ruby/lib/rdoc/markup/to_rdoc.rb
Hartley McGuire 5ac6194c2b [ruby/rdoc] Fix ToRdoc generating incorrect {label,name}-lists
Previously, trying to round-trip label-list and name-lists with the
ToRdoc converter was not possible:

```ruby
doc = <<~RDOC
foo ::
bar ::
  hi
RDOC

markup = RDoc::Markup.parse(doc)
markup # => [doc: [list: NOTE [item: ["foo ", "bar"]; [para: "hi"]]]]

rt = RDoc::Markup::ToRdoc.new.convert(markup)
rt # => "foo\nbar:\n  hi\n\n"

rt_markup = RDoc::Markup.parse(rt)
rt_markup # => [doc: [para: "foo ", "bar:"], [verb: "hi\n"]]
```

This commit addresses the issue by fixing ToRdoc to generate output that
can be properly reparsed by RDoc. ToRdoc tests additionally needed to be
updated for the new output.

The old implementation of `accept_list_item_start` was copied to ToBs
because those tests did not pass with the new changes and I am
unfamiliar with the `backspace` format.

After:

```ruby
doc = <<~RDOC
foo ::
bar ::
  hi
RDOC

markup = RDoc::Markup.parse(doc)
markup # => [doc: [list: NOTE [item: ["foo ", "bar"]; [para: "hi"]]]]

rt = RDoc::Markup::ToRdoc.new.convert(markup)
rt # => "foo::\nbar::\n  hi\n\n"

rt_markup = RDoc::Markup.parse(rt)
rt_markup # => [doc: [list: NOTE [item: ["foo", "bar"]; [para: "hi"], blankline]]]
```

