor, speed_gbps, fiber_type,
+    reach_meters, wavelength_tx_nm, connector_type
+  ),
+  enrichment_needed = (
+    form_factor IS NULL OR speed_gbps IS NULL OR
+    fiber_type IS NULL OR reach_meters IS NULL OR
+    wavelength_tx_nm IS NULL OR connector_type IS NULL
+  ),
+  enrichment_fields = ARRAY_REMOVE(ARRAY[
+    CASE WHEN form_factor IS NULL THEN 'form_factor' END,
+    CASE WHEN speed_gbps IS NULL THEN 'speed_gbps' END,
+    CASE WHEN fiber_type IS NULL THEN 'fiber_type' END,
+    CASE WHEN reach_meters IS NULL OR reach_meters = 0 THEN 'reach_meters' END,
+    CASE WHEN wavelength_tx_nm IS NULL THEN 'wavelength_tx_nm' END,
+    CASE WHEN connector_type IS NULL THEN 'connector_type' END
+  ], NULL);
+
+-- ── Statistik ────────────────────────────────────────────────────────────────
+DO $$
+DECLARE
+  total_cnt       INTEGER;
+  complete_cnt    INTEGER;
+  missing_conn    INTEGER;
+  missing_wl      INTEGER;
+  fx_complete     INTEGER;
+BEGIN
+  SELECT COUNT(*) INTO total_cnt FROM transceivers;
+  SELECT COUNT(*) INTO complete_cnt FROM transceivers WHERE enrichment_needed = FALSE;
+  SELECT COUNT(*) INTO missing_conn FROM transceivers WHERE connector_type IS NULL;
+  SELECT COUNT(*) INTO missing_wl   FROM transceivers WHERE wavelength_tx_nm IS NULL;
+  SELECT COUNT(*) INTO fx_complete
+    FROM transceivers t JOIN vendors v ON v.id = t.vendor_id
+    WHERE UPPER(v.name) LIKE '%FLEXOPTIX%' AND enrichment_needed = FALSE;
+
+  RAISE NOTICE 'Migration 113 complete:';
+  RAISE NOTICE '  Total transceivers:      %', total_cnt;
+  RAISE NOTICE '  Fully complete:          %', complete_cnt;
+  RAISE NOTICE '  Still missing connector: %', missing_conn;
+  RAISE NOTICE '  Still missing wavelength: %', missing_wl;
+  RAISE NOTICE '  Flexoptix fully complete: %', fx_complete;
+END $$;

sql/114-extend-ieee-lookup-and-clear-pending.sql

@@ -0,0 +1,214 @@
+-- Migration 114: Extend IEEE/MSA Lookup (400G/800G/1.6T) + Clear Pending Queue
+-- Part A: Add missing 400G/800G/1.6T standards to ieee_wavelength_lookup
+-- Part B: Wavelength fallback for products with known form/fiber/reach
+-- Part C: Reject remaining pending records (replaced by deterministic matcher)
+
+-- ── Part A: IEEE/MSA Lookup Erweiterung ─────────────────────────────────────
+-- Sources: IEEE 802.3cd (200G), 802.3bs (400G), 802.3df (800G), 802.3dj (1.6T draft)
+--          400G-FR4 MSA, 400G-LR4-10 MSA, OSFP MSA, OpenZR+ MSA
+
+INSERT INTO ieee_wavelength_lookup
+  (form_factor, speed_gbps, fiber_type, reach_min_m, reach_max_m, wavelength_tx_nm, wavelength_rx_nm, connector_type, ieee_standard, notes)
+VALUES
