qiita_marker 0.23.2.0

data/ext/qiita_marker/xml.c

@@ -0,0 +1,181 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <assert.h>
+
+#include "config.h"
+#include "cmark-gfm.h"
+#include "node.h"
+#include "buffer.h"
+#include "houdini.h"
+#include "syntax_extension.h"
+
+#define BUFFER_SIZE 100
+
+// Functions to convert cmark_nodes to XML strings.
+
+static void escape_xml(cmark_strbuf *dest, const unsigned char *source,
+                       bufsize_t length) {
+  houdini_escape_html0(dest, source, length, 0);
+}
+
+struct render_state {
+  cmark_strbuf *xml;
+  int indent;
+};
+
+static CMARK_INLINE void indent(struct render_state *state) {
+  int i;
+  for (i = 0; i < state->indent; i++) {
+    cmark_strbuf_putc(state->xml, ' ');
+  }
+}
+
+static int S_render_node(cmark_node *node, cmark_event_type ev_type,
+                         struct render_state *state, int options) {
+  cmark_strbuf *xml = state->xml;
+  bool literal = false;
+  cmark_delim_type delim;
+  bool entering = (ev_type == CMARK_EVENT_ENTER);
+  char buffer[BUFFER_SIZE];
+
+  if (entering) {
+    indent(state);
+    cmark_strbuf_putc(xml, '<');
+    cmark_strbuf_puts(xml, cmark_node_get_type_string(node));
+
+    if (options & CMARK_OPT_SOURCEPOS && node->start_line != 0) {
+      snprintf(buffer, BUFFER_SIZE, " sourcepos=\"%d:%d-%d:%d\"",
+               node->start_line, node->start_column, node->end_line,
+               node->end_column);
+      cmark_strbuf_puts(xml, buffer);
+    }
+
+    if (node->extension && node->extension->xml_attr_func) {
+      const char* r = node->extension->xml_attr_func(node->extension, node);
+      if (r != NULL)
+        cmark_strbuf_puts(xml, r);
+    }
+
+    literal = false;
+
+    switch (node->type) {
+    case CMARK_NODE_DOCUMENT:
+      cmark_strbuf_puts(xml, " xmlns=\"http://commonmark.org/xml/1.0\"");
+      break;
+    case CMARK_NODE_TEXT:
+    case CMARK_NODE_CODE:
+    case CMARK_NODE_HTML_BLOCK:
+    case CMARK_NODE_HTML_INLINE:
+      cmark_strbuf_puts(xml, " xml:space=\"preserve\">");
+      escape_xml(xml, node->as.literal.data, node->as.literal.len);
+      cmark_strbuf_puts(xml, "</");
+      cmark_strbuf_puts(xml, cmark_node_get_type_string(node));
+      literal = true;
+      break;
+    case CMARK_NODE_LIST:
+      switch (cmark_node_get_list_type(node)) {
+      case CMARK_ORDERED_LIST:
+        cmark_strbuf_puts(xml, " type=\"ordered\"");
+        snprintf(buffer, BUFFER_SIZE, " start=\"%d\"",
+                 cmark_node_get_list_start(node));
+        cmark_strbuf_puts(xml, buffer);
+        delim = cmark_node_get_list_delim(node);
+        if (delim == CMARK_PAREN_DELIM) {
+          cmark_strbuf_puts(xml, " delim=\"paren\"");
+        } else if (delim == CMARK_PERIOD_DELIM) {
+          cmark_strbuf_puts(xml, " delim=\"period\"");
+        }
+        break;
+      case CMARK_BULLET_LIST:
+        cmark_strbuf_puts(xml, " type=\"bullet\"");
+        break;
+      default:
+        break;
+      }
