aiex-cli 0.1.1-beta.8 → 0.1.1-beta.9

package/dist/cli.mjs

@@ -1,4 +1,4 @@
-import { A as seedConfig, C as CORRECTION_SYSTEM_PROMPT, D as PLACEHOLDER_TEXT, E as PLACEHOLDER_SCHEMA, M as name, N as package_default, O as buildCorrectionUserPrompt, P as version, S as DEFAULT_MINERU_CONFIG, T as EVIDENCE_INSTRUCTIONS, _ as doctorDiagnosticsSeverityRows, a as recognizeImageText, b as DEFAULT_LITEPARSE_CONFIG, c as t, d as readAIConfig, f as writeAIConfig, h as parseJsonSchema, j as description, k as createConfig, l as createProjectDatabase, m as JsonSchemaDefinitionSchema, n as collectDoctorDiagnostics, o as shouldUseImageOcrFallback, p as AIConfigSchema, r as createMigrationConfig, s as initI18n, t as generateDrizzleSchema, u as getDefaultAIConfig, v as doctorDiagnosticsTableRows, w as DEFAULT_PROMPT_CONFIG, x as DEFAULT_MINERU_API_CONFIG, y as formatDoctorDiagnosticsJson } from "./generate-drizzle-schema-BcStmKUc.mjs";
+import { A as seedConfig, C as CORRECTION_SYSTEM_PROMPT, D as PLACEHOLDER_TEXT, E as PLACEHOLDER_SCHEMA, M as name, N as package_default, O as buildCorrectionUserPrompt, P as version, S as DEFAULT_MINERU_CONFIG, T as EVIDENCE_INSTRUCTIONS, _ as doctorDiagnosticsSeverityRows, a as recognizeImageText, b as DEFAULT_LITEPARSE_CONFIG, c as t, d as readAIConfig, f as writeAIConfig, h as parseJsonSchema, j as description, k as createConfig, l as createProjectDatabase, m as JsonSchemaDefinitionSchema, n as collectDoctorDiagnostics, o as shouldUseImageOcrFallback, p as AIConfigSchema, r as createMigrationConfig, s as initI18n, t as generateDrizzleSchema, u as getDefaultAIConfig, v as doctorDiagnosticsTableRows, w as DEFAULT_PROMPT_CONFIG, x as DEFAULT_MINERU_API_CONFIG, y as formatDoctorDiagnosticsJson } from "./generate-drizzle-schema-BdkbSP5F.mjs";
 import { createRequire } from "node:module";
 import fs from "node:fs/promises";
 import os from "node:os";
@@ -1256,6 +1256,19 @@ function verifyFieldEvidence(input) {
 	}
 	return Object.keys(verified).length > 0 ? verified : void 0;
 }
+function findInvalidFieldEvidence(input) {
+	if (!input.text || !isRecord$3(input.data) || !input.rawEvidence) return [];
+	const invalid = [];
+	for (const [field, raw] of Object.entries(input.rawEvidence)) {
+		const property = input.schema.properties[field];
+		if (!property) continue;
+		if (property.type !== "string" && property.type !== "number" && property.type !== "integer") continue;
+		if (typeof raw.quote !== "string" || raw.quote.trim().length === 0) continue;
+		if (!input.text.includes(raw.quote)) continue;
+		if (!quoteContainsValue(raw.quote, input.data[field], property)) invalid.push(field);
+	}
+	return invalid;
+}
 function isEvidenceEligibleProperty(property) {
 	return property.type === "string" || property.type === "number" || property.type === "integer";
 }
@@ -1266,6 +1279,7 @@ function buildFieldEvidenceQuality(input) {
 	const unsupportedFields = [];
 	const missingFields = [];
 	const invalidFields = [];
+	const invalidEvidenceFields = new Set(input.invalidEvidenceFields ?? []);
 	for (const [field, property] of Object.entries(input.schema.properties)) {
 		if (!isEvidenceEligibleProperty(property)) continue;
 		const value = input.data[field];
@@ -1273,12 +1287,15 @@ function buildFieldEvidenceQuality(input) {
 		if (value === null || value === void 0) {
 			status = "missing";
 			missingFields.push(field);
+		} else if (invalidEvidenceFields.has(field)) {
+			status = "invalid";
+			invalidFields.push(field);
 		} else if (input.verifiedEvidence?.[field]) {
 			status = "supported";
 			supportedFields.push(field);
 		} else if (input.rawEvidence?.[field]) {
-			status = "invalid";
-			invalidFields.push(field);
+			status = "unsupported";
+			unsupportedFields.push(field);
 		} else {
 			status = "unsupported";
 			unsupportedFields.push(field);
@@ -1900,11 +1917,18 @@ async function extractStructuredData(input) {
 						data: businessData,
 						rawEvidence: stripped.rawEvidence
 					}) : void 0;
+					const invalidEvidenceFields = canLocateEvidence ? findInvalidFieldEvidence({
+						schema,
+						text: text$1,
+						data: businessData,
+						rawEvidence: stripped.rawEvidence
+					}) : [];
 					const evidenceQuality = canLocateEvidence ? buildFieldEvidenceQuality({
 						schema,
 						data: businessData,
 						rawEvidence: stripped.rawEvidence,
-						verifiedEvidence: evidence
+						verifiedEvidence: evidence,
+						invalidEvidenceFields
 					}) : void 0;
 					return {
 						success: true,

