I am trying to understand the best ways in which to bring native library code into a NativeScript plugin for iOS. I've had success in past bringing in a CocoaPod and accessing the symbols from that. So I want to create my own.
I follow the process for pod lib create TestPod to generate my library project. (https://guides.cocoapods.org/making/using-pod-lib-create.html)
The 'library' is a trivial test: it simply creates a class with a function that returns
a recognizable greeting string.
The associated "Example app" demo produces a text label that displays this string.
This all works as expected, all running in pure iOS world (written in Swift).
At the Nativescript side, I'm using the Nativescript Plugin Seed and I'm declaring my Podfile in the src/platforms/iOS folder as directed. My "plugin" (trivial as it is as a test), has the iOS-specific parts in the pluginName.ios.js file. I have a separate test method in here that verifies I can reach and use native iOS platform symbols (e.g. NSMutableString) and that works as expected. I want to do the same thing with native code imported from my library.
But when I bring the Podfile into Nativescript, it builds okay, but I'm not able to see any of the symbols as I would expect to see.
I generate typings and I don't find them either, but I do find a "TestPod.d.ts" typings file that declares some version info and a mysterious class named "UITest" that I did not define and bears no resemblance to my "SimpleTest" class, which I can't find anywhere.
I'm sure I'm missing something here that is probably obvious to the knowledgeable. But I'm unable to guess what it is. Any ideas?
import Foundation
#objc public class SimpleTest : NSObject {
public func announce() -> String {
return "Greetings from Swift code in a library"
}
}
I'm calling it in the plugin using the following:
public testNativeLib() : string {
let str:string;
try {
const testClass = new SimpleTest()
str = testClass.announce()
} catch(e) {
str = e.message;
}
return str;
}
and what I get returned for str is the catch case error message: "Can't find variable: SimpleTest"
Without seeing the source to the pod file; the only guess would be that you didn't use the #objc on anything that you wanted exposed. Without you exposing anything with #objc (or descending from a native objc class, so it is tagged by swift automagically) NativeScript cannot see it.
Please note their are some types in Swift that currently cannot be consumed by NativeScript (Or ObjC) and you have to write some wrapper code in swift around it using types that are compatible with ObjC so that NativeScript can use it.
Please see: https://docs.nativescript.org/guides/ios-source-code for more info on requirements...
The code needs to be:
import Foundation
public class SimpleTest : NSObject {
#objc public func announce() -> String {
return "Greetings from Swift code in a library"
}
}
Move the #objc to the actual property/function you want exposed. The NSObject will already cause the class to be exposed if it has exposed members.
The metadata generated:
A complete demo using a swift source file in a plugin is now located at
https://github.com/NathanaelA/demo-swift-plugin
If I have full control over the source, then I won't normally add another moving piece (i.e. cocoapod) to the mix; just let xcode compile the swift code and expose it, no need to add any additional places something can break.
However, if you want to see how to do an actual swift CocoaPod plugin; checkout any of these plugin repos that use swift code and cocoapod:
https://github.com/tomvardasca/nativescript-crypto
https://github.com/arpit2438735/nativescript-tglib
https://github.com/Daltron/NotificationBanner
In the case of an actual cocoapod; you need to have a valid Podfile, and a valid podspec file! Once you have those; when you build the application, you can open up xcode and verify that the cocoapod is linked in. If it isn't linked in then your podfile/podspec is messed up and has to do with a issue with Cocoapod and it is not a NativeScript issue. So in that case; you need to follow some Cocoapod tutorials to get it to work.
Please note; nuking your platform folder frequently while you are testing with cocoapods is highly recommended. Occasionally Nativescript does not detect the changes to Native code properly and so it then doesn't rebuild the xcode project/workspace files. So nuking your platforms folder will of course force it to rebuild them. If you don't you WILL waste a lot of time while messing around with cocoapods.
When running the sample project you will see this:
In addition, in the demo, the metadata generated from the code I saved in the demo/metadata folder so you can look at it.
Related
basically I have some Objective-c class already existing and I want to use them in my NativeScript projet. Currently I have added those files to the Xcode project target and I want to be able to call my objective-c code from nativeScript js. I've read the doc but I don't understand it. It seems so complicated. basically currently all I want to do is be able to present my custom view controller by calling probably something along
const vc = MyCustomViewController.alloc.initWithNibName("xib file")
page.frame.ios.controller.present(vc,true, nil)
Am I obligated to create a plugin for that? Am I obligated to use my objective-c class to build a framework in Xcode and then import the framework?
So I found out.
Actually what you need to do is first to compile your native iOS code into a framework. As per the documentation all your classes must inherit NSObject and all your function must be marked with #objc to be exposed to the objective-c Nativescript side if you write in swift. You will notice as well that in a framework your bundle is not the main bundle. In this example you can see how you can retrieve the bundle from the framework and load a xib from it.
