LibMarkdown: Implement introspection of the document tree

Userland/Libraries/LibMarkdown/Block.h

@@ -6,8 +6,10 @@
 
 #pragma once
 
+#include <AK/RecursionDecision.h>
 #include <AK/StringView.h>
 #include <AK/Vector.h>
+#include <LibMarkdown/Forward.h>
 
 namespace Markdown {
 
@@ -17,6 +19,7 @@ public:
 
     virtual String render_to_html(bool tight = false) const = 0;
     virtual String render_for_terminal(size_t view_width = 0) const = 0;
+    virtual RecursionDecision walk(Visitor&) const = 0;
 };
 
 }

Userland/Libraries/LibMarkdown/BlockQuote.cpp

@@ -6,6 +6,7 @@
 
 #include <AK/StringBuilder.h>
 #include <LibMarkdown/BlockQuote.h>
+#include <LibMarkdown/Visitor.h>
 
 namespace Markdown {
 
@@ -24,6 +25,15 @@ String BlockQuote::render_for_terminal(size_t view_width) const
     return m_contents->render_for_terminal(view_width);
 }
 
+RecursionDecision BlockQuote::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    return m_contents->walk(visitor);
+}
+
 OwnPtr<BlockQuote> BlockQuote::parse(LineIterator& lines)
 {
     lines.push_context(LineIterator::Context::block_quote());

Userland/Libraries/LibMarkdown/BlockQuote.h

@@ -22,6 +22,7 @@ public:
 
     virtual String render_to_html(bool tight = false) const override;
     virtual String render_for_terminal(size_t view_width = 0) const override;
+    virtual RecursionDecision walk(Visitor&) const override;
 
     static OwnPtr<BlockQuote> parse(LineIterator& lines);
 

Userland/Libraries/LibMarkdown/CodeBlock.cpp

@@ -7,6 +7,7 @@
 #include <AK/StringBuilder.h>
 #include <LibJS/MarkupGenerator.h>
 #include <LibMarkdown/CodeBlock.h>
+#include <LibMarkdown/Visitor.h>
 #include <LibRegex/Regex.h>
 
 namespace Markdown {
@@ -54,6 +55,22 @@ String CodeBlock::render_for_terminal(size_t) const
     return builder.build();
 }
 
+RecursionDecision CodeBlock::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    rd = visitor.visit(m_code);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    // Don't recurse on m_language and m_style.
+
+    // Normalize return value.
+    return RecursionDecision::Continue;
+}
+
 static Regex<ECMA262> style_spec_re("\\s*([\\*_]*)\\s*([^\\*_\\s]*).*");
 
 OwnPtr<CodeBlock> CodeBlock::parse(LineIterator& lines)

Userland/Libraries/LibMarkdown/CodeBlock.h

@@ -25,6 +25,7 @@ public:
 
     virtual String render_to_html(bool tight = false) const override;
     virtual String render_for_terminal(size_t view_width = 0) const override;
+    virtual RecursionDecision walk(Visitor&) const override;
     static OwnPtr<CodeBlock> parse(LineIterator& lines);
 
 private:

Userland/Libraries/LibMarkdown/ContainerBlock.cpp

@@ -12,6 +12,7 @@
 #include <LibMarkdown/List.h>
 #include <LibMarkdown/Paragraph.h>
 #include <LibMarkdown/Table.h>
+#include <LibMarkdown/Visitor.h>
 
 namespace Markdown {
 
@@ -50,6 +51,21 @@ String ContainerBlock::render_for_terminal(size_t view_width) const
     return builder.build();
 }
 
+RecursionDecision ContainerBlock::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    for (auto const& block : m_blocks) {
+        rd = block.walk(visitor);
+        if (rd == RecursionDecision::Break)
+            return rd;
+    }
+
+    return RecursionDecision::Continue;
+}
+
 template<typename BlockType>
 static bool try_parse_block(LineIterator& lines, NonnullOwnPtrVector<Block>& blocks)
 {

