diff --git a/lib/html/proofer.rb b/lib/html/proofer.rb
index fcbd08ba..967e0f7f 100644
--- a/lib/html/proofer.rb
+++ b/lib/html/proofer.rb
@@ -1,4 +1,6 @@
require 'nokogiri'
+require 'find'
+require 'html/proofer/checkable'
require 'html/proofer/checks'
module HTML
@@ -24,7 +26,10 @@ def run
end
end
- if !@failedTests.empty?
+ if @failedTests.empty?
+ puts "Tests executed sucesfully.".green
+ exit 0
+ else
# make the hash default to 0 so that += will work correctly
count = Hash.new(0)
@@ -41,7 +46,7 @@ def run
end
def self.create_nokogiri(path)
- path << "index.html" if File.directory? path #support for Jekyll-style links
+ path << "/index.html" if File.directory? path #support for Jekyll-style links
Nokogiri::HTML(File.read(path))
end
diff --git a/lib/html/proofer/check.rb b/lib/html/proofer/check.rb
index c92ffb15..b49d1a39 100644
--- a/lib/html/proofer/check.rb
+++ b/lib/html/proofer/check.rb
@@ -10,7 +10,7 @@ class HTML::Proofer::Checks
class Check
- attr_reader :issues, :hydra
+ attr_reader :issues, :hydra, :src, :path, :options, :additional_href_ignores
def initialize(src, path, html, opts={})
@src = src
@@ -28,31 +28,13 @@ def run
end
def add_issue(desc)
- @issues << desc
+ @issues << "#{@path.blue}: #{desc}"
end
def output_filenames
Dir[@site.config[:output_dir] + '/**/*'].select{ |f| File.file?(f) }
end
- def external_href?(href)
- uri = URI.parse(href)
- %w( http https ).include?(uri.scheme)
- rescue URI::BadURIError
- false
- rescue URI::InvalidURIError
- false
- end
-
- def ignore_href?(href)
- uri = URI.parse(href)
- %w( mailto ).include?(uri.scheme) || @additional_href_ignores.include?(href)
- rescue URI::BadURIError
- false
- rescue URI::InvalidURIError
- false
- end
-
def validate_url(href, issue_text)
request = Typhoeus::Request.new(href, {:followlocation => true})
request.on_complete do |response|
@@ -93,34 +75,5 @@ def self.subclasses
classes
end
- def resolve_path(path)
-
- #Strip anchor, not needed to resolve file
- path = path.split('#').first
-
- if path =~ /^\// #path relative to root
- base = @src
- elsif File.exist? File.join Dir.pwd, @path #relative links, path is a file
- base = File.dirname(@path)
- else #relative link, path is a directory
- base = @path
- end
-
- if path =~ /^#/ #anchor link, no trailing slash
- path = "#{base}#{path}"
- else # relative path, resolve trailing slashes automatically
- path = File.join base, path
- end
-
- # implicit /index.html support, with support for tailing slashes
- path = File.join path, "index.html" if File.directory? File.join Dir.pwd, path
-
- path
- end
-
- # checks if a file exists relative to the current pwd
- def file?(path)
- File.exist? File.join Dir.pwd, resolve_path(path)
- end
end
end
diff --git a/lib/html/proofer/checkable.rb b/lib/html/proofer/checkable.rb
new file mode 100644
index 00000000..9922dd26
--- /dev/null
+++ b/lib/html/proofer/checkable.rb
@@ -0,0 +1,106 @@
+module HTML
+ class Proofer
+ class Checkable
+
+ def initialize(obj, check)
+ @src = obj['src']
+ @href = obj['href']
+ @alt = obj['alt']
+ @name = obj['name']
+ @id = obj['id']
+ @check = check
+
+ if @href && @check.options[:href_swap]
+ @options[:href_swap].each do |link, replace|
+ @href = @href.gsub(link, replace)
+ end
+ end
+
+ end
+
+ def url
+ @src || @href || ""
+ end
+
+ def valid?
