Files
kotlin-fork/kotlin-native/samples/python_extension/README.md
T
Stanislav Erokhin f624800b84 Move everything under kotlin-native folder
I was forced to manually do update the following files, because otherwise
they would be ignored according .gitignore settings. Probably they
should be deleted from repo.

Interop/.idea/compiler.xml
Interop/.idea/gradle.xml
Interop/.idea/libraries/Gradle__org_jetbrains_kotlin_kotlin_runtime_1_0_3.xml
Interop/.idea/libraries/Gradle__org_jetbrains_kotlin_kotlin_stdlib_1_0_3.xml
Interop/.idea/modules.xml
Interop/.idea/modules/Indexer/Indexer.iml
Interop/.idea/modules/Runtime/Runtime.iml
Interop/.idea/modules/StubGenerator/StubGenerator.iml
backend.native/backend.native.iml
backend.native/bc.frontend/bc.frontend.iml
backend.native/cli.bc/cli.bc.iml
backend.native/cli.bc/src/org/jetbrains/kotlin/cli/bc/K2Native.kt
backend.native/cli.bc/src/org/jetbrains/kotlin/cli/bc/K2NativeCompilerArguments.kt
backend.native/tests/link/lib/foo.kt
backend.native/tests/link/lib/foo2.kt
backend.native/tests/teamcity-test.property
2020-10-27 21:00:28 +03:00

115 lines
4.7 KiB
Markdown

# C to Kotlin/Native interoperability example
This example shows how to use Kotlin/Native programs from other execution environments, such as Python.
Python has C native interface, which could be used to organize a bridge with the
Kotlin/Native application. File `kotlin_bridge.c` contains translation code between Kotlin and Python
lands. This demo works on Linux, Windows (64-bit) or macOS hosts.
To build and run the sample do the following:
* Install Python with development headers, i.e. on Linux
```
sudo apt install python-dev
```
* Run `build.sh`, it will ask for superuser password to install Python extension.
Use build.bat on Windows.
* On macOS copy Kotlin binary to extension's directory and change install name with
`install_name_tool` tool, i.e.
```
sudo cp ./build/libserver.dylib /Library/Python/2.7/site-packages/
sudo install_name_tool /Library/Python/2.7/site-packages/kotlin_bridge.so \
-change libserver.dylib @loader_path/libserver.dylib
```
* On Linux copy Kotlin binary in some place where libraries could be loaded from, i.e.
```
cp ./build/libserver.so /usr/local/lib/
ldconfig
```
or modify dynamic loader search path with
```
export LD_LIBRARY_PATH=`pwd`
```
* run Python code using Kotlin functionality with
```
python src/main/python/main.py
```
* it will show you result of using several Kotlin/Native APIs, accepting and returning both objects and
primitive types
The example works as following. Kotlin/Native API is implemented in `Server.kt`, and we run Kotlin/Native compiler
with `-produce dynamic` option. Compiler produces two artifacts: `server_api.h` which is C language API
to all public functions and classes available in the application. `libserver.dylib` or `libserver.so` or `server.dll`
shared object contains C bridge to all above APIs.
This C bridge looks like a C struct, reflecting all scopes in program, with operations available. For example,
for class Server
```c_cpp
class Server(val prefix: String) {
fun greet(session: Session) = "$prefix: Hello from Kotlin/Native in ${session}"
fun concat(session: Session, a: String, b: String) = "$prefix: $a $b in ${session}"
fun add(session: Session, a: Int, b: Int) = a + b + session.number
}
```
following C API is produced
```c_cpp
typedef struct {
server_KNativePtr pinned;
} server_kref_demo_Session;
typedef struct {
server_KNativePtr pinned;
} server_kref_demo_Server;
typedef struct {
/* Service functions. */
void (*DisposeStablePointer)(server_KNativePtr ptr);
void (*DisposeString)(const char* string);
server_KBoolean (*IsInstance)(server_KNativePtr ref, const server_KType* type);
/* User functions. */
struct {
struct {
struct {
server_KType* (*_type)(void);
server_kref_demo_Session (*Session)(const char* name, server_KInt number);
} Session;
struct {
server_KType* (*_type)(void);
server_kref_demo_Server (*Server)(const char* prefix);
const char* (*greet)(server_kref_demo_Server thiz, server_kref_demo_Session session);
const char* (*concat)(server_kref_demo_Server thiz, server_kref_demo_Session session, const char* a, const char* b);
server_KInt (*add)(server_kref_demo_Server thiz, server_kref_demo_Session session, server_KInt a, server_KInt b);
} Server;
} demo;
} kotlin;
} server_ExportedSymbols;
extern server_ExportedSymbols* server_symbols(void);
```
So every class instance is represented with a single element structure, encapsulating stable pointer to an instance.
Once no longer needed, `DisposeStablePointer()` with that stable pointer shall be called, and if value is not stored
somewhere else - it is disposed. For primitive types and `kotlin.String` smart bridges converting to C primitive types
or to C strings (which has to be manually freed with `DisposeString()`) are implemented.
For example, running constructor of class Server taking a string will look like
server_kref_demo_Server server = server_symbols()->kotlin.demo.Server.Server("the server");
And disposing no longer needed instance will look like
server_symbols()->DisposeStablePointer(server.pinned);
To make code easier readable, macro definitions like
#define T_(name) server_kref_demo_ ## name
#define __ server_symbols()->
will transform above, overly verbose lines to more readable
T_(Server) server = __ kotlin.demo.Server.Server("the server");
`_type()` function will return opaque type pointer, which could be checked with `IsInstance()` operation, like
__ IsInstance(ref.pinned, __ kotlin.demo.Server._type())