Userland/Libraries/LibMarkdown/ContainerBlock.h

@@ -27,6 +27,7 @@ public:
 
     virtual String render_to_html(bool tight = false) const override;
     virtual String render_for_terminal(size_t view_width = 0) const override;
+    virtual RecursionDecision walk(Visitor&) const override;
 
     static OwnPtr<ContainerBlock> parse(LineIterator& lines);
 

Userland/Libraries/LibMarkdown/Document.cpp

@@ -8,6 +8,7 @@
 #include <AK/StringBuilder.h>
 #include <LibMarkdown/Document.h>
 #include <LibMarkdown/LineIterator.h>
+#include <LibMarkdown/Visitor.h>
 
 namespace Markdown {
 
@@ -41,6 +42,15 @@ String Document::render_for_terminal(size_t view_width) const
     return m_container->render_for_terminal(view_width);
 }
 
+RecursionDecision Document::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    return m_container->walk(visitor);
+}
+
 OwnPtr<Document> Document::parse(const StringView& str)
 {
     const Vector<StringView> lines_vec = str.lines();

Userland/Libraries/LibMarkdown/Document.h

@@ -23,6 +23,17 @@ public:
     String render_to_inline_html() const;
     String render_for_terminal(size_t view_width = 0) const;
 
+    /*
+     * Walk recursively through the document tree. Returning `RecursionDecision::Recurse` from
+     * `Visitor::visit` proceeds with the next element of the pre-order walk, usually a child element.
+     * Returning `RecursionDecision::Continue` skips the subtree, and usually proceeds with the next
+     * sibling. Returning `RecursionDecision::Break` breaks the recursion, with no further calls to
+     * any of the `Visitor::visit` methods.
+     *
+     * Note that `walk()` will only return `RecursionDecision::Continue` or `RecursionDecision::Break`.
+     */
+    RecursionDecision walk(Visitor&) const;
+
     static OwnPtr<Document> parse(const StringView&);
 
 private:

Userland/Libraries/LibMarkdown/Forward.h

@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2021, Ben Wiederhake <BenWiederhake.GitHub@gmx.de>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+namespace Markdown {
+
+class Block;
+class Document;
+class Text;
+
+class BlockQuote;
+class CodeBlock;
+class ContainerBlock;
+class Heading;
+class HoriziontalRule;
+class List;
+class Paragraph;
+class Table;
+
+class Visitor;
+
+}

Userland/Libraries/LibMarkdown/Heading.cpp

@@ -6,6 +6,7 @@
 
 #include <AK/StringBuilder.h>
 #include <LibMarkdown/Heading.h>
+#include <LibMarkdown/Visitor.h>
 
 namespace Markdown {
 
@@ -35,6 +36,15 @@ String Heading::render_for_terminal(size_t) const
     return builder.build();
 }
 
+RecursionDecision Heading::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    return m_text.walk(visitor);
+}
+
 OwnPtr<Heading> Heading::parse(LineIterator& lines)
 {
     if (lines.is_end())

Userland/Libraries/LibMarkdown/Heading.h

@@ -27,6 +27,7 @@ public:
 
     virtual String render_to_html(bool tight = false) const override;
     virtual String render_for_terminal(size_t view_width = 0) const override;
+    virtual RecursionDecision walk(Visitor&) const override;
     static OwnPtr<Heading> parse(LineIterator& lines);
 
 private:

Userland/Libraries/LibMarkdown/HorizontalRule.cpp

@@ -7,6 +7,7 @@
 #include <AK/String.h>
 #include <AK/StringBuilder.h>
 #include <LibMarkdown/HorizontalRule.h>
+#include <LibMarkdown/Visitor.h>
 
 namespace Markdown {
 
@@ -24,6 +25,15 @@ String HorizontalRule::render_for_terminal(size_t view_width) const
     return builder.to_string();
 }
 
+RecursionDecision HorizontalRule::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+    // Normalize return value.
+    return RecursionDecision::Continue;
+}
+
 OwnPtr<HorizontalRule> HorizontalRule::parse(LineIterator& lines)
 {
     if (lines.is_end())

