class Asciidoctor::PreprocessorReader
directives as each line is read off the Array of lines.
Public: Methods for retrieving lines from AsciiDoc source files, evaluating preprocessor
def create_include_cursor file, path, lineno
def create_include_cursor file, path, lineno if ::String === file dir = ::File.dirname file elsif RUBY_ENGINE_OPAL dir = ::File.dirname(file = file.to_s) else dir = (dir = ::File.dirname file.path) == '' ? '/' : dir file = file.to_s end Cursor.new file, dir, path, lineno end
def empty?
def empty? peek_line ? false : true end
def exceeds_max_depth?
Returns nil if no max depth is set and includes are disabled (max-include-depth=0), false if the current max depth
Public: Reports whether pushing an include on the include stack exceeds the max include depth.
def exceeds_max_depth? @maxdepth && @include_stack.size >= @maxdepth[:curr] && @maxdepth[:rel] end
def has_more_lines?
def has_more_lines? peek_line ? true : false end
def include_depth
def include_depth @include_stack.size end
def include_processors?
def include_processors? if @include_processor_extensions.nil? if @document.extensions? && @document.extensions.include_processors? !!(@include_processor_extensions = @document.extensions.include_processors) else @include_processor_extensions = false end else @include_processor_extensions != false end end
def initialize document, data = nil, cursor = nil, opts = {}
def initialize document, data = nil, cursor = nil, opts = {} @document = document super data, cursor, opts if (default_include_depth = (document.attributes['max-include-depth'] || 64).to_i) > 0 # track absolute max depth, current max depth for comparing to include stack size, and relative max depth for reporting @maxdepth = { abs: default_include_depth, curr: default_include_depth, rel: default_include_depth } else # if @maxdepth is not set, built-in include functionality is disabled @maxdepth = nil end @include_stack = [] @includes = document.catalog[:includes] @skipping = false @conditional_stack = [] @include_processor_extensions = nil end
def peek_line direct = false
in the current include context or a parent include context.
Returns the next line of the source data as a String if there are lines remaining
one include on the stack.
stack if the last line has been reached and there's at least
Public: Override the Reader#peek_line method to pop the include
def peek_line direct = false if (line = super) line elsif @include_stack.empty? nil else pop_include peek_line direct end end
def pop_include
def pop_include if @include_stack.size > 0 @lines, @file, @dir, @path, @lineno, @maxdepth, @process_lines = @include_stack.pop # FIXME kind of a hack #Document::AttributeEntry.new('infile', @file).save_to_next_block @document #Document::AttributeEntry.new('indir', ::File.dirname(@file)).save_to_next_block @document @look_ahead = 0 nil end end
def prepare_lines data, opts = {}
def prepare_lines data, opts = {} result = super # QUESTION should this work for AsciiDoc table cell content? Currently it does not. if @document && @document.attributes['skip-front-matter'] if (front_matter = skip_front_matter! result) @document.attributes['front-matter'] = front_matter.join LF end end if opts.fetch :condense, true result.shift && @lineno += 1 while (first = result[0]) && first.empty? result.pop while (last = result[-1]) && last.empty? end Parser.adjust_indentation! result, opts[:indent].to_i, (@document.attr 'tabsize').to_i if opts[:indent] result end
def preprocess_conditional_directive keyword, target, delimiter, text
ifndef directives, and for the conditional expression for the ifeval directive.
Used for a single-line conditional block in the case of the ifdef or
text - The text associated with this directive (occurring between the square brackets)
can be defined or undefined.
attributes must be defined or undefined, ',' means any of the attributes
delimiter - The conditional delimiter for multiple attributes ('+' means all
used in the condition (blank in the case of the ifeval directive)
target - The target, which is the name of one or more attributes that are
keyword - The conditional inclusion directive (ifdef, ifndef, ifeval, endif)
found.
preprocessing recursively until the next line of available content is
not skipping, mark whether the condition is satisfied and continue
open and close delimiters of any nested conditional blocks. If Reader is
the cursor. If Reader is currently skipping content, then simply track the
Preprocess the conditional directive (ifdef, ifndef, ifeval, endif) under
Internal: Preprocess the directive to conditionally include or exclude content.
