@sembl/core 0.2.0 → 0.3.0

package/dist/index.cjs

@@ -45,6 +45,7 @@ __export(index_exports, {
   field: () => field,
   isCoerceInput: () => isCoerceInput,
   isSource: () => isSource,
+  normalizeInstructions: () => normalizeInstructions,
   partialCoerce: () => partialCoerce,
   partialCoerceWithProvenance: () => partialCoerceWithProvenance,
   provenanceInstructions: () => provenanceInstructions,
@@ -507,6 +508,14 @@ var SOURCE_INSTRUCTIONS = [
 ].join("\n");
 
 // src/coerce/prompt-builder.ts
+function normalizeInstructions(instructions) {
+  if (instructions === void 0) return [];
+  const list = typeof instructions === "string" ? [instructions] : instructions;
+  if (!Array.isArray(list) || list.some((entry) => typeof entry !== "string")) {
+    throw new RangeError("instructions must be a string or an array of strings");
+  }
+  return list.map((entry) => entry.trim()).filter((entry) => entry.length > 0);
+}
 function describeConstraints(constraints) {
   if (!constraints) {
     return [];
@@ -629,6 +638,14 @@ function buildPrompt(schema, bundle, options = {}) {
   lines.push("- Interpret the user's input semantically \u2014 infer meaning, don't just pattern match.");
   lines.push("- Respect every stated limit exactly; truncate or drop lower-priority content to stay within it.");
   lines.push("- Return only the structured JSON output matching the schema.");
+  const instructions = normalizeInstructions(options.instructions);
+  if (instructions.length > 0) {
+    lines.push("");
+    lines.push("Additional guidance for this extraction:");
+    for (const instruction of instructions) {
+      lines.push(`- ${instruction}`);
+    }
+  }
   return lines.join("\n");
 }
 
@@ -1303,6 +1320,7 @@ async function runCoercion(input, options, { mode, provenance }) {
       `maxRepairAttempts must be a non-negative integer, got ${String(options.maxRepairAttempts)}`
     );
   }
+  const instructions = normalizeInstructions(options.instructions);
   const onInvalidField = options.onInvalidField ?? "throw";
   if (!INVALID_FIELD_POLICIES.includes(onInvalidField)) {
     throw new RangeError(
@@ -1333,11 +1351,12 @@ async function runCoercion(input, options, { mode, provenance }) {
       rootSpan
     );
     const promptSpan = tracer.startSpan("buildPrompt", {}, rootSpan);
-    const basePrompt = buildPrompt(schema, bundle, { resolvedEnums });
+    const basePrompt = buildPrompt(schema, bundle, { resolvedEnums, instructions });
     const systemPrompt = provenance ? `${basePrompt}
 ${provenanceInstructions({ sourceLabels })}` : basePrompt;
     tracer.addEvent(promptSpan, "promptBuilt", {
-      promptLength: systemPrompt.length
+      promptLength: systemPrompt.length,
+      instructionCount: instructions.length
     });
     tracer.endSpan(promptSpan);
     const request = provenance ? toProvenanceSchema(schema, bundle, { sourceLabels }) : { schema, bundle };
@@ -1623,6 +1642,7 @@ function resolveConfig(callConfig) {
     traceSinks: callConfig?.traceSinks ?? global.traceSinks,
     maxRepairAttempts: callConfig?.maxRepairAttempts ?? global.maxRepairAttempts,
     onInvalidField: callConfig?.onInvalidField ?? global.onInvalidField,
+    instructions: callConfig?.instructions ?? global.instructions,
     maxInputChars: callConfig?.maxInputChars ?? global.maxInputChars,
     truncate: callConfig?.truncate ?? global.truncate,
     preprocess: callConfig?.preprocess ?? global.preprocess
@@ -1656,6 +1676,7 @@ var Coercible = class _Coercible {
       traceSinks: this._config.traceSinks,
       maxRepairAttempts: this._config.maxRepairAttempts,
       onInvalidField: this._config.onInvalidField,
+      instructions: this._config.instructions,
       maxInputChars: this._config.maxInputChars,
       truncate: this._config.truncate,
       preprocess: this._config.preprocess
@@ -1741,6 +1762,7 @@ var ConsoleSink = class {
   field,
   isCoerceInput,
   isSource,
+  normalizeInstructions,
   partialCoerce,
   partialCoerceWithProvenance,
   provenanceInstructions,