sitediff 1.0.0 → 1.1.1

data/lib/sitediff/config.rb

@@ -20,8 +20,8 @@ class SiteDiff
       'settings' => {
         'depth' => 3,
         'interval' => 0,
-        'whitelist' => '',
-        'blacklist' => '',
+        'include' => '',
+        'exclude' => '',
         'concurrency' => 3,
         'preset' => nil
       },
@@ -43,6 +43,8 @@ class SiteDiff
       after_url
       ignore_whitespace
       export
+      output
+      report
     ]
 
     ##
@@ -52,8 +54,8 @@ class SiteDiff
     ALLOWED_SETTINGS_KEYS = %w[
       preset
       depth
-      whitelist
-      blacklist
+      include
+      exclude
       concurrency
       interval
       curl_opts
@@ -62,6 +64,8 @@ class SiteDiff
     class InvalidConfig < SiteDiffException; end
     class ConfigNotFound < SiteDiffException; end
 
+    attr_reader :directory
+
     # Takes a Hash and normalizes it to the following form by merging globals
     # into before and after. A normalized config Hash looks like this:
     #
@@ -71,6 +75,12 @@ class SiteDiff
     #     before:
     #       url: http://before
     #       selector: body
+    #       ## Note: use either `selector` or `regions`, but not both
+    #       regions:
+    #         - name: title
+    #           selector: .field-name-title h2
+    #         - name: body
+    #           selector: .field-name-field-news-description .field-item
     #       dom_transform:
     #       - type: remove
     #         selector: script
@@ -79,6 +89,13 @@ class SiteDiff
     #       url: http://after
     #       selector: body
     #
+    #     ## Note: use `output` only with `regions`
+    #     output:
+    #       - title
+    #       - author
+    #       - source
+    #       - body
+    #
     def self.normalize(conf)
       tools = Sanitizer::TOOLS
 
@@ -117,6 +134,7 @@ class SiteDiff
       result = {
         'before' => {},
         'after' => {},
+        'output' => [],
         'settings' => {}
       }
 
@@ -149,12 +167,26 @@ class SiteDiff
         end
       end
 
+      # Merge output array.
+      result['output'] += (first['output'] || []) + (second['output'] || [])
+
+      # Merge url_report keys.
+      %w[before_url_report after_url_report].each do |pos|
+        result[pos] = first[pos] || second[pos]
+      end
+
       # Merge settings.
       result['settings'] = merge_deep(
         first['settings'] || {},
         second['settings'] || {}
       )
 
+      # Merge report labels.
+      result['report'] = merge_deep(
+        first['report'] || {},
+        second['report'] || {}
+      )
+
       result
     end
 
@@ -162,9 +194,10 @@ class SiteDiff
     # Merges 2 iterable objects deeply.
     def self.merge_deep(first, second)
       first.merge(second) do |_key, val1, val2|
-        if val1.is_a? Hash
+        case val1.class
+        when Hash
           self.class.merge_deep(val1, val2 || {})
-        elsif val1.is_a? Array
+        when Array
           val1 + (val2 || [])
         else
           val2
@@ -280,6 +313,33 @@ class SiteDiff
       @config['export'] = export
     end
 
+    # Get output option
+    def output
+      @config['output']
+    end
+
+    # Set output option
+    def output=(output)
+      raise 'Output must be an Array' unless output.is_a? Array
+
+      @config['output'] = output
+    end
+
+    # Return report display settings.
+    def report
+      @config['report']
+    end
+
+    # Set crawl time for 'before'
+    def before_time=(time)
+      @config['report']['before_time'] = time
+    end
+
+    # Set crawl time for 'after'
+    def after_time=(time)
+      @config['report']['after_time'] = time
+    end
+
     ##
     # Writes an array of paths to a file.
     #
@@ -407,6 +467,18 @@ class SiteDiff
       @return_value
     end
 
+    ##
+    # Return merged CURL options.
+    def curl_opts
+      # We do want string keys here
+      bool_hash = { 'true' => true, 'false' => false }
+      curl_opts = UriWrapper::DEFAULT_CURL_OPTS
+                  .clone
+                  .merge(settings['curl_opts'] || {})
+      curl_opts.each { |k, v| curl_opts[k] = bool_hash.fetch(v, v) }
+      curl_opts
+    end
+
     private
 
