A utility library for working with Table Schema

v1.1.1 2022-01-21 01:11 UTC

This package is auto-updated.

Last update: 2022-11-21 03:00:29 UTC


Tests Coveralls Scrutinizer-ci Packagist SemVer Codebase Support

A utility library for working with Table Schema in PHP.

Features summary and Usage guide


$ composer require frictionlessdata/tableschema


Table class allows to iterate over data conforming to a table schema

Instantiate a Table object based on a data source and a table schema.

use frictionlessdata\tableschema\Table;

$table = new Table("tests/fixtures/data.csv", ["fields" => [
    ["name" => "first_name"],
    ["name" => "last_name"],
    ["name" => "order"]

Schema can be any parameter valid for the Schema object (See below), so you can use a url or filename which contains the schema

$table = new Table("tests/fixtures/data.csv", "tests/fixtures/data.json");

iterate over the data, all the values are cast and validated according to the schema

foreach ($table as $row) {
    print($row["order"]." ".$row["first_name"]." ".$row["last_name"]."\n");

validate function will validate the schema and get some sample of the data itself to validate it as well

Table::validate(new CsvDataSource(""), $schema);

You can instantiate a table object without schema, in this case the schema will be inferred automatically based on the data

$table = new Table("tests/fixtures/data.csv");
$table->schema()->fields();  // ["first_name" => StringField, "last_name" => StringField, "order" => IntegerField]

Optionally, specify a CSV Dialect:

$table = new Table("tests/fixtures/data.csv", null, ["delimiter" => ";"]);

Table::read method allows to get all data as an array, it also supports options to modify reader behavior

$table->read()  // returns all the data as an array

read accepts an options parameter, for example:

$table->read(["cast" => false, "limit": 5])

The following options are available (the values are the default values):

    "keyed" => true,  // flag to emit keyed rows
    "extended" => false,  // flag to emit extended rows
    "cast" => true,  //flag to disable data casting if false
    "limit" => null,  // integer limit of rows to return

Additional methods and functionality

$table->headers()  // ["first_name", "last_name", "order"]
$table->save("output.csv")  // iterate over all the rows and save the to a csv file
$table->schema()  // get the Schema object
$table->read()  // returns all the data as an array


Schema class provides helpful methods for working with a table schema and related data.

use frictionlessdata\tableschema\Schema;

Schema objects can be constructed using any of the following:

  • php array (or object)
$schema = new Schema([
    'fields' => [
            'name' => 'id', 'title' => 'Identifier', 'type' => 'integer', 
            'constraints' => [
                "required" => true,
                "minimum" => 1,
                "maximum" => 500
        ['name' => 'name', 'title' => 'Name', 'type' => 'string'],
    'primaryKey' => 'id'
  • string containing json
$schema = new Schema("{
    \"fields\": [
        {\"name\": \"id\"},
        {\"name\": \"height\", \"type\": \"integer\"}
$schema = new Schema("");

The schema is loaded, parsed and validated and will raise exceptions in case of any problems.

access the schema data, which is ensured to conform to the specs.

$schema->missingValues(); // [""]
$schema->primaryKey();  // ["id"]
$schema->foreignKeys();  // []
$schema->fields(); // ["id" => IntegerField, "name" => StringField]
$field = $schema->field("id");  // Field object (See Field reference below)

validate function accepts the same arguemnts as the Schema constructor but returns a list of errors instead of raising exceptions

// validate functions accepts the same arguments as the Schema constructor
$validationErrors = Schema::validate("http://invalid.schema.json");
foreach ($validationErrors as $validationError) {

validate and cast a row of data according to the schema

$row = $schema->castRow(["id" => "1", "name" => "First Name"]);

will raise exception if row fails validation

it returns the row with all native values

$row  // ["id" => 1, "name" => "First Name"];

validate the row to get a list of errors

$schema->validateRow(["id" => "foobar"]);  // ["id is not numeric", "name is required" .. ]

Infer schema based on source data:

$schema = Schema::infer("tests/fixtures/data.csv");
$table->schema()->fields();  // ["first_name" => StringField, "last_name" => StringField, "order" => IntegerField]

You can also create a new empty schema for editing

$schema = new Schema();

set fields

    "id" => (object)["type" => "integer"],
    "name" => (object)["type" => "string"],

appropriate Field object is created according to the given descriptor (see below for Field class reference)

$schema->field("id");  // IntegerField object

add / update or remove fields

$schema->field("email", ["type" => "string", "format" => "email"]);
$schema->field("name", ["type" => "string"]);

set or update other table schema attributes


after every change - schema is validated and will raise Exception in case of validation errors

Finally, you can get the full validated descriptor


And, save it to a json file



Field class represents a single table schema field descriptor

Create a field from a descriptor

use frictionlessdata\tableschema\Fields\FieldsFactory;
$field = FieldsFactory::field([
    "name" => "id", "type" => "integer",
    "constraints" => ["required" => true, "minimum" => 5]

Cast and validate values using the field

$field->castValue("3");  // exception: value is below minimum
$field->castValue("7");  // 7

Additional method to access field data

$field("id")->format();  // "default"
$field("id")->name();  // "id"
$field("id")->type(); // "integer"
$field("id")->constraints();  // (object)["required"=>true, "minimum"=>1, "maximum"=>500]
$field("id")->enum();  // []
$field("id")->required();  // true
$field("id")->unique();  // false
$field("id")->title();  // "Id" (or null if not provided in descriptor)
$field("id")->description();  // "The ID" (or null if not provided in descriptor)
$field("id")->rdfType();  // "" (or null if not provided in descriptor)


Please read the contribution guidelines: How to Contribute