# openapi
This repo contains helper-scripts to generate API client/server code from an OpenAPI specification using [openapi-generator](https://github.com/OpenAPITools/openapi-generator).
It solves a common problem: merging newly generated versions of an API with existing implementations.
## Workflow
1. use `openapi-generator` to generate client/server code
2. merge or replace new files with existing ones
3. replace content of generated files (based on config)
## Repo structure
```
examples/ example configurations
src/ source files
```
## Examples
Some examples for the configuration and usage can be found [here](examples/).
## Installation
The script expect a local installation of the [openapi-generator](https://github.com/OpenAPITools/openapi-generator) tool.
For details on the installation, se [here](https://github.com/OpenAPITools/openapi-generator#1---installation).
We recommend installing it via brew:
```
brew install openapi-generator
```
To execute our wrapper script, you must install python (3.7+) and the required modules:
```
brew install python3
cd src/
pip3 install -r requirements.txt
```
## Execution
To execute the script and generate all components specified in the configuration file, simply call the python script as follows:
```
python openapi.py [openapi.json]
```
Note that the specification of the path to the configuration file is optional and defaults to `./openapi.json`.
## Configuration
The configuration is stored in a `.json` file.
It is used to specify paths and client/server stubs to be generated using the `openapi-generator`.
### Structure
A basic configuration file must contain the following fields:
```
{
"defaults": {},
"generators": []
}
```
`generators` is a list of generator configurations.
`defaults` contains the default configuration shared by all generators.
These default properties are merged with specific ones such that values are replaced, lists are appended, and dicts are merged.
### Generator
Each generator configuration must contain the following fields, specified directly or via `defaults`:
```
{
"spec": "../backend.yml",
"tempDir": "temp",
"tempSuffix": ".tmp",
"updatedFileSuffix": ".updated",
"updatedDirSuffix": "-updated",
"generator": "python-flask",
"srcSubDir": "openapi_server",
"outputDir": "backend",
"replaceInFiles": {},
"mode": "merge",
"filesToKeep": [],
"dirsToKeep": []
}
```
The fields have the following meaning:
- `spec`: openapi specification to use
- `tempDir `: path to use for the temporary code generation
- `tempSuffix `: suffix to use for temporary files
- `updatedFileSuffix `: suffix for updated files
- `updatedDirSuffix `: suffix for updated dirs
- `generator`: the generator to use (see [the official documentation](https://github.com/OpenAPITools/openapi-generator) for a complete list
- `srcSubDir`: the sub dir of generated code to use as output
- `outputDir`: the target path where to move the generated files
- `replaceInFiles`: dicts of in-file replacements (see below)
- `mode`: how to proceed if `outputDir` is not empty (`overwrite` or `merge`)
- `filesToKeep`: files that should not be overwritten in `mode` `merge`
- `dirsToKeep`: dirs that should not be overwritten in `mode` `merge`
Before processing, each configuration is merged with the default values (`defaults `) as described above.
#### `replaceInFiles`
The keys of this dict are regular expressions (see [here](https://www.w3schools.com/python/python_regex.asp), [here](https://docs.python.org/3/library/re.html), and [here](https://pythex.org/) for documentation) that describe the filenames of files to be processed.
The values are dicts which map strings to their replacement in the corresponding file(s).
An example of this field is:
```
{
"(.*\\.)(yml|yaml|py)": {
"openapi_server": "backend"
},
"__main__.py": {
"import connexion": "import connexion, os",
"app\\.run\\(port=8080\\)": "app.run(port=int(os.getenv('PORT', 8080)), debug=bool(os.getenv('DEBUG', False)))",
"openapi_server": "backend"
}
}
```
This results in the replacement of `openapi_server` with `backend` in all `.py` and `.yaml` files as well as three replacements in `__main__.py`.
#### `mode`
The `mode` specifies how to proceed with the generated files if the target dir `outputDir` is not empty, i.e., the respective component has been generated before.
So far, two modes are available:
- `overwrite`: delete all files and dirs in `outputDir` and move content from `tempDir`/`srcSubDir` there
- `merge`: replace all files/dirs not in `filesToKeep` or `dirsToKeep` (handle those as described below)
All newly created files are processed as specified in `replaceInFiles` before replacing or being merged with existing ones.
#### `mode`: `overwrite`
In this mode, all existing files and dirs in `outputDir` are deleted before the generated files from `tempDir`/`srcSubDir` are moved to `outputDir`.
#### `mode`: `merge`
In this mode, all files in `outputDir` are replaced by the files in `tempDir`/`srcSubDir` except for those specifeid in `dirsToKeep` and `filesToKeep`.
For those firs/files, the newly generated ones are suffixed by `updatedDirSuffix` or `updatedFileSuffix` resp. and moved next to their existing version in `targetDir`.
As an example, consider the following configuration and file structure after generation:
```
{
"mode": "merge",
"filesToKeep": [
"a/a1.txt",
"a/a2.txt"
],
"dirsToKeep": [
"b",
"c",
]
}
```
```
backend/ # the existing files (targetDir)
a/
a1.txt
b/
b1.txt
temp/ # the newly generated files (tempDir)
relevant/ # the files to consider (srcSubDir)
a/
a1.txt
a2.txt
b/
b1.txt
b2.txt
c/
c1.txt
c2.txt
```
This would result in the following file structure:
```
backend/
a/
a1.txt # old
a1.txt.updated # new
a2.txt # new
b/
b1.txt # old
b-updated/
b1.txt # new
b2.txt # new
c/
c1.txt # new
c2.txt # new
```
Note that `temp/relevant/b/b2.txt` is not moved to the new files from `baclend//b/b2.txt` because the whole dir is specified in `dirsToKeep` and it exists already in `backend/`.
In contrast, `c/` does not exist in `backend/` yet and is therefore moved to the target position.