I tested out some code in a playground and it works as I would expect.
I built an extremely basic (one function, one button, one textfield) project to test the code in and it doesn't work – in fact it hangs up (beach balling).
What might cause this to happen?
Both the playground and the project import Cocoa and Foundation.
The code is below.
It appears to get hung up on this line:
let data = pipe.fileHandleForReading.readDataToEndOfFile()
Here's the code as it is written in the playground (and copied into the project):
import Cocoa
import Foundation
// *** Getting exiftool version number
func exiftoolVersion() -> String {
let task = Process()
let pipe = Pipe()
task.standardOutput = pipe
task.arguments = ["-ver"]
task.executableURL = URL(fileURLWithPath: "/usr/local/bin/exiftool")
do {
try task.run()
task.waitUntilExit()
}
catch {
}
let data = pipe.fileHandleForReading.readDataToEndOfFile()
var output = String(data: data, encoding: .utf8)!
output = output.filter { !$0.isWhitespace }
return output
}
The one way that I know would solve your issue is to go to your .entitlements file and switch turn off App Sandbox, by setting the value to false, as shown below.
<key>com.apple.security.app-sandbox</key>
<false/>
This is because in sandboxed mode you aren't allowed to execute an external program.
Edit: Since it didn't work for you, can you try replacing catch {} with
catch {
print(error)
}
and see what/if there's any error?
Related
I'm trying to write a small app to start/stop and display data from a command line "app" someone else wrote. The command executable is installed in '/usr/local/bin'. It outputs status text data to standardOutput while running. I can execute this "command" from the Terminal.app without issue. From swiftUI code I can successfully execute "built-in" commands like ls. However, when (in swiftUI code) I attempt to execute Process.run WITH the new command it throws the exception 'The file “” doesn’t exist.'
Anyone have any ideas? Thanks in advance!
Here's a code snip:
// NOTE: "installedCommand" is just a placeholder for the actual command.
let task = Process()
let connection = Pipe()
let exeUrl = URL(fileURLWithPath: "/usr/local/bin/installedCommand")
//let exeUrl = URL(fileURLWithPath: "/bin/ls") <--works fine
task.executableURL = exeUrl
task.standardOutput = connection
do
{
try task.run()
}
catch
{
print("Error: \(error.localizedDescription)")
return
}
I found article describing how to create plugin using Swift and Cocoa. It uses NSBundle to load plugin, but that, as far as I know, is not available in pure swift (no Cocoa). Is there way how to achieve same result without using Cocoa?
More info:
In case it's relevant, here is what I want to achieve. I create app in swift that runs on linux server. User can connect to it using their browser. I want to be able to have other people write "plugins" that will implement functionality itself (what user can see and do once they connect), from printing out hello world, through chat programs to games without having to worry about low level stuff provided by my app. Some sort of dll, that my server application loads and runs.
Solution to this is not trivial, but it's not impossible to do either. I prefer to use swift package manager to manage dependencies and Xcode as IDE. This combination is not perfect as it needs a lot of tinkering but there is not any other useable free swift IDE as of now.
You will need to set up two projects, let's call them Plugin (3rd party library) and PluginConsumer (app that uses other people plugins). You will also need to decide on API, for now we will use simple
TestPluginFunc()
Create Plugin.swift file with TestPluginFunc implementation in your Plugin project:
public func TestPluginFunc() {
print("Hooray!")
}
Set the project to build framework, not executable and build[1]. You will get Plugin.framework file which contains your plugin.
Now switch to your PluginConsumer project
Copy Plugin.framework from your Plugin project somewhere where you can easily find it. To actually load the framework and use it:
// we need to define how our plugin function looks like
typealias TestPluginFunc = #convention(c) ()->()
// and what is its name
let pluginFuncName = "TestPluginFunc"
func loadPlugin() {
let pluginName = "Plugin"
let openRes = dlopen("./\(pluginName).framework/\(pluginName)", RTLD_NOW|RTLD_LOCAL)
if openRes != nil {
// this is fragile
let symbolName = "_TF\(pluginName.utf8.count)\(pluginName)\(initFuncName.utf8.count)\(initFuncName)FT_T_"
let sym = dlsym(openRes, symbolName)
if sym != nil {
// here we load func from framework based on the name we constructed in "symbolName" variable
let f: TestPluginFunc = unsafeBitCast(sym, to: TestPluginFunc.self)
// and now all we need to do is execute our plugin function
f()
} else {
print("Error loading \(realPath). Symbol \(symbolName) not found.")
dlclose(openRes)
}
} else {
print("error opening lib")
}
}
If done correctly, you should see "Hooray!" being printed to your log.