+      snprintf(buffer, BUFFER_SIZE, " tight=\"%s\"",
+               (cmark_node_get_list_tight(node) ? "true" : "false"));
+      cmark_strbuf_puts(xml, buffer);
+      break;
+    case CMARK_NODE_HEADING:
+      snprintf(buffer, BUFFER_SIZE, " level=\"%d\"", node->as.heading.level);
+      cmark_strbuf_puts(xml, buffer);
+      break;
+    case CMARK_NODE_CODE_BLOCK:
+      if (node->as.code.info.len > 0) {
+        cmark_strbuf_puts(xml, " info=\"");
+        escape_xml(xml, node->as.code.info.data, node->as.code.info.len);
+        cmark_strbuf_putc(xml, '"');
+      }
+      cmark_strbuf_puts(xml, " xml:space=\"preserve\">");
+      escape_xml(xml, node->as.code.literal.data, node->as.code.literal.len);
+      cmark_strbuf_puts(xml, "</");
+      cmark_strbuf_puts(xml, cmark_node_get_type_string(node));
+      literal = true;
+      break;
+    case CMARK_NODE_CUSTOM_BLOCK:
+    case CMARK_NODE_CUSTOM_INLINE:
+      cmark_strbuf_puts(xml, " on_enter=\"");
+      escape_xml(xml, node->as.custom.on_enter.data,
+                 node->as.custom.on_enter.len);
+      cmark_strbuf_putc(xml, '"');
+      cmark_strbuf_puts(xml, " on_exit=\"");
+      escape_xml(xml, node->as.custom.on_exit.data,
+                 node->as.custom.on_exit.len);
+      cmark_strbuf_putc(xml, '"');
+      break;
+    case CMARK_NODE_LINK:
+    case CMARK_NODE_IMAGE:
+      cmark_strbuf_puts(xml, " destination=\"");
+      escape_xml(xml, node->as.link.url.data, node->as.link.url.len);
+      cmark_strbuf_putc(xml, '"');
+      cmark_strbuf_puts(xml, " title=\"");
+      escape_xml(xml, node->as.link.title.data, node->as.link.title.len);
+      cmark_strbuf_putc(xml, '"');
+      break;
+    default:
+      break;
+    }
+    if (node->first_child) {
+      state->indent += 2;
+    } else if (!literal) {
+      cmark_strbuf_puts(xml, " /");
+    }
+    cmark_strbuf_puts(xml, ">\n");
+
+  } else if (node->first_child) {
+    state->indent -= 2;
+    indent(state);
+    cmark_strbuf_puts(xml, "</");
+    cmark_strbuf_puts(xml, cmark_node_get_type_string(node));
+    cmark_strbuf_puts(xml, ">\n");
+  }
+
+  return 1;
+}
+
+char *cmark_render_xml(cmark_node *root, int options) {
+  return cmark_render_xml_with_mem(root, options, cmark_node_mem(root));
+}
+
+char *cmark_render_xml_with_mem(cmark_node *root, int options, cmark_mem *mem) {
+  char *result;
+  cmark_strbuf xml = CMARK_BUF_INIT(mem);
+  cmark_event_type ev_type;
+  cmark_node *cur;
+  struct render_state state = {&xml, 0};
+
+  cmark_iter *iter = cmark_iter_new(root);
+
+  cmark_strbuf_puts(state.xml, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
+  cmark_strbuf_puts(state.xml,
+                    "<!DOCTYPE document SYSTEM \"CommonMark.dtd\">\n");
+  while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
+    cur = cmark_iter_get_node(iter);
+    S_render_node(cur, ev_type, &state, options);
+  }
+  result = (char *)cmark_strbuf_detach(&xml);
+
+  cmark_iter_free(iter);
+  return result;
+}

