require "erubis"
module ActionView
class Template
module Handlers
class Erubis < ::Erubis::Eruby
def add_preamble(src)
@newline_pending = 0
src << "@output_buffer = output_buffer || ActionView::OutputBuffer.new;"
end
def add_text(src, text)
return if text.empty?
if text == "\n"
@newline_pending += 1
else
src << "@output_buffer.safe_append='"
src << "\n" * @newline_pending if @newline_pending > 0
src << escape_text(text)
src << "'.freeze;"
@newline_pending = 0
end
end
# Erubis toggles <%= and <%== behavior when escaping is enabled.
# We override to always treat <%== as escaped.
def add_expr(src, code, indicator)
case indicator
when "=="
add_expr_escaped(src, code)
else
super
end
end
BLOCK_EXPR = /\s*((\s+|\))do|\{)(\s*\|[^|]*\|)?\s*\Z/
def add_expr_literal(src, code)
flush_newline_if_pending(src)
if BLOCK_EXPR.match?(code)
src << "@output_buffer.append= " << code
else
src << "@output_buffer.append=(" << code << ");"
end
end
def add_expr_escaped(src, code)
flush_newline_if_pending(src)
if BLOCK_EXPR.match?(code)
src << "@output_buffer.safe_expr_append= " << code
else
src << "@output_buffer.safe_expr_append=(" << code << ");"
end
end
def add_stmt(src, code)
flush_newline_if_pending(src)
super
end
def add_postamble(src)
flush_newline_if_pending(src)
src << "@output_buffer.to_s"
end
def flush_newline_if_pending(src)
if @newline_pending > 0
src << "@output_buffer.safe_append='#{"\n" * @newline_pending}'.freeze;"
@newline_pending = 0
end
end
end
class ERB
# Specify trim mode for the ERB compiler. Defaults to '-'.
# See ERB documentation for suitable values.
class_attribute :erb_trim_mode
self.erb_trim_mode = "-"
# Default implementation used.
class_attribute :erb_implementation
self.erb_implementation = Erubis
# Do not escape templates of these mime types.
class_attribute :escape_whitelist
self.escape_whitelist = ["text/plain"]
ENCODING_TAG = Regexp.new("\\A(<%#{ENCODING_FLAG}-?%>)[ \\t]*")
def self.call(template)
new.call(template)
end
def supports_streaming?
true
end
def handles_encoding?
true
end
def call(template)
# First, convert to BINARY, so in case the encoding is
# wrong, we can still find an encoding tag
# (<%# encoding %>) inside the String using a regular
# expression
template_source = template.source.dup.force_encoding(Encoding::ASCII_8BIT)
erb = template_source.gsub(ENCODING_TAG, "")
encoding = $2
erb.force_encoding valid_encoding(template.source.dup, encoding)
# Always make sure we return a String in the default_internal
erb.encode!
self.class.erb_implementation.new(
erb,
escape: (self.class.escape_whitelist.include? template.type),
trim: (self.class.erb_trim_mode == "-")
).src
end
private
def valid_encoding(string, encoding)
# If a magic encoding comment was found, tag the
# String with this encoding. This is for a case
# where the original String was assumed to be,
# for instance, UTF-8, but a magic comment
# proved otherwise
string.force_encoding(encoding) if encoding
# If the String is valid, return the encoding we found
return string.encoding if string.valid_encoding?
# Otherwise, raise an exception
raise WrongEncodingError.new(string, string.encoding)
end
end
end
end
end