def preprocess_conditional_directive keyword, target, delimiter, text # attributes are case insensitive target = target.downcase unless (no_target = target.empty?) if keyword == 'endif' if text logger.error message_with_context %(malformed preprocessor directive - text not permitted: endif::#{target}[#{text}]), source_location: cursor elsif @conditional_stack.empty? logger.error message_with_context %(unmatched preprocessor directive: endif::#{target}[]), source_location: cursor elsif no_target || target == (pair = @conditional_stack[-1])[:target] @conditional_stack.pop @skipping = @conditional_stack.empty? ? false : @conditional_stack[-1][:skipping] else logger.error message_with_context %(mismatched preprocessor directive: endif::#{target}[], expected endif::#{pair[:target]}[]), source_location: cursor end return true elsif @skipping skip = false else # QUESTION any way to wrap ifdef & ifndef logic up together? case keyword when 'ifdef' if no_target logger.error message_with_context %(malformed preprocessor directive - missing target: ifdef::[#{text}]), source_location: cursor return true end case delimiter when ',' # skip if no attribute is defined skip = target.split(',', -1).none? {|name| @document.attributes.key? name } when '+' # skip if any attribute is undefined skip = target.split('+', -1).any? {|name| !@document.attributes.key? name } else # if the attribute is undefined, then skip skip = !@document.attributes.key?(target) end when 'ifndef' if no_target logger.error message_with_context %(malformed preprocessor directive - missing target: ifndef::[#{text}]), source_location: cursor return true end case delimiter when ',' # skip if any attribute is defined skip = target.split(',', -1).any? {|name| @document.attributes.key? name } when '+' # skip if all attributes are defined skip = target.split('+', -1).all? {|name| @document.attributes.key? name } else # if the attribute is defined, then skip skip = @document.attributes.key?(target) end when 'ifeval' if no_target # the text in brackets must match a conditional expression if text && EvalExpressionRx =~ text.strip lhs = $1 op = $2 rhs = $3 # regex enforces a restricted set of math-related operations (==, !=, <=, >=, <, >) skip = ((resolve_expr_val lhs).send op, (resolve_expr_val rhs)) ? false : true else logger.error message_with_context %(malformed preprocessor directive - #{text ? 'invalid expression' : 'missing expression'}: ifeval::[#{text}]), source_location: cursor return true end else logger.error message_with_context %(malformed preprocessor directive - target not permitted: ifeval::#{target}[#{text}]), source_location: cursor return true end end end # conditional inclusion block if keyword == 'ifeval' || !text @skipping = true if skip @conditional_stack << { target: target, skip: skip, skipping: @skipping } # single line conditional inclusion else unless @skipping || skip replace_next_line text.rstrip # HACK push dummy line to stand in for the opening conditional directive that's subsequently dropped unshift '' # NOTE force line to be processed again if it looks like an include directive # QUESTION should we just call preprocess_include_directive here? @look_ahead -= 1 if text.start_with? 'include::' end end true end
def preprocess_include_directive target, attrlist
Returns a [Boolean] indicating whether the line under the cursor was changed. To skip over the
include directive.
attrlist - An attribute list String, which is the text between the square brackets of the
target slot of the include directive.
target - The unsubstituted String name of the target document to include as specified in the
If none of the above apply, emit the include directive line verbatim.
of the Array of source data.
stack size, normalize the target path and read the lines onto the beginning
Otherwise, if the max depth is greater than 0, and is not exceeded by the
attributes to that processor and expect an Array of String lines in return.
Otherwise, if an include processor is specified pass the target and
directive line is emitted verbatim.
