middleman/middleman-core/lib/middleman-core/util.rb

311 lines
10 KiB
Ruby

# For instrumenting
require 'active_support/notifications'
# Using Thor's indifferent hash access
require 'thor'
# Core Pathname library used for traversal
require 'pathname'
# Template and Mime detection
require 'tilt'
require 'rack/mime'
module Middleman
module Util
class << self
# Whether the source file is binary.
#
# @param [String] filename The file to check.
# @return [Boolean]
def binary?(filename)
ext = File.extname(filename)
# We hardcode detecting of gzipped SVG files
return true if ext == '.svgz'
return false if Tilt.registered?(ext.sub('.', ''))
dot_ext = (ext.to_s[0] == '.') ? ext.dup : ".#{ext}"
if mime = ::Rack::Mime.mime_type(dot_ext, nil)
!nonbinary_mime?(mime)
else
file_contents_include_binary_bytes?(filename)
end
end
# Facade for ActiveSupport/Notification
def instrument(name, payload={}, &block)
suffixed_name = (name =~ /\.middleman$/) ? name.dup : "#{name}.middleman"
::ActiveSupport::Notifications.instrument(suffixed_name, payload, &block)
end
# Recursively convert a normal Hash into a HashWithIndifferentAccess
#
# @private
# @param [Hash] data Normal hash
# @return [Thor::CoreExt::HashWithIndifferentAccess]
def recursively_enhance(data)
if data.is_a? Hash
data = ::Thor::CoreExt::HashWithIndifferentAccess.new(data)
data.each do |key, val|
data[key] = recursively_enhance(val)
end
data
elsif data.is_a? Array
data.each_with_index do |val, i|
data[i] = recursively_enhance(val)
end
data
else
data
end
end
# Normalize a path to not include a leading slash
# @param [String] path
# @return [String]
def normalize_path(path)
# The tr call works around a bug in Ruby's Unicode handling
URI.decode(path).sub(%r{^/}, '').tr('', '')
end
# This is a separate method from normalize_path in case we
# change how we normalize paths
def strip_leading_slash(path)
path.sub(%r{^/}, '')
end
# Extract the text of a Rack response as a string.
# Useful for extensions implemented as Rack middleware.
# @param response The response from #call
# @return [String] The whole response as a string.
def extract_response_text(response)
# The rack spec states all response bodies must respond to each
result = ''
response.each do |part, _|
result << part
end
result
end
# Takes a matcher, which can be a literal string
# or a string containing glob expressions, or a
# regexp, or a proc, or anything else that responds
# to #match or #call, and returns whether or not the
# given path matches that matcher.
#
# @param matcher A matcher string/regexp/proc/etc
# @param path A path as a string
# @return [Boolean] Whether the path matches the matcher
def path_match(matcher, path)
case
when matcher.is_a?(String)
path.match(matcher)
when matcher.respond_to?(:match)
matcher.match(path)
when matcher.respond_to?(:call)
matcher.call(path)
else
File.fnmatch(matcher.to_s, path)
end
end
# Get a recursive list of files inside a path.
# Works with symlinks.
#
# @param path Some path string or Pathname
# @param ignore A proc/block that returns true if a given path should be ignored - if a path
# is ignored, nothing below it will be searched either.
# @return [Array<Pathname>] An array of Pathnames for each file (no directories)
def all_files_under(path, &ignore)
path = Pathname(path)
return [] if ignore && ignore.call(path)
if path.directory?
path.children.flat_map do |child|
all_files_under(child, &ignore)
end.compact
elsif path.file?
[path]
else
[]
end
end
# Given a source path (referenced either absolutely or relatively)
# or a Resource, this will produce the nice URL configured for that
# path, respecting :relative_links, directory indexes, etc.
def url_for(app, path_or_resource, options={})
# Handle Resources and other things which define their own url method
url = if path_or_resource.respond_to?(:url)
path_or_resource.url
else
path_or_resource.dup
end
# Try to parse URL
begin
uri = URI(url)
rescue URI::InvalidURIError
# Nothing we can do with it, it's not really a URI
return url
end
relative = options[:relative]
raise "Can't use the relative option with an external URL" if relative && uri.host