Userland/Libraries/LibMarkdown/HorizontalRule.h

@@ -23,6 +23,7 @@ public:
 
     virtual String render_to_html(bool tight = false) const override;
     virtual String render_for_terminal(size_t view_width = 0) const override;
+    virtual RecursionDecision walk(Visitor&) const override;
     static OwnPtr<HorizontalRule> parse(LineIterator& lines);
 };
 

Userland/Libraries/LibMarkdown/List.cpp

@@ -8,6 +8,7 @@
 #include <AK/StringBuilder.h>
 #include <LibMarkdown/List.h>
 #include <LibMarkdown/Paragraph.h>
+#include <LibMarkdown/Visitor.h>
 
 namespace Markdown {
 
@@ -54,6 +55,21 @@ String List::render_for_terminal(size_t) const
     return builder.build();
 }
 
+RecursionDecision List::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    for (auto const& block : m_items) {
+        rd = block->walk(visitor);
+        if (rd == RecursionDecision::Break)
+            return rd;
+    }
+
+    return RecursionDecision::Continue;
+}
+
 OwnPtr<List> List::parse(LineIterator& lines)
 {
     Vector<OwnPtr<ContainerBlock>> items;

Userland/Libraries/LibMarkdown/List.h

@@ -26,6 +26,7 @@ public:
 
     virtual String render_to_html(bool tight = false) const override;
     virtual String render_for_terminal(size_t view_width = 0) const override;
+    virtual RecursionDecision walk(Visitor&) const override;
 
     static OwnPtr<List> parse(LineIterator& lines);
 

Userland/Libraries/LibMarkdown/Paragraph.cpp

@@ -6,6 +6,7 @@
 
 #include <AK/StringBuilder.h>
 #include <LibMarkdown/Paragraph.h>
+#include <LibMarkdown/Visitor.h>
 
 namespace Markdown {
 
@@ -34,4 +35,13 @@ String Paragraph::render_for_terminal(size_t) const
     return builder.build();
 }
 
+RecursionDecision Paragraph::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    return m_text.walk(visitor);
+}
+
 }

Userland/Libraries/LibMarkdown/Paragraph.h

@@ -24,6 +24,7 @@ public:
 
     virtual String render_to_html(bool tight = false) const override;
     virtual String render_for_terminal(size_t view_width = 0) const override;
+    virtual RecursionDecision walk(Visitor&) const override;
 
 private:
     Text m_text;

Userland/Libraries/LibMarkdown/Table.cpp

@@ -7,6 +7,7 @@
 #include <AK/Debug.h>
 #include <AK/StringBuilder.h>
 #include <LibMarkdown/Table.h>
+#include <LibMarkdown/Visitor.h>
 
 namespace Markdown {
 
@@ -96,6 +97,21 @@ String Table::render_to_html(bool) const
     return builder.to_string();
 }
 
+RecursionDecision Table::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    for (auto const& column : m_columns) {
+        rd = column.walk(visitor);
+        if (rd == RecursionDecision::Break)
+            return rd;
+    }
+
+    return RecursionDecision::Continue;
+}
+
 OwnPtr<Table> Table::parse(LineIterator& lines)
 {
     auto peek_it = lines;
@@ -207,4 +223,23 @@ OwnPtr<Table> Table::parse(LineIterator& lines)
     return table;
 }
 
+RecursionDecision Table::Column::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    rd = header.walk(visitor);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    for (auto const& row : rows) {
+        rd = row.walk(visitor);
+        if (rd == RecursionDecision::Break)
+            return rd;
+    }
+
+    return RecursionDecision::Continue;
+}
+
 }

Userland/Libraries/LibMarkdown/Table.h

@@ -27,6 +27,8 @@ public:
         Vector<Text> rows;
         Alignment alignment { Alignment::Left };
         size_t relative_width { 0 };
+
+        RecursionDecision walk(Visitor&) const;
     };
 