If SafeMode is SECURE or greater, the directive is ignore and the include
are as follows:
Preprocess the directive to include the target document. The scenarios
Internal: Preprocess the directive to include lines from another document.
def preprocess_include_directive target, attrlist doc = @document if ((expanded_target = target).include? ATTR_REF_HEAD) && (expanded_target = doc.sub_attributes target, attribute_missing: ((attr_missing = doc.attributes['attribute-missing'] || Compliance.attribute_missing) == 'warn' ? 'drop-line' : attr_missing)).empty? if attr_missing == 'drop-line' && (doc.sub_attributes target + ' ', attribute_missing: 'drop-line', drop_line_severity: :ignore).empty? logger.info { message_with_context %(include dropped due to missing attribute: include::#{target}[#{attrlist}]), source_location: cursor } shift true elsif (doc.parse_attributes attrlist, [], sub_input: true)['optional-option'] logger.info { message_with_context %(optional include dropped #{attr_missing == 'warn' && (doc.sub_attributes target + ' ', attribute_missing: 'drop-line', drop_line_severity: :ignore).empty? ? 'due to missing attribute' : 'because resolved target is blank'}: include::#{target}[#{attrlist}]), source_location: cursor } shift true else logger.warn message_with_context %(include dropped #{attr_missing == 'warn' && (doc.sub_attributes target + ' ', attribute_missing: 'drop-line', drop_line_severity: :ignore).empty? ? 'due to missing attribute' : 'because resolved target is blank'}: include::#{target}[#{attrlist}]), source_location: cursor # QUESTION should this line include target or expanded_target (or escaped target?) replace_next_line %(Unresolved directive in #{@path} - include::#{target}[#{attrlist}]) end elsif include_processors? && (ext = @include_processor_extensions.find {|candidate| candidate.instance.handles? expanded_target }) shift # FIXME parse attributes only if requested by extension ext.process_method[doc, self, expanded_target, (doc.parse_attributes attrlist, [], sub_input: true)] true # if running in SafeMode::SECURE or greater, don't process this directive # however, be friendly and at least make it a link to the source document elsif doc.safe >= SafeMode::SECURE # FIXME we don't want to use a link macro if we are in a verbatim context replace_next_line %(link:#{expanded_target}[]) elsif @maxdepth if @include_stack.size >= @maxdepth[:curr] logger.error message_with_context %(maximum include depth of #{@maxdepth[:rel]} exceeded), source_location: cursor return end parsed_attrs = doc.parse_attributes attrlist, [], sub_input: true inc_path, target_type, relpath = resolve_include_path expanded_target, attrlist, parsed_attrs if target_type == :file reader = ::File.method :open read_mode = FILE_READ_MODE elsif target_type == :uri reader = ::OpenURI.method :open_uri read_mode = URI_READ_MODE else # NOTE if target_type is not set, inc_path is a boolean to skip over (false) or reevaluate (true) the current line return inc_path end inc_linenos = inc_tags = nil if attrlist if parsed_attrs.key? 'lines' inc_linenos = [] (split_delimited_value parsed_attrs['lines']).each do |linedef| if linedef.include? '..' from, _, to = linedef.partition '..' inc_linenos += (to.empty? || (to = to.to_i) < 0) ? [from.to_i, 1.0/0.0] : (from.to_i..to).to_a else inc_linenos << linedef.to_i end end inc_linenos = inc_linenos.empty? ? nil : inc_linenos.sort.uniq elsif parsed_attrs.key? 'tag' unless (tag = parsed_attrs['tag']).empty? || tag == '!' inc_tags = (tag.start_with? '!') ? { (tag.slice 1, tag.length) => false } : { tag => true } end elsif parsed_attrs.key? 'tags' inc_tags = {} (split_delimited_value parsed_attrs['tags']).each do |tagdef| if tagdef.start_with? '!' inc_tags[tagdef.slice 1, tagdef.length] = false else inc_tags[tagdef] = true end unless tagdef.empty? || tagdef == '!' end inc_tags = nil if inc_tags.empty? end end if inc_linenos inc_lines, inc_offset, inc_lineno = [], nil, 0 begin reader.call inc_path, read_mode do |f| select_remaining = nil f.each_line do |l| inc_lineno += 1 if select_remaining || (::Float === (select = inc_linenos[0]) && (select_remaining = select.infinite?)) # NOTE record line where we started selecting inc_offset ||= inc_lineno inc_lines << l else if select == inc_lineno # NOTE record line where we started selecting inc_offset ||= inc_lineno inc_lines << l inc_linenos.shift end break if inc_linenos.empty? end end end rescue logger.error message_with_context %(include #{target_type} not readable: #{inc_path}), source_location: cursor return replace_next_line %(Unresolved directive in #{@path} - include::#{expanded_target}[#{attrlist}]) end shift # FIXME not accounting for skipped lines in reader line numbering if inc_offset parsed_attrs['partial-option'] = '' push_include inc_lines, inc_path, relpath, inc_offset, parsed_attrs end elsif inc_tags inc_lines, inc_offset, inc_lineno, tag_stack, tags_used, active_tag = [], nil, 0, [], ::Set.new, nil if inc_tags.key? '**' if inc_tags.key? '*' select = base_select = inc_tags.delete '**' wildcard = inc_tags.delete '*' else select = base_select = wildcard = inc_tags.delete '**' end else select = base_select = !(inc_tags.value? true) wildcard = inc_tags.delete '*' end begin reader.call inc_path, read_mode do |f| dbl_co, dbl_sb = '::', '[]' f.each_line do |l| inc_lineno += 1 if (l.include? dbl_co) && (l.include? dbl_sb) && TagDirectiveRx =~ l this_tag = $2 if $1 # end tag if this_tag == active_tag tag_stack.pop active_tag, select = tag_stack.empty? ? [nil, base_select] : tag_stack[-1] elsif inc_tags.key? this_tag include_cursor = create_include_cursor inc_path, expanded_target, inc_lineno if (idx = tag_stack.rindex {|key, _| key == this_tag }) idx == 0 ? tag_stack.shift : (tag_stack.delete_at idx) logger.warn message_with_context %(mismatched end tag (expected '#{active_tag}' but found '#{this_tag}') at line #{inc_lineno} of include #{target_type}: #{inc_path}), source_location: cursor, include_location: include_cursor else logger.warn message_with_context %(unexpected end tag '#{this_tag}' at line #{inc_lineno} of include #{target_type}: #{inc_path}), source_location: cursor, include_location: include_cursor end end elsif inc_tags.key? this_tag tags_used << this_tag # QUESTION should we prevent tag from being selected when enclosing tag is excluded? tag_stack << [(active_tag = this_tag), (select = inc_tags[this_tag]), inc_lineno] elsif !wildcard.nil? select = active_tag && !select ? false : wildcard tag_stack << [(active_tag = this_tag), select, inc_lineno] end elsif select # NOTE record the line where we started selecting inc_offset ||= inc_lineno inc_lines << l end end end rescue logger.error message_with_context %(include #{target_type} not readable: #{inc_path}), source_location: cursor return replace_next_line %(Unresolved directive in #{@path} - include::#{expanded_target}[#{attrlist}]) end unless tag_stack.empty? tag_stack.each do |tag_name, _, tag_lineno| logger.warn message_with_context %(detected unclosed tag '#{tag_name}' starting at line #{tag_lineno} of include #{target_type}: #{inc_path}), source_location: cursor, include_location: (create_include_cursor inc_path, expanded_target, tag_lineno) end end unless (missing_tags = inc_tags.keys - tags_used.to_a).empty? logger.warn message_with_context %(tag#{missing_tags.size > 1 ? 's' : ''} '#{missing_tags.join ', '}' not found in include #{target_type}: #{inc_path}), source_location: cursor end shift if inc_offset parsed_attrs['partial-option'] = '' unless base_select && wildcard && inc_tags.empty? # FIXME not accounting for skipped lines in reader line numbering push_include inc_lines, inc_path, relpath, inc_offset, parsed_attrs end else begin # NOTE read content before shift so cursor is only advanced if IO operation succeeds inc_content = reader.call(inc_path, read_mode) {|f| f.read } shift push_include inc_content, inc_path, relpath, 1, parsed_attrs rescue logger.error message_with_context %(include #{target_type} not readable: #{inc_path}), source_location: cursor return replace_next_line %(Unresolved directive in #{@path} - include::#{expanded_target}[#{attrlist}]) end end true end end
def process_line line
def process_line line return line unless @process_lines if line.empty? @look_ahead += 1 return line end # NOTE highly optimized if line.end_with?(']') && !line.start_with?('[') && line.include?('::') if (line.include? 'if') && ConditionalDirectiveRx =~ line # if escaped, mark as processed and return line unescaped if $1 == '\\' @unescape_next_line = true @look_ahead += 1 line.slice 1, line.length elsif preprocess_conditional_directive $2, $3, $4, $5 # move the pointer past the conditional line shift # treat next line as uncharted territory nil else # the line was not a valid conditional line # mark it as visited and return it @look_ahead += 1 line end elsif @skipping shift nil elsif (line.start_with? 'inc', '\\inc') && IncludeDirectiveRx =~ line # if escaped, mark as processed and return line unescaped if $1 == '\\' @unescape_next_line = true @look_ahead += 1 line.slice 1, line.length # QUESTION should we strip whitespace from raw attributes in Substitutors#parse_attributes? (check perf) elsif preprocess_include_directive $2, $3 # peek again since the content has changed nil else # the line was not a valid include line and is unchanged # mark it as visited and return it @look_ahead += 1 line end else # NOTE optimization to inline super @look_ahead += 1 line end elsif @skipping shift nil else # NOTE optimization to inline super @look_ahead += 1 line end end