package/dist/{generate-drizzle-schema-BcStmKUc.mjs → generate-drizzle-schema-BdkbSP5F.mjs}

@@ -12,7 +12,7 @@ import { Kysely, SqliteDialect, sql } from "kysely";
 
 //#region package.json
 var name = "aiex-cli";
-var version = "0.1.1-beta.8";
+var version = "0.1.1-beta.9";
 var description = "JSON Schema → SQLite with AI-powered data extraction";
 var package_default = {
 	name,
@@ -167,6 +167,9 @@ const EVIDENCE_INSTRUCTIONS = `Evidence requirements:
 - Also return a top-level "_evidence" object.
 - For each top-level scalar field you extracted from the text, include "_evidence.<field>.quote".
 - The quote must be an exact contiguous substring copied from the input text.
+- Prefer the shortest quote that still uniquely identifies the field in the document.
+- Include the field label and nearby context when a value is repeated, for example "考试年份：2017年" instead of "2017", or "语文 106 150 71%" instead of "150".
+- Do not use a quote that supports a different field with the same repeated value.
 - Do not invent offsets. Only provide quotes.
 - If no exact quote supports a field, omit that field from "_evidence".`;
 const CORRECTION_SYSTEM_PROMPT = `You are a precise data correction assistant. Your task is to correct validation errors in a previously generated JSON object to make it comply with the provided JSON Schema.

package/dist/index.mjs

@@ -1,3 +1,3 @@
-import { _ as doctorDiagnosticsSeverityRows, g as buildDoctorDiagnostics, h as parseJsonSchema, i as generateDrizzleConfig, m as JsonSchemaDefinitionSchema, n as collectDoctorDiagnostics, r as createMigrationConfig, t as generateDrizzleSchema, v as doctorDiagnosticsTableRows, y as formatDoctorDiagnosticsJson } from "./generate-drizzle-schema-BcStmKUc.mjs";
+import { _ as doctorDiagnosticsSeverityRows, g as buildDoctorDiagnostics, h as parseJsonSchema, i as generateDrizzleConfig, m as JsonSchemaDefinitionSchema, n as collectDoctorDiagnostics, r as createMigrationConfig, t as generateDrizzleSchema, v as doctorDiagnosticsTableRows, y as formatDoctorDiagnosticsJson } from "./generate-drizzle-schema-BdkbSP5F.mjs";
 
 export { JsonSchemaDefinitionSchema, buildDoctorDiagnostics, collectDoctorDiagnostics, createMigrationConfig, doctorDiagnosticsSeverityRows, doctorDiagnosticsTableRows, formatDoctorDiagnosticsJson, generateDrizzleConfig, generateDrizzleSchema, parseJsonSchema };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "aiex-cli",
   "type": "module",
-  "version": "0.1.1-beta.8",
+  "version": "0.1.1-beta.9",
   "description": "JSON Schema → SQLite with AI-powered data extraction",
   "author": "OSpoon <zxin088@gmail.com>",
   "license": "MIT",