@setzkasten-cms/astro-admin 0.6.0

package/src/init/field-label-enricher.ts

@@ -0,0 +1,72 @@
+/**
+ * Phase 2: Cosmetic label enrichment for inner fields of repeated elements.
+ *
+ * This module is purely cosmetic — it sets `.label` on InnerFieldInfo objects.
+ * It NEVER changes keys, types, positions, or structural data.
+ *
+ * Rules are fuzzy and best-effort. If no rule matches, the generic key remains.
+ * All text analysis ignores special characters (€, /, <, >) and works only
+ * with letters, digits, and whitespace.
+ */
+
+import type { InnerFieldInfo } from './analyzer-types.js'
+
+/**
+ * Enrich fields with human-friendly labels based on tag context and text patterns.
+ * Mutates `field.label` in place.
+ */
+export function enrichFieldLabels(fields: InnerFieldInfo[]): void {
+  for (const field of fields) {
+    const label = inferLabel(field)
+    if (label) field.label = label
+  }
+}
+
+function inferLabel(field: InnerFieldInfo): string | undefined {
+  // 1. HTML tag context (most reliable)
+  if (/^h[1-6]$/.test(field.tag)) return 'Überschrift'
+  if (field.type === 'array') return 'Liste'
+  if (field.tag === 'a' && field.type === 'link') return 'Link'
+  if (field.tag === 'a' && field.type === 'text') return 'Link-Text'
+
+  // 2. Text content analysis — use ORIGINAL values (with special chars) for some checks
+  const rawValues = field.defaultValues.filter((v): v is string => typeof v === 'string')
+  const cleanValues = rawValues.map(v => stripSpecialChars(v))
+
+  if (rawValues.length === 0) return undefined
+
+  // Short single-word text (all instances) → Label / Overline / Name
+  if (rawValues.every(v => v.trim().length > 0 && v.trim().length < 20 && !v.includes(' '))) {
+    return 'Label'
+  }
+
+  // Contains currency symbols in ANY value + all short → Preis
+  // (e.g. "9 €", "29 €", "Auf Anfrage" — the last is a price placeholder)
+  if (rawValues.some(v => /[€$£¥₹]/.test(v)) && rawValues.every(v => v.length < 25)) {
+    return 'Preis'
+  }
+  // All values contain digits and are short → Preis
+  if (rawValues.every(v => /\d/.test(v) && v.length < 25)) {
+    return 'Preis'
+  }
+
+  // Long text → Beschreibung
+  if (cleanValues.every(v => v.length > 40)) {
+    return 'Beschreibung'
+  }
+
+  // Medium text in <p> → Beschreibung
+  if (field.tag === 'p' && cleanValues.every(v => v.length > 15)) {
+    return 'Beschreibung'
+  }
+
+  return undefined
+}
+
+/**
+ * Strip everything except letters, digits, umlauts, and whitespace.
+ * This makes text comparison robust against currency symbols, slashes, etc.
+ */
+function stripSpecialChars(str: string): string {
+  return str.replace(/[^a-zA-Z0-9äöüÄÖÜß\s]/g, '').trim()
+}