Changelog

9.0.0

💥 BREAKING CHANGES

• Minimum supported Node version is v14.18; we’re dropping v12 support

8.1.0

✨ Features

• export types (11b5fb9)

8.0.16

🔧 Fixed

• correct the types (3ad3026)
• remove unused opts key and tweak regexes (3cf7a4c)
• tweak types (a890de1)

8.0.0

✨ Features

• migrate to ES Modules (8c9d95d)

💥 BREAKING CHANGES

• programs now are in ES Modules and won’t work with Common JS require()

7.15.0

✨ Features

• config file based major bump blacklisting (e15f9bb)

7.14.0

✨ Features

• support undefined as a value within plain objects (6cdea34)

7.13.15

⏪ Reverts

• Revert “chore: setup refresh” (23cf206)

7.13.1

🔧 Fixed

• add testStats to npmignore (f3c84e9)

7.13.0

✨ Features

• rewrite in TS (136fc5f)

7.12.0

Accidental version bump during migration to SourceHut. Sorry about that.

7.11.0

✨ Features

• remove two dependencies and “info” mode which never worked (89c06f4)

7.10.0

• Various documentation and setup tweaks after we migrated to monorepo
• Setup refresh: updated dependencies and all config files using automated tools

7.6.0

• Updated all dependencies and restored coverage tracking both via terminal when testing and through coveralls.io

7.5.0

• Moved to BitBucket. GitHub sold us out.
• Removed Travis and BitHound (RIP), enabled Codacy (for code quality audit)
• Removed package-lock

7.4.0

✨ Features

• Set up Prettier
• Removed package.lock and .editorconfig
• Wired Rollup to remove comments from non-dev builds. This means we can now leave the console.logs in the source code — Rollup will remove from production code.

7.3.0

✨ Features

• Updated to the latest release of ast-monkey-traverse (npm, GitLab) which gives the innerObj.path in the callback object. The path is exactly the same notation as per popular object-path and you can feed innerObj.path to object-path, you’d get current. However, this enables us to traverse up the tree, following the path. That’s the reason why I added this feature — to traverse the AST up to root when resolving variables in json-variables.

7.2.0

✈️ Changes

• Separated traverse() into a standalone library, ast-monkey-traverse (npm, GitLab) and then tapped it

7.1.0

✨ Features

• find() and del() methods previously were using string-to-string comparisons. I replaced lodash.isequal with ast-compare (npm, GitLab) running in strict mode with wildcards enabled. Now you can use matcher API when querying the keys or values.
• Some rebasing done to improve the algorithm’s performance. For example, find() and del() previously matched the actual content first, then checked opts.only conditions. It was not effective because why perform a check if opts.only is not satisfied anyway? Now, opts.only checks, is it of a desired type, and if so, continues to compare the values.

7.0.0

✈️ Changes

• The main source now is in ES2015 modules with import/export.
• Implemented Rollup to generate 3 flavours of this package: CommonJS, UMD and ESM module with import/export.

6.4.0

🔧 Fixed

• 🔧 JS Standard and replaced it with raw ESLint on airbnb-base config, with override to ban semicolons.

6.3.0

🔧 Fixed

• 🔧 Removed object-assign and replaced it with ES6 native Object.assign.

6.2.0

🔧 Fixed

• 🔧 Removed monkey.existy() from the exported methods list. It’s not used any more as checks are done stricly against undefined. Now null can be a valid value, as per JSON data types spec.

6.1.0

✨ Features

• Added checkTypes() onto drop(). Missed it this morning. Now all sorted.

6.0.0

BREAKING CHANGES

JSON spec allows objects to have null in values. However, the traversal algotithm has not (until now) considered that. The key in an array would have val reported as null — same as (theoretical) object that has null as key’s value.

This now changes.

Now, the absence of value will be marked with undefined. Checking does a key/value pair have a value should be done checking if value is not undefined.

This changes how you interact with traversal() function. Sorry about this breaking change, but it’s part of my major drive to have all my libraries to support JSON spec. And in JSON, objects can have null values.

✈️ Changes

• 🔧 Absence of value during traversal (as is the case when traversing arrays) is now marked as undefined. Previously it was null. This will surely break all the algorithms that use monkey.traverse(). On a positive side, once you migrate your code to ast-monkey v6, you’ll be able to support null in object key values, as per JSON spec.

✨ Features

• opts.only is now present on all relevant ast-monkey‘s methods and allows you to describe in natural language do you want to query only objects, or arrays or both. Previously it was only both. Supported values for opts.only are quite wide:

// keywords for `opts.only` to query only objects:
[“object”, “objects”, “obj”, “ob”, “o”][
  // keywords for `opts.only` to query only arrays:
  (“array”, “arrays”, “arr”, “aray”, “arr”, “a”)
][
  // keywords for `opts.only` to query both:
  (“any”,
  “all”,
  “everything”,
  “both”,
  “either”,
  “each”,
  “whatever”,
  “whatevs”,
  “e”)
];

5.5.0

Hardened the API, namely, all added more validations to options object key values.

✨ Features

• Recoded all options objects’ validations, on each method. Preparing for next big secret feature. We’ll need this.

5.4.0

✨ Features

• Holes in arrays were skipped in traverse() as if they didn’t exist. Now I push it further, traverse() will silently delete any holes in arrays it encounters. I think this feature a no-brainer since array holes have no use in JS.

5.3.0

✨ Features

• Set standard to be consumer under normal semver range, not the latest in order to prevent surprises in the future. Which happened as late as v10.

5.2.0

✨ Features

• Tighetened the API in cases when monkey() inputs are missing or falsey.

