parse table cell spans in cmark and use the information (#74)

QuietMisdreavus · web-flow · commit 395fbf23fc09 · 2022-09-21T09:09:44.000-06:00
rdar://98017807
diff --git a/Sources/Markdown/Base/RawMarkup.swift b/Sources/Markdown/Base/RawMarkup.swift
@@ -49,7 +49,18 @@ enum RawMarkupData: Equatable {
     case tableHead
     case tableBody
     case tableRow
-    case tableCell
+    case tableCell(colspan: UInt, rowspan: UInt)
+}
+
+extension RawMarkupData {
+    func isTableCell() -> Bool {
+        switch self {
+        case .tableCell:
+            return true
+        default:
+            return false
+        }
+    }
 }
 
 /// The header for the `RawMarkup` managed buffer.
@@ -297,12 +308,12 @@ final class RawMarkup: ManagedBuffer<RawMarkupHeader, RawMarkup> {
     }
 
     static func tableRow(parsedRange: SourceRange?, _ columns: [RawMarkup]) -> RawMarkup {
-        precondition(columns.allSatisfy { $0.header.data == .tableCell })
+        precondition(columns.allSatisfy { $0.header.data.isTableCell() })
         return .create(data: .tableRow, parsedRange: parsedRange, children: columns)
     }
 
     static func tableHead(parsedRange: SourceRange?, columns: [RawMarkup]) -> RawMarkup {
-        precondition(columns.allSatisfy { $0.header.data == .tableCell })
+        precondition(columns.allSatisfy { $0.header.data.isTableCell() })
         return .create(data: .tableHead, parsedRange: parsedRange, children: columns)
     }
 
@@ -311,8 +322,8 @@ final class RawMarkup: ManagedBuffer<RawMarkupHeader, RawMarkup> {
         return .create(data: .tableBody, parsedRange: parsedRange, children: rows)
     }
 
-    static func tableCell(parsedRange: SourceRange?, _ children: [RawMarkup]) -> RawMarkup {
-        return .create(data: .tableCell, parsedRange: parsedRange, children: children)
+    static func tableCell(parsedRange: SourceRange?, colspan: UInt, rowspan: UInt, _ children: [RawMarkup]) -> RawMarkup {
+        return .create(data: .tableCell(colspan: colspan, rowspan: rowspan), parsedRange: parsedRange, children: children)
     }
 }
 
diff --git a/Sources/Markdown/Block Nodes/Tables/TableCell.swift b/Sources/Markdown/Block Nodes/Tables/TableCell.swift
@@ -30,10 +30,48 @@ extension Table {
 
 public extension Table.Cell {
 
+    /// The number of columns this cell spans over.
+    ///
+    /// A normal, non-spanning table cell has a `colspan` of 1. A value greater than one indicates
+    /// that this cell has expanded to cover up that number of columns. A value of zero means that
+    /// this cell is being covered up by a previous cell in the same row.
+    var colspan: UInt {
+        get {
+            guard case let .tableCell(colspan, _) = _data.raw.markup.data else {
+                fatalError("\(self) markup wrapped unexpected \(_data.raw)")
+            }
+            return colspan
+        }
+        set {
+            _data = _data.replacingSelf(.tableCell(parsedRange: nil, colspan: newValue, rowspan: rowspan, _data.raw.markup.copyChildren()))
+        }
+    }
+
+    /// The number of rows this cell spans over.
+    ///
+    /// A normal, non-spanning table cell has a `rowspan` of 1. A value greater than one indicates
+    /// that this cell has expanded to cover up that number of rows. A value of zero means that
+    /// this cell is being covered up by another cell in a row above it.
+    var rowspan: UInt {
+        get {
+            guard case let .tableCell(_, rowspan) = _data.raw.markup.data else {
+                fatalError("\(self) markup wrapped unexpected \(_data.raw)")
+            }
+            return rowspan
+        }
+        set {
+            _data = _data.replacingSelf(.tableCell(parsedRange: nil, colspan: colspan, rowspan: newValue, _data.raw.markup.copyChildren()))
+        }
+    }
+
     // MARK: BasicInlineContainer
 
     init<Children>(_ children: Children) where Children : Sequence, Children.Element == InlineMarkup {
-        try! self.init(RawMarkup.tableCell(parsedRange: nil, children.map { $0.raw.markup }))
+        self.init(colspan: 1, rowspan: 1, children)
+    }
+
+    init<Children>(colspan: UInt, rowspan: UInt, _ children: Children) where Children : Sequence, Children.Element == InlineMarkup {
+        try! self.init(RawMarkup.tableCell(parsedRange: nil, colspan: colspan, rowspan: rowspan, children.map { $0.raw.markup }))
     }
 