data/lib/qiita_marker/config.rb

@@ -0,0 +1,52 @@
+# frozen_string_literal: true
+
+module QiitaMarker
+  # For Ruby::Enum, these must be classes, not modules
+  module Config
+    # See https://github.com/github/cmark-gfm/blob/master/src/cmark-gfm.h#L673
+    OPTS = {
+      parse: {
+        DEFAULT: 0,
+        SOURCEPOS: (1 << 1),
+        UNSAFE: (1 << 17),
+        VALIDATE_UTF8: (1 << 9),
+        SMART: (1 << 10),
+        LIBERAL_HTML_TAG: (1 << 12),
+        FOOTNOTES: (1 << 13),
+        STRIKETHROUGH_DOUBLE_TILDE: (1 << 14)
+      }.freeze,
+      render: {
+        DEFAULT: 0,
+        SOURCEPOS: (1 << 1),
+        HARDBREAKS: (1 << 2),
+        UNSAFE: (1 << 17),
+        NOBREAKS: (1 << 4),
+        VALIDATE_UTF8: (1 << 9),
+        SMART: (1 << 10),
+        GITHUB_PRE_LANG: (1 << 11),
+        LIBERAL_HTML_TAG: (1 << 12),
+        FOOTNOTES: (1 << 13),
+        STRIKETHROUGH_DOUBLE_TILDE: (1 << 14),
+        TABLE_PREFER_STYLE_ATTRIBUTES: (1 << 15),
+        FULL_INFO_STRING: (1 << 16)
+      }.freeze,
+      format: %i[html xml commonmark plaintext].freeze
+    }.freeze
+
+    def self.process_options(option, type)
+      case option
+      when Symbol
+        OPTS.fetch(type).fetch(option)
+      when Array
+        raise TypeError if option.none?
+
+        # neckbearding around. the map will both check the opts and then bitwise-OR it
+        OPTS.fetch(type).fetch_values(*option).inject(0, :|)
+      else
+        raise TypeError, "option type must be a valid symbol or array of symbols within the #{name}::OPTS[:#{type}] context"
+      end
+    rescue KeyError => e
+      raise TypeError, "option ':#{e.key}' does not exist for #{name}::OPTS[:#{type}]"
+    end
+  end
+end

data/lib/qiita_marker/node/inspect.rb

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+require 'pp'
+
+module QiitaMarker
+  class Node
+    module Inspect
+      PP_INDENT_SIZE = 2
+
+      def inspect
+        PP.pp(self, +'', Float::INFINITY)
+      end
+
+      # @param printer [PrettyPrint] pp
+      def pretty_print(printer)
+        printer.group(PP_INDENT_SIZE, "#<#{self.class}(#{type}):", '>') do
+          printer.breakable
+
+          attrs = %i[
+            sourcepos
+            string_content
+            url
+            title
+            header_level
+            list_type
+            list_start
+            list_tight
+            fence_info
+          ].map do |name|
+            [name, __send__(name)]
+          rescue NodeError
+            nil
+          end.compact
+
+          printer.seplist(attrs) do |name, value|
+            printer.text "#{name}="
+            printer.pp value
+          end
+
+          if first_child
+            printer.breakable
+            printer.group(PP_INDENT_SIZE) do
+              children = []
+              node = first_child
+              while node
+                children << node
+                node = node.next
+              end
+              printer.text 'children='
+              printer.pp children
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/qiita_marker/node.rb

@@ -0,0 +1,83 @@
+# frozen_string_literal: true
+
+require 'qiita_marker/node/inspect'
+
+module QiitaMarker
+  class Node
+    include Enumerable
+    include Inspect
+
+    # Public: An iterator that "walks the tree," descending into children recursively.
+    #
+    # blk - A {Proc} representing the action to take for each child
+    def walk(&block)
+      return enum_for(:walk) unless block
+
+      yield self
+      each do |child|
+        child.walk(&block)
+      end
+    end
+
+    # Public: Convert the node to an HTML string.
+    #
+    # options - A {Symbol} or {Array of Symbol}s indicating the render options
+    # extensions - An {Array of Symbol}s indicating the extensions to use
+    #
+    # Returns a {String}.
+    def to_html(options = :DEFAULT, extensions = [])
+      opts = Config.process_options(options, :render)
+      _render_html(opts, extensions).force_encoding('utf-8')
+    end
+
+    # Public: Convert the node to an XML string.
+    #
+    # options - A {Symbol} or {Array of Symbol}s indicating the render options
+    #
+    # Returns a {String}.
+    def to_xml(options = :DEFAULT)
+      opts = Config.process_options(options, :render)
+      _render_xml(opts).force_encoding('utf-8')
+    end
+
+    # Public: Convert the node to a CommonMark string.
+    #
+    # options - A {Symbol} or {Array of Symbol}s indicating the render options
+    # width - Column to wrap the output at
+    #
+    # Returns a {String}.
+    def to_commonmark(options = :DEFAULT, width = 120)
+      opts = Config.process_options(options, :render)
+      _render_commonmark(opts, width).force_encoding('utf-8')
+    end
+
+    # Public: Convert the node to a plain text string.
+    #
+    # options - A {Symbol} or {Array of Symbol}s indicating the render options
+    # width - Column to wrap the output at
+    #
+    # Returns a {String}.
+    def to_plaintext(options = :DEFAULT, width = 120)
+      opts = Config.process_options(options, :render)
+      _render_plaintext(opts, width).force_encoding('utf-8')
+    end
+
+    # Public: Iterate over the children (if any) of the current pointer.
+    def each
+      return enum_for(:each) unless block_given?
+
+      child = first_child
+      while child
+        nextchild = child.next
+        yield child
+        child = nextchild
+      end
+    end
+
+    # Deprecated: Please use `each` instead
+    def each_child(&block)
+      warn '[DEPRECATION] `each_child` is deprecated.  Please use `each` instead.'
+      each(&block)
+    end
+  end
+end