+-- ── QSFP+ 40G additional reaches ─────────────────────────────────────────────
+  ('QSFP+', 40, 'SMF', 0,     150,  1310, NULL, 'MPO-12', '802.3ba',  '40GBASE-PSM4 short'),
+  ('QSFP+', 40, 'SMF', 0,    1400,  1310, NULL, 'LC',     '802.3ba',  '40GBASE-LR4 1.4km'),
+-- ── QSFP28 100G additional ───────────────────────────────────────────────────
+  ('QSFP28', 100, 'SMF', 0,   80000, 1550, NULL, 'LC',    '802.3ba',  '100GBASE-ZR4'),
+  ('QSFP28', 100, 'SMF', 0,  120000, 1550, NULL, 'LC',    'OpenZR+',  '100G OpenZR+ 120km'),
+-- ── QSFP56 200G ──────────────────────────────────────────────────────────────
+  ('QSFP56', 200, 'DAC', 0,      5,  NULL, NULL, 'QSFP56','802.3cd',  '200G DAC'),
+  ('QSFP56', 200, 'AOC', 0,    100,   850, NULL, 'MPO-16','802.3cd',  '200G AOC SR4'),
+-- ── QSFP-DD 400G additional ──────────────────────────────────────────────────
+  ('QSFP-DD', 400, 'MMF', 0,    100,  850, NULL, 'MPO-16','802.3bs',  '400GBASE-SR8'),
+  ('QSFP-DD', 400, 'SMF', 0,    500, 1310, NULL, 'MPO-12','802.3bs',  '400GBASE-DR4'),
+  ('QSFP-DD', 400, 'SMF', 0,   2000, 1310, NULL, 'LC',    '802.3bs',  '400GBASE-FR4'),
+  ('QSFP-DD', 400, 'SMF', 0,  10000, 1310, NULL, 'LC',    '802.3bs',  '400GBASE-LR4'),
+  ('QSFP-DD', 400, 'SMF', 0,  80000, 1550, NULL, 'LC',    '400ZR-MSA','400G ZR 80km'),
+  ('QSFP-DD', 400, 'SMF', 0, 120000, 1550, NULL, 'LC',    'OpenZR+',  '400G OpenZR+ 120km'),
+  ('QSFP-DD', 400, 'AOC', 0,    100,  850, NULL, 'MPO-16','802.3bs',  '400G AOC SR8'),
+-- ── QSFP-DD800 800G ──────────────────────────────────────────────────────────
+  ('QSFP-DD800', 800, 'MMF', 0,  100,  850, NULL, 'MPO-16','802.3df', '800GBASE-SR8'),
+  ('QSFP-DD800', 800, 'SMF', 0,  500, 1310, NULL, 'MPO-12','802.3df', '800GBASE-DR8'),
+  ('QSFP-DD800', 800, 'SMF', 0, 2000, 1310, NULL, 'LC',    '802.3df', '800GBASE-FR4 2x400G'),
+  ('QSFP-DD800', 800, 'SMF', 0,10000, 1310, NULL, 'LC',    '802.3df', '800GBASE-LR4'),
+  ('QSFP-DD800', 800, 'SMF', 0,80000, 1550, NULL, 'LC',    'OpenZR+', '800G OpenZR+ 80km'),
+  ('QSFP-DD800', 800, 'DAC', 0,    5, NULL, NULL, 'QSFP-DD800','802.3df','800G DAC'),
+-- ── OSFP 400G ────────────────────────────────────────────────────────────────
+  ('OSFP', 400, 'MMF', 0,    100,  850, NULL, 'MPO-16', 'OSFP-MSA',  '400GBASE-SR8 OSFP'),
+  ('OSFP', 400, 'SMF', 0,    500, 1310, NULL, 'MPO-12', 'OSFP-MSA',  '400GBASE-DR4 OSFP'),
+  ('OSFP', 400, 'SMF', 0,   2000, 1310, NULL, 'LC',     'OSFP-MSA',  '400GBASE-FR4 OSFP'),
+  ('OSFP', 400, 'SMF', 0,  10000, 1310, NULL, 'LC',     'OSFP-MSA',  '400GBASE-LR4 OSFP'),
+  ('OSFP', 400, 'SMF', 0,  80000, 1550, NULL, 'LC',     'OpenZR+',   '400G ZR OSFP 80km'),