     // MARK: Visitation
diff --git a/Sources/Markdown/Parser/CommonMarkConverter.swift b/Sources/Markdown/Parser/CommonMarkConverter.swift
@@ -570,17 +570,22 @@ struct MarkupParser {
         precondition(state.nodeType == .tableCell)
         let parsedRange = state.range(state.node)
         let childConversion = convertChildren(state)
+        let colspan = UInt(cmark_gfm_extensions_get_table_cell_colspan(state.node))
+        let rowspan = UInt(cmark_gfm_extensions_get_table_cell_rowspan(state.node))
         precondition(childConversion.state.node == state.node)
         precondition(childConversion.state.event == CMARK_EVENT_EXIT)
-        return MarkupConversion(state: childConversion.state.next(), result: .tableCell(parsedRange: parsedRange, childConversion.result))
+        return MarkupConversion(state: childConversion.state.next(), result: .tableCell(parsedRange: parsedRange, colspan: colspan, rowspan: rowspan, childConversion.result))
     }
 
     static func parseString(_ string: String, source: URL?, options: ParseOptions) -> Document {
         cmark_gfm_core_extensions_ensure_registered()
+
+        var cmarkOptions = CMARK_OPT_TABLE_SPANS
+        if !options.contains(.disableSmartOpts) {
+            cmarkOptions |= CMARK_OPT_SMART
+        }
         
-        let parser = cmark_parser_new(options.contains(.disableSmartOpts)
-                                      ? CMARK_OPT_DEFAULT
-                                      : CMARK_OPT_SMART)
+        let parser = cmark_parser_new(cmarkOptions)
         
         cmark_parser_attach_syntax_extension(parser, cmark_find_syntax_extension("table"))
         cmark_parser_attach_syntax_extension(parser, cmark_find_syntax_extension("strikethrough"))
diff --git a/Sources/Markdown/Walker/Walkers/MarkupFormatter.swift b/Sources/Markdown/Walker/Walkers/MarkupFormatter.swift
@@ -899,15 +899,37 @@ public struct MarkupFormatter: MarkupWalker {
             $0.formatIndependently(options: cellFormattingOptions)
         }).ensuringCount(atLeast: uniformColumnCount, filler: "")
 
+        /// All of the column-span values from the head cells, adding cells as
+        /// needed to meet the uniform `uniformColumnCount`.
+        let headCellSpans = Array(table.head.cells.map {
+            $0.colspan
+        }).ensuringCount(atLeast: uniformColumnCount, filler: 1)
+
         /// All of the independently formatted body cells' text by row, adding
         /// cells to each row to meet the `uniformColumnCount`.
         let bodyRowTexts = Array(table.body.rows.map { row -> [String] in
             return Array(row.cells.map {
-                $0.formatIndependently(options: cellFormattingOptions)
+                if $0.rowspan == 0 {
+                    // If this cell is being spanned over, replace its text
+                    // (which should be the empty string anyway) with the
+                    // rowspan marker.
+                    return "^"
+                } else {
+                    return $0.formatIndependently(options: cellFormattingOptions)
+                }
             }).ensuringCount(atLeast: uniformColumnCount,
                              filler: "")
         })
 
+        /// All of the column- and row-span information for the body cells,
+        /// cells to each row to meet the `uniformColumnCount`.
+        let bodyRowSpans = Array(table.body.rows.map { row in
+            return Array(row.cells.map {
+                (colspan: $0.colspan, rowspan: $0.rowspan)
+            }).ensuringCount(atLeast: uniformColumnCount,
+                             filler: (colspan: 1, rowspan: 1))
+        })
+
         // Next, calculate the maximum width of each column.
 
         /// The column alignments of the table, filled out to `uniformColumnCount`.
@@ -952,15 +974,32 @@ public struct MarkupFormatter: MarkupWalker {
             }
         }
 
+        /// Calculate the width of the given column and colspan.
+        ///
+        /// This adds up the appropriate column widths based on the given column span, including
+        /// the default span of 1, where it will only return the `finalColumnWidths` value for the
+        /// given `column`.
+        func columnWidth(column: Int, colspan: Int) -> Int {
+            let lastColumn = column + colspan
+            return (column..<lastColumn).map({ finalColumnWidths[$0] }).reduce(0, { $0 + $1 })
+        }
+
         // We now know the width that each printed column will be.
 