Then you need to add your framework file to a Nativescript plugin. For that, you want to add the framework to the plugin's iOS folder yourPlugin/platform/iOS/yourFramework.framework.
then, you need to add your plugin to your app. You can add your local plugin by using the next command line. Notice the path end with the /src folder.
tns plugin add /path/to/yourplugin/src
Now, you can then call your native functions and classes without even importing them. Of course this works only on iOS. If you run your app on android, calling those methods will crash.
To show this viewController on your Nativescript side you will need to call the following code. By the way You can find documentation elsewhere to get a reference to the current page or frame object.
const controller = page.frame.ios.controller
const vc = IOMediaViewController.create()
vc.modalPresentationStyle = UIModalPresentationFullScreen
controller.presentViewControllerAnimatedCompletion(vc,true,null)
I have a custom Swift framework in a Cocoapod, wrapped in a NativeScript plugin. My NativeScript TypeScript code can do everything it needs to access classes and methods in the classes in the framework. That in itself was a lot of work to figure out all the nuances. Now I'm at the last part, which is to downcast a custom viewController in the framework from UIViewController. The following code is working, in that NativeScript actually loads my storyboard and controller:
var podBundle = NSBundle.bundleForClass(CustomViewController)
var bundleUrl = podBundle.URLForResourceWithExtension("MyFramework", "bundle")
var bundle = NSBundle.bundleWithURL(bundleUrl)
var storyboard = UIStoryboard.storyboardWithNameBundle("Main", bundle)
var viewController = storyboard.instantiateViewControllerWithIdentifier("CustomViewControllerID")
console.log(viewController)
console.log(viewController instanceof CustomViewController)
application.ios.rootController.presentViewControllerAnimatedCompletion(viewController, true, null)
The first console.log actually logs the full class, as in MyFramework.CustomViewController: hexcodehere
The second console.log logs true as well, meaning it knows the type of the viewController variable is indeed CustomViewController.
The problem is, I need to call a function in CustomViewController, so I need to downcast:
storyboard.instantiateViewControllerWithIdentifier("CustomViewControllerID") as CustomViewController
But the compiler says "Cannot find name" to CustomViewController. If I try MyFramework.CustomViewController, it says "Cannot find name" to MyFramework. Notice NSBundle.bundleForClass(CustomViewController) works. So that means CustomViewController is known to the compiler.
If I import the plugin at the top of the file:
import { CustomViewController } from "myplugin", it says "cannot find module" to myplugin. Mind you I can actually access all the classes in this plugin. So the plugin is definitely properly installed. It is installed from a local file path, not git.
It seems I'm missing something devilishly simple. But I can't figure it out for the life of me.
I'm happy to help others who want to use native iOS storyboard and Swift classes, etc.
Attesting to the brain's ability to work on a problem in the background, it suddenly came to me. Since the runtime is aware of the type, all I have to do is to cast to "any" to get around the compiler issue:
var viewController = storyboard.instantiateViewControllerWithIdentifier("CustomViewControllerID") as any
Then the compiler allows me to call whatever function on my CustomViewController viewcontroller instance without a complaint, and runtime is fine too. To be extra safe, wrap the calls to subclass methods in a if instanceof block.
I've got an iOS specific code (class definition) and would like to include in a separate ios specific file.
So I created two files
utils2.ios.ts <-- added my code here
utils2.android.ts
And then from another file, I did:
import utils2 = require("./utils2");
But I'm getting an error message that the module is not found.
I had to go this route because I've got a class definition in
the code and if I have just one file, I get a run-time error on Android.
Does typescript support platform specific files?
Or another option - is something like this possible in ts file
if ios
do this
end
Similar to C preprocesser
You can get by with just a utils.ts file and inside it you can do platform specific logic that you mention. Your issue about the module not found is likely just a path issue to that module, not that the module isn't actually there. Also note that Typescript doesn't really "support" platform specific files. Nativescript takes those .android.ts and .ios.ts files and at runtime loads the file you need on the platform running. So just FYI that's not really TS related :)
Nativescript provides a platform module in the tns-core-modules. Import the { isAndroid } or { isIOS } from the platform module in your code and use it as the
if (isAndroid) {
// do android only here
} else {
// ios
}
I'm trying to use Hyperloop in order to use a third-party open source library. The library I want to use is Sweet Alert iOS.
The library has a swift file. I have put this file inside an src directory inside my project's main folder (does it have to be inside src folder? can I use other folder in my project? what about sub-directories?).
