inou/lib/normalize.go

package lib

import (
	"encoding/json"
	"fmt"
	"log"
	"strings"
)

// Normalize resolves hospital test names to official LOINC codes and updates entries.
// Flow: hospital name + specimen + unit → LOINC (via cache or Gemini lookup) → official name from loinc_lab.
// No Fireworks LLM. Original Type field is never modified.
func Normalize(dossierID string, category int, progress ...func(processed, total int)) error {
	reportProgress := func(p, t int) {
		if len(progress) > 0 && progress[0] != nil {
			progress[0](p, t)
		}
	}

	// 1. Load all entries, build parent map for specimen lookup
	entries, err := EntryQueryOld(dossierID, category, "")
	if err != nil {
		return fmt.Errorf("load entries: %w", err)
	}

	parentMap := make(map[string]*Entry)
	for _, e := range entries {
		if e.ParentID == "" || e.Type == "lab_order" {
			parentMap[e.EntryID] = e
		}
	}

	// 2. Collect unique type|specimen|unit combos, resolve each to LOINC
	type testKey struct {
		name, specimen, unit string
	}
	type resolved struct {
		loinc string
		info  *LoincInfo
		abbr  string
	}
	cache := make(map[testKey]*resolved)
	var lookupCount, cacheHits, misses int

	for _, e := range entries {
		if e.ParentID == "" || e.Type == "lab_order" || e.Type == "" {
			continue
		}
		var data map[string]interface{}
		json.Unmarshal([]byte(e.Data), &data)
		unit, _ := data["unit"].(string)

		specimen := ""
		if parent, ok := parentMap[e.ParentID]; ok {
			var pdata map[string]interface{}
			if json.Unmarshal([]byte(parent.Data), &pdata) == nil {
				specimen, _ = pdata["specimen"].(string)
			}
		}

		tk := testKey{e.Type, specimen, unit}
		if _, ok := cache[tk]; ok {
			continue
		}
		lookupCount++

		loinc := LoincLookup(e.Type, specimen, unit)
		if loinc == "" {
			cache[tk] = &resolved{}
			misses++
			continue
		}

		info := LoincGet(loinc)
		abbr := LoincAbbr(info)
		cache[tk] = &resolved{loinc: loinc, info: info, abbr: abbr}
		if info != nil {
			cacheHits++
		}
	}

	reportProgress(lookupCount, lookupCount)
	log.Printf("normalize: %d unique combos, %d resolved, %d unresolved", lookupCount, cacheHits, misses)

	// 3. Apply to entries
	var toSave []Entry
	for _, e := range entries {
		if e.ParentID == "" || e.Type == "lab_order" || e.Type == "" {
			continue
		}
		var data map[string]interface{}
		if json.Unmarshal([]byte(e.Data), &data) != nil {
			data = make(map[string]interface{})
		}
		unit, _ := data["unit"].(string)

		specimen := ""
		if parent, ok := parentMap[e.ParentID]; ok {
			var pdata map[string]interface{}
			if json.Unmarshal([]byte(parent.Data), &pdata) == nil {
				specimen, _ = pdata["specimen"].(string)
			}
		}

		r := cache[testKey{e.Type, specimen, unit}]
		if r == nil || r.loinc == "" || r.info == nil {
			continue
		}

		// Check if already up to date
		existingLoinc, _ := data["loinc"].(string)
		existingName, _ := data["normalized_name"].(string)
		if existingLoinc == r.loinc && existingName == r.info.LongName && e.SearchKey == r.loinc && e.SearchKey2 != "" {
			continue
		}

		// Update Data JSON
		data["loinc"] = r.loinc
		data["normalized_name"] = r.info.LongName
		data["abbreviation"] = r.abbr
		b, _ := json.Marshal(data)
		e.Data = string(b)

		// Update search keys
		e.SearchKey = r.loinc
		e.SearchKey2 = strings.ToLower(r.info.LongName)

		// Rebuild Summary: "Abbr: value unit"
		summary := r.abbr + ": " + e.Value
		if unit != "" {
			summary += " " + unit
		}
		e.Summary = summary

		toSave = append(toSave, *e)
	}

	if len(toSave) == 0 {
		log.Printf("normalize: no changes needed")
		return nil
	}

	log.Printf("normalize: updating %d entries", len(toSave))
	ptrs := make([]*Entry, len(toSave))
	for i := range toSave {
		ptrs[i] = &toSave[i]
	}
	return EntryWrite("", ptrs...)
}