def push_include data, file = nil, path = nil, lineno = 1, attributes = {}
reader.push_include data, file, path
data = File.read file
file = File.expand_path path
path = 'partial.adoc'
Examples
read from the target specified.
This method is typically used in an IncludeProcessor to add source
based on the file, document-relative path and line information given.
Public: Push source onto the front of the reader and switch the context
def push_include data, file = nil, path = nil, lineno = 1, attributes = {} @include_stack << [@lines, @file, @dir, @path, @lineno, @maxdepth, @process_lines] if (@file = file) # NOTE if file is not a string, assume it's a URI if ::String === file @dir = ::File.dirname file elsif RUBY_ENGINE_OPAL @dir = ::URI.parse ::File.dirname(file = file.to_s) else # NOTE this intentionally throws an error if URI has no path (@dir = file.dup).path = (dir = ::File.dirname file.path) == '/' ? '' : dir file = file.to_s end path ||= ::File.basename file # only process lines in AsciiDoc files @process_lines = ASCIIDOC_EXTENSIONS[::File.extname file] else @dir = '.' # we don't know what file type we have, so assume AsciiDoc @process_lines = true end if path @path = path @includes[Helpers.rootname path] = attributes['partial-option'] ? nil : true if @process_lines else @path = '<stdin>' end @lineno = lineno if @maxdepth && (attributes.key? 'depth') if (rel_maxdepth = attributes['depth'].to_i) > 0 if (curr_maxdepth = @include_stack.size + rel_maxdepth) > (abs_maxdepth = @maxdepth[:abs]) # if relative depth exceeds absolute max depth, effectively ignore relative depth request curr_maxdepth = rel_maxdepth = abs_maxdepth end @maxdepth = { abs: abs_maxdepth, curr: curr_maxdepth, rel: rel_maxdepth } else @maxdepth = { abs: @maxdepth[:abs], curr: @include_stack.size, rel: 0 } end end # effectively fill the buffer if (@lines = prepare_lines data, normalize: true, condense: false, indent: attributes['indent']).empty? pop_include else # FIXME we eventually want to handle leveloffset without affecting the lines if attributes.key? 'leveloffset' @lines.unshift '' @lines.unshift %(:leveloffset: #{attributes['leveloffset']}) @lines << '' if (old_leveloffset = @document.attr 'leveloffset') @lines << %(:leveloffset: #{old_leveloffset}) else @lines << ':leveloffset!:' end # compensate for these extra lines @lineno -= 2 end # FIXME kind of a hack #Document::AttributeEntry.new('infile', @file).save_to_next_block @document #Document::AttributeEntry.new('indir', @dir).save_to_next_block @document @look_ahead = 0 end self end
def resolve_expr_val val
# => "value"
resolve_expr_val expr
expr = '"{name}"'
@document.attributes['name'] = 'value'
# => 2
resolve_expr_val expr
expr = '2'
# => nil
resolve_expr_val expr
expr = '{undefined}'
# => ""
resolve_expr_val expr
expr = '"{undefined}"'
# => "\"value"
resolve_expr_val expr
expr = '"value'
# => "value"
resolve_expr_val expr
expr = '"value"'
Examples
Private: Resolve the value of one side of the expression
def resolve_expr_val val if ((val.start_with? '"') && (val.end_with? '"')) || ((val.start_with? '\'') && (val.end_with? '\'')) quoted = true val = val.slice 1, (val.length - 1) else quoted = false end # QUESTION should we substitute first? # QUESTION should we also require string to be single quoted (like block attribute values?) val = @document.sub_attributes val, attribute_missing: 'drop' if val.include? ATTR_REF_HEAD if quoted val elsif val.empty? nil elsif val == 'true' true elsif val == 'false' false elsif val.rstrip.empty? ' ' elsif val.include? '.' val.to_f else # fallback to coercing to integer, since we # require string values to be explicitly quoted val.to_i end end
def resolve_include_path target, attrlist, attributes
Returns An Array containing the resolved (absolute) include path, the target type, and the path
attributes - A Hash of attributes parsed from attrlist.