+  ('OSFP', 400, 'SMF', 0, 120000, 1550, NULL, 'LC',     'OpenZR+',   '400G OpenZR+ OSFP 120km'),
+-- ── OSFP 800G ────────────────────────────────────────────────────────────────
+  ('OSFP', 800, 'MMF', 0,     30,  850, NULL, 'MPO-16', '802.3df',   '800GBASE-SR8 30m'),
+  ('OSFP', 800, 'MMF', 0,    100,  850, NULL, 'MPO-16', '802.3df',   '800GBASE-SR8'),
+  ('OSFP', 800, 'SMF', 0,    500, 1310, NULL, 'MPO-12', '802.3df',   '800GBASE-DR8 OSFP'),
+  ('OSFP', 800, 'SMF', 0,   2000, 1310, NULL, 'LC',     '802.3df',   '800GBASE-FR4 OSFP'),
+  ('OSFP', 800, 'SMF', 0,  10000, 1310, NULL, 'LC',     '802.3df',   '800GBASE-LR4 OSFP'),
+  ('OSFP', 800, 'SMF', 0,  80000, 1550, NULL, 'LC',     'OpenZR+',   '800G ZR OSFP 80km'),
+-- ── OSFP 1.6T (IEEE 802.3dj draft) ──────────────────────────────────────────
+  ('OSFP', 1600, 'SMF', 0,   500, 1310, NULL, 'MPO-16', '802.3dj',   '1.6TBASE-DR16 OSFP'),
+  ('OSFP', 1600, 'SMF', 0,  2000, 1310, NULL, 'LC',     '802.3dj',   '1.6TBASE-FR4 OSFP'),
+  ('OSFP', 1600, 'SMF', 0, 10000, 1310, NULL, 'LC',     '802.3dj',   '1.6TBASE-LR4 OSFP'),
+  ('OSFP112', 800, 'SMF', 0, 10000, 1310, NULL, 'LC',   '802.3df',   '800GBASE-LR4 OSFP112'),
+  ('OSFP112', 800, 'SMF', 0, 80000, 1550, NULL, 'LC',   'OpenZR+',   '800G ZR OSFP112 80km'),
+  ('OSFP112', 800, 'SMF', 0, 120000,1550, NULL, 'LC',   'OpenZR+',   '800G OpenZR+ OSFP112 120km'),
+-- ── CFP2 100G coherent ───────────────────────────────────────────────────────
+  ('CFP2', 100, 'SMF', 0,  10000, 1310, NULL, 'LC',     'OIF-100G',  '100GBASE-LR4 CFP2'),
+  ('CFP2', 100, 'SMF', 0,  80000, 1550, NULL, 'LC',     'OIF-100G',  '100G ZR CFP2 80km'),
+  ('CFP2', 100, 'SMF', 0, 120000, 1550, NULL, 'LC',     'OpenZR+',   '100G OpenZR+ CFP2'),
+-- ── SFP+ / SFP 1G non-standard reaches ───────────────────────────────────────
+  ('SFP', 1, 'SMF', 0, 20000, 1310, NULL, 'LC', '802.3z',    '1000BASE-LH 20km'),
+  ('SFP', 1, 'SMF', 0, 60000, 1310, NULL, 'LC', '802.3z',    '1000BASE-LH 60km'),
+  ('SFP', 1, 'SMF', 0, 80000, 1550, NULL, 'LC', '802.3z',    '1000BASE-ZX 80km'),
+  ('SFP', 1, 'SMF', 0,100000, 1550, NULL, 'LC', '802.3z',    '1000BASE-ZX 100km'),
+-- ── SFP+ 10G non-standard reaches ────────────────────────────────────────────
+  ('SFP+', 10, 'SMF', 0, 20000, 1310, NULL, 'LC', '802.3ae',   '10GBASE-LR 20km variant'),
+  ('SFP+', 10, 'SMF', 0, 60000, 1550, NULL, 'LC', '802.3ae',   '10GBASE-ZR 60km'),
+  ('SFP+', 10, 'SMF', 0, 80000, 1550, NULL, 'LC', '802.3ae',   '10GBASE-ZR 80km'),
+  ('SFP+', 10, 'SMF', 0,100000, 1550, NULL, 'LC', '802.3ae',   '10GBASE-ZR 100km'),
+-- ── XFP 10G ──────────────────────────────────────────────────────────────────
