Skip to content

Latest commit

 

History

History
339 lines (265 loc) · 13.6 KB

serialization.md

File metadata and controls

339 lines (265 loc) · 13.6 KB

Serialization

Overview

Serialization is the process of translating a data structure or object state into a format that can be stored or transmitted and reconstructed later (Wikipedia). All aspects related to how to ensure the object state is consistently serialized and deserialized should be covered in this content.

  • Resource request/response entities SHOULD always be represented in JavaScript Object Notation (JSON) format by default.
  • If an Accept header is sent and application/json is not an acceptable response, a 406 Not Acceptable error MUST be returned.
// REQUEST
POST /articles
Accept: application/json
Content-Type: application/json
{
    "foo": "bar"
}
  
// RESPONSE
200 OK
Content-Type: application/json
{
    "fooResult": "not-bar"
}

Query Parameters

  • Query parameter serialization MUST follow HTTP standards and URL query parameter guidelines.
  • Query parameter values MUST NOT represent serialized JSON.
// INCORRECT
// JSON strings as params values
https://api.spscommerce.com/documents?q={query: { type: "Order", startDate: "2021-05-16T14:12:07−05:00Z", endDate: "2021-05-17T14:12:07−05:00Z"}}
 
// URL encoded string is not acceptable
https://api.spscommerce.com/documents?q=%7Bquery%3A%20%7B%20type%3A%20%22Order%22%2C%20startDate%3A%20%222021-05-16T14%3A12%3A07%E2%88%9205%3A00Z%22%2C%20endDate%3A%20%222021-05-17T14%3A12%3A07%E2%88%9205%3A00Z%22%7D%7D
 
// CORRECT
// JSON strings as params values
https://api.spscommerce.com/documents?type=Order&startDate=2021-05-16T14:12:07−05:00Z&endDate=2021-05-17T14:12:07−05:00Z
  • When advanced query language is required in URL query parameters, the guidelines of using FIQL/RSQL SHOULD be followed in Collections
APIs **MUST** be strict in the information they produce, and they **SHOULD** be strict in what they consume as well (where tolerance cannot be applied). Since we are dealing with programming interfaces, we need to avoid guessing the meaning of what is being sent to us as much as possible. Given that integration is typically a one-time task for a developer and we provide good documentation, we need to be strict with using the data that is being received. [Postel's law](https://en.wikipedia.org/wiki/Robustness_principle) must be weighed against the many dangers of permissive parsing.

Quantities

  • Quantities SHOULD use representation by the entity that includes a Unit of Measure (UOM) field along with a quantity field.
  • Quantities represented by an integer type MUST include the unit of measurement.
  • Quantities represented by an integer type SHOULD be represented in a 32-bit signed integer in the schema.
  • If the quantity is a number of items, then the field SHOULD have the suffix Count, for example nodeCount.
{
  // Quantity represented by Quantity object that represents the amount and Unit of Measure (UOM) fields
  productQuantity: {
    amount: 10,
    uom: "boxes"
  },
 
  // Quantity represented by two fields of amount and UOM next to each other and follow naming convention
  productQuantity: 10,
  productQuantityUom: "boxes",
 
  // Quantity represented by integer should include UOM in the field name, xxx{Bytes|Pixels|Cartons}
  quantityOfBoxes: 15,
 
  // Quantity that indicates number of items/elements have Count suffix
  documentCount: 10,
  pageCount: 25
}

Types

NULL

null is a primitive type in JSON. When validating a JSON document against JSON Schema, a property's value can be null only when it is explicitly allowed by the schema, using the type keyword (e.g. {"type": "null"}).

  • null values SHOULD be present in a response payload for an attribute value as a whole.