         /// Each of the header cells expanded to the right dimensions by
         /// extending each line with spaces to fit the uniform column width.
         let expandedHeaderCellTexts = (0..<uniformColumnCount)
             .map { column -> String in
-                let minLineLength = finalColumnWidths[column]
-                return headCellTexts[column]
-                    .ensuringCount(atLeast: minLineLength, filler: " ")
+                let colspan = headCellSpans[column]
+                if colspan == 0 {
+                    // If this cell is being spanned over, collapse it so it
+                    // can be filled with the spanning cell.
+                    return ""
+                } else {
+                    let minLineLength = columnWidth(column: column, colspan: Int(colspan))
+                    return headCellTexts[column]
+                        .ensuringCount(atLeast: minLineLength, filler: " ")
+                }
             }
 
         /// Rendered delimter row cells with the correct width.
@@ -988,10 +1027,18 @@ public struct MarkupFormatter: MarkupWalker {
         /// appropriately for their row and column.
         let expandedBodyRowTexts = bodyRowTexts.enumerated()
             .map { (row, rowCellTexts) -> [String] in
+                let rowSpans = bodyRowSpans[row]
                 return (0..<uniformColumnCount).map { column -> String in
-                    let minLineLength = finalColumnWidths[column]
-                    return rowCellTexts[column]
-                        .ensuringCount(atLeast: minLineLength, filler: " ")
+                    let colspan = rowSpans[column].colspan
+                    if colspan == 0 {
+                        // If this cell is being spanned over, collapse it so it
+                        // can be filled with the spanning cell.
+                        return ""
+                    } else {
+                        let minLineLength = columnWidth(column: column, colspan: Int(colspan))
+                        return rowCellTexts[column]
+                            .ensuringCount(atLeast: minLineLength, filler: " ")
+                    }
                 }
             }
 
diff --git a/Sources/Markdown/Walker/Walkers/MarkupTreeDumper.swift b/Sources/Markdown/Walker/Walkers/MarkupTreeDumper.swift
@@ -258,4 +258,20 @@ struct MarkupTreeDumper: MarkupWalker {
     mutating func visitSymbolLink(_ symbolLink: SymbolLink) {
         dump(symbolLink, customDescription: symbolLink.destination.map { "destination: \($0)" })
     }
+
+    mutating func visitTableCell(_ tableCell: Table.Cell) {
+        var desc = ""
+        if tableCell.colspan != 1 {
+            desc += " colspan: \(tableCell.colspan)"
+        }
+        if tableCell.rowspan != 1 {
+            desc += " rowspan: \(tableCell.rowspan)"
+        }
+        desc = desc.trimmingCharacters(in: .whitespaces)
+        if !desc.isEmpty {
+            dump(tableCell, customDescription: desc)
+        } else {
+            dump(tableCell)
+        }
+    }
 }
diff --git a/Tests/MarkdownTests/Block Nodes/TableTests.swift b/Tests/MarkdownTests/Block Nodes/TableTests.swift
@@ -205,4 +205,40 @@ class TableTests: XCTestCase {
         """
         XCTAssertEqual(expectedDump, document.debugDescription(options: .printSourceLocations))
     }
+
+    func testParseCellSpans() {
+        let source = """
+        | one | two | three |
+        | --- | --- | ----- |
+        | big      || small |
+        | ^        || small |
+        """
+
+        let document = Document(parsing: source)
+
+        let expectedDump = """
+        Document @1:1-4:22
+        └─ Table @1:1-4:22 alignments: |-|-|-|
+           ├─ Head @1:1-1:22
+           │  ├─ Cell @1:2-1:7
+           │  │  └─ Text @1:3-1:6 "one"
+           │  ├─ Cell @1:8-1:13
+           │  │  └─ Text @1:9-1:12 "two"
+           │  └─ Cell @1:14-1:21
+           │     └─ Text @1:15-1:20 "three"
+           └─ Body @3:1-4:22
+              ├─ Row @3:1-3:22
+              │  ├─ Cell @3:2-3:12 colspan: 2 rowspan: 2
+              │  │  └─ Text @3:3-3:6 "big"
+              │  ├─ Cell @3:13-3:14 colspan: 0
+              │  └─ Cell @3:14-3:21
+              │     └─ Text @3:15-3:20 "small"
+              └─ Row @4:1-4:22
+                 ├─ Cell @4:2-4:12 colspan: 2 rowspan: 0
+                 ├─ Cell @4:13-4:14 colspan: 0
+                 └─ Cell @4:14-4:21
+                    └─ Text @4:15-4:20 "small"
+        """
+        XCTAssertEqual(expectedDump, document.debugDescription(options: .printSourceLocations))
+    }
 }
diff --git a/Tests/MarkdownTests/Visitors/MarkupFormatterTests.swift b/Tests/MarkdownTests/Visitors/MarkupFormatterTests.swift
@@ -1265,9 +1265,9 @@ class MarkupFormatterTableTests: XCTestCase {
               │     └─ Link destination: "https://swift.org"
               │        └─ Text "https://swift.org"
               └─ Row
-                 ├─ Cell
+                 ├─ Cell colspan: 2
                  │  └─ InlineHTML <br/>
-                 ├─ Cell
+                 ├─ Cell colspan: 0
                  └─ Cell
         """
         XCTAssertEqual(expectedDump, document.debugDescription())
@@ -1277,7 +1277,58 @@ class MarkupFormatterTableTests: XCTestCase {
         |*A*                       |**B**                 |~C~                |
         |:-------------------------|:--------------------:|------------------:|
         |[Apple](https://apple.com)|![image](image.png "")|<https://swift.org>|
-        |<br/>                     |                      |                   |
+        |<br/>                                           ||                   |
+        """
+
+        XCTAssertEqual(expected, formatted)
+        print(formatted)
+
+        let reparsed = Document(parsing: formatted)
+        print(reparsed.debugDescription())
+        XCTAssertTrue(document.hasSameStructure(as: reparsed))
+    }
+
+    func testRoundTripRowspan() {
+        let source = """
+        | one | two | three |
+        | --- | --- | ----- |
+        | big      || small |
+        | ^        || small |
+        """
+
+        let document = Document(parsing: source)
+
+        let expectedDump = """
+        Document
+        └─ Table alignments: |-|-|-|
+           ├─ Head
+           │  ├─ Cell
+           │  │  └─ Text "one"
+           │  ├─ Cell
+           │  │  └─ Text "two"
+           │  └─ Cell
+           │     └─ Text "three"
+           └─ Body
+              ├─ Row
+              │  ├─ Cell colspan: 2 rowspan: 2
+              │  │  └─ Text "big"
+              │  ├─ Cell colspan: 0
+              │  └─ Cell
+              │     └─ Text "small"
+              └─ Row
+                 ├─ Cell colspan: 2 rowspan: 0
+                 ├─ Cell colspan: 0
+                 └─ Cell
+                    └─ Text "small"
+        """
+        XCTAssertEqual(expectedDump, document.debugDescription())
+
+        let formatted = document.format()
+        let expected = """
+        |one|two|three|
+        |---|---|-----|
+        |big   ||small|
+        |^     ||small|
         """
 
