semverist

Intelligent Semver-based hierarchy processing

The Semverist provides a uniform interface for understanding and manipulating directories, config, and objects with a semver based hierarchy.

Read our full API docs at https://thebruce.github.io/semverist/

The Semverist combines tree hierarchy organization and semver enabling a richly expressive object or directory structure with less hassle. redundancy accompanied by several utilities to assist and empower maintenance.

Semantic Version upgrades for objects and directories

The Semverist combines tree hierarchy and semver enabling a richly expressive object or directory structure without redundancy accompanied by several utilities to assist and empower maintenance. This can be useful for things like api schema descriptions, functions, or configuration where traditional declarative methods would either have a great deal of repetition or reference includes which can be frustrating to track down. Overhead is lessened by adopting semverish hierarchy, an organizational structure that marries semver to traditional tree based hierarchy. By using semver implied inheritence within an object or directory you can compose a powerful final product with less repetition and a consistent and declarative approach. Alternatively, you can use the lazy semverist style of authoring to mirror API best practices where the last instance of an element in a semver structure will carry through a major version from its definition until overridden.

The semverist can be used with the config module (we recommend it) or you can pass your own in and has a wide variety of helpful tools to keep your semver source files as Dry as possible, such as defaults, groups, file pattern matching, object or directory input sources with configurable object or directory output sources.

Contents

• Usage
• Semverist Components
• What is a semverish object?
• Capabilities: defaults
• Capabilities: groups
• Capabilities: attributes

Usage

Let's look at two of our tests for inspiration, an object based one and a directory based one.

Create a semver object from a semverish object source

The semverist requires that valid configuration be past to it or live in the config module's configuration files.

A example configuration file is fully commented in the /examples directory: Example Config

We will be using the following configuration for our script:

{
  "semverist": {
   "semverImpliedOrchestraObject": {
      "semveristBehaviors": {
        "inheritence": "semverImplied",
        "lazySemverist": {
          "preReleaseForwards": false,
          "attributes": true
        },
        "default": true,
        "defaultName": "orchestraDefault",
        "groups": true,
        "mergeStrategy": "merge"
      },
      "groups": {
        "strings": {
          "members": [
            "violin",
            "viola"
          ]
        },
        "winds": {
          "members": [
            "flute",
            "clarinet"
          ]
        },
        "brass": {
          "members": [
            "trumpet",
            "trombone"
          ]
        }
      },
      "directoryFileIgnorePattern": "json$",
      "converterType": "default",
      "composer": {
        "composerType": "default",
        "destination": null,
        "priority": "default"
      }
    }
  }
}

This configuration lists the groups we have and their members, our custom default name, as well as the kind of source material we have (converterType: default - an object), and what we want to return (composerType:default - an object).

As indicated in the configuration we will be using an object as our source. You can take a look at that object here: semveristOrchestraExampleObject

Take a look at that file and notice that we have attributes within typical semver hierarchies. Since we've indicated that we would like a semver implied approach to our composition we will be merging our groups, defaults, and attributes in, hierarchically, with like elements overwriting as we become more specific in our semver places (i.e. minor attributes will overwrite major attributes of the same name, just as prerelease attributes will overwrite minor items of the same name.)

Our script to turn this 'semverish object' into a semverist object is as follows:

const nestedConfig = 'path/to/the/json/config/above'; // note this could also live in the config modules config files.

const schoenberg = semverist.composer(
  semverishObject,  // A semverish object
  nestedConfig.semverist,
  'semverImpliedOrchestraObject'  // Namespace within the semverist config that applies.
)
.then((composer) => {
  // Now you have a composer class and can assemble your semverish objects.
  // Like so.
  composer.assembleManifest();
  return composer.getComposition(); // returns semver object.
})

This results in a semver object which is quite different than the originating semverish object compare:

Original semverish Object: semverish Object

-vs-

Transformed Semver Object: semver Object

Semverist Components

Schoenberg, the composer

The Semverist composer builds a fleshed out object from the Semverist inspector with support for defaults, arbitrary groups and attribute/file overides or stubs. It also features lazy semver, a companion to API versioning best practices and lazy configuration inheritence that allows for minimal definition and maintenance but maximum impact.

Poirot, the inspector (COMING SOON)

The Semverist inspector can tell you about the attributes/files in the sember shaped object/directory you pass to the Semverist. It can:

• Tell you the semver number of the last occurence of an attribute in a semver object/directory.
• Tell you whether an attribute/file exists for a given semver range/value.
• Tell you what elements form a given attribute file for a given semver.
• Tell you the location w/in the semver hierarchy of the attribute for a given semver range (in the case of lazy semver)
• Tell you the semver range for an attribute, default, or group.
• Tell you the first occurence of a given attribute /file in this semver object/directory.

What is a semverish object?

The Semverish Object: What is a semver shaped hierarchy?

Semver is an extremely useful versioning standard whose declarative syntax helps machine and human readers to set package and software version use in a sane format. Read about semver at http://semver.org/. Semver is not only useful for package versions however, using its declarative ranges within configuration, objects and directories is an awesome way to relate configuration, metadata, scripts, and schemas to constantly evolving services and software. The semverist was born out of a need to reliably deal with these realizations in objects and directories.

