Collapse whitespace according to CSS spec after converting HTML to

attributed string

Fixes #27

Tusker.xcodeproj/project.pbxproj

@@ -104,6 +104,7 @@
 		D62D2422217AA7E1005076CC /* UserActivityManager.swift in Sources */ = {isa = PBXBuildFile; fileRef = D62D2421217AA7E1005076CC /* UserActivityManager.swift */; };
 		D62D2424217ABF3F005076CC /* NSUserActivity+Extensions.swift in Sources */ = {isa = PBXBuildFile; fileRef = D62D2423217ABF3F005076CC /* NSUserActivity+Extensions.swift */; };
 		D62D2426217ABF63005076CC /* UserActivityType.swift in Sources */ = {isa = PBXBuildFile; fileRef = D62D2425217ABF63005076CC /* UserActivityType.swift */; };
+		D62FF04823D7CDD700909D6E /* AttributedStringHelperTests.swift in Sources */ = {isa = PBXBuildFile; fileRef = D62FF04723D7CDD700909D6E /* AttributedStringHelperTests.swift */; };
 		D6333B372137838300CE884A /* AttributedString+Helpers.swift in Sources */ = {isa = PBXBuildFile; fileRef = D6333B362137838300CE884A /* AttributedString+Helpers.swift */; };
 		D6333B772138D94E00CE884A /* ComposeMediaView.swift in Sources */ = {isa = PBXBuildFile; fileRef = D6333B762138D94E00CE884A /* ComposeMediaView.swift */; };
 		D6333B792139AEFD00CE884A /* Date+TimeAgo.swift in Sources */ = {isa = PBXBuildFile; fileRef = D6333B782139AEFD00CE884A /* Date+TimeAgo.swift */; };
@@ -376,6 +377,7 @@
 		D62D2421217AA7E1005076CC /* UserActivityManager.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = UserActivityManager.swift; sourceTree = "<group>"; };
 		D62D2423217ABF3F005076CC /* NSUserActivity+Extensions.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = "NSUserActivity+Extensions.swift"; sourceTree = "<group>"; };
 		D62D2425217ABF63005076CC /* UserActivityType.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = UserActivityType.swift; sourceTree = "<group>"; };
+		D62FF04723D7CDD700909D6E /* AttributedStringHelperTests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = AttributedStringHelperTests.swift; sourceTree = "<group>"; };
 		D6333B362137838300CE884A /* AttributedString+Helpers.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = "AttributedString+Helpers.swift"; sourceTree = "<group>"; };
 		D6333B762138D94E00CE884A /* ComposeMediaView.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = ComposeMediaView.swift; sourceTree = "<group>"; };
 		D6333B782139AEFD00CE884A /* Date+TimeAgo.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = "Date+TimeAgo.swift"; sourceTree = "<group>"; };
@@ -1206,6 +1208,7 @@
 			isa = PBXGroup;
 			children = (
 				D6D4DDE4212518A200E1C4BB /* TuskerTests.swift */,
+				D62FF04723D7CDD700909D6E /* AttributedStringHelperTests.swift */,
 				D6D4DDE6212518A200E1C4BB /* Info.plist */,
 			);
 			path = TuskerTests;
@@ -1743,6 +1746,7 @@
 			isa = PBXSourcesBuildPhase;
 			buildActionMask = 2147483647;
 			files = (
+				D62FF04823D7CDD700909D6E /* AttributedStringHelperTests.swift in Sources */,
 				D6D4DDE5212518A200E1C4BB /* TuskerTests.swift in Sources */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;

