[File] Fixed out-of-bounds exception and inaccurate parsed declaratio…

…ns (fixes #35)
jpsim · Mar 26, 2015 · 73941ec · 73941ec
1 parent 9622e6b
commit 73941ec
Show file tree

Hide file tree

Showing 5 changed files with 52 additions and 71 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -15,6 +15,12 @@ None.
   [JP Simard](https://github.com/jpsim)
   [#30](https://github.com/jpsim/SourceKitten/issues/30)
 
+* Fixed out-of-bounds exception and inaccurate parsed declarations when using
+  multibyte characters.  
+  [JP Simard](https://github.com/jpsim)
+  [#35](https://github.com/jpsim/SourceKitten/issues/35)
+
+
 ## 0.3.0
 
 ##### Breaking

diff --git a/Source/SourceKittenFramework/File.swift b/Source/SourceKittenFramework/File.swift
@@ -15,8 +15,6 @@ public struct File {
     public let path: String?
     /// File contents.
     public let contents: NSString
-    /// Array of newline offsets in contents.
-    public let lineBreaks: [Int]
 
     /**
     Failable initializer by path. Fails if file contents could not be read as a UTF8 string.
@@ -27,7 +25,6 @@ public struct File {
         self.path = path
         if let contents = NSString(contentsOfFile: path, encoding: NSUTF8StringEncoding, error: nil) {
             self.contents = contents
-            lineBreaks = contents.lineBreaks()
         } else {
             fputs("Could not read contents of `\(path)`\n", stderr)
             return nil
@@ -42,7 +39,6 @@ public struct File {
     public init(contents: NSString) {
         path = nil
         self.contents = contents
-        lineBreaks = contents.lineBreaks()
     }
 
     /**
@@ -52,18 +48,28 @@ public struct File {
 
     :returns: Source declaration if successfully parsed.
     */
-    private func parseDeclaration(dictionary: XPCDictionary) -> String? {
+    public func parseDeclaration(dictionary: XPCDictionary) -> String? {
         if !shouldParseDeclaration(dictionary) {
             return nil
         }
-        let offset = Int(SwiftDocKey.getOffset(dictionary)!)
-        let lineBreakIndexBeforeOffset = indexBefore(offset, inSequence: lineBreaks) ?? 0
-        let previousLineBreakOffset = lineBreaks[lineBreakIndexBeforeOffset] + 1
-        if let bodyOffset = SwiftDocKey.getBodyOffset(dictionary) {
-            return contents.filteredSubstring(previousLineBreakOffset, end: Int(bodyOffset))
+        if let offset = flatMap(SwiftDocKey.getOffset(dictionary), {Int($0)}) {
+            let stringContents = contents as String
+            let bodyOffsetIndex = flatMap(SwiftDocKey.getBodyOffset(dictionary)) {
+                advance(stringContents.utf8.startIndex, Int($0))
+            }
+            var lineStart = stringContents.startIndex
+            var lineEnd = stringContents.endIndex
+            let declarationStart = advance(stringContents.utf8.startIndex, offset).samePositionIn(stringContents)!
+            let declarationEnd = (bodyOffsetIndex ?? advance(stringContents.utf8.startIndex, offset+1)).samePositionIn(stringContents)!
+            stringContents.getLineStart(&lineStart,
+                end: &lineEnd,
+                contentsEnd: nil,
+                forRange: declarationStart..<declarationEnd)
+            let unwantedSet = NSCharacterSet.whitespaceAndNewlineCharacterSet().mutableCopy() as! NSMutableCharacterSet
+            unwantedSet.addCharactersInString("{")
+            return stringContents[lineStart..<lineEnd].stringByTrimmingCharactersInSet(unwantedSet)
         }
-        let nextLineBreakOffset = (lineBreakIndexBeforeOffset + 1 < lineBreaks.count) ? lineBreaks[lineBreakIndexBeforeOffset + 1] : lineBreaks.last!
-        return contents.filteredSubstring(previousLineBreakOffset, end: nextLineBreakOffset + 1)
+        return nil
     }
 
     /**
@@ -250,21 +256,22 @@ public struct File {
     internal func shouldTreatAsSameFile(dictionary: XPCDictionary) -> Bool {
         return path == SwiftDocKey.getFilePath(dictionary)
     }
-}
 
-/**
-Returns true if the input dictionary contains a parseable declaration.
+    /**
+    Returns true if the input dictionary contains a parseable declaration.
 
-:param: dictionary Dictionary to parse.
+    :param: dictionary Dictionary to parse.
 
-:returns: True if the input dictionary contains a parseable declaration.
-*/
-private func shouldParseDeclaration(dictionary: XPCDictionary) -> Bool {
-    let hasTypeName = SwiftDocKey.getTypeName(dictionary) != nil
-    let hasAnnotatedDeclaration = SwiftDocKey.getAnnotatedDeclaration(dictionary) != nil
-    let hasOffset = SwiftDocKey.getOffset(dictionary) != nil
-    let isntExtension = SwiftDocKey.getKind(dictionary) != SwiftDeclarationKind.Extension.rawValue
-    return hasTypeName && hasAnnotatedDeclaration && hasOffset && isntExtension
+    :returns: True if the input dictionary contains a parseable declaration.
+    */
+    private func shouldParseDeclaration(dictionary: XPCDictionary) -> Bool {
+        let sameFile = shouldTreatAsSameFile(dictionary)
+        let hasTypeName = SwiftDocKey.getTypeName(dictionary) != nil
+        let hasAnnotatedDeclaration = SwiftDocKey.getAnnotatedDeclaration(dictionary) != nil
+        let hasOffset = SwiftDocKey.getOffset(dictionary) != nil
+        let isntExtension = SwiftDocKey.getKind(dictionary) != SwiftDeclarationKind.Extension.rawValue
+        return sameFile && hasTypeName && hasAnnotatedDeclaration && hasOffset && isntExtension
+    }
 }
 