There is a lot of room for improvement, first thing you should do is replace Plugin.framework string with parameter, preferably using some file library (I am using PerfectLib). Another thing to look at is defining plugin API in your PluginConsumer project as a protocol or base class, creating framework out of that, importing that framework in your plugin project and basing your implementation on that protocol/base class. I am trying to figure out exactly how to do that. I will update this post if I mange to do it properly.
[1]: I usually do this by creating Package.swift file and creating xcode project out of it using swift package generate-xcodeproj. If your project doesn't contain main.swift, xcode will create framework instead of executable
What you will want to do is create a folder your program will look in. Let's say it's called 'plugins'. It should make a list of names from the files in there, and then iterate through using them, passing parameters to the files and getting the output and making use of that in some way.
Activating a program and getting output:
func runCommand(cmd : String, args : String...) -> (output: [String], error: [String], exitCode: Int32) {
var output : [String] = []
var error : [String] = []
let task = Process()
task.launchPath = cmd
task.arguments = args
let outpipe = Pipe()
task.standardOutput = outpipe
let errpipe = Pipe()
task.standardError = errpipe
task.launch()
let outdata = outpipe.fileHandleForReading.readDataToEndOfFile()
if var string = String(data: outdata, encoding: .utf8) {
string = string.trimmingCharacters(in: .newlines)
output = string.components(separatedBy: "\n")
}
let errdata = errpipe.fileHandleForReading.readDataToEndOfFile()
if var string = String(data: errdata, encoding: .utf8) {
string = string.trimmingCharacters(in: .newlines)
error = string.components(separatedBy: "\n")
}
task.waitUntilExit()
let status = task.terminationStatus
return (output, error, status)
}
`
Here is how a swift plugin would accept arguments:
for i in 1..C_ARGC {
let index = Int(i);
let arg = String.fromCString(C_ARGV[index])
switch arg {
case 1:
println("1");
case 2:
println("2")
default:
println("3)
}
}
So once you have the program and plugin communicating you just have to add handling in your program based on the output so the plugins output can do something meaningful. Without cocoa libraries this seems the way to go, though if you use C there are a couple of other options available there as well. Hope this helps.
I have a command-line app A, and in A I execute an executable script B, in B I'm expecting an input from stdin.
I wrote a demo, implementing A in Swift, using Foundation's Process api, finding that B, no matter implemented in whatever language, cannot get user input from stdin.
Code:
// `A`'s main.swift
import Foundation
let process = Process()
process.launchPath = PATH_TO_SCRIPT_B
process.launch()
process.waitUntilExit()
// `B`
#!/usr/bin/swift
print("intpu something")
let input = readLine()
print("input: \(input)")
I did not set the process's input since according to the doc:
If this method isn’t used, the standard input is inherited from the process that created the receiver.
UPDATE:
A is an executable package created using Swift Package Manager. I used swift package generate-xcodeproj to generate an Xcode project file. I confirmed that if I run the executable built using swift build or xcodebuild in a shell, the problem with getting input from stdin from B arose. However if I run it directly inside Xcode, by pressing command + R inside Xcode, then it worked. So if I understand the difference between running an executable in a shell and Xcode, I can probably make everything work.
func task() {
print("input here")
let x = input()
print ("inputed:" + x)
}
func input() -> String {
let keyboard = FileHandle.standardInput
let inputData = keyboard.availableData
let strData = String(data: inputData, encoding: .utf8)!
let string = strData.trimmingCharacters(in: .newlines)
return string
}
task()
Hope it helps
I try to start a command file from my app, so I need to use "posix_spawn". But usage of this requires strange use of pointers. I didn't find any example of Swift (3.0), only C++ or Objective C which I couldn't translate.
I simply need something like that (in old system call):
let err = system("ls -param >file.txt")
Any ideas?
Edit:
The linked solution didn't match.
First it does not use posix_spawn function, which was mentioned by the complier. It uses NSTask, which seems also be abandoned. But I tried this example and ended up in:
func shell(launchPath: String, arguments: [String]) -> String
{
let task = Process()
task.launchPath = launchPath
task.arguments = arguments
let pipe = Pipe()
task.standardOutput = pipe
task.launch()
let data = pipe.fileHandleForReading.readDataToEndOfFile()
let output: String = NSString(data: data, encoding: String.Encoding.utf8.rawValue)! as String
return output
}
Here all changes needed for XCode 8 were made. But when calling, it will never return from ".launch()".
