The JSON standard recommends that a JSON object not have duplicate field names. Oracle Database enforces this for JSON type data by raising an error. If stored textually, Oracle recommends that you do not allow duplicate field names, by using an is json check constraint with keywords WITH UNIQUE KEYS.

If stored textually (VARCHAR2, CLOB, BLOB column), JSON data is, by default, allowed to have duplicate field names, simply because checking for duplicate names takes additional time. This default behavior for JSON data stored textually can result in inconsistent behavior, so Oracle recommends against relying on it.

You can override this default behavior, to instead raise an error if an attempt is made to insert data containing an object with duplicate fields. You do this by using an is json check constraint with the keywords WITH UNIQUE KEYS. (These keywords have no effect for data inserted into a JSON type column.)

Whether duplicate field names are allowed in well-formed textual JSON data is orthogonal to whether Oracle uses strict or lax syntax to determine well-formedness.

The Oracle default syntax for JSON is lax. In particular: it reflects the JavaScript syntax for object fields; the Boolean and null values are not case-sensitive; and it is more permissive with respect to numerals, whitespace, and escaping of Unicode characters.

Standard ECMA-404, the JSON Data Interchange Format, and ECMA-262, the ECMAScript Language Specification, define JSON syntax.

According to these specifications, each JSON field and each string value must be enclosed in double quotation marks ("). Oracle supports this strict JSON syntax, but it is not the default syntax.

In JavaScript notation, a field used in an object literal can be, but need not be, enclosed in double quotation marks. It can also be enclosed in single quotation marks ('). Oracle also supports this lax JSON syntax, and it is the default syntax.

In addition, in practice, some JavaScript implementations (but not the JavaScript standard) allow one or more of the following:

• Case variations for keywords true, false, and null (for example, TRUE, True, TrUe, fALSe, NulL).

• An extra comma (,) after the last element of an array or the last member of an object (for example, [a, b, c,], {a:b, c:d,}).

• Numerals with one or more leading zeros (for example, 0042.3).

• Fractional numerals that lack 0 before the decimal point (for example, .14 instead of 0.14).

• Numerals with no fractional part after the decimal point (for example, 342. or 1.e27).

• A plus sign (+) preceding a numeral, meaning that the number is non-negative (for example, +1.3).

This syntax too is allowed as part of the Oracle default (lax) JSON syntax. (See the JSON standard for the strict numeral syntax.)

In addition to the ASCII space character (U+0020), the JSON standard defines the following characters as insignificant (ignored) whitespace when used outside a quoted field or a string value:

• Tab, horizontal tab (HT, ^I, decimal 9, U+0009, \t)

• Line feed, newline (LF, ^J, decimal 10, U+000A, \n)

• Carriage return (CR, ^M, decimal 13, U+000D, \r)

The lax JSON syntax, however, treats all of the ASCII control characters (Control+0 through Control+31), as well as the ASCII space character (decimal 32, U+0020), as (insignificant) whitespace characters. The following are among the control characters:

• Null (NUL, ^@, decimal 0, U+0000, \0)

• Bell (NEL, ^G, decimal 7, U+0007, \a)

• Vertical tab (VT, ^K, decimal 11, U+000B)

• Escape (ESC, ^[, decimal 27, U+001B, \e)

• Delete (DEL, ^?, decimal 127, U+007F)

An ASCII space character (U+0020) is the only whitespace character allowed, unescaped, within a quoted field or a string value. This is true for both the lax and strict JSON syntaxes.

For both strict and lax JSON syntax, quoted object field and string values can contain any Unicode character, but some of them must be escaped, as follows:

• ASCII control characters are not allowed, except for those represented by the following escape sequences: \b (backspace), \f (form feed), \n (newline, line feed), \r (carriage return), and \t (tab, horizontal tab).

• Double quotation mark ("), slash (/), and backslash (\) characters must also be escaped (preceded by a backslash): \", \/, and \\, respectively.

In the lax JSON syntax, an object field that is not quoted can contain any Unicode character except whitespace and the JSON structural characters — left and right brackets ([, ]) and curly braces ({, }), colon (:), and comma (,), but escape sequences are not allowed.

Any Unicode character can also be included in a name or string by using the ASCII escape syntax \u followed by the four ASCII hexadecimal digits that represent the Unicode code point.

Note that other Unicode characters that are not printable or that might appear as whitespace, such as a no-break space character (U+00A0), are not considered whitespace for either the strict or the lax JSON syntax.

Table 5-1 shows some examples of JSON syntax.

The default JSON syntax for Oracle Database is lax. Strict or lax syntax matters only for SQL/JSON conditions is json and is not json. All other SQL/JSON functions and conditions use lax syntax for interpreting input and strict syntax when returning output.

CREATE TABLE j_purchaseorder
  (id          VARCHAR2 (32) NOT NULL PRIMARY KEY,
   date_loaded TIMESTAMP (6) WITH TIME ZONE,
   po_document VARCHAR2 (32767)
   CONSTRAINT ensure_json CHECK (po_document is json (STRICT)));