     Table() { }
@@ -34,6 +36,7 @@ public:
 
     virtual String render_to_html(bool tight = false) const override;
     virtual String render_for_terminal(size_t view_width = 0) const override;
+    virtual RecursionDecision walk(Visitor&) const override;
     static OwnPtr<Table> parse(LineIterator& lines);
 
 private:

Userland/Libraries/LibMarkdown/Text.cpp

@@ -9,6 +9,7 @@
 #include <AK/ScopeGuard.h>
 #include <AK/StringBuilder.h>
 #include <LibMarkdown/Text.h>
+#include <LibMarkdown/Visitor.h>
 #include <ctype.h>
 #include <string.h>
 
@@ -39,6 +40,15 @@ size_t Text::EmphasisNode::terminal_length() const
     return child->terminal_length();
 }
 
+RecursionDecision Text::EmphasisNode::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    return child->walk(visitor);
+}
+
 void Text::CodeNode::render_to_html(StringBuilder& builder) const
 {
     builder.append("<code>");
@@ -58,6 +68,15 @@ size_t Text::CodeNode::terminal_length() const
     return code->terminal_length();
 }
 
+RecursionDecision Text::CodeNode::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    return code->walk(visitor);
+}
+
 void Text::BreakNode::render_to_html(StringBuilder& builder) const
 {
     builder.append("<br />");
@@ -72,6 +91,15 @@ size_t Text::BreakNode::terminal_length() const
     return 0;
 }
 
+RecursionDecision Text::BreakNode::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+    // Normalize return value
+    return RecursionDecision::Continue;
+}
+
 void Text::TextNode::render_to_html(StringBuilder& builder) const
 {
     builder.append(escape_html_entities(text));
@@ -95,6 +123,18 @@ size_t Text::TextNode::terminal_length() const
     return text.length();
 }
 
+RecursionDecision Text::TextNode::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+    rd = visitor.visit(text);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+    // Normalize return value
+    return RecursionDecision::Continue;
+}
+
 void Text::LinkNode::render_to_html(StringBuilder& builder) const
 {
     if (is_image) {
@@ -134,6 +174,17 @@ size_t Text::LinkNode::terminal_length() const
     return text->terminal_length();
 }
 
+RecursionDecision Text::LinkNode::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    // Don't recurse on href.
+
+    return text->walk(visitor);
+}
+
 void Text::MultiNode::render_to_html(StringBuilder& builder) const
 {
     for (auto& child : children) {
@@ -157,6 +208,21 @@ size_t Text::MultiNode::terminal_length() const
     return length;
 }
 
+RecursionDecision Text::MultiNode::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    for (auto const& child : children) {
+        rd = child.walk(visitor);
+        if (rd == RecursionDecision::Break)
+            return rd;
+    }
+
+    return RecursionDecision::Continue;
+}
+
 size_t Text::terminal_length() const
 {
     return m_node->terminal_length();
@@ -176,6 +242,15 @@ String Text::render_for_terminal() const
     return builder.build().trim(" \n\t");
 }
 
+RecursionDecision Text::walk(Visitor& visitor) const
+{
+    RecursionDecision rd = visitor.visit(*this);
+    if (rd != RecursionDecision::Recurse)
+        return rd;
+
+    return m_node->walk(visitor);
+}
+
 Text Text::parse(StringView const& str)
 {
     Text text;

Userland/Libraries/LibMarkdown/Text.h

@@ -10,7 +10,9 @@
 #include <AK/Noncopyable.h>
 #include <AK/NonnullOwnPtrVector.h>
 #include <AK/OwnPtr.h>
+#include <AK/RecursionDecision.h>
 #include <AK/String.h>
+#include <LibMarkdown/Forward.h>
 
 namespace Markdown {
 
@@ -21,6 +23,7 @@ public:
         virtual void render_to_html(StringBuilder& builder) const = 0;
         virtual void render_for_terminal(StringBuilder& builder) const = 0;
         virtual size_t terminal_length() const = 0;
+        virtual RecursionDecision walk(Visitor&) const = 0;
 