Tusker/Extensions/AttributedString+Helpers.swift

@@ -8,28 +8,95 @@
 
 import Foundation
 
-extension NSMutableAttributedString {
+private let ASCII_NEWLINE: unichar = 10
-    
+private let ASCII_SPACE: unichar = 32
+
+extension NSAttributedString {
+
     var fullRange: NSRange {
         return NSRange(location: 0, length: self.length)
     }
-    
+
+    /// Creates a new string with the whitespace collapsed according to the CSS Text Module Level 3 rules.
+    /// See https://www.w3.org/TR/css-text-3/#white-space-phase-1
+    func collapsingWhitespace() -> NSAttributedString {
+        let mut = NSMutableAttributedString(attributedString: self)
+        mut.collapseWhitespace()
+        return mut
+    }
+
+}
+
+extension NSMutableAttributedString {
+
     func trimLeadingCharactersInSet(_ charSet: CharacterSet) {
         var range = (string as NSString).rangeOfCharacter(from: charSet)
-        
+
         while range.length != 0 && range.location == 0 {
             replaceCharacters(in: range, with: "")
             range = (string as NSString).rangeOfCharacter(from: charSet)
         }
     }
-    
+
     func trimTrailingCharactersInSet(_ charSet: CharacterSet) {
         var range = (string as NSString).rangeOfCharacter(from: charSet, options: .backwards)
-        
+
         while range.length != 0 && range.length + range.location == length {
             replaceCharacters(in: range, with: "")
             range = (string as NSString).rangeOfCharacter(from: charSet, options: .backwards)
         }
     }
+
+    /// Collapses whitespace in this string according to the CSS Text Module Level 3 rules.
+    /// See https://www.w3.org/TR/css-text-3/#white-space-phase-1
+    func collapseWhitespace() {
+        let str = self.mutableString
+        
+        var i = 0
+        while i < str.length {
+            if str.character(at: i) == ASCII_NEWLINE {
+                var j: Int
+                if i > 0 {
+                    // scan backwards to find beginning of space characters preceeding newline
+                    j = i - 1
+                    while j >= 0 {
+                        if str.character(at: j) != ASCII_SPACE {
+                            break
+                        }
+                        j -= 1
+                    }
+                    // add one after loop completes because start of range is _inclusive_
+                    j += 1
+                } else {
+                    j = 0
+                }
+                
+                var k: Int
+                if i < str.length - 1 {
+                    // scan forwards to find end of space characters following newline
+                    k = i + 1
+                    while k < str.length {
+                        if str.character(at: k) != ASCII_SPACE {
+                            break
+                        }
+                        k += 1
+                    }
+                    // don't need to subtract one before breaking out of loop, because end of range is _exclusive_
+                } else {
+                    // range end is _exclusive_, so use whole string length that way last character is included
+                    k = str.length
+                }
+                
+                // if there's only one character to be replaced, that means we'd be replacing the newline with a newline, so don't bother
+                if k - j > 1 {
+                    str.replaceCharacters(in: NSRange(location: j, length: k - j), with: "\n")
+                    
+                    // continue scanning through the string starting after the newline we just inserted
+                    i = j
+                }
+            }
+            i += 1
+        }
+    }
     
 }

Tusker/Views/ContentTextView.swift

@@ -109,11 +109,12 @@ class ContentTextView: LinkTextView {
         let attributedText = attributedTextForHTMLNode(body)
         let mutAttrString = NSMutableAttributedString(attributedString: attributedText)
         mutAttrString.trimTrailingCharactersInSet(.whitespacesAndNewlines)
+        mutAttrString.collapseWhitespace()
         
         self.attributedText = mutAttrString
     }
     
-    func attributedTextForHTMLNode(_ node: Node, usePreformattedText: Bool = false) -> NSAttributedString {
+    private func attributedTextForHTMLNode(_ node: Node, usePreformattedText: Bool = false) -> NSAttributedString {
         switch node {
         case let node as TextNode:
             let text: String

TuskerTests/AttributedStringHelperTests.swift

@@ -0,0 +1,46 @@
+//
+//  AttributedStringHelperTests.swift
+//  TuskerTests
+//
+//  Created by Shadowfacts on 1/21/20.
+//  Copyright © 2020 Shadowfacts. All rights reserved.
+//
+
+import XCTest
+@testable import Tusker
+
+class AttributedStringHelperTests: XCTestCase {
+
+    override func setUp() {
+    }
+
+    override func tearDown() {
+    }
+
+    func testCollapsingWhitespace() {
+        var str = NSAttributedString(string: "test 1\n")
+        XCTAssertEqual(str.collapsingWhitespace(), NSAttributedString(string: "test 1\n"))
+        
+        str = NSAttributedString(string: "test 2   \n")
+        XCTAssertEqual(str.collapsingWhitespace(), NSAttributedString(string: "test 2\n"))
+        
+        str = NSAttributedString(string: "test 3\n    ")
+        XCTAssertEqual(str.collapsingWhitespace(), NSAttributedString(string: "test 3\n"))
+        
+        str = NSAttributedString(string: "test 4     \n    ")
+        XCTAssertEqual(str.collapsingWhitespace(), NSAttributedString(string: "test 4\n"))
+        
+        str = NSAttributedString(string: "test 5     \n    blah")
+        XCTAssertEqual(str.collapsingWhitespace(), NSAttributedString(string: "test 5\nblah"))
+        
+        str = NSAttributedString(string: "\ntest 6")
+        XCTAssertEqual(str.collapsingWhitespace(), NSAttributedString(string: "\ntest 6"))
+        
+        str = NSAttributedString(string: "   \ntest 7")
+        XCTAssertEqual(str.collapsingWhitespace(), NSAttributedString(string: "\ntest 7"))
+        
+        str = NSAttributedString(string: "   \n    test 8")
+        XCTAssertEqual(str.collapsingWhitespace(), NSAttributedString(string: "\ntest 8"))
+    }
+
+}