According to the Titanium documentation:
Any *.swift files found in your src directories will automatically be compiled
In my app when I do something like this:
var UIView = require('UIKit/UIView');
Ti.API.info('UIView => ' + UIView);
It works. But if I try to require any of the classes in the swift file I get an error that it can't find architecture x86_64 module.
var sweetAlert = require('SweetAlert');
var cancelAnimatedView = require('CancelAnimatedView');
When looking at the SweetAlert source code, I also notice that SweetAlert class is open while other classes are just regular classes:
open class SweetAlert
I'm not a swift developer so I don't really know what it means.
What am I missing here?
EDIT
Some progress... So I found that I need to have an appc.js file that's defining the src frame work and a name for MyFramework. And that the require I do should match the filename of the swift file. Now when I compile I see some more interesting output about the swift compilation, but getting an error:
[INFO] Generating metabase for swift MyFramework /Users/ophir/Documents/Appcelerator_Studio_Workspace/HyperloopTest1/src/SweetAlert.swift
2017-06-29T14:43:44.061Z | ERROR | An uncaught exception was thrown!
Cannot read property '1' of null
2017-06-29T14:43:44.064Z | ERROR | Cannot read property '1' of null
EDIT 2
I've cleared all the script in the swift file, and narrowed it down to the following failing script:
import Foundation
import UIKit
import QuartzCore
public enum AlertStyle {
case success,error,warning,none
case customImag(imageFile:String)
}
class SweetAlert: UIViewController {
}
class AnimatableView: UIView {
func animate(){
}
}
If I remove the func animate() { } it will not fail. Even if this function is inside that one SweetAlert class then it will fail. And from what I've read this is a pretty standard swift code.
So this makes me wonder - is Hyperloop 2.0.0(!) ready for production apps?
I receive a similar error when creating a simple Swift class and including it in a Titanium/Appcelerator project with the appc.js file set up just how they show in the sparse documentation. The project will compile with no issues and run in the Xcode Simulator, as long as I don't reference the class from the Swift file. If I reference that class in the project, the error arises. Based on the compile log, Swift files do compile into the project, but the classes don't seem to properly link in the end for "require"-ing into the JavaScript code.
When I search online for solutions, this same issue shows up all over, and the Titanium/Appcelerator team never presents a solution. I've tried Titanium SDK 6.0.1, 6.1.2, and 6.2.0, with all the same results; I've tried adding Swift code to existing projects and to new ones; I've also uninstalled and reinstalled Titanium/Appcelerator. (I've even tried running their sample app with no luck!)
Hyperloop does work when I use native frameworks, like UIKit, but it doesn't seem ready for production apps for 3rd party frameworks and classes.
I hope my discoveries help you and others to debug projects, but, unfortunately, I don't think I will be able to offer much help outside that.
I'm having problems importing a class (which is written in Swift and from a pod) into my Objective-C code. I think the problem is due to the class failing to appear in my autogenerated -Swift.h header file, but I don't know how to fix this.
My Podfile has the use_frameworks! declaration and I have the pod installed and I'm able to use the pod's class in any of my Swift code but not in Obj-C.
The Swift class I'm trying to import is an extension of UITableViewCell, which is declared as a public class. It didn't have the class declared with the #objc annotation in its source but adding it myself didn't seem to make a difference either.
In my Obj-C source I have #import "ProjectName-Swift.h", (with my actual project name) but my Obj-C file doesn't still doesn't recognise the class after a clean and build. I have checked that the Swift header file exists (and it does) but upon inspection it doesn't seem to have any mention of the Swift class that I'm trying to import.
I tried putting in a forward #class declaration to the class in my Obj-C source, but this only removed the warning and delayed the error until I try to build. Modules are enabled in my build settings. If anyone has some insight, your help would be much appreciated.
Edit:
If I manually add the source file to Build Phases -> Compile Sources for my project, then everything works - but of course this defeats the purpose of using a pod. So this seems to be an XCode search path issue.
Only public classes and methods are published to be used from other modules. To ensure Objective-C compatibility, you can annotate your classes with #objc and inherit NSObject.
For example, change your class definition from:
class MyClass {
func sayHello() {}
}
To:
#objc public class MyClass: NSObject {
public func sayHello() {}
}
To use Swift frameworks (iOS 8+) in Objective-C on XCode 7, you just have to import them like you would in Swift using the #import statement:
#import MyModule;
And all your public classes will be available.
More information in 'Importing External Frameworks' and 'Using Swift from Objective-C' sections of the Swift and Objective-C in the Same Project apple docs.
I ended up finding a solution. It seems that Swift pods aren't added to the Project-Name-Swift.h, but instead a separate Swift header is generated for each module.
What I needed to do was import the following Swift header file.
#import "Module_Name/Module_Name-Swift.h"
An alternate solution that eluded me was pointed out by redent84. You just need to put the following in your Objective-C source as you would in Swift.
#import Module_Name;
which is probably equivalent to the swift header import once the sugar is stripped away.