data/lib/qiita_marker/renderer/html_renderer.rb

@@ -0,0 +1,252 @@
+# frozen_string_literal: true
+
+module QiitaMarker
+  class HtmlRenderer < Renderer
+    def document(_)
+      super
+      out("</ol>\n</section>\n") if @written_footnote_ix
+    end
+
+    def header(node)
+      block do
+        out('<h', node.header_level, "#{sourcepos(node)}>", :children,
+            '</h', node.header_level, '>')
+      end
+    end
+
+    def paragraph(node)
+      if @in_tight && node.parent.type != :blockquote
+        out(:children)
+      else
+        block do
+          container("<p#{sourcepos(node)}>", '</p>') do
+            out(:children)
+            if node.parent.type == :footnote_definition && node.next.nil?
+              out(' ')
+              out_footnote_backref
+            end
+          end
+        end
+      end
+    end
+
+    def list(node)
+      old_in_tight = @in_tight
+      @in_tight = node.list_tight
+
+      block do
+        if node.list_type == :bullet_list
+          container("<ul#{sourcepos(node)}>\n", '</ul>') do
+            out(:children)
+          end
+        else
+          start = if node.list_start == 1
+                    "<ol#{sourcepos(node)}>\n"
+                  else
+                    "<ol start=\"#{node.list_start}\"#{sourcepos(node)}>\n"
+                  end
+          container(start, '</ol>') do
+            out(:children)
+          end
+        end
+      end
+
+      @in_tight = old_in_tight
+    end
+
+    def list_item(node)
+      block do
+        tasklist_data = tasklist(node)
+        container("<li#{sourcepos(node)}#{tasklist_data}>#{' ' if tasklist?(node)}", '</li>') do
+          out(:children)
+        end
+      end
+    end
+
+    def tasklist(node)
+      return '' unless tasklist?(node)
+
+      state = if checked?(node)
+                'checked="" disabled=""'
+              else
+                'disabled=""'
+              end
+      "><input type=\"checkbox\" #{state} /"
+    end
+
+    def blockquote(node)
+      block do
+        container("<blockquote#{sourcepos(node)}>\n", '</blockquote>') do
+          out(:children)
+        end
+      end
+    end
+
+    def hrule(node)
+      block do
+        out("<hr#{sourcepos(node)} />")
+      end
+    end
+
+    def code_block(node)
+      block do
+        if option_enabled?(:GITHUB_PRE_LANG)
+          out("<pre#{sourcepos(node)}")
+          out(' lang="', node.fence_info.split(/\s+/)[0], '"') if node.fence_info && !node.fence_info.empty?
+          out('><code>')
+        else
+          out("<pre#{sourcepos(node)}><code")
+          if node.fence_info && !node.fence_info.empty?
+            out(' class="language-', node.fence_info.split(/\s+/)[0], '">')
+          else
+            out('>')
+          end
+        end
+        out(escape_html(node.string_content))
+        out('</code></pre>')
+      end
+    end
+
+    def html(node)
+      block do
+        if option_enabled?(:UNSAFE)
+          out(tagfilter(node.string_content))
+        else
+          out('<!-- raw HTML omitted -->')
+        end
+      end
+    end
+
+    def inline_html(node)
+      if option_enabled?(:UNSAFE)
+        out(tagfilter(node.string_content))
+      else
+        out('<!-- raw HTML omitted -->')