+  ('XFP', 10, 'MMF', 0,    300,  850, NULL, 'LC', '802.3ae',   '10GBASE-SR XFP'),
+  ('XFP', 10, 'SMF', 0,  10000, 1310, NULL, 'LC', '802.3ae',   '10GBASE-LR XFP'),
+  ('XFP', 10, 'SMF', 0,  40000, 1310, NULL, 'LC', '802.3ae',   '10GBASE-ER XFP'),
+  ('XFP', 10, 'SMF', 0,  80000, 1550, NULL, 'LC', '802.3ae',   '10GBASE-ZR XFP')
+ON CONFLICT DO NOTHING;
+
+-- ── Re-run IEEE lookup for wavelength after new entries ──────────────────────
+UPDATE transceivers t SET
+  wavelength_tx_nm = (
+    SELECT il.wavelength_tx_nm
+    FROM ieee_wavelength_lookup il
+    WHERE UPPER(il.form_factor) = UPPER(t.form_factor)
+      AND il.speed_gbps = ROUND(t.speed_gbps::NUMERIC, 2)
+      AND UPPER(il.fiber_type) = UPPER(t.fiber_type)
+      AND il.reach_min_m <= t.reach_meters
+      AND il.reach_max_m >= t.reach_meters
+      AND il.wavelength_tx_nm IS NOT NULL
+    ORDER BY il.reach_max_m ASC
+    LIMIT 1
+  ),
+  wavelength_rx_nm = COALESCE(
+    wavelength_rx_nm,
+    (
+      SELECT il.wavelength_rx_nm
+      FROM ieee_wavelength_lookup il
+      WHERE UPPER(il.form_factor) = UPPER(t.form_factor)
+        AND il.speed_gbps = ROUND(t.speed_gbps::NUMERIC, 2)
+        AND UPPER(il.fiber_type) = UPPER(t.fiber_type)
+        AND il.reach_min_m <= t.reach_meters
+        AND il.reach_max_m >= t.reach_meters
+      ORDER BY il.reach_max_m ASC
+      LIMIT 1
+    )
+  ),
+  connector_type = COALESCE(
+    connector_type,
+    (
+      SELECT il.connector_type
+      FROM ieee_wavelength_lookup il
+      WHERE UPPER(il.form_factor) = UPPER(t.form_factor)
+        AND il.speed_gbps = ROUND(t.speed_gbps::NUMERIC, 2)
+        AND UPPER(il.fiber_type) = UPPER(t.fiber_type)
+        AND il.reach_min_m <= t.reach_meters
+        AND il.reach_max_m >= t.reach_meters
+      ORDER BY il.reach_max_m ASC
+      LIMIT 1
+    )
+  )
+WHERE t.wavelength_tx_nm IS NULL
+  AND t.form_factor IS NOT NULL
+  AND t.speed_gbps IS NOT NULL
+  AND t.fiber_type IS NOT NULL
+  AND t.fiber_type NOT IN ('Copper', 'DAC', 'AOC', 'COPPER')
+  AND t.reach_meters IS NOT NULL
+  AND t.reach_meters > 0;
+
+-- ── Part B: Fallback wavelength by fiber_type for remaining ──────────────────
+-- Conservative rule: SMF products with reach > 80km → 1550nm (ZR/coherent)
+-- All other SMF → 1310nm (covers ER/LR/DR/FR/LH etc.)
+-- All MMF → 850nm (SR variants)
+-- Products with DAC fiber_type: no optical wavelength (leave NULL)
+
+UPDATE transceivers SET
+  wavelength_tx_nm = CASE
+    -- Long-reach SMF: reach > 80km → 1550nm (ZR, coherent)
+    WHEN UPPER(fiber_type) = 'SMF' AND reach_meters > 80000 THEN 1550
+    -- Standard SMF: 1310nm (LR/ER/DR/FR/LH etc.)
