store ordered list start index from cmark (#22)

Author: QuietMisdreavus

Sources/Markdown/Base/RawMarkup.swift

@@ -24,7 +24,7 @@ enum RawMarkupData: Equatable {
     case thematicBreak
     case htmlBlock(String)
     case listItem(checkbox: Checkbox?)
-    case orderedList
+    case orderedList(startIndex: UInt = 1)
     case unorderedList
     case paragraph
     case blockDirective(name: String, nameLocation: SourceLocation?, arguments: DirectiveArgumentText)
@@ -228,8 +228,8 @@ final class RawMarkup: ManagedBuffer<RawMarkupHeader, RawMarkup> {
         return .create(data: .listItem(checkbox: checkbox), parsedRange: parsedRange, children: children)
     }
 
-    static func orderedList(parsedRange: SourceRange?, _ children: [RawMarkup]) -> RawMarkup {
-        return .create(data: .orderedList, parsedRange: parsedRange, children: children)
+    static func orderedList(parsedRange: SourceRange?, _ children: [RawMarkup], startIndex: UInt = 1) -> RawMarkup {
+        return .create(data: .orderedList(startIndex: startIndex), parsedRange: parsedRange, children: children)
     }
 
     static func unorderedList(parsedRange: SourceRange?, _ children: [RawMarkup]) -> RawMarkup {

Sources/Markdown/Block Nodes/Block Container Blocks/OrderedList.swift

@@ -32,6 +32,26 @@ public extension OrderedList {
         try! self.init(.orderedList(parsedRange: nil, items.map { $0.raw.markup }))
     }
 
+    /// The starting index for the list.
+    ///
+    /// The default starting index in CommonMark is 1. In this case, clients may use the default
+    /// ordered-list start index of their desired rendering format. For example, when rendering to
+    /// HTML, clients may omit the `start` attribute of the rendered list when this returns 1.
+    var startIndex: UInt {
+        get {
+            guard case let .orderedList(start) = _data.raw.markup.data else {
+                fatalError("\(self) markup wrapped unexpected \(_data.raw)")
+            }
+            return start
+        }
+        set {
+            guard startIndex != newValue else {
+                return
+            }
+            _data = _data.replacingSelf(.orderedList(parsedRange: nil, _data.raw.markup.copyChildren(), startIndex: newValue))
+        }
+    }
+
     // MARK: Visitation
 
     func accept<V: MarkupVisitor>(_ visitor: inout V) -> V.Result {

Sources/Markdown/Parser/CommonMarkConverter.swift

@@ -299,7 +299,8 @@ struct MarkupParser {
         case CMARK_BULLET_LIST:
             return MarkupConversion(state: childConversion.state.next(), result: .unorderedList(parsedRange: parsedRange, childConversion.result))
         case CMARK_ORDERED_LIST:
-            return MarkupConversion(state: childConversion.state.next(), result: .orderedList(parsedRange: parsedRange, childConversion.result))
+            let cmarkStart = UInt(cmark_node_get_list_start(state.node))
+            return MarkupConversion(state: childConversion.state.next(), result: .orderedList(parsedRange: parsedRange, childConversion.result, startIndex: cmarkStart))
         default:
             fatalError("cmark reported a list node but said its list type is CMARK_NO_LIST?")
         }

Sources/Markdown/Walker/Walkers/MarkupFormatter.swift

@@ -496,6 +496,7 @@ public struct MarkupFormatter: MarkupWalker {
             return nil
         }
         let numeral: UInt
+        // FIXME: allow `orderedListNumerals` to defer to the user-authored starting index (#76, rdar://99970544)
         switch formattingOptions.orderedListNumerals {
         case let .allSame(n):
             numeral = n

Sources/Markdown/Walker/Walkers/MarkupTreeDumper.swift

@@ -189,6 +189,14 @@ struct MarkupTreeDumper: MarkupWalker {
         dump(heading, customDescription: "level: \(heading.level)")
     }
 