// CORRECT
GET /articles/1
{
    "id": "1",
    "name": "my article name",
    "displayName": null             // display name is provided, even when null
}
 
// INCORRECT
GET /articles/1
{
    "id": "1",
    "name": "my article name"
                                    // display name was left out because it was null.
}
  • An entire nested object SHOULD be represented as a single null instead of an object with all null attributes.
// CORRECT
GET /articles/1
{
    "id": "1",
    "name": "my article name",
    "relatedObject": null         // related object is not always present and should be treated as an object that is null when absent
}
 
// INCORRECT
{
    "id": "1",
    "name": "my article name",
    "relatedObject": {
        "attributeOne": null,
        "attributeTwo": null,
        "attributeThree": null
    }
}
  • null MUST be specified in PATCH request to indicate that value should be removed from the specified field of the object.
PATCH /articles/1
{   
   author: "John Doe",  // sets value of 'John Doe' to the object of article
   description: null   // removes description from the object of article
}

Boolean

  • Primitive values MUST be serialized to JSON following the rules of RFC 4627.
  • Boolean values MUST be true/false values, not represented as a string.
// INCORRECT
{
  completed: "true",
  required: "0"
}
 
// CORRECT
{
  completed: true,
  required: false
}

String

  • string SHOULD always explicitly define a minLength and maxLength in the associated API schema.
  • string SHOULD use the pattern property in the schema as appropriate.
{
    name: "John Doe"
}

Number

  • Numbers SHOULD be referenced in JSON as integer types and be represented in a 32-bit signed integer, values between ((2^31) - 1) and -(2^31) in the associated API schema.
    • When an integer type is used as a 32-bit integer it SHOULD provide an explicit minimum and a maximum in the associated API schema.
    • When requiring a 64-bit integer (a.k.a "long") it SHOULD be serialized as a string and specified as a formatted "int64" to ensure maximum compatibility across programming languages, particularly with JavaScript.
    • In the circumstance where serializing a number as a 64-bit integer is required, you MAY provide the 64-bit value as both a string and a number to ensure its accessibility.
{
    "largeValue": 10765432100123656789,
    "largeValueString": "10765432100123456799"
}
  • Resource identifiers SHOULD NOT be serialized as numbers or integers (of any precision) as they should be considered opaque values to the API consumer.
  • String MUST be used to represent a decimal numeric value to avoid possible JSON interpreter precision loss across different languages. Some languages may interpret a JSON "number" as a fixed point and others as a floating-point.
  • Percentage values SHOULD be represented as a string of fixed-point decimals. The property name MUST indicate that this is a percentage field.
// INCORRECT
{
    id: 234,                // IDs should be opaque string values.
    number: "10",           // 32-bit integers should be serialized as numbers.
    decimal: 10.2          // floating-point numbers should be serialized as strings to avoid precision loss.
    taxAddition: "8.75%",   // symbols should be left out of the string serialized percentage values.
    percentage: 8           // floating-point numbers should be serialized as strings to avoid precision loss.
}
 
// CORRECT
{
    id: "234",
    number: 10,
    decimal: "10.2",
    percentage: "9.75"
}

Enums

Enum values SHOULD be composed of only upper-case alphanumeric characters and the underscore character, ( _ ).

  • FIELD_10
  • NOT_EQUAL
This modification is knowingly in conflict with general naming rules, as it applies specifically to enumerations.

If there is an industry standard that requires us to do otherwise, enums MAY contain other characters.

// INCORRECT, but acceptable when required by the industry standard
{
    type: "PurchaseOrder",  // enum of the type of the document
    id: "6be295c6-8977-4dce-bddc-8c943456725f",
    ...
}
 
// CORRECT
{
    type: "PURCHASE_ORDER",  // enum of the type of the document
    id: "6be295c6-8977-4dce-bddc-8c943456725f",
    ...
}

Dates

JSON itself does not specify how dates should be represented, but most organizations use the date specification described in JavaScript.

  • All dates posted in requests and returned in response MUST conform to ISO 8601 standard (format YYYY-MM-DDTHH:mm:ss.sssZ).
  • All dates MUST include the time zone, and SHOULD be represented in UTC time zone.
  • All date fields SHOULD have the level of specificity and accuracy in their name:
    • xxxDateTime for fields with accuracy to time level
    • xxxDate for fields with accuracy to date level
// INCORRECT
{
    createdDate: "May 16 2021 14:12:07",        // date format incorrect, name of the field indicates Date but value has DateTime
    createdDateTime: "May 16 2021",             // specified value does not contain time portion
    created: "2021-05-16T14:12:07"              // name of the field does not indicate that it is DateTime field
    localDateTime: "2021-05-16T14:12:07",        // while valid ISO 8601 format for local time zone, it is not appropriate for these API standards.
    zoneDateTime: "2021-05-16T14:12:07−05:00",  // should use UTC time zone specified by Z at the end of the string
}
 
