This commit is contained in:
2026-08-23 13:39:56 +02:00
parent 89ae7e5fa8
commit 815e26d359
7 changed files with 283 additions and 8 deletions
+73 -5
View File
@@ -8,7 +8,9 @@ import (
"sort"
"strings"
"time"
"unicode"
"github.com/descrybe/descrybe-v2/apps/api/internal/aiprompts"
"github.com/descrybe/descrybe-v2/apps/api/internal/company"
"github.com/descrybe/descrybe-v2/apps/api/internal/eprel"
)
@@ -739,6 +741,10 @@ func appendStepLog(gpt map[string]any, name string, raw any) {
// category description_template (short prose / title echo instead of multi-section HTML).
const descriptionFormulaRetrySuffix = "\n\nINVALID DESCRIPTION. Your JSON ignored the Description formula. Reply with ONLY one JSON object; \"description\" must be ONE HTML string covering each formula section in order with matching tags (h1/h2/h3/h4, p, ul)."
// titleRewriteRetrySuffix is appended when the Title role requires a new retail
// name but the model echoed the supplier Name unchanged.
const titleRewriteRetrySuffix = "\n\nINVALID NAME. Category Title instructions require a NEW product name (formula / rewrite). \"name\" must NOT equal the supplier Name. Reply with ONLY one JSON object; rebuild \"name\" from type + brand + model + color (or the Title formula) in {{language}}."
// descriptionNeedsEnhanceRepair is true when desc is weak/empty/title-echo or
// fails an active category description_template (A1 multi-section HTML).
// Heuristic invent/synth is intentionally excluded here — enhanceHashForceReason
@@ -750,6 +756,47 @@ func descriptionNeedsEnhanceRepair(desc string, template any, titles ...string)
return !descriptionSatisfiesFormula(desc, template)
}
// titlesAreEquivalent treats supplier vs enriched names as the same when only
// case/spacing/punctuation differ — used to detect Title formula copy-paste.
func titlesAreEquivalent(a, b string) bool {
na := normalizeTitleCompare(a)
nb := normalizeTitleCompare(b)
if na == "" || nb == "" {
return false
}
return na == nb
}
func normalizeTitleCompare(s string) string {
s = strings.TrimSpace(strings.ToLower(s))
if s == "" || s == "<nil>" {
return ""
}
var b strings.Builder
b.Grow(len(s))
prevSpace := false
for _, r := range s {
switch {
case unicode.IsLetter(r) || unicode.IsDigit(r):
b.WriteRune(r)
prevSpace = false
case unicode.IsSpace(r) || r == '-' || r == '_' || r == '/' || r == ',':
if !prevSpace && b.Len() > 0 {
b.WriteByte(' ')
prevSpace = true
}
}
}
return strings.TrimSpace(b.String())
}
func titleNeedsRewriteRetry(in ProductInput, name string) bool {
if !aiprompts.CategoryTitlePromptRequiresRewrite(in.CategoryEnhancePrompt) {
return false
}
return titlesAreEquivalent(name, in.Name)
}
// enhanceHashForceReason returns why a matching prior enhance_input_hash must not
// skip the LLM (empty = safe to reuse as ai_enhance_unchanged).
func enhanceHashForceReason(in ProductInput) string {
@@ -762,10 +809,18 @@ func enhanceHashForceReason(in ProductInput) string {
if !descriptionSatisfiesFormula(in.PriorProcessedDescription, in.DescriptionTemplate) {
return "formula-mismatch"
}
if titleNeedsRewriteRetry(in, in.PriorProcessedName) {
return "title-copy-paste"
}
return ""
}
func (e *Engine) enhance(ctx context.Context, in ProductInput, category string, attrs map[string]any) (string, string, int, any, error) {
// Prompt Description HTML overlays count as the formula when description_template
// is empty — otherwise enhance accepts supplier-like prose and skips formula retry.
if effective := aiprompts.EffectiveDescriptionTemplateAny(in.DescriptionTemplate, in.CategoryEnhancePrompt); effective != nil {
in.DescriptionTemplate = effective
}
catUID := strings.TrimSpace(in.CategoryUniqueID)
catName := strings.TrimSpace(category)
sysTpl, userTpl := resolveProductPromptTemplates(in)
@@ -885,11 +940,19 @@ func (e *Engine) enhance(ctx context.Context, in ProductInput, category string,
// Quality-gate on LLM description alone (do not absorb originals yet).
desc := preferredProductDescription(name, llmDesc)
didFormulaRetry := false
// Fast garbage that ignores A1 description_template: one formula-aware retry, then synthesize.
if !descriptionSatisfiesFormula(desc, in.DescriptionTemplate) &&
FormatDescriptionFormulaConstraint(in.DescriptionTemplate) != "" {
needsDescRetry := !descriptionSatisfiesFormula(desc, in.DescriptionTemplate) &&
FormatDescriptionFormulaConstraint(in.DescriptionTemplate) != ""
needsTitleRetry := titleNeedsRewriteRetry(in, name)
// One retry when Description HTML formula and/or Title rewrite were ignored.
if needsDescRetry || needsTitleRetry {
didFormulaRetry = true
retryUser := user + descriptionFormulaRetrySuffix
retryUser := user
if needsDescRetry {
retryUser += descriptionFormulaRetrySuffix
}
if needsTitleRetry {
retryUser += titleRewriteRetrySuffix
}
retryStarted := time.Now()
comp2, obj2, err2 := CompleteJSON(ctx, e.Completer, system, retryUser, CompleteOptions{
MaxTokens: MaxTokensEnhance,
@@ -904,7 +967,12 @@ func (e *Engine) enhance(ctx context.Context, in ProductInput, category string,
n2 := SanitizeOutput(fmt.Sprint(obj2["name"]))
d2 := SanitizeOutput(fmt.Sprint(obj2["description"]))
if llmEnhanceHardRefuseReason(n2, d2) == "" {
name = preferredProductTitle(in.GTIN, n2, name, in.Name, in.PriorProcessedName)
// Prefer rewritten title over supplier echo when Title role requires it.
if needsTitleRetry && n2 != "" && !titlesAreEquivalent(n2, in.Name) {
name = preferredProductTitle(in.GTIN, n2)
} else {
name = preferredProductTitle(in.GTIN, n2, name, in.Name, in.PriorProcessedName)
}
desc = preferredProductDescription(name, d2)
mt2, md2 := metaFieldsFromEnhanceObj(obj2)
if mt2 != "" {