+ begin
+ URI.parse url
+ rescue
+ false
+ end
+ end
+
+ def parts
+ URI.parse url
+ end
+
+ def path
+ parts.path
+ end
+
+ def hash
+ parts.fragment
+ end
+
+ # path is to an external server
+ def remote?
+ uri = URI.parse url
+ %w( http https ).include?(uri.scheme)
+ rescue URI::BadURIError
+ false
+ rescue URI::InvalidURIError
+ false
+ end
+
+ def ignore?
+ uri = URI.parse url
+ %w( mailto ).include?(uri.scheme) || @check.additional_href_ignores.include?(href)
+ rescue URI::BadURIError
+ false
+ rescue URI::InvalidURIError
+ false
+ end
+
+ # path is external to the file
+ def external?
+ !internal?
+ end
+
+ # path is an anchor
+ def internal?
+ url[0] == "#"
+ end
+
+ def file_path
+
+ return if path.nil?
+
+ if path =~ /^\// #path relative to root
+ base = @check.src
+ elsif File.exist? File.expand_path path, @check.src #relative links, path is a file
+ base = File.dirname @check.path
+ else #relative link, path is a directory
+ base = @check.path
+ end
+
+ file = File.join base, path
+
+ # implicit /index.html support, with support for tailing slashes
+ file = File.join path, "index.html" if File.directory? File.expand_path file, @check.src
+
+ file
+ end
+
+ # checks if a file exists relative to the current pwd
+ def exists?
+ File.exist? absolute_path
+ end
+
+ def absolute_path
+ path = file_path || @check.path
+ File.expand_path path, Dir.pwd
+ end
+
+ end
+ end
+end
diff --git a/lib/html/proofer/checks/images.rb b/lib/html/proofer/checks/images.rb
index 0e18ac4a..cd0d538c 100644
--- a/lib/html/proofer/checks/images.rb
+++ b/lib/html/proofer/checks/images.rb
@@ -1,30 +1,50 @@
# encoding: utf-8
+class Image < ::HTML::Proofer::Checkable
+
+ SCREEN_SHOT_REGEX = /Screen(?: |%20)Shot(?: |%20)\d+-\d+-\d+(?: |%20)at(?: |%20)\d+.\d+.\d+/
+
+ def valid_alt_tag?
+ @alt and !@alt.empty?
+ end
+
+ def terrible_filename?
+ @src =~ SCREEN_SHOT_REGEX
+ end
+
+ def src
+ @src unless @src.nil? || @src.empty?
+ end
+
+ def missing_src?
+ !src
+ end
+
+end
+
class Images < ::HTML::Proofer::Checks::Check
def run
@html.css('img').each do |img|
- src = img['src']
-
- # check image sources
- if src && src.length > 0
- if !external_href?(src)
- self.add_issue("#{@path}".blue + ": internal image #{src} does not exist") unless file? src
- else
- validate_url(src, "#{@path}".blue + ": external image #{src} does not exist")
- end
+
+ img = Image.new img, self
+
+ # screenshot filenames, return because invalid URL
+ return self.add_issue "image has a terrible filename (#{img.src})" if img.terrible_filename?
+
+ # does the image exist?
+ if img.missing_src?
+ self.add_issue "image has no src attribute"
+ elsif img.remote?
+ validate_url img.src, "external image #{img.src} does not exist"
else
- self.add_issue("#{@path}".blue + ": image has no src attribute")
+ self.add_issue("internal image #{img.src} does not exist") unless img.exists?
end
# check alt tag
- self.add_issue("#{@path}".blue + ": image #{src} does not have an alt attribute") unless img['alt'] and !img['alt'].empty?
+ self.add_issue "image #{img.src} does not have an alt attribute" unless img.valid_alt_tag?