+      end
+    end
+
+    def emph(_)
+      out('<em>', :children, '</em>')
+    end
+
+    def strong(_)
+      out('<strong>', :children, '</strong>')
+    end
+
+    def link(node)
+      out('<a href="', node.url.nil? ? '' : escape_href(node.url), '"')
+      out(' title="', escape_html(node.title), '"') if node.title && !node.title.empty?
+      out('>', :children, '</a>')
+    end
+
+    def image(node)
+      out('<img src="', escape_href(node.url), '"')
+      plain do
+        out(' alt="', :children, '"')
+      end
+      out(' title="', escape_html(node.title), '"') if node.title && !node.title.empty?
+      out(' />')
+    end
+
+    def text(node)
+      out(escape_html(node.string_content))
+    end
+
+    def code(node)
+      out('<code>')
+      out(escape_html(node.string_content))
+      out('</code>')
+    end
+
+    def linebreak(_node)
+      out("<br />\n")
+    end
+
+    def softbreak(_)
+      if option_enabled?(:HARDBREAKS)
+        out("<br />\n")
+      elsif option_enabled?(:NOBREAKS)
+        out(' ')
+      else
+        out("\n")
+      end
+    end
+
+    def table(node)
+      @alignments = node.table_alignments
+      @needs_close_tbody = false
+      out("<table#{sourcepos(node)}>\n", :children)
+      out("</tbody>\n") if @needs_close_tbody
+      out("</table>\n")
+    end
+
+    def table_header(node)
+      @column_index = 0
+
+      @in_header = true
+      out("<thead>\n<tr#{sourcepos(node)}>\n", :children, "</tr>\n</thead>\n")
+      @in_header = false
+    end
+
+    def table_row(node)
+      @column_index = 0
+      if !@in_header && !@needs_close_tbody
+        @needs_close_tbody = true
+        out("<tbody>\n")
+      end
+      out("<tr#{sourcepos(node)}>\n", :children, "</tr>\n")
+    end
+
+    def table_cell(node)
+      align = case @alignments[@column_index]
+              when :left then ' align="left"'
+              when :right then ' align="right"'
+              when :center then ' align="center"'
+              else; ''
+              end
+      out(@in_header ? "<th#{align}#{sourcepos(node)}>" : "<td#{align}#{sourcepos(node)}>", :children, @in_header ? "</th>\n" : "</td>\n")
+      @column_index += 1
+    end
+
+    def strikethrough(_)
+      out('<del>', :children, '</del>')
+    end
+
+    def footnote_reference(node)
+      out("<sup class=\"footnote-ref\"><a href=\"#fn#{node.string_content}\" id=\"fnref#{node.string_content}\">#{node.string_content}</a></sup>")
+    end
+
+    def footnote_definition(_)
+      unless @footnote_ix
+        out("<section class=\"footnotes\">\n<ol>\n")
+        @footnote_ix = 0
+      end
+
+      @footnote_ix += 1
+      out("<li id=\"fn#{@footnote_ix}\">\n", :children)
+      out("\n") if out_footnote_backref
+      out("</li>\n")
+      # </ol>
+      # </section>
+    end
+
+    private
+
+    def out_footnote_backref
+      return false if @written_footnote_ix == @footnote_ix
+
+      @written_footnote_ix = @footnote_ix
+
+      out("<a href=\"#fnref#{@footnote_ix}\" class=\"footnote-backref\">↩</a>")
+      true
+    end
+
+    def tasklist?(node)
+      node.type_string == 'tasklist'
+    end
+
+    def checked?(node)
+      node.tasklist_item_checked?
+    end
+  end
+end