 /**

diff --git a/Source/SourceKittenFramework/NSString+SourceKitten.swift b/Source/SourceKittenFramework/NSString+SourceKitten.swift
@@ -35,41 +35,6 @@ extension NSString {
         return NSString.pathWithComponents([rootDirectory, self]).stringByStandardizingPath
     }
 
-    /**
-    Returns offsets of all the line breaks in the current string.
-    */
-    public func lineBreaks() -> [Int] {
-        var lineBreaks = [-1] // Count start of file
-        var searchRange = NSRange(location: 0, length: length)
-        while searchRange.length > 0 {
-            searchRange.length = length - searchRange.location
-            let foundRange = rangeOfString("\n", options: nil, range: searchRange)
-            if foundRange.location != NSNotFound {
-                lineBreaks.append(foundRange.location)
-                searchRange.location = foundRange.location + foundRange.length
-            } else {
-                break
-            }
-        }
-        return lineBreaks
-    }
-
-    /**
-    Filter self from start to end while trimming unwanted characters (whitespace & '{').
-
-    :param: start Starting offset to filter.
-    :param: end Ending offset to filter.
-
-    :returns: Filtered string.
-    */
-    public func filteredSubstring(start: Int, end: Int) -> String {
-        precondition(end > start, "Range should be positive")
-        let range = NSRange(location: start, length: end - start - 1)
-        let unwantedSet = NSCharacterSet.whitespaceAndNewlineCharacterSet().mutableCopy() as! NSMutableCharacterSet
-        unwantedSet.addCharactersInString("{")
-        return substringWithRange(range).stringByTrimmingCharactersInSet(unwantedSet)
-    }
-
     /**
     Find integer offsets of documented Swift tokens in self.
 

diff --git a/Source/SourceKittenFrameworkTests/Fixtures/Commandant.json b/Source/SourceKittenFrameworkTests/Fixtures/Commandant.json
@@ -517,7 +517,7 @@
         {
           "key.kind" : "source.lang.swift.decl.var.static",
           "key.offset" : 3222,
-          "key.parsed_declaration" : "static var name: String",
+          "key.parsed_declaration" : "static var name: String { get }",
           "key.namelength" : 4,
           "key.bodyoffset" : 3236,
           "key.bodylength" : 5,
@@ -1126,7 +1126,7 @@
         {
           "key.kind" : "source.lang.swift.decl.var.instance",
           "key.offset" : 384,
-          "key.parsed_declaration" : "var verb: String",
+          "key.parsed_declaration" : "var verb: String { get }",
           "key.namelength" : 4,
           "key.bodyoffset" : 398,
           "key.bodylength" : 5,
@@ -1151,7 +1151,7 @@
         {
           "key.kind" : "source.lang.swift.decl.var.instance",
           "key.offset" : 496,
-          "key.parsed_declaration" : "var function: String",
+          "key.parsed_declaration" : "var function: String { get }",
           "key.namelength" : 8,
           "key.bodyoffset" : 514,
           "key.bodylength" : 5,
@@ -1329,7 +1329,7 @@
         {
           "key.kind" : "source.lang.swift.decl.function.method.instance",
           "key.offset" : 1225,
-          "key.parsed_declaration" : "public init()",
+          "key.parsed_declaration" : "public init() {}",
           "key.namelength" : 6,
           "key.bodyoffset" : 1233,
           "key.bodylength" : 0,

diff --git a/Source/SourceKittenFrameworkTests/StringTests.swift b/Source/SourceKittenFrameworkTests/StringTests.swift
@@ -8,6 +8,7 @@
 
 import Foundation
 import SourceKittenFramework
+import SwiftXPC
 import XCTest
 
 class StringTests: XCTestCase {
@@ -42,15 +43,17 @@ class StringTests: XCTestCase {
         XCTAssertEqual(__FILE__.absolutePathRepresentation(), __FILE__, "absolutePathRepresentation() should return the caller if it's already an absolute path")
     }
 
-    func testParseLineBreaks() {
-        XCTAssertEqual("a\nbc\nd\n".lineBreaks(), [-1, 1, 4, 6], "should parse line breaks")
-    }
-
     func testFilteredSubstring() {
-        let expected = "public func myFunc()"
-        let end = count(expected) + 4 // 4 == 2 spaces before + 2 characters after (until newline)
-        let actual = ("  \(expected) {\n}" as NSString).filteredSubstring(0, end: end)
-        XCTAssertEqual(expected, actual, "should extract function declaration from source text")
+        let dict = [
+            "key.kind": "source.lang.swift.decl.class",
+            "key.offset": Int64(24),
+            "key.bodyoffset": Int64(32),
+            "key.annotated_decl": "",
+            "key.typename": "ClassA.Type"
+        ] as XPCDictionary
+        // This string is a regression test for https://github.com/jpsim/SourceKitten/issues/35.
+        let file = File(contents: "/**\n　ほげ\n*/\nclass ClassA {\n}\n")
+        XCTAssertEqual("class ClassA", file.parseDeclaration(dict)!, "should extract declaration from source text")
     }
 
     func testGenerateDocumentedTokenOffsets() {