// CORRECT
{
    preciseDateTime: "2021-05-16T14:12:07.123Z",     // standard date time with sub-second precision
    createdDateTime: "2021-05-16T14:12:07Z",         // standard date time with second precision
}

Intervals and Durations

Reference: Microsoft API Guidelines for Durations

Durations

Durations MUST be serialized in conformance with ISO 8601. In general durations format contains parts of data and time separated from each other P<date>T<time>. P indicates the beginning of the duration value, and T is a separation between date and time parts to avoid ambiguity. Durations are "represented by the format P[n]Y[n]M[n]DT[n]H[n]M[n]S" where after each numeric value followed by designator of that value:

  • P is the duration designator (historically called "period") placed at the start of the duration representation.
  • Y is the year designator that follows the value for the number of years.
  • M is the month designator that follows the value for the number of months.
  • W is the week designator that follows the value for the number of weeks.
  • D is the day designator that follows the value for the number of days.
  • T is the time designator that precedes the time components of the representation.
  • H is the hour designator that follows the value for the number of hours.
  • M is the minute designator that follows the value for the number of minutes.
  • S is the second designator that follows the value for the number of seconds.

For example, P3Y6M4DT12H30M5S represents a duration of "three years, six months, four days, twelve hours, thirty minutes, and five seconds."

// INCORRECT
{
    duration: "3Y6M4DT12H30M5S", // missing "P" - duration designator at the beginning of the duration string
    duration: "15 minutes", // string does not follow ISO standard
    duration: "PTH30MS" // designators should not exist without actual value, and vice versa, each value should have designator associated to it
}
 
// CORRECT
{
    duration: "P3Y6M4DT12H30M5S", // 3 years, 6 months, 4 days, 12 hours, 30 minutes, 5 seconds
    duration: "PT1H30M5S", // only time duration since there is no date value between P and T: 1 hour, 30 minutes, 5 seconds
    duration: "P3MT30M", // example where ambiguity between "month" and "minute" resolved by separating date and time parts. This period is 3 months and 30 minutes
    duration: "P3Y", // date only: 3 years
    duration: "PT30M" // time only: 30 minutes period
}

Intervals

Intervals are defined as part of ISO 8601 and MUST follow one of the following formats:

  • Start and end dates: <startDate>/<endDate>
  • Start and duration: <startDate>/<duration>
  • Duration and end date: <duration>/<endDate>
{
    interval: "2007-03-01T13:00:00Z/2008-05-11T15:30:00Z", // <startDate>/<endDate> interval that starts at 1pm of March 1st, 2007 and ends at 3:30pm of May 11th 2008.
 
    interval: "2007-03-01T13:00:00Z/P1Y2M10DT2H30M", // <startDate>/<duration> interval that starts 1pm of March 1st, 2007
                                                     // and ends after specified period of 1 year, 2 months, 10 days, 2 hours and 30 minutes
 
    interval: "P1Y2M10DT2H30M/2008-05-11T15:30:00Z" // <duration>/<endDate> interval that defined by specified period
                                                    // of 1 year, 2 months, 10 days, 2 hours and 30 minutes and ends 3:30pm of May 11th 2008
}

Repeating Intervals MUST follow R[n]/<interval> format where R[n] indicates amount of repetitions and interval value itself as specified earlier, separated by slash /:

{
    repeat: "R5/2007-03-01T13:00:00Z/2008-05-11T15:30:00Z", // repeat 5 times interval that starts at 1pm of March 1st, 2007 and ends at 3:30pm of May 11th 2008
}

Schedules

When there is a need to represent a schedule-based string that represents a repetitive or one-time scheduled operation, that string SHOULD follow cron format, given that service that executes those schedules support cron as well.

schedule:    "30       4       1      1         *"
           (minute)  (hour)  (day)  (month) (weekday)
 
// EXAMPLE
{
    schedule: "30 4 1 1 *" // At 04:30 on day-of-month 1 in January.
}