The semverish object is declarative and meant to be useful to both Inspector, Poirot and Composer, Schoenberg. Additional utility is provided by defaults and groups and lazy semver funcitonality but even without these concepts you will still be able to inspect and learn about the semverist object you have and build it out with the correct composition.

A semver object might take the shape of:

{
  1: {
    0: {
      0: {
        attributeOfVersion100: 'value',
        'thing': 'thingValue100'
      },
      1: {
        attributeOfVersion101: 'value',
        'thing': 'thingValue101'
      }
    }
    1: {
      0: {
        attributeOfVersion110: 'value',
        'thing': 'thingValue110'
      }
    }
  }
  2: {
    0: {
      0: {
        attributeOfVersion200: 'value',
        'thing': 'thingValue200'
      }
    }
  }
}

A semver shaped directory with similarly named children might look like the following:

|-- 1
|   |-- 0
|   |   |-- 0
|   |   |   |-- file100.json
|   |   |   |-- thing.json
|   |   |-- 1
|   |   |   |-- file101.json
|   |   |   |-- thing.json
|   |-- 1
|   |   |-- 0
|   |   |   |-- file110.json
|   |   |   |-- thing.json
|-- 2
|   |-- 0
|   |   |-- 0
|   |   |   |-- file200.json
|   |   |   |-- thing.json

Capabilities: Defaults

The Semverist Object: Defaults

Semverist Objects can utilize default overrides for all children objects/files. When combined with lazy semver it can create powerful inheritence structures with less effort or mirror best practice api behavior for documentation or schemas. Defaults in objects can be used as base definitions for attributes when used with lazy semver and can act as stubs, when used with directories of json files they again can provide defaults. Defaults can be used to override or stub out empty attributes or empty files as well increasing the possiblities for richly inherited data shapes in the semverist object.

Defaults confer some inheritence based on the level of the hierarchy in which they are placed.

Overriding defaults with other defaults

Defaults are merged into the resulting attribute/file according to the merge strategy you have chosen. Thus a default that shares the same shape of a default set at a higher level will overwrite the higher level default for the children it affects. Example: We start with a default in 1, which applies to all minor and patch children.

{
  1: {
    "default": {
      "test": "value"
    },
    0: {
      0:{
        "item":{}
      }
    }
  }
}

In the above case with a mergeStrategy of last in we would expect that item would have the following value when built by the composer, Schoenberg:

item: {
  "test": "value"
}

However, lets override the default at level 1, with one at level 1.0 like so:

{
  1: {
    "default": {
      "test": "value"
    },
    0: {
      "default": {
        "test": "otherValue"
      }
      0:{
        "item":{}
      }
    }
  }
}

In this new case with a mergeStrategy of last in we would expect that item would have the following value when built by the composer, Schoenberg:

item: {
  "test": "otherValue"
}

This is so because the minor version's defaults apply to patch version children and with the last in merge strategy the default at the minor level overrides the default at the major level.

In a directory situation this takes a similar shape but with files. Where the contents of a file will be merged according to your merge strategy. Let's take a look at an example for directories:

Let's imagine a default text file with the following contents: This is a {{text}} file represented by default.txt in the following directory structure:

|-- 1
|   |-- default.txt (a file with the text `This is a {{text}} file`)
|   |-- 0
|   |   |-- 0
|   |   |   |-- file1.txt (an empty file)
|   |   |-- 1
|   |-- 1
|   |   |-- 0
|   |   |   |-- file2.txt (a file with the text `I already have text`)

With a merge strategy of replace we would expect the following output:

File 2 does not take the default.txt values because it already had values in it and replace will use the last in to overwrite.

Lets add an additional default.txt file and another empty file to see a more complex example.

|-- 1
|   |-- default.txt (a file with the text `This is a {{text}} file`)
|   |-- 0
|   |   |-- default.txt (a file with the text `This is a {{otherText}} file`)
|   |   |-- 0
|   |   |   |-- file1.txt (an empty file)
|   |   |-- 1
|   |-- 1
|   |   |-- 0
|   |   |   |-- file2.txt (a file with the text `I already have text`)
|   |   |   |-- file3.txt (an empty file)

File 1 inherits the default.txt contents from the nearest parent, in this case the minor version where we added the new default. File 2 remains unchanged in this merge strategy and File 3 inherits from the nearest default.txt located in a parent, which is the original default.txt we placed in the major version folder.

Capabilities: Groups

The Semverist Object: Schoenberg Groups

Semverist objects can utilize aribitrary groups of overrides for select objects/directories. This works a lot like defaults but instead of applying to every item underneath it in the hierarchy it only applies to items below it that are indicated for the group. For config and directories this is typically indicated inside the group object which functions fairly similarly to defaults. For directories groups are folders nested in the semver hierarchy including the files that belong to the group.

Capabilities: Attributes

The Semverist Object: attributes

Named item overides applying to specific object/directories within a semver object/directory.

The Semverist will inspect your directories/objects & tell you about them with Poirot and masterfully compose them with Schoenberg.

Lazy Semver

The power of laziness now brought to your semver experience.

Lazy Semver allows you to create schema items once and inherit them until they change mirroring API versioning best practices. The lazy semver configuration option allows you to inherit properties up through a major version overriding and overwriting with defaults, groups, and redeclarations. This can be very useful for setting attributes once per major version of semver and lowering maintenance tasks and copy/paste type errors.

License

MIT @ David Diers

Contributors

David Diers	Elliott Foster	Flip