fixes

apps/api/internal/company/weak_desc.go

@@ -0,0 +1,151 @@
+package company
+
+import "strings"
+
+// minUsableProductDescRunes is the floor for a prior description that may hash-skip
+// the enhance LLM. Shorter copy is treated as thin and forced through enhance.
+const minUsableProductDescRunes = 40
+
+// shortBoilerplateDescRunes: short copy without product-fact token overlap is weak
+// even when longer than minUsableProductDescRunes (generic one-liners).
+const shortBoilerplateDescRunes = 96
+
+// weakFillerPhrases are known heuristic / placeholder snippets that must never
+// hash-skip enhance (mock-llm / invent fallbacks historically produced these).
+var weakFillerPhrases = []string{
+	"ready for retail listing",
+	"quality product ready",
+	"product description",
+	"based on available specifications",
+	"with available specifications",
+	"available catalog details",
+	"pripravljeno za prodajo",
+	"na podlagi razpoložljivih specifikacij",
+}
+
+// IsWeakPriorEnhanceDescription reports empty, too-short, title-echo, known filler,
+// or short boilerplate without overlapping tokens from title/fact sources.
+func IsWeakPriorEnhanceDescription(priorDesc string, titles ...string) bool {
+	priorDesc = strings.TrimSpace(priorDesc)
+	if priorDesc == "" || priorDesc == "<nil>" {
+		return true
+	}
+	if len([]rune(priorDesc)) < minUsableProductDescRunes {
+		return true
+	}
+	for _, title := range titles {
+		if DescriptionEchoesTitle(priorDesc, title) {
+			return true
+		}
+	}
+	if ContainsWeakFillerPhrase(priorDesc) {
+		return true
+	}
+	if len([]rune(priorDesc)) <= shortBoilerplateDescRunes &&
+		!descriptionOverlapsProductFacts(priorDesc, titles...) {
+		return true
+	}
+	return false
+}
+
+// ContainsWeakFillerPhrase reports known placeholder / invent-fallback snippets.
+func ContainsWeakFillerPhrase(desc string) bool {
+	lower := strings.ToLower(desc)
+	for _, p := range weakFillerPhrases {
+		if p != "" && strings.Contains(lower, p) {
+			return true
+		}
+	}
+	return false
+}
+
+var weakDescStopTokens = map[string]struct{}{
+	"a": {}, "an": {}, "the": {}, "and": {}, "or": {}, "of": {}, "in": {}, "on": {}, "for": {},
+	"to": {}, "with": {}, "from": {}, "by": {}, "is": {}, "are": {}, "this": {}, "that": {},
+	"product": {}, "products": {}, "category": {}, "description": {}, "specs": {}, "spec": {},
+	"key": {}, "je": {}, "v": {}, "za": {}, "z": {}, "iz": {}, "ter": {}, "ali": {},
+}
+
+func significantDescTokens(s string) map[string]struct{} {
+	s = strings.ToLower(s)
+	out := map[string]struct{}{}
+	var b strings.Builder
+	flush := func() {
+		tok := b.String()
+		b.Reset()
+		if len(tok) < 3 {
+			return
+		}
+		if _, stop := weakDescStopTokens[tok]; stop {
+			return
+		}
+		out[tok] = struct{}{}
+	}
+	for _, r := range s {
+		if (r >= 'a' && r <= 'z') || (r >= '0' && r <= '9') || r == '_' {
+			b.WriteRune(r)
+			continue
+		}
+		// Keep Latin-extended letters (Slovenian čšž etc.) via unicode letter-ish: non-space punctuation splits.
+		if r > 127 && ((r >= 'à' && r <= 'ÿ') || r == 'č' || r == 'š' || r == 'ž' || r == 'ć' || r == 'đ') {
+			b.WriteRune(r)
+			continue
+		}
+		flush()
+	}
+	flush()
+	return out
+}
+
+// descriptionOverlapsProductFacts is true when desc shares significant tokens with
+// any fact source (title, brand, model, …).
+func descriptionOverlapsProductFacts(desc string, factSources ...string) bool {
+	descToks := significantDescTokens(desc)
+	if len(descToks) == 0 {
+		return false
+	}
+	overlap := 0
+	seen := map[string]struct{}{}
+	for _, src := range factSources {
+		for tok := range significantDescTokens(src) {
+			if _, ok := descToks[tok]; !ok {
+				continue
+			}
+			if _, dup := seen[tok]; dup {
+				continue
+			}
+			seen[tok] = struct{}{}
+			overlap++
+			if overlap >= 2 || len(tok) >= 5 {
+				return true
+			}
+		}
+	}
+	return overlap >= 1 && len(seen) >= 1 && len([]rune(desc)) >= minUsableProductDescRunes+20
+}
+
+func normalizeForEchoCompare(s string) string {
+	return strings.Join(strings.Fields(strings.ToLower(strings.TrimSpace(s))), " ")
+}
+
+// DescriptionEchoesTitle is true when desc is EqualFold or near-equal to title
+// (whitespace-normalized, or one contains the other with only a tiny length delta).
+func DescriptionEchoesTitle(desc, title string) bool {
+	d := normalizeForEchoCompare(desc)
+	t := normalizeForEchoCompare(title)
+	if d == "" || t == "" {
+		return false
+	}
+	if d == t {
+		return true
+	}
+	shorter, longer := d, t
+	if len(d) > len(t) {
+		shorter, longer = t, d
+	}
+	if !strings.Contains(longer, shorter) {
+		return false
+	}
+	delta := len(longer) - len(shorter)
+	return delta <= 8 && float64(len(shorter))/float64(len(longer)) >= 0.85
+}