     ##

data/lib/sitediff/config/creator.rb

@@ -14,10 +14,10 @@ class SiteDiff
     class Creator
       ##
       # Creates a Creator object.
-      def initialize(debug, *urls)
+      def initialize(debug, before, after)
         @config = nil
-        @after = urls.pop
-        @before = urls.pop # May be nil
+        @before = before
+        @after = after
         @debug = debug
       end
 
@@ -25,15 +25,17 @@ class SiteDiff
       # Determine if we're dealing with one or two URLs.
       def roots
         @roots = { 'after' => @after }
-        @roots['before'] = @before if @before
+        @roots['before'] = @before || @after
         @roots
       end
 
       ##
       # Build a config structure, return it.
-      def create(options, &block)
+      def create(options)
         @config = {}
-        @callback = block
+
+        # @callback = block
+
         @dir = Pathname.new(options[:directory])
 
         # Setup instance vars

data/lib/sitediff/crawler.rb

@@ -17,8 +17,8 @@ class SiteDiff
     # Create a crawler with a base URL
     def initialize(hydra, base,
                    interval,
-                   whitelist,
-                   blacklist,
+                   include_regex,
+                   exclude_regex,
                    depth = DEFAULT_DEPTH,
                    curl_opts = UriWrapper::DEFAULT_CURL_OPTS,
                    debug = true,
@@ -27,8 +27,8 @@ class SiteDiff
       @base_uri = Addressable::URI.parse(base)
       @base = base
       @interval = interval
-      @whitelist = whitelist
-      @blacklist = blacklist
+      @include_regex = include_regex
+      @exclude_regex = exclude_regex
       @found = Set.new
       @callback = block
       @curl_opts = curl_opts
@@ -119,12 +119,12 @@ class SiteDiff
                      u.path.start_with?(@base_uri.path)
         next unless is_sub_uri
 
-        is_whitelisted = @whitelist.nil? ? false : @whitelist.match(u.path)
-        is_blacklisted = @blacklist.nil? ? false : @blacklist.match(u.path)
-        if is_blacklisted && !is_whitelisted
-          SiteDiff.log "Ignoring blacklisted URL #{u.path}", :info
+        is_included = @include_regex.nil? ? false : @include_regex.match(u.path)
+        is_excluded = @exclude_regex.nil? ? false : @exclude_regex.match(u.path)
+        if is_excluded && !is_included
+          SiteDiff.log "Ignoring excluded URL #{u.path}", :info
         end
-        is_whitelisted || !is_blacklisted
+        is_included || !is_excluded
       end
     end
   end

data/lib/sitediff/diff.rb

@@ -57,7 +57,11 @@ class SiteDiff
     ##
     # Generates an HTML report.
     # TODO: Generate the report in SiteDif::Report instead.
-    def generate_html(results, before, after, cache, relative = false)
+    def generate_html(results, before, after, cache, config)
+      relative = config.export
+      report = config.report || {}
+      before_url_report = report['before_url_report']
+      after_url_report = report['after_url_report']
       erb_path = File.join(SiteDiff::FILES_DIR, 'report.html.erb')
       ERB.new(File.read(erb_path)).result(binding)
     end

data/lib/sitediff/files/report.html.erb

@@ -23,6 +23,12 @@
   </head>
   <body class="page-overview" data-page="overview">
     <div id="layout">
+      <div class="container">
+        <div class="heading">
+          <h1><%= report["title"] %></h1>
+          <p><%= report["details"] %></p>
+        </div>
+      </div>
       <div class="container">
         <div class="statistical-info">
           <div class="changed-pages">
@@ -48,11 +54,26 @@
             <div class="site site-<%= tag %>">
               <% notes = ['base url']
                 notes << 'cached' if cache.read_tags.include?(tag.to_sym) %>
-              <h3 class="site__tag"><%= tag.capitalize %></h3>
-              <a href="<%= eval(tag) %>" class="site__url"><%= eval(tag) %></a>
-              <% if cache.read_tags.include?(tag.to_sym) %>
-                (<%= 'cached' if cache.read_tags.include?(tag.to_sym) %>)
-              <% end %>
+              <div>
+                <h3 class="site__tag"><%= tag.capitalize %></h3>
+              </div>
+              <div>
+                <% if display_url = report[tag + '_url_report'] %>
+                  <a href="<%= display_url %>" class="site__url"><%= display_url %></a>
+                <% else %>
+                  <a href="<%= eval(tag) %>" class="site__url"><%= eval(tag) %></a>
+                  <% if cache.read_tags.include?(tag.to_sym) %>
+                    (<%= 'cached' if cache.read_tags.include?(tag.to_sym) %>)
+                  <% end %>
+                <% end %>
+              </div>
+              <div>
+                Crawled on: <%= report[tag + '_time'] %>
+              </div>
+              <div>
+                <%= report[tag + '_note'] %>
+              </div>
+
             </div>
           <% end %>
         </div>
@@ -112,9 +133,15 @@
                 <span class="path"><%= result.path %></span>
                 <div class="buttons">
                   <% unless relative %>
-                    <a href="<%= result.url(:before, before, cache) %>" class="button-before" target="_blank">Before</a>
-                    <a href="<%= result.url(:after, after, cache) %>" class="button-after" target="_blank">After</a>
-                    <a href="/sidebyside<%= result.path %>" class="button-both">Both</a>
+                    <% unless report['before_url_report'] === false %>
+                      <a href="<%= result.url(:before, before_url_report || before, cache) %>" class="button-before" target="_blank">Before</a>
+                    <% end %>
+                    <% unless report['after_url_report'] === false %>
+                      <a href="<%= result.url(:after, after_url_report || after, cache) %>" class="button-after" target="_blank">After</a>
+                    <% end %>
+                    <% unless report['before_url_report'] === false || report['after_url_report'] === false %>
+                      <a href="/sidebyside<%= result.path %>" class="button-both">Both</a>
+                    <% end %>
                   <% end %>
                   <% unless result.diff_url.nil? %>
                     <a href="<%= result.diff_url(relative) %>" class="button button-diff">View diff</a>