- screenShotRegExp = /Screen(?: |%20)Shot(?: |%20)\d+-\d+-\d+(?: |%20)at(?: |%20)\d+.\d+.\d+/
- if src =~ screenShotRegExp
- self.add_issue("#{@path}".blue + ": image has a terrible filename (#{src})")
- end
end
end
end
diff --git a/lib/html/proofer/checks/links.rb b/lib/html/proofer/checks/links.rb
index b0b6772f..37182959 100644
--- a/lib/html/proofer/checks/links.rb
+++ b/lib/html/proofer/checks/links.rb
@@ -1,56 +1,48 @@
# encoding: utf-8
+class Link < ::HTML::Proofer::Checkable
+
+ def href
+ @href unless @href.nil? || @href.empty?
+ end
+
+ def missing_href?
+ href.nil? and @name.nil? and @id.nil?
+ end
+
+end
+
class Links < ::HTML::Proofer::Checks::Check
def run
- @html.css('a').each do |a|
- href = a['href']
-
- if href && href.length > 0
- if @options[:href_swap]
- @options[:href_swap].each do |link, replace|
- href = href.gsub(link, replace)
- end
- end
+ @html.css('a').each do |link|
- return if ignore_href?(href)
+ link = Link.new link, self
+ return if link.ignore?
- if href.include? '#'
- href_split = href.split('#')
- end
- if !external_href?(href)
-
- # an internal link, with a hash
- if href_split && !href_split.empty?
- href_file = href_split[0]
- href_hash = href_split[1]
-
- # it's not an internal hash; it's pointing to some other file
- if href_file.length > 0
- if !file?(href_file)
- self.add_issue("#{@path}".blue + ": internal link #{href_file} does not exist")
- else
- href_html = HTML::Proofer.create_nokogiri(resolve_path(href_file))
- found_hash_match = false
- unless hash_check(href_html, href_hash)
- self.add_issue("#{@path}".blue + ": linking to #{href}, but #{href_hash} does not exist")
- end
- end
- # it is an internal link, with an internal hash
- else
- unless hash_check(@html, href_hash)
- self.add_issue("#{@path}".blue + ": linking to an internal hash called #{href_hash} that does not exist")
- end
- end
- # internal link, no hash
- else
- self.add_issue("#{@path}".blue + ": internally linking to #{href}, which does not exist") unless file?(href)
- end
- else
- validate_url(href, "#{@path}".blue + ": externally linking to #{href}, which does not exist")
- end
+ # is there even a href?
+ return self.add_issue("link has no href attribute") if link.missing_href?
+
+ # is it even a valid URL?
+ return self.add_issue "#{link.href} is an invalid URL" unless link.valid?
+
+ # does the file even exist?
+ if link.remote?
+ validate_url link.href, "externally linking to #{link.href}, which does not exist"
else
- self.add_issue("#{@path}".blue + ": link has no href attribute") unless a['name'] || a['id']
+ self.add_issue "internally linking to #{link.href}, which does not exist" unless link.exists?
+ end
+
+ # verify the target hash
+ if link.hash
+ if link.remote?
+ #not yet checked
+ elsif link.internal?
+ self.add_issue "linking to internal hash ##{link.hash} that does not exist" unless hash_check @html, link.hash
+ elsif link.external?
+ target_html = HTML::Proofer.create_nokogiri link.absolute_path
+ self.add_issue "linking to #{link.href}, but #{link.hash} does not exist" unless hash_check target_html, link.hash
+ end
end
end
end
diff --git a/spec/html/proofer/fixtures/brokenInternalLink.html b/spec/html/proofer/fixtures/brokenInternalLink.html
new file mode 100644
index 00000000..e69de29b
diff --git a/spec/html/proofer/fixtures/brokenLinkWithNumber.html b/spec/html/proofer/fixtures/brokenLinkWithNumber.html
index 1e316b14..31a75f94 100644
--- a/spec/html/proofer/fixtures/brokenLinkWithNumber.html
+++ b/spec/html/proofer/fixtures/brokenLinkWithNumber.html
@@ -2,7 +2,7 @@
- Blah blah blah. 25 Method not allowed
+ Blah blah blah. 25 Method not allowed
-