handling-validation-errors（处理验证错误）

Handling Validation Errors in z-schema

z-schema reports validation errors as ValidateError objects containing a .details array of SchemaErrorDetail. This skill covers inspecting, filtering, mapping, and presenting these errors.

Error structure

import { ValidateError } from 'z-schema';
import type { SchemaErrorDetail } from 'z-schema';

ValidateError extends Error:

Each SchemaErrorDetail:

Capturing errors

Try/catch (default mode)

import ZSchema from 'z-schema';

const validator = ZSchema.create();

try {
  validator.validate(data, schema);
} catch (err) {
  if (err instanceof ValidateError) {
    for (const detail of err.details) {
      console.log(`[${detail.code}] ${detail.path}: ${detail.message}`);
    }
  }
}

Safe mode (no try/catch)

const validator = ZSchema.create();
const { valid, err } = validator.validateSafe(data, schema);

if (!valid && err) {
  for (const detail of err.details) {
    console.log(`[${detail.code}] ${detail.path}: ${detail.message}`);
  }
}

Walking nested errors

Combinators (anyOf, oneOf, not) produce nested inner errors. A recursive walker handles any depth:

function walkErrors(details: SchemaErrorDetail[], depth = 0): void {
  for (const detail of details) {
    const indent = '  '.repeat(depth);
    console.log(`${indent}[${detail.code}] ${detail.path}: ${detail.message}`);
    if (detail.inner) {
      walkErrors(detail.inner, depth + 1);
    }
  }
}

const { valid, err } = validator.validateSafe(data, schema);
if (!valid && err) {
  walkErrors(err.details);
}

Collecting all leaf errors

Flatten the tree to get every concrete error, skipping combinator wrappers:

function collectLeafErrors(details: SchemaErrorDetail[]): SchemaErrorDetail[] {
  const leaves: SchemaErrorDetail[] = [];
  for (const detail of details) {
    if (detail.inner && detail.inner.length > 0) {
      leaves.push(...collectLeafErrors(detail.inner));
    } else {
      leaves.push(detail);
    }
  }
  return leaves;
}

Mapping errors to form fields

Convert JSON Pointer paths to field names for UI form validation:

function pathToFieldName(path: string | Array<string | number>): string {
  if (Array.isArray(path)) {
    return path.join('.');
  }
  // JSON Pointer string: "#/address/city" → "address.city"
  return path.replace(/^#\/?/, '').replace(/\//g, '.');
}

function errorsToFieldMap(details: SchemaErrorDetail[]): Record<string, string[]> {
  const map: Record<string, string[]> = {};
  const leaves = collectLeafErrors(details);
  for (const detail of leaves) {
    const field = pathToFieldName(detail.path) || '_root';
    (map[field] ??= []).push(detail.message);
  }
  return map;
}

// Usage
const { valid, err } = validator.validateSafe(formData, schema);
if (!valid && err) {
  const fieldErrors = errorsToFieldMap(err.details);
  // { "email": ["Expected type string but found type number"],
  //   "age": ["Value 150 is greater than maximum 120"] }
}

Using array paths

Enable reportPathAsArray for easier programmatic access:

const validator = ZSchema.create({ reportPathAsArray: true });
const { valid, err } = validator.validateSafe(data, schema);
// err.details[0].path → ["address", "city"] instead of "#/address/city"

Filtering errors

Per-call filtering

Pass includeErrors or excludeErrors as the third argument:

// Only report type mismatches
validator.validate(data, schema, { includeErrors: ['INVALID_TYPE'] });

// Suppress string-length errors
validator.validate(data, schema, { excludeErrors: ['MIN_LENGTH', 'MAX_LENGTH'] });

Programmatic post-filtering

const { valid, err } = validator.validateSafe(data, schema);
if (!valid && err) {
  const typeErrors = err.details.filter((d) => d.code === 'INVALID_TYPE');
  const requiredErrors = err.details.filter((d) => d.code === 'OBJECT_MISSING_REQUIRED_PROPERTY');
}

Custom error messages

Map error codes to user-friendly messages:

const friendlyMessages: Record<string, (detail: SchemaErrorDetail) => string> = {
  INVALID_TYPE: (d) => `${pathToFieldName(d.path)} must be a ${d.params[0]}`,
  OBJECT_MISSING_REQUIRED_PROPERTY: (d) => `${d.params[0]} is required`,
  MINIMUM: (d) => `${pathToFieldName(d.path)} must be at least ${d.params[1]}`,
  MAXIMUM: (d) => `${pathToFieldName(d.path)} must be at most ${d.params[1]}`,
  MIN_LENGTH: (d) => `${pathToFieldName(d.path)} must be at least ${d.params[1]} characters`,
  MAX_LENGTH: (d) => `${pathToFieldName(d.path)} must be at most ${d.params[1]} characters`,
  PATTERN: (d) => `${pathToFieldName(d.path)} has an invalid format`,
  ENUM_MISMATCH: (d) => `${pathToFieldName(d.path)} must be one of the allowed values`,
  INVALID_FORMAT: (d) => `${pathToFieldName(d.path)} is not a valid ${d.params[0]}`,
  ARRAY_UNEVALUATED_ITEMS: () => `Array contains items not covered by any schema`,
  OBJECT_UNEVALUATED_PROPERTIES: (d) => `${d.params[0]} is not allowed by the schema`,
  COLLECT_EVALUATED_DEPTH_EXCEEDED: (d) => `Schema nesting too deep (max ${d.params[0]})`,
  MAX_RECURSION_DEPTH_EXCEEDED: (d) => `Recursion depth exceeded (max ${d.params[0]})`,
};

function toFriendlyMessage(detail: SchemaErrorDetail): string {
  const fn = friendlyMessages[detail.code];
  return fn ? fn(detail) : detail.message;
}

Stopping at first error

For fail-fast scenarios:

const validator = ZSchema.create({ breakOnFirstError: true });

This reports only the first error encountered, reducing noise during iterative fixing.

Using the keyword field

The keyword field tells you which schema keyword triggered the error — useful for categorizing errors programmatically:

const { valid, err } = validator.validateSafe(data, schema);
if (!valid && err) {
  for (const detail of err.details) {
    switch (detail.keyword) {
      case 'required':
        // handle missing field
        break;
      case 'type':
        // handle type mismatch
        break;
      case 'format':
        // handle format failure
        break;
    }
  }
}

Reference files

For the full error code list with descriptions, see the validating-json-data skill's error-codes reference.