5.1.0

✨ Features

• innerObj.parent to traverse(). Now you can query sibling elements. I needed this for json-variables (npm, GitLab) to allow variables lookup at deeper levels, not only at the root. 🦄

5.0.0

After spending nearly whole Sunday testing [v4], I discovered that passing undefined as an instruction to delete is wrong, because how do you pass the message that the current item is an array? Previously, when there were no null values allowed, null in the value meant array, but also, when received as a result of traverse() it meant an instruction to delete. Now we can’t touch null because it’s a legitimate value! So we switched to undefined. But we can’t use it for both as an instruction to delete AND as a marker of an array, because that way we will not be able to delete from arrays.

✈️ Changes

• 🔧 Internally, the message to delete in traverse() is now NaN.

🔧 Fixed

• 🔧 All the methods stay the same. I just rewired all internal messaging to use NaN instead of undefined as an instruction for traverse() to delete.

4.0.0

The good thing about being not popular is you can make breaking changes and very few (if anybody) will care. I will make use of this privilege and do some cardinal yet necessary API changes.

✈️ Changes

• 🔧 Removing options from traverse(). It’s not necessary any more. See below why.
• 🔧 When particular node is to be deleted, the message (function’s return value) previously was null. This is not effective as JSON objects can have null values and this means monkey.traverse() does not know, is it value null being returned recursively, or is it an instruction coming from deeper resursions to delete current thing. That’s why I decided to move onto undefined as a deletion message — it can’t be a JSON value, and it does not belong among the object values — it’s perfect format for a deletion message.

🔧 Fixed

• 🔧 All the methods stay the same. I just rewired all internal messaging to use undefined instead of null as an instruction for traverse() to delete.

3.3.0

✨ Features

• 🐒.traverse() gets options! ✨ Optional opts.nullDeletes===false now let’s you to write null values during traversal. Previously on all cases (and currently during default opts.nullDeletes===true) null would be interpreted as an instruction to delete the current piece of AST. Now you can essentially turn off the deletion in favor of being able to write null as value. For the record, null is a valid JSON value type. 🦄

3.2.0

✨ Features

• Imagine, you’re using 🐒.traverse() (ast-monkey on npm/ast-monkey on GitLab) on the following piece of AST:

{
  title: [‘something’, ‘anything’],
  title_data: {
    subtitle: ‘text’,
    submarine: ‘ship’
  }
}

When you’ll be traversing the array, [‘something’, ‘anything’], you’ll have access to the key name, title, via innerObj.topmostKey on the callback. ✨

I needed this feature for json-variables (npm, GitLab) where I wanted to access title_data key, same-named key except with appended string, at the same level as parent. This does not affect any unit tests, it’s a handy extra information piece which was always there, only just now tapped. 👍

3.1.0 (2017-04-01) International Fools day No tricks here though

🏗️ Improvements

• 🔧 All this Saturday morning I worked on 🐒.traverse(). Yesterday night I discovered that when you delete something on traverse(), the traversal reports extra non-existing nodes. The solution is not so elementary. Yes, the iterator was not being reduced in the for loop — i-- was missing — but there were also more fixes necessary to implement for this to work. Now when you want to instruct traverse() to delete current node, you have to pass null (undefined won’t work). I believe that’s how everybody were using it anyway, so it doesn’t warrant major semver bump. ✨

✨ Features

• Added traverse() unit tests, namely, 09.x group.

🔧 Fixed

• Unit test coverage stays solid 100% lines.

3.0.0

✈️ Changes

BREAKING API CHANGES.

• 🔧 flatten() method renamed to arrayFirstOnly() to reflect better what this does. The real “flatten” is object-flatten-all-arrays and while it could be rewritten in ast-monkey, it goes against the overall flow of the ast-monkey‘s algorithm — 🐒 goes horizontal, by branch, while flatten-all-arrays goes vertically, by array, all keys at once. The new arrayFirstOnly() is easy feature because it simply filters the first element of each array encountered during the traversal.

✨ Features

• Exposed .traverse() too; shielded its inner API with another function (one input arguement-less now)

2.9.0 (2017-03-09) International recursive alrorithms day

✨ Features

• .flatten()
• related unit tests.

✈️ Changes

• .info() now returns the input, not undefined. This doesn’t warrant major version bump because method was for logging only and nothing changes in this aspect.

2.8.0

✨ Features

• Now cloning all arguments in main 🐒() and auxiliary traverse() functions’ input object args to prevent any accidental mutation. This is big and very important.

2.7.0

🔧 Fixed

• Replaced spread operator with lodash equivalent to avoid unnecessary Babel use 😌

2.6.0 (2017-02-19) Actual day of NTFS invention

✨ Features

• Incoming input is cloned upon receiving and clone is used instead, so that original input is not mutated. This is very important. ✨

2.5.0

✨ Features

• Rebased the requirements for opts.key or opts.val to exist, now find() and del() are combined.

2.4.0

✨ Features

• Enforcing the {index: ?} to be provided for drop(). ✨

2.3.0

✨ Features

• Added index key to each of find() result object. 👌

2.2.0 (2017-02-16) International software testers commemoration day

✨ Features

• Added del() method which deletes pieces from AST’s by key or by value or by both. It leaves empty stumps and does not clean after deletion.

2.0.0

✈️ Changes

• 🔧 Major API change. Initial release’s get() didn’t make sense. It was returning a “synthetic” object with a separate keys containing info about fetched piece of AST, not the piece itself. This meant, it was not possible to actually get the whole intact piece! Now, I am simply returning the whole finding from get(). That’s it. 😌