+    mutating func visitOrderedList(_ orderedList: OrderedList) {
+        if orderedList.startIndex != 1 {
+            dump(orderedList, customDescription: "startIndex: \(orderedList.startIndex)")
+        } else {
+            defaultVisit(orderedList)
+        }
+    }
+
     mutating func visitCodeBlock(_ codeBlock: CodeBlock) {
         let lines = indentLiteralBlock(codeBlock.code, from: codeBlock, countLines: false)
         dump(codeBlock, customDescription: "language: \(codeBlock.language ?? "none")\n\(lines)")

Tests/MarkdownTests/Visitors/Everything.md

@@ -12,6 +12,9 @@
 
 > BlockQuote
 
+2. flour
+2. sugar
+
 ```swift
 func foo() {
     let x = 1

Tests/MarkdownTests/Visitors/MarkupFormatterTests.swift

@@ -144,6 +144,33 @@ class MarkupFormatterSingleElementTests: XCTestCase {
         }
     }
 
+    func testPrintOrderedListCustomStart() {
+        let options = MarkupFormatter.Options(orderedListNumerals: .allSame(2))
+        do { // no checkbox
+            let expected = "2. A list item."
+            var renderedList = OrderedList(ListItem(Paragraph(Text("A list item."))))
+            renderedList.startIndex = 2
+            let printed = renderedList.format(options: options)
+            XCTAssertEqual(expected, printed)
+        }
+        do { // unchecked
+            let expected = "2. [ ] A list item."
+            var renderedList = OrderedList(ListItem(checkbox: .unchecked,
+                                                    Paragraph(Text("A list item."))))
+            renderedList.startIndex = 2
+            let printed = renderedList.format(options: options)
+            XCTAssertEqual(expected, printed)
+        }
+        do { // checked
+            let expected = "2. [x] A list item."
+            var renderedList = OrderedList(ListItem(checkbox: .checked,
+                                                    Paragraph(Text("A list item."))))
+            renderedList.startIndex = 2
+            let printed = renderedList.format(options: options)
+            XCTAssertEqual(expected, printed)
+        }
+    }
+
     func testPrintParagraph() {
         let expected = "A paragraph."
         let printed = Paragraph(Text("A paragraph.")).format()
@@ -426,13 +453,13 @@ class MarkupFormatterOptionsTests: XCTestCase {
         3. C
         """
         let allSame = """
-        0. A
-        0. B
-        0. C
+        1. A
+        1. B
+        1. C
         """
         do {
             let document = Document(parsing: incrementing)
-            let printed = document.format(options: .init(orderedListNumerals: .allSame(0)))
+            let printed = document.format(options: .init(orderedListNumerals: .allSame(1)))
             XCTAssertEqual(allSame, printed)
         }
 
@@ -917,7 +944,7 @@ class MarkupFormatterLineSplittingTests: XCTestCase {
 
         let expectedTreeDump = """
         Document
-        └─ OrderedList
+        └─ OrderedList startIndex: 1000
            └─ ListItem
               └─ Paragraph
                  ├─ Text "Really really"

Tests/MarkdownTests/Visitors/MarkupTreeDumperTests.swift

@@ -14,7 +14,7 @@ import XCTest
 final class MarkupTreeDumperTests: XCTestCase {
     func testDumpEverything() {
         let expectedDump = """
