library.json

library.json is a manifest file of development library. It allows developers to keep project in own structure and define:

PlatformIO Library Crawler uses library.json manifest to extract source code from developer’s location and keeps a cleaned library in own Library Registry.

A data in library.json should be represented in JSON-style via associative array (name/value pairs). An order doesn’t matter. The allowable fields (names from pairs) are described below.

name

Required | Type: String | Max. Length: 50

A name of the library.

  • Must be unique.

  • Should be slug style for simplicity, consistency and compatibility. Example: Arduino-SPI

  • Title Case, Aa-z, can contain digits and dashes (but not start/end with them).

  • Consecutive dashes are not allowed.

description

Required | Type: String | Max. Length: 255

The field helps users to identify and search for your library with a brief description. Describe the hardware devices (sensors, boards and etc.) which are suitable with it.

keywords

Required | Type: String | Max. Length: 255

Used for search by keyword. Helps to make your library easier to discover without people needing to know its name.

The keyword should be lowercased, can contain a-z, digits and dash (but not start/end with them). A list from the keywords can be specified with separator ,

authors

Required if repository field is not defined | Type: Object or Array

An author contact information

  • name Full name (Required)

  • email

  • url An author’s contact page

  • maintainer Specify “maintainer” status

Examples:

"authors":
{
    "name": "John Smith",
    "email": "me@john-smith.com",
    "url": "http://www.john-smith/contact"
}

...

"authors":
[
    {
        "name": "John Smith",
        "email": "me@john-smith.com",
        "url": "http://www.john-smith/contact"
    },
    {
        "name": "Andrew Smith",
        "email": "me@andrew-smith.com",
        "url": "http://www.andrew-smith/contact",
        "maintainer": true
    }
]

Note

You can omit authors field and define repository field. Only GitHub-based repository is supported now. In this case PlatformIO Library Registry Crawler will use information from GitHub API Users.

repository

Required if downloadUrl field is not defined | Type: Object

The repository in which the source code can be found. The field consists of the next items:

  • type the only “git”, “hg” or “svn” are supported

  • url

  • branch if is not specified, default branch will be used. This field will be ignored if tag/release exists with the value of version.

Example:

"repository":
{
    "type": "git",
    "url": "https://github.com/foo/bar.git"
}

version

Required if repository field is not defined | Type: String | Max. Length: 20

A version of the current library source code. Can contain a-z, digits, dots or dash. Semantic Versioning IS RECOMMENDED.

Case 1:

version and repository fields are defined. The repository is hosted on GitHub or Bitbucket.

PlatformIO Library Registry Crawler will lookup for release tag named as value of version or with v prefix (you do not need to pass this v prefix to the version field).

Case 2:

version and repository fields are defined and repository does not contain tag/release with value of version.

PlatformIO Library Registry Crawler will use the latest source code from repository and link it with specified version. If repository branch is not specified, then default branch will be used. Also, if you push new commits to repository and do not update version field, the library will not be updated until you change the version.

Case 3:

version field is not defined and repository field is defined.

PlatformIO Library Registry Crawler will use the VCS revision from the latest commit as “current version”. For example, 13 (SVN) or first 10 chars of SHA digest e4564b7da4 (Git). If repository branch is not specified, then default branch will be used.

We recommend to use version field and specify the real release version and make appropriate tag in the repository. In other case, users will receive updates for library with each new commit to repository.

Note

PlatformIO Library Registry Crawler updates library only if:
  • the version is changed

  • library.json is modified

Example:

"repository":
{
    "type": "git",
    "url": "https://github.com/foo/bar.git"
},
"version": "1.0.0"

license

Optional | Type: String

A license of the library. You can check the full list of SPDX license IDs. Ideally you should pick one that is OSI approved.

"license": "Apache-2.0"

downloadUrl

Required if repository field is not defined | Type: String

It is the HTTP URL to the archived source code of library. It should end with the type of archive (.zip or .tar.gz).

Note

downloadUrl has higher priority than repository.

Example with detached release/tag on GitHub:

"version": "1.0.0",
"downloadUrl": "https://github.com/foo/bar/archive/v1.0.0.tar.gz",
"include": "bar-1.0.0"

See more library.json Examples.

homepage

Optional | Type: String | Max. Length: 255

Home page of library (if is different from repository url).

export

Optional | Type: Object

Explain PlatformIO Library Crawler which content from the repository/archive should be exported as “source code” of the library. This option is useful if need to exclude extra data (test code, docs, images, PDFs, etc). It allows to reduce size of the final archive.

Possible options:

include

Optional | Type: String or Array | Glob Pattern

If include field is a type of String, then PlatformIO Library Registry Crawler will recognize it like a “relative path inside repository/archive to library source code”. See example below where the only source code from the relative directory LibrarySourceCodeHere will be included.

"include": "some/child/dir/LibrarySourceCodeHere"

If include field is a type of Array, then PlatformIO Library Registry Crawler firstly will apply exclude filter and then include only directories/files which match with include patterns.

Example:

"export": {
    "include":
    [
        "dir/*.[ch]pp",
        "dir/examples/*",
        "*/*/*.h"
    ]
}

Pattern Meaning

Pattern

Meaning

*

matches everything

?

matches any single character

[seq]

matches any character in seq

[!seq]

matches any character not in seq

See more library.json Examples.

exclude

Optional | Type: String or Array | Glob Pattern

Exclude the directories and files which match with exclude patterns.

frameworks

Optional | Type: String or Array

A list with compatible frameworks. The available framework types are defined in the Frameworks section.

If the library is compatible with the all frameworks, then you can use * symbol:

"frameworks": "*"

platforms

Optional | Type: String or Array

A list with compatible platforms. The available platform types are defined in Development Platforms section.