# Allow people to turn on relative paths for all links with
# set :relative_links, true
# but still override on a case by case basis with the :relative parameter.
effective_relative = relative || false
effective_relative = true if relative.nil? && app.config[:relative_links]
# Try to find a sitemap resource corresponding to the desired path
this_resource = options[:current_resource]
if path_or_resource.is_a?(::Middleman::Sitemap::Resource)
resource = path_or_resource
resource_url = url
elsif this_resource && uri.path && !uri.host
# Handle relative urls
url_path = Pathname(uri.path)
current_source_dir = Pathname('/' + this_resource.path).dirname
url_path = current_source_dir.join(url_path) if url_path.relative?
resource = app.sitemap.find_resource_by_path(url_path.to_s)
resource_url = resource.url if resource
elsif options[:find_resource] && uri.path && !uri.host
resource = app.sitemap.find_resource_by_path(uri.path)
resource_url = resource.url if resource
end
if resource
uri.path = URI.encode(relative_path_from_resource(this_resource, resource_url, effective_relative))
else
# If they explicitly asked for relative links but we can't find a resource...
raise "No resource exists at #{url}" if relative
end
# Support a :query option that can be a string or hash
if query = options[:query]
uri.query = query.respond_to?(:to_param) ? query.to_param : query.to_s
end
# Support a :fragment or :anchor option just like Padrino
fragment = options[:anchor] || options[:fragment]
uri.fragment = fragment.to_s if fragment
# Finally make the URL back into a string
uri.to_s
end
# Expand a path to include the index file if it's a directory
#
# @param [String] path Request path/
# @param [Middleman::Application] app The requesting app.
# @return [String] Path with index file if necessary.
def full_path(path, app)
resource = app.sitemap.find_resource_by_destination_path(path)
unless resource
# Try it with /index.html at the end
indexed_path = File.join(path.sub(%r{/$}, ''), app.config[:index_file])
resource = app.sitemap.find_resource_by_destination_path(indexed_path)
end
if resource
'/' + resource.destination_path
else
'/' + normalize_path(path)
end
end
# Glob a directory and try to keep path encoding consistent.
#
# @param [String] path The glob path.
# @return [Array<String>]
def glob_directory(path)
results = ::Dir[path]
return results unless RUBY_PLATFORM =~ /darwin/
results.map { |r| r.encode('UTF-8', 'UTF-8-MAC') }
end
# Get the PWD and try to keep path encoding consistent.
#
# @param [String] path The glob path.
# @return [Array<String>]
def current_directory
result = ::Dir.pwd
return result unless RUBY_PLATFORM =~ /darwin/
result.encode('UTF-8', 'UTF-8-MAC')
end
private
# Is mime type known to be non-binary?
#
# @param [String] mime The mimetype to check.
# @return [Boolean]
def nonbinary_mime?(mime)
case
when mime.start_with?('text/')
true
when mime.include?('xml')
true
when mime.include?('json')
true
when mime.include?('javascript')
true
else
false
end
end
# Read a few bytes from the file and see if they are binary.
#
# @param [String] filename The file to check.
# @return [Boolean]
def file_contents_include_binary_bytes?(filename)
binary_bytes = [0, 1, 2, 3, 4, 5, 6, 11, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 28, 29, 30, 31]
s = File.read(filename, 4096) || ''
s.each_byte do |c|
return true if binary_bytes.include?(c)
end
false
end
# Get a relative path to a resource.
#
# @param [Middleman::Sitemap::Resource] curr_resource The resource.
# @param [String] resource_url The target url.
# @param [Boolean] relative If the path should be relative.
# @return [String]
def relative_path_from_resource(curr_resource, resource_url, relative)
# Switch to the relative path between resource and the given resource
# if we've been asked to.
if relative && curr_resource
# Output urls relative to the destination path, not the source path
current_dir = Pathname('/' + curr_resource.destination_path).dirname
relative_path = Pathname(resource_url).relative_path_from(current_dir).to_s
# Put back the trailing slash to avoid unnecessary Apache redirects
if resource_url.end_with?('/') && !relative_path.end_with?('/')
relative_path << '/'
end
relative_path
else
resource_url
end
end
end
end
end