         virtual ~Node() { }
     };
@@ -39,6 +42,7 @@ public:
         virtual void render_to_html(StringBuilder& builder) const override;
         virtual void render_for_terminal(StringBuilder& builder) const override;
         virtual size_t terminal_length() const override;
+        virtual RecursionDecision walk(Visitor&) const override;
     };
 
     class CodeNode : public Node {
@@ -53,6 +57,7 @@ public:
         virtual void render_to_html(StringBuilder& builder) const override;
         virtual void render_for_terminal(StringBuilder& builder) const override;
         virtual size_t terminal_length() const override;
+        virtual RecursionDecision walk(Visitor&) const override;
     };
 
     class BreakNode : public Node {
@@ -60,6 +65,7 @@ public:
         virtual void render_to_html(StringBuilder& builder) const override;
         virtual void render_for_terminal(StringBuilder& builder) const override;
         virtual size_t terminal_length() const override;
+        virtual RecursionDecision walk(Visitor&) const override;
     };
 
     class TextNode : public Node {
@@ -82,6 +88,7 @@ public:
         virtual void render_to_html(StringBuilder& builder) const override;
         virtual void render_for_terminal(StringBuilder& builder) const override;
         virtual size_t terminal_length() const override;
+        virtual RecursionDecision walk(Visitor&) const override;
     };
 
     class LinkNode : public Node {
@@ -100,6 +107,7 @@ public:
         virtual void render_to_html(StringBuilder& builder) const override;
         virtual void render_for_terminal(StringBuilder& builder) const override;
         virtual size_t terminal_length() const override;
+        virtual RecursionDecision walk(Visitor&) const override;
     };
 
     class MultiNode : public Node {
@@ -109,12 +117,14 @@ public:
         virtual void render_to_html(StringBuilder& builder) const override;
         virtual void render_for_terminal(StringBuilder& builder) const override;
         virtual size_t terminal_length() const override;
+        virtual RecursionDecision walk(Visitor&) const override;
     };
 
     size_t terminal_length() const;
 
     String render_to_html() const;
     String render_for_terminal() const;
+    RecursionDecision walk(Visitor&) const;
 
     static Text parse(StringView const&);
 

Userland/Libraries/LibMarkdown/Visitor.h

@@ -0,0 +1,50 @@
+/*
+ * Copyright (c) 2021, Ben Wiederhake <BenWiederhake.GitHub@gmx.de>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <AK/RecursionDecision.h>
+#include <LibMarkdown/BlockQuote.h>
+#include <LibMarkdown/CodeBlock.h>
+#include <LibMarkdown/Document.h>
+#include <LibMarkdown/Heading.h>
+#include <LibMarkdown/HorizontalRule.h>
+#include <LibMarkdown/List.h>
+#include <LibMarkdown/Paragraph.h>
+#include <LibMarkdown/Table.h>
+
+namespace Markdown {
+
+class Visitor {
+public:
+    Visitor() = default;
+    virtual ~Visitor() = default;
+
+    virtual RecursionDecision visit(Document const&) { return RecursionDecision::Recurse; }
+
+    virtual RecursionDecision visit(BlockQuote const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(CodeBlock const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(ContainerBlock const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(Heading const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(HorizontalRule const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(List const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(Paragraph const&) { return RecursionDecision::Recurse; }
+
+    virtual RecursionDecision visit(Table const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(Table::Column const&) { return RecursionDecision::Recurse; }
+
+    virtual RecursionDecision visit(Text const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(Text::BreakNode const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(Text::CodeNode const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(Text::EmphasisNode const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(Text::LinkNode const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(Text::MultiNode const&) { return RecursionDecision::Recurse; }
+    virtual RecursionDecision visit(Text::TextNode const&) { return RecursionDecision::Recurse; }
+
+    virtual RecursionDecision visit(String const&) { return RecursionDecision::Recurse; }
+};
+
+}