c6c51aa900
2024-03-09 03:50:09 +00:00

352 lines
6.7 KiB
Ruby

# frozen_string_literal: true
##
# Outputs RDoc markup as RDoc markup! (mostly)
class RDoc::Markup::ToRdoc < RDoc::Markup::Formatter
##
# Current indent amount for output in characters
attr_accessor :indent
##
# Output width in characters
attr_accessor :width
##
# Stack of current list indexes for alphabetic and numeric lists
attr_reader :list_index
##
# Stack of list types
attr_reader :list_type
##
# Stack of list widths for indentation
attr_reader :list_width
##
# Prefix for the next list item. See #use_prefix
attr_reader :prefix
##
# Output accumulator
attr_reader :res
##
# Creates a new formatter that will output (mostly) \RDoc markup
def initialize markup = nil
super nil, markup
@markup.add_regexp_handling(/\\\S/, :SUPPRESSED_CROSSREF)
@width = 78
init_tags
@headings = {}
@headings.default = []
@headings[1] = ['= ', '']
@headings[2] = ['== ', '']
@headings[3] = ['=== ', '']
@headings[4] = ['==== ', '']
@headings[5] = ['===== ', '']
@headings[6] = ['====== ', '']
@hard_break = "\n"
end
##
# Maps attributes to HTML sequences
def init_tags
add_tag :BOLD, "<b>", "</b>"
add_tag :TT, "<tt>", "</tt>"
add_tag :EM, "<em>", "</em>"
end
##
# Adds +blank_line+ to the output
def accept_blank_line blank_line
@res << "\n"
end
##
# Adds +paragraph+ to the output
def accept_block_quote block_quote
@indent += 2
block_quote.parts.each do |part|
@prefix = '> '
part.accept self
end
@indent -= 2
end
##
# Adds +heading+ to the output
def accept_heading heading
use_prefix or @res << ' ' * @indent
@res << @headings[heading.level][0]
@res << attributes(heading.text)
@res << @headings[heading.level][1]
@res << "\n"
end
##
# Finishes consumption of +list+
def accept_list_end list
@list_index.pop
@list_type.pop
@list_width.pop
end
##
# Finishes consumption of +list_item+
def accept_list_item_end list_item
width = case @list_type.last
when :BULLET then
2
when :NOTE, :LABEL then
if @prefix then
@res << @prefix.strip
@prefix = nil
end
@res << "\n"
2
else
bullet = @list_index.last.to_s
@list_index[-1] = @list_index.last.succ
bullet.length + 2
end
@indent -= width
end
##
# Prepares the visitor for consuming +list_item+
def accept_list_item_start list_item
type = @list_type.last
case type
when :NOTE, :LABEL then
stripped_labels = Array(list_item.label).map do |label|
attributes(label).strip
end
bullets = case type
when :NOTE
stripped_labels.map { |b| "#{b}::" }
when :LABEL
stripped_labels.map { |b| "[#{b}]" }
end
bullets = bullets.join("\n")
bullets << "\n" unless stripped_labels.empty?
@prefix = ' ' * @indent
@indent += 2
@prefix << bullets + (' ' * @indent)
else
bullet = type == :BULLET ? '*' : @list_index.last.to_s + '.'
@prefix = (' ' * @indent) + bullet.ljust(bullet.length + 1)
width = bullet.length + 1
@indent += width
end
end
##
# Prepares the visitor for consuming +list+
def accept_list_start list
case list.type
when :BULLET then
@list_index << nil
@list_width << 1
when :LABEL, :NOTE then
@list_index << nil
@list_width << 2
when :LALPHA then
@list_index << 'a'
@list_width << list.items.length.to_s.length
when :NUMBER then
@list_index << 1
@list_width << list.items.length.to_s.length
when :UALPHA then
@list_index << 'A'
@list_width << list.items.length.to_s.length
else
raise RDoc::Error, "invalid list type #{list.type}"
end
@list_type << list.type
end
##
# Adds +paragraph+ to the output
def accept_paragraph paragraph
text = paragraph.text @hard_break
wrap attributes text
end
##
# Adds +paragraph+ to the output
def accept_indented_paragraph paragraph
@indent += paragraph.indent
text = paragraph.text @hard_break
wrap attributes text
@indent -= paragraph.indent
end
##
# Adds +raw+ to the output
def accept_raw raw
@res << raw.parts.join("\n")
end
##
# Adds +rule+ to the output
def accept_rule rule
use_prefix or @res << ' ' * @indent
@res << '-' * (@width - @indent)
@res << "\n"
end
##
# Outputs +verbatim+ indented 2 columns
def accept_verbatim verbatim
indent = ' ' * (@indent + 2)
verbatim.parts.each do |part|
@res << indent unless part == "\n"
@res << part
end
@res << "\n"
end
##
# Adds +table+ to the output
def accept_table header, body, aligns
widths = header.zip(body) do |h, b|
[h.size, b.size].max
end
aligns = aligns.map do |a|
case a
when nil
:center
when :left
:ljust
when :right
:rjust
end
end
@res << header.zip(widths, aligns) do |h, w, a|
h.__send__(a, w)
end.join("|").rstrip << "\n"
@res << widths.map {|w| "-" * w }.join("|") << "\n"
body.each do |row|
@res << row.zip(widths, aligns) do |t, w, a|
t.__send__(a, w)
end.join("|").rstrip << "\n"
end
end
##
# Applies attribute-specific markup to +text+ using RDoc::AttributeManager
def attributes text
flow = @am.flow text.dup
convert_flow flow
end
##
# Returns the generated output
def end_accepting
@res.join
end
##
# Removes preceding \\ from the suppressed crossref +target+
def handle_regexp_SUPPRESSED_CROSSREF target
text = target.text
text = text.sub('\\', '') unless in_tt?
text
end
##
# Adds a newline to the output
def handle_regexp_HARD_BREAK target
"\n"
end
##
# Prepares the visitor for text generation
def start_accepting
@res = [""]
@indent = 0
@prefix = nil
@list_index = []
@list_type = []
@list_width = []
end
##
# Adds the stored #prefix to the output and clears it. Lists generate a
# prefix for later consumption.
def use_prefix
prefix, @prefix = @prefix, nil
@res << prefix if prefix
prefix
end
##
# Wraps +text+ to #width
def wrap text
return unless text && !text.empty?
text_len = @width - @indent
text_len = 20 if text_len < 20
next_prefix = ' ' * @indent
prefix = @prefix || next_prefix
@prefix = nil
text.scan(/\G(?:([^ \n]{#{text_len}})(?=[^ \n])|(.{1,#{text_len}})(?:[ \n]|\z))/) do
@res << prefix << ($1 || $2) << "\n"
prefix = next_prefix
end
end
end