Swift: loading data from url causes semaphore_wait_trap to freeze
In my application, clicking a button allows me to download data from an internet site. The site is a list of links containing binary data. Sometimes the first link may not contain the correct data. In this case, the application takes the next reference in the array and gets data from there. The links are correct.
The problem is that often (not always) the app freezes for a few seconds when I press the button. After 5-30 seconds, it defrosts and loads the instruments normally. I understand something is blocking the main thread. When you stop the process in xCode I get this (marked by semaphore_wait_trap):
This is how I do it:
// Button Action
@IBAction func downloadWindNoaa(_ sender: UIButton)
{
// Starts activity indicator
startActivityIndicator()
// Starts downloading and processing data
// Either use this
DispatchQueue.global(qos: .default).async
{
DispatchQueue.main.async
{
self.downloadWindsAloftData()
}
}
// Or this - no difference.
//downloadWindsAloftData()
}
}
func downloadWindsAloftData()
{
// Creates a list of website addresses to request data: CHECKED.
self.listOfLinks = makeGribWebAddress()
// Extract and save the data
saveGribFile()
}
// This downloads the data and saves it in a required format. I suspect, this is the culprit
func saveGribFile()
{
// Check if the links have been created
if (!self.listOfLinks.isEmpty)
{
/// Instance of OperationQueue
queue = OperationQueue()
// Convert array of Strings to array of URL links
let urls = self.listOfLinks.map { URL(string: $0)! }
guard self.urlIndex != urls.count else
{
NSLog("report failure")
return
}
// Current link
let url = urls[self.urlIndex]
// Increment the url index
self.urlIndex += 1
// Add operation to the queue
queue.addOperation { () -> Void in
// Variables for Request, Queue, and Error
let request = URLRequest(url: url)
let session = URLSession.shared
// Array of bytes that will hold the data
var dataReceived = [UInt8]()
// Read data
let task = session.dataTask(with: request) {(data, response, error) -> Void in
if error != nil
{
print("Request transport error")
}
else
{
let response = response as! HTTPURLResponse
let data = data!
if response.statusCode == 200
{
//Converting data to String
dataReceived = [UInt8](data)
}
else
{
print("Request server-side error")
}
}
// Main thread
OperationQueue.main.addOperation(
{
// If downloaded data is less than 2 KB in size, repeat the operation
if dataReceived.count <= 2000
{
self.saveGribFile()
}
else
{
self.setWindsAloftDataFromGrib(gribData: dataReceived)
// Reset the URL Index back to 0
self.urlIndex = 0
}
}
)
}
task.resume()
}
}
}
// Processing data further
func setWindsAloftDataFromGrib(gribData: [UInt8])
{
// Stops spinning activity indicator
stopActivityIndicator()
// Other code to process data...
}
// Makes Web Address
let GRIB_URL = "http://xxxxxxxxxx"
func makeGribWebAddress() -> [String]
{
var finalResult = [String]()
// Main address site
let address1 = "http://xxxxxxxx"
// Address part with type of data
let address2 = "file=gfs.t";
let address4 = "z.pgrb2.1p00.anl&lev_250_mb=on&lev_450_mb=on&lev_700_mb=on&var_TMP=on&var_UGRD=on&var_VGRD=on"
let leftlon = "0"
let rightlon = "359"
let toplat = "90"
let bottomlat = "-90"
// Address part with coordinates
let address5 = "&leftlon="+leftlon+"&rightlon="+rightlon+"&toplat="+toplat+"&bottomlat="+bottomlat
// Vector that includes all Grib files available for download
let listOfFiles = readWebToString()
if (!listOfFiles.isEmpty)
{
for i in 0..<listOfFiles.count
{
// Part of the link that includes the file
let address6 = "&dir=%2F"+listOfFiles[i]
// Extract time: last 2 characters
let address3 = listOfFiles[i].substring(from:listOfFiles[i].index(listOfFiles[i].endIndex, offsetBy: -2))
// Make the link
let addressFull = (address1 + address2 + address3 + address4 + address5 + address6).trimmingCharacters(in: .whitespacesAndNewlines)
finalResult.append(addressFull)
}
}
return finalResult;
}
func readWebToString() -> [String]
{
// Final array to return
var finalResult = [String]()
guard let dataURL = NSURL(string: self.GRIB_URL)
else
{
print("IGAGribReader error: No URL identified")
return []
}
do
{
// Get contents of the page
let contents = try String(contentsOf: dataURL as URL)
// Regular expression
let expression : String = ">gfs\\.\\d+<"
let range = NSRange(location: 0, length: contents.characters.count)
do
{
// Match the URL content with regex expression
let regex = try NSRegularExpression(pattern: expression, options: NSRegularExpression.Options.caseInsensitive)
let contentsNS = contents as NSString
let matches = regex.matches(in: contents, options: [], range: range)
for match in matches
{
for i in 0..<match.numberOfRanges
{
let resultingNS = contentsNS.substring(with: (match.rangeAt(i))) as String
finalResult.append(resultingNS)
}
}
// Remove "<" and ">" from the strings
if (!finalResult.isEmpty)
{
for i in 0..<finalResult.count
{
finalResult[i].remove(at: finalResult[i].startIndex)
finalResult[i].remove(at: finalResult[i].index(before: finalResult[i].endIndex))
}
}
}
catch
{
print("IGAGribReader error: No regex match")
}
}
catch
{
print("IGAGribReader error: URL content is not read")
}
return finalResult;
}
I've been trying to fix this for the past few weeks, but to no avail. Any help would be much appreciated!
source to share
The stack trace tells you what it stops at String(contentsOf:)
, gets called readWebToString
, gets called makeGribWebAddress
.
The problem is what String(contentsOf:)
makes a synchronous network request. If this request takes some time, it will block this thread. And if you call this from the main thread, your application might hang.
In theory, you could just send this process to the background, but that just hides a deeper problem: you are making a network request with a synchronous, non-invalidate API and you are not providing meaningful error reporting.
You should really be making asynchronous requests from URLSession
, as elsewhere. Avoid using String(contentsOf:)
with a remote url.
source to share
let contents = try String(contentsOf: dataURL as URL)
You are calling String(contentsOf: url)
on the main thread (main queue). This loads the content of the url into the string synchronously. The main thread is used to control the user interface, running synchronous network code will freeze the user interface. It's a big no, no .
You don't have to call readWebToString()
in the main queue. Execution DispatchQueue.main.async { self.downloadWindsAloftData() }
precisely puts the block on the main queue, which we should avoid. ( async
just means "do it later", it runs on anyway Dispatch.main
.)
You should just run downloadWindsAloftData
on the global queue instead of the main queue
DispatchQueue.global(qos: .default).async {
self.downloadWindsAloftData()
}
Run DispatchQueue.main.async
if you want to update the interface.
source to share