data/lib/sitediff/files/sitediff.css

@@ -1,6 +1,5 @@
 * {
   box-sizing: border-box;
-  outline: none;
 }
 
 a {
@@ -247,3 +246,81 @@ table .status-col {
   top: 0;
   width: 100%;
 }
+
+/*** Overlay */
+.overlay {
+  background-color: rgba(0, 0, 0, 0.25);
+  height: 100vh;
+  left: 0;
+  padding: 2em;
+  position: fixed;
+  top: 0;
+  width: 100vw;
+}
+
+.overlay__inner {
+  background-color: #fff;
+  height: 100%;
+  margin: 0;
+  position: relative;
+  width: 100%;
+}
+
+.overlay header {
+  background-color: #f9f9fb;
+  border-bottom: 2px solid #eaecf3;
+  display: flex;
+  min-height: 3em;
+  left: 0;
+  position: absolute;
+  top: 0;
+  width: 100%;
+}
+
+.overlay header .prev,
+.overlay header .next,
+.overlay header .exit {
+  padding: 0.5em 0;
+  margin: 0.5em;
+}
+
+.overlay header .prev a,
+.overlay header .next a,
+.overlay header .exit a {
+  background-color: #eaecf3;
+  font-size: .9em;
+  padding: 0.5em;
+  text-decoration: none;
+}
+
+.overlay header .prev a:hover,
+.overlay header .next a:hover,
+.overlay header .exit a:hover {
+  background-color: #d2d7ef;
+}
+
+.overlay header .prev.disabled a,
+.overlay header .next.disabled a,
+.overlay header .exit.disabled a,
+.overlay header .prev.disabled a:hover,
+.overlay header .next.disabled a:hover,
+.overlay header .exit.disabled a:hover {
+  background-color: eaecf3;
+  color: #9b9db8;
+  cursor: not-allowed;
+}
+.overlay header .path {
+  flex-grow: 1;
+  padding: 1em;
+}
+
+.overlay article {
+  background-color: #fff;
+  height: 100%;
+  width: 100%;
+}
+
+.overlay article iframe {
+  height: 100%;
+  width: 100%;
+}

data/lib/sitediff/files/sitediff.js

@@ -3,11 +3,21 @@
  * SiteDiff report behaviors.
  */
 
+'use strict';
+
+/* global $ */
 /**
  * SiteDiff namespace.
  */
 var SiteDiff = SiteDiff || {};
 
+/**
+ * SiteDiff global map of diffs.
+ */
+SiteDiff.diffs = SiteDiff.diffs || {};
+
+SiteDiff.currentDiff = -1;
+
 /**
  * Scrolls the document to the said position.
  *
@@ -75,23 +85,42 @@ SiteDiff.scrollToElement = function (el, options) {
 SiteDiff.init = function () {
     // On the overview page.
     switch ($(document.body).data('page')) {
-        case 'overview':
-            SiteDiff.initFilterForm();
-            break;
+    case 'overview':
+        SiteDiff.initFilterForm();
+        break;
 