+    WHEN UPPER(fiber_type) = 'SMF' AND reach_meters > 0 THEN 1310
+    -- Short MMF: 850nm (SR variants)
+    WHEN UPPER(fiber_type) = 'MMF' AND reach_meters > 0 THEN 850
+    ELSE wavelength_tx_nm
+  END
+WHERE wavelength_tx_nm IS NULL
+  AND fiber_type IS NOT NULL
+  AND UPPER(fiber_type) IN ('SMF', 'MMF')
+  AND reach_meters IS NOT NULL
+  AND reach_meters > 0
+  AND form_factor IS NOT NULL
+  AND UPPER(form_factor) NOT IN ('LC', 'SC', 'DAC', 'TRANSCEIVER', 'PLUGGABLE', 'VARIES');
+
+-- ── Completeness final update ─────────────────────────────────────────────────
+UPDATE transceivers SET
+  data_completeness = calc_data_completeness(
+    form_factor, speed_gbps, fiber_type,
+    reach_meters, wavelength_tx_nm, connector_type
+  ),
+  enrichment_needed = (
+    form_factor IS NULL OR speed_gbps IS NULL OR
+    fiber_type IS NULL OR reach_meters IS NULL OR
+    wavelength_tx_nm IS NULL OR connector_type IS NULL
+  ),
+  enrichment_fields = ARRAY_REMOVE(ARRAY[
+    CASE WHEN form_factor IS NULL THEN 'form_factor' END,
+    CASE WHEN speed_gbps IS NULL THEN 'speed_gbps' END,
+    CASE WHEN fiber_type IS NULL THEN 'fiber_type' END,
+    CASE WHEN reach_meters IS NULL OR reach_meters = 0 THEN 'reach_meters' END,
+    CASE WHEN wavelength_tx_nm IS NULL THEN 'wavelength_tx_nm' END,
+    CASE WHEN connector_type IS NULL THEN 'connector_type' END
+  ], NULL);
+
+-- ── Part C: Clear pending queue ───────────────────────────────────────────────
+-- All pending records from confidence-based matcher are superseded.
+-- Deterministic matcher (maintenance:find-equivalences) will re-generate
+-- correct matches at confidence=1.0 for products with complete data.
+UPDATE transceiver_equivalences
+SET status          = 'rejected',
+    reject_reason   = 'Superseded by deterministic matcher — confidence-based pending removed in migration 114',
+    reviewed_at     = NOW(),
+    reviewed_by     = 'system:migration-114'
+WHERE status = 'pending';
+
+-- ── Final Statistics ─────────────────────────────────────────────────────────
+DO $$
+DECLARE
+  total_cnt        INTEGER;
+  complete_cnt     INTEGER;
+  missing_conn     INTEGER;
+  missing_wl       INTEGER;
+  fx_complete      INTEGER;
+  fx_total         INTEGER;
+  pending_cnt      INTEGER;
+BEGIN
+  SELECT COUNT(*) INTO total_cnt FROM transceivers;
+  SELECT COUNT(*) INTO complete_cnt FROM transceivers WHERE enrichment_needed = FALSE;
+  SELECT COUNT(*) INTO missing_conn FROM transceivers WHERE connector_type IS NULL;
+  SELECT COUNT(*) INTO missing_wl   FROM transceivers WHERE wavelength_tx_nm IS NULL;
+  SELECT COUNT(*) INTO pending_cnt  FROM transceiver_equivalences WHERE status = 'pending';
+  SELECT COUNT(*) INTO fx_total
+    FROM transceivers t JOIN vendors v ON v.id = t.vendor_id
+    WHERE UPPER(v.name) LIKE '%FLEXOPTIX%';
+  SELECT COUNT(*) INTO fx_complete
+    FROM transceivers t JOIN vendors v ON v.id = t.vendor_id
+    WHERE UPPER(v.name) LIKE '%FLEXOPTIX%' AND enrichment_needed = FALSE;
+
+  RAISE NOTICE 'Migration 114 complete:';
+  RAISE NOTICE '  Total transceivers:       %', total_cnt;
+  RAISE NOTICE '  Fully complete:           %', complete_cnt;
+  RAISE NOTICE '  Still missing connector:  %', missing_conn;
+  RAISE NOTICE '  Still missing wavelength: %', missing_wl;
+  RAISE NOTICE '  Flexoptix fully complete: % / %', fx_complete, fx_total;
+  RAISE NOTICE '  Pending queue:            % (target: 0)', pending_cnt;
+END $$;