-        Document @1:1-39:90 Root #\(everythingDocument.raw.metadata.id.rootId) #0
+        Document @1:1-42:90 Root #\(everythingDocument.raw.metadata.id.rootId) #0
         ├─ Heading @1:1-1:9 #1 level: 1
         │  └─ Text @1:3-1:9 #2 "Header"
         ├─ Paragraph @3:1-3:65 #3
@@ -55,37 +55,44 @@ final class MarkupTreeDumperTests: XCTestCase {
         ├─ BlockQuote @13:1-13:13 #38
         │  └─ Paragraph @13:3-13:13 #39
         │     └─ Text @13:3-13:13 #40 "BlockQuote"
-        ├─ CodeBlock @15:1-19:4 #41 language: swift
+        ├─ OrderedList @15:1-17:1 #41 startIndex: 2
+        │  ├─ ListItem @15:1-15:9 #42
+        │  │  └─ Paragraph @15:4-15:9 #43
+        │  │     └─ Text @15:4-15:9 #44 "flour"
+        │  └─ ListItem @16:1-17:1 #45
+        │     └─ Paragraph @16:4-16:9 #46
+        │        └─ Text @16:4-16:9 #47 "sugar"
+        ├─ CodeBlock @18:1-22:4 #48 language: swift
         │  func foo() {
         │      let x = 1
         │  }
-        ├─ CodeBlock @21:5-22:1 #42 language: none
+        ├─ CodeBlock @24:5-25:1 #49 language: none
         │  // Is this real code? Or just fantasy?
-        ├─ Paragraph @23:1-23:31 #43
-        │  ├─ Text @23:1-23:12 #44 "This is an "
-        │  ├─ Link @23:12-23:30 #45 destination: "topic://autolink"
-        │  │  └─ Text @23:13-23:29 #46 "topic://autolink"
-        │  └─ Text @23:30-23:31 #47 "."
-        ├─ ThematicBreak @25:1-26:1 #48
-        ├─ HTMLBlock @27:1-29:5 #49
+        ├─ Paragraph @26:1-26:31 #50
+        │  ├─ Text @26:1-26:12 #51 "This is an "
+        │  ├─ Link @26:12-26:30 #52 destination: "topic://autolink"
+        │  │  └─ Text @26:13-26:29 #53 "topic://autolink"
+        │  └─ Text @26:30-26:31 #54 "."
+        ├─ ThematicBreak @28:1-29:1 #55
+        ├─ HTMLBlock @30:1-32:5 #56
         │  <a href="foo.png">
         │  An HTML Block.
         │  </a>
-        ├─ Paragraph @31:1-31:33 #50
-        │  ├─ Text @31:1-31:14 #51 "This is some "
-        │  ├─ InlineHTML @31:14-31:17 #52 <p>
-        │  ├─ Text @31:17-31:28 #53 "inline html"
-        │  ├─ InlineHTML @31:28-31:32 #54 </p>
-        │  └─ Text @31:32-31:33 #55 "."
-        ├─ Paragraph @33:1-34:6 #56
-        │  ├─ Text @33:1-33:7 #57 "line"
-        │  ├─ LineBreak #58
-        │  └─ Text @34:1-34:6 #59 "break"
-        ├─ Paragraph @36:1-37:6 #60
-        │  ├─ Text @36:1-36:5 #61 "soft"
-        │  ├─ SoftBreak #62
-        │  └─ Text @37:1-37:6 #63 "break"
-        └─ HTMLBlock @39:1-39:90 #64
+        ├─ Paragraph @34:1-34:33 #57
+        │  ├─ Text @34:1-34:14 #58 "This is some "
+        │  ├─ InlineHTML @34:14-34:17 #59 <p>
+        │  ├─ Text @34:17-34:28 #60 "inline html"
+        │  ├─ InlineHTML @34:28-34:32 #61 </p>
+        │  └─ Text @34:32-34:33 #62 "."
+        ├─ Paragraph @36:1-37:6 #63
+        │  ├─ Text @36:1-36:7 #64 "line"
+        │  ├─ LineBreak #65
+        │  └─ Text @37:1-37:6 #66 "break"
+        ├─ Paragraph @39:1-40:6 #67
+        │  ├─ Text @39:1-39:5 #68 "soft"
+        │  ├─ SoftBreak #69
+        │  └─ Text @40:1-40:6 #70 "break"
+        └─ HTMLBlock @42:1-42:90 #71
            <!-- Copyright (c) 2021 Apple Inc and the Swift Project authors. All Rights Reserved. -->
         """
         print(everythingDocument.debugDescription(options: [.printEverything]))