Somewhere in Output debug window I found this line:
2016-09-15 15:06:36.793 DSRenamer[96562:2569582] launch path not accessible
Same command works fine in terminal window:
/usr/local/bin/ExifTool -ext .CR2
I use swift to call the objective-c menthod(though not the best way)
define an objective-c util menthod
#import "SystemTaskHelper.h"
#implementation SystemTaskHelper
+(void)performSystemTask:(NSString *)task
{
system([task UTF8String]);
}
#end
import objective-c util.h into Bridging-Header.h , and in swift use
SystemTaskHelper.performSystemTask("ls -param >file.txt")
I have a simple AppleScript that sends an email. How can I call it from within a Swift application?
(I wasn't able to find the answer via Google.)
As Kamaros suggests, you can call NSApplescript directly without having to launch a separate process via NSTask (as CRGreen suggests.)
Swift Code
let myAppleScript = "..."
var error: NSDictionary?
if let scriptObject = NSAppleScript(source: myAppleScript) {
if let output: NSAppleEventDescriptor = scriptObject.executeAndReturnError(
&error) {
print(output.stringValue)
} else if (error != nil) {
print("error: \(error)")
}
}
Tested: one can do something like this (arbitrary script path added):
import Foundation
let task = Process()
task.launchPath = "/usr/bin/osascript"
task.arguments = ["~/Desktop/testscript.scpt"]
task.launch()
For anyone who is getting the warning below for Swift 4, for the line while creating an NSAppleEventDescriptor from zekel's answer
Non-optional expression of type 'NSAppleEventDescriptor' used in a check for optionals
You can get rid of it with this edited short version:
let myAppleScript = "..."
var error: NSDictionary?
if let scriptObject = NSAppleScript(source: myAppleScript) {
if let outputString = scriptObject.executeAndReturnError(&error).stringValue {
print(outputString)
} else if (error != nil) {
print("error: ", error!)
}
}
However, you may have also realized; with this method, system logs this message to console everytime you run the script:
AppleEvents: received mach msg which wasn't complex type as expected
in getMemoryReference.
Apparently it is a declared bug by an Apple staff developer, and is said to be 'just' a harmless log spam and is scheduled to be removed on future OS updates, as you can see in this very long apple developer forum post and SO question below:
AppleEvents: received mach msg which wasn't complex type as expected in getMemoryReference
Thanks Apple, for those bazillions of junk console logs thrown around.
You can try NSAppleScript, from Apple's Technical Note TN2084
Using AppleScript Scripts in Cocoa Applications
https://developer.apple.com/library/mac/technotes/tn2084/_index.html
NSAppleScript* scriptObject = [[NSAppleScript alloc] initWithSource:
#"\
set app_path to path to me\n\
tell application \"System Events\"\n\
if \"AddLoginItem\" is not in (name of every login item) then\n\
make login item at end with properties {hidden:false, path:app_path}\n\
end if\n\
end tell"];
returnDescriptor = [scriptObject executeAndReturnError: &errorDict];
I struggled few hours, but nothing worked. Finally I managed to run AppleScript through shell:
let proc = Process()
proc.launchPath = "/usr/bin/env"
proc.arguments = ["/usr/bin/osascript", "scriptPath"]
proc.launch()
Dunno is this the best way to do it, but at least it works.
As of March 2018, I think the strongest answer on this thread is still the accepted answer from 2011. The implementations that involved using NSAppleScript or OSAScript suffered the drawbacks having some minor, but highly unpleasant, memory leaks without really providing any additional benefits. Anyone struggling with getting that answer to execute properly (in Swift 4) may want to try this:
let manager = FileManager()
// Note that this assumes your .scpt file is located somewhere in the Documents directory
let script: URL? = try? manager.url(for: .documentDirectory, in: .userDomainMask, appropriateFor: nil, create: false)
if let scriptPath = script?.appendingPathComponent("/path/to/scriptName").appendingPathExtension("scpt").path {
let process = Process()
if process.isRunning == false {
let pipe = Pipe()
process.launchPath = "/usr/bin/osascript"
process.arguments = [scriptPath]
process.standardError = pipe
process.launch()
}
}
Update
The simple, accepted answer from 2011 has gotten more complex. Apple has deprecated the launch() function as of as of 10.14, suggesting that we "use run() instead". Unfortunately, it's not a direct replacement. Fortunately, most of the original answer still holds, with a simple change.
Instead of the original line:
task.launch()
when you use run() you have to use it with try to catch possible errors. The line becomes:
try task.run()
The accepted answer above then becomes:
let task = Process()
task.launchPath = "/usr/bin/osascript"
task.arguments = ["~/Desktop/testscript.scpt"]
try task.run()
However, this only works if the code is at the top level, not if it is inside a function. run() throws errors while launch() does not. If you use try inside a function, the function also has to be declared to throw errors. Or you can handle possible errors from run() inside the function.
This page from "Hacking With Swift" has examples of how to catch STDOUT and STDERR from the process.
I hope this helps. I'm very new at Swift and this is what worked for me.