attrlist - An attribute list String (i.e., the text between the square brackets).
(Attribute references in target value have already been resolved).
target - A String containing the unresolved include target.
environment.
This method is overridden in Asciidoctor.js to resolve the target of an include in the browser
cursor should be advanced beyond this line (true) or the line should be reprocessed (false).
return a boolean to halt processing of the include directive line and to indicate whether the
and the path of the target relative to the outermost document. Alternately, the method may
Array containing the resolved (absolute) path of the target, the target type (:file or :uri),
An internal method to resolve the target of an include directive. This method must return an
Internal: Resolve the target of an include directive.
def resolve_include_path target, attrlist, attributes doc = @document if (Helpers.uriish? target) || (::String === @dir ? nil : (target = %(#{@dir}/#{target}))) return replace_next_line %(link:#{target}[#{attrlist}]) unless doc.attr? 'allow-uri-read' if doc.attr? 'cache-uri' # caching requires the open-uri-cached gem to be installed # processing will be automatically aborted if these libraries can't be opened Helpers.require_library 'open-uri/cached', 'open-uri-cached' unless defined? ::OpenURI::Cache elsif !RUBY_ENGINE_OPAL # autoload open-uri ::OpenURI end [(::URI.parse target), :uri, target] else # include file is resolved relative to dir of current include, or base_dir if within original docfile inc_path = doc.normalize_system_path target, @dir, nil, target_name: 'include file' unless ::File.file? inc_path if attributes['optional-option'] logger.info { message_with_context %(optional include dropped because include file not found: #{inc_path}), source_location: cursor } shift return true else logger.error message_with_context %(include file not found: #{inc_path}), source_location: cursor return replace_next_line %(Unresolved directive in #{@path} - include::#{target}[#{attrlist}]) end end # NOTE relpath is the path relative to the root document (or base_dir, if set) # QUESTION should we move relative_path method to Document relpath = doc.path_resolver.relative_path inc_path, doc.base_dir [inc_path, :file, relpath] end end
def shift
also, we now have the field in the super class, so perhaps
TODO Document this override
def shift if @unescape_next_line @unescape_next_line = false (line = super).slice 1, line.length else super end end
def skip_front_matter! data, increment_linenos = true
def skip_front_matter! data, increment_linenos = true front_matter = nil if data[0] == '---' original_data = data.drop 0 data.shift front_matter = [] @lineno += 1 if increment_linenos while !data.empty? && data[0] != '---' front_matter << data.shift @lineno += 1 if increment_linenos end if data.empty? data.unshift(*original_data) @lineno = 0 if increment_linenos front_matter = nil else data.shift @lineno += 1 if increment_linenos end end front_matter end
def split_delimited_value val
def split_delimited_value val (val.include? ',') ? (val.split ',') : (val.split ';') end
def to_s
def to_s %(#<#{self.class}@#{object_id} {path: #{@path.inspect}, line: #{@lineno}, include depth: #{@include_stack.size}, include stack: [#{@include_stack.map {|inc| inc.to_s }.join ', '}]}>) end