         XCTAssertEqual(expected, formatted)

Original file line number	Diff line number	Diff line change
`@@ -49,7 +49,18 @@ enum RawMarkupData: Equatable {`
`49`	`49`	`case tableHead`
`50`	`50`	`case tableBody`
`51`	`51`	`case tableRow`
`52`		`- case tableCell`
	`52`	`+ case tableCell(colspan: UInt, rowspan: UInt)`
	`53`	`+}`
	`54`	`+`
	`55`	`+extension RawMarkupData {`
	`56`	`+ func isTableCell() -> Bool {`
	`57`	`+ switch self {`
	`58`	`+ case .tableCell:`
	`59`	`+ return true`
	`60`	`+ default:`
	`61`	`+ return false`
	`62`	`+ }`
	`63`	`+ }`
`53`	`64`	`}`
`54`	`65`
`55`	`66`	/// The header for the `RawMarkup` managed buffer.
`@@ -297,12 +308,12 @@ final class RawMarkup: ManagedBuffer<RawMarkupHeader, RawMarkup> {`
`297`	`308`	`}`
`298`	`309`
`299`	`310`	`static func tableRow(parsedRange: SourceRange?, _ columns: [RawMarkup]) -> RawMarkup {`
`300`		`- precondition(columns.allSatisfy { $0.header.data == .tableCell })`
	`311`	`+ precondition(columns.allSatisfy { $0.header.data.isTableCell() })`
`301`	`312`	`return .create(data: .tableRow, parsedRange: parsedRange, children: columns)`
`302`	`313`	`}`
`303`	`314`
`304`	`315`	`static func tableHead(parsedRange: SourceRange?, columns: [RawMarkup]) -> RawMarkup {`
`305`		`- precondition(columns.allSatisfy { $0.header.data == .tableCell })`
	`316`	`+ precondition(columns.allSatisfy { $0.header.data.isTableCell() })`
`306`	`317`	`return .create(data: .tableHead, parsedRange: parsedRange, children: columns)`
`307`	`318`	`}`
`308`	`319`
`@@ -311,8 +322,8 @@ final class RawMarkup: ManagedBuffer<RawMarkupHeader, RawMarkup> {`
`311`	`322`	`return .create(data: .tableBody, parsedRange: parsedRange, children: rows)`
`312`	`323`	`}`
`313`	`324`
`314`		`- static func tableCell(parsedRange: SourceRange?, _ children: [RawMarkup]) -> RawMarkup {`
`315`		`- return .create(data: .tableCell, parsedRange: parsedRange, children: children)`
	`325`	`+ static func tableCell(parsedRange: SourceRange?, colspan: UInt, rowspan: UInt, _ children: [RawMarkup]) -> RawMarkup {`
	`326`	`+ return .create(data: .tableCell(colspan: colspan, rowspan: rowspan), parsedRange: parsedRange, children: children)`
`316`	`327`	`}`
`317`	`328`	`}`
`318`	`329`