If the library is compatible with the all platforms, then you can use * symbol:

"platforms": "*"

dependencies

Optional | Type: Array or Object

A list of dependent libraries. They will be installed automatically with platformio lib install command.

Allowed requirements for dependent library:

  • name | Type: String

  • version | Type: String

  • authors | Type: String or Array

  • frameworks | Type: String or Array

  • platforms | Type: String or Array

The version supports Semantic Versioning ( <major>.<minor>.<patch>) and can take any of the following forms:

  • 1.2.3 - an exact version number. Use only this exact version

  • ^1.2.3 - any compatible version (exact version for 1.x.x versions

  • ~1.2.3 - any version with the same major and minor versions, and an equal or greater patch version

  • >1.2.3 - any version greater than 1.2.3. >=, <, and <= are also possible

  • >0.1.0,!=0.2.0,<0.3.0 - any version greater than 0.1.0, not equal to 0.2.0 and less than 0.3.0

The rest possible values including VCS repository URLs are documented in platformio lib install command.

Example:

"dependencies":
[
    {
        "name": "Library-Foo",
        "authors":
        [
            "Jhon Smith",
            "Andrew Smith"
        ]
    },
    {
        "name": "Library-Bar",
        "version": "~1.2.3"
    },
    {
        "name": "lib-from-repo",
        "version": "https://github.com/user/package.git#1.2.3"
    }
]

A short definition of dependencies is allowed:

"dependencies": {
    "mylib": "1.2.3",
    "lib-from-repo": "githubuser/package"
}

See more library.json Examples.

examples

Optional | Type: String or Array | Glob Pattern

A list of example patterns. This field is predefined with default value:

"examples": [
    "[Ee]xamples/*.c",
    "[Ee]xamples/*.cpp",
    "[Ee]xamples/*.ino",
    "[Ee]xamples/*.pde",
    "[Ee]xamples/*/*.c",
    "[Ee]xamples/*/*.cpp",
    "[Ee]xamples/*/*.ino",
    "[Ee]xamples/*/*.pde",
    "[Ee]xamples/*/*/*.c",
    "[Ee]xamples/*/*/*.cpp",
    "[Ee]xamples/*/*/*.ino",
    "[Ee]xamples/*/*/*.pde"
]

build

Optional | Type: Object

Specify advanced settings, options and flags for the build system. Possible options:

flags

Optional | Type: String or Array

Extra flags to control preprocessing, compilation, assembly and linking processes. More details build_flags.

unflags

Optional | Type: String or Array

Remove base/initial flags which were set by development platform. More details build_unflags.

srcFilter

Optional | Type: String or Array

Specify which source files should be included/excluded from build process. The path in filter should be relative from a root of library.

See syntax in src_filter.

Please note that you can generate source filter “on-the-fly” using extraScript (see below)

extraScript

Optional | Type: String

Launch extra script before build process. More details extra_scripts.

Example (HAL-based library)

This example demonstrates how to build HAL-dependent source files and exclude other source files from a build process.

Project structure

├── lib
│   ├── README
│   └── SomeLib
│       ├── extra_script.py
│       ├── hal
│       │   ├── bar
│       │   │   ├── hal.c
│       │   │   └── hal.h
│       │   ├── foo
│       │       ├── hal.c
│       │       └── hal.h
│       ├── library.json
│       ├── SomeLib.c
│       └── SomeLib.h
├── platformio.ini
└── src
    └── test.c

platformio.ini

[env:foo]
platform = native
build_flags = -DHAL=foo

[env:bar]
platform = native
build_flags = -DHAL=bar

library.json

{
    "name": "SomeLib",
    "version": "0.0.0",
    "build": {
        "extraScript": "extra_script.py"
    }
}

extra_script.py

Import('env')
from os.path import join, realpath


for item in env.get("CPPDEFINES", []):
    if isinstance(item, tuple) and item[0] == "HAL":
        env.Append(CPPPATH=[realpath(join("hal", item[1]))])
        env.Replace(SRC_FILTER=["+<*>", "-<hal>", "+<%s>" % join("hal", item[1])])
        break

libArchive

Optional | Type: Boolean

Create an archive (*.a, static library) from the object files and link it into a firmware (program). This is default behavior of PlatformIO Build System ("libArchive": true).

Setting "libArchive": false will instruct PIO Build System to link object files directly (in-line). This could be useful if you need to override weak symbols defined in framework or other libraries.

You can disable library archiving globally using lib_archive option in “platformio.ini” (Project Configuration File).

libLDFMode

Optional | Type: String

Specify Library Dependency Finder Mode. See Dependency Finder Mode for details.

libCompatMode

Optional | Type: Integer

Specify Library Compatibility Mode. See Compatibility Mode for details.

Examples

  1. Custom macros/defines

"build": {
    "flags": "-D MYLIB_REV=1.2.3 -DRELEASE"
}
  1. Extra includes for C preprocessor

"build": {
    "flags": [
        "-I inc",
        "-I inc/target_x13"
    ]
}
  1. Force to use C99 standard instead of C11

"build": {
    "unflags": "-std=gnu++11",
    "flags": "-std=c99"
}
  1. Build source files (c, cpp, h) at the top level of the library

"build": {
    "srcFilter": [
        "+<*.c>",
        "+<*.cpp>",
        "+<*.h>"
    ]
}
  1. Extend PlatformIO Build System with own extra script

"build": {
    "extraScript": "generate_headers.py"
}

generate_headers.py

Import('env')
# print env.Dump()
env.Append(
    CPPDEFINES=["HELLO=WORLD", "TAG=1.2.3", "DEBUG"],
    CPPPATH=["inc", "inc/devices"]
)

# some python code that generates header files "on-the-fly"