-        case 'diff':
-            SiteDiff.jumpToFirstDiff();
-            // TODO: Create a prev-next mechanism.
-            break;
+    case 'diff':
+        SiteDiff.jumpToFirstDiff();
+        break;
     }
 };
 
 /**
- * Initializes report filters.
+ * Initializes report filters and overlay.
  */
 SiteDiff.initFilterForm = function () {
+    SiteDiff.initDiffArray();
     SiteDiff.initStatusFilter();
     SiteDiff.initSearchFilter();
+    SiteDiff.initOverlay();
+    SiteDiff.initClickHandlers();
+};
+
+/**
+ * Initialize global diff array
+ *
+ */
+SiteDiff.initDiffArray = function() {
+    SiteDiff.diffs = $('.button-diff').map(function (i, element) {
+        var $el = $(element);
+        $el.data('diffindex', i);
+        return {
+            diff: $el.attr('href'),
+            element: $el,
+            index: i,
+            path: $el.parents('.description').find('.path').text()
+        };
+    });
 };
 
 /**
@@ -102,11 +131,10 @@ SiteDiff.initStatusFilter = function () {
         .on('change', function () {
             // Get a list of applied filters.
             var appliedFilters = $('.form-item--status input:checked')
-               .map(function () {
-                   return this.getAttribute('value');
-                   // applied.push(this.getAttribute('value'));
-                   // console.log(applied);
-               });
+                .map(function () {
+                    return this.getAttribute('value');
+                    // applied.push(this.getAttribute('value'));
+                });
             // Show only matching results, hide the rest.
             $('#sitediff-report')
                 .find('.sitediff-result')
@@ -155,6 +183,169 @@ SiteDiff.initSearchFilter = function () {
         });
 };
 
+/**
+ * Set up the diff overlay to be displayed.
+ */
+SiteDiff.initOverlay = function () {
+    if (SiteDiff.diffs.length <= 0) return;
+
+    // add overlay
+    $('body').append($(
+        '<div class="overlay" style="display: none;"><div class="overlay__inner"><header>' +
+        '<div class="path"></div>' +
+        '<div class="prev"><a href="#" title="Previous diff (left arrow)">< Prev</a></div>' +
+        '<div class="next"><a href="#" title="Next diff (right arrow)">Next ></a></div>' +
+        '<div class="exit"><a href="#" title="Close diff display (Esc)">Close</a></div>' +
+        '</header><article></article></div></div>'));
+    // add header click handlers
+    $('.overlay header .exit').click(function (event) {
+        event.preventDefault();
+        SiteDiff.destroyOverlay();
+    });
+    $('.overlay header .prev').click(function (event) {
+        event.preventDefault();
+        SiteDiff.prevDiff();
+    });
+    $('.overlay header .next').click(function (event) {
+        event.preventDefault();
+        SiteDiff.nextDiff();
+    });
+
+};
+
+/**
+ * Set up click handlers for all diff buttons
+ */
+SiteDiff.initClickHandlers = function () {
+    SiteDiff.diffs.each( function (i, diff) {
+        diff.element.click({index: i}, function (event) {
+            event.preventDefault();
+            SiteDiff.openOverlay(event.data.index);
+        });
+    });
+};
+
+/**
+ * Set up key handlers for overlay.
+ */
+SiteDiff.initKeyHandlers = function () {
+    $(document).keyup(function (event) {
+        switch (event.which) {
+        case 37:
+            SiteDiff.prevDiff();
+            break;
+        case 39:
+            SiteDiff.nextDiff();
+            break;
+        case 27:
+            SiteDiff.destroyOverlay();
+            break;
+        }
+    });
+};
+
+/**
+ * Remove overlay key handlers.
+ */
+SiteDiff.removeKeyHandlers = function () {
+    $(document).off('keyup');
+};
+
+/**
+ * Open overlay for the diff identified by the `index`.
+ *
+ * @param integer index
+ *      The index of the diff to be viewed.
+ */
+SiteDiff.openOverlay = function (index) {
+    SiteDiff.currentDiff = index;
+    SiteDiff.showDiff();
+    SiteDiff.initKeyHandlers();
+    $('.overlay').fadeIn(300);
+};
+
+/**
+ * Create the iframe to display the current diff.
+ */
+SiteDiff.showDiff = function () {
+    var diff = SiteDiff.diffs[SiteDiff.currentDiff];
+    var iframe = '<iframe src="' + diff.diff + '"></iframe>';
+    $('.overlay header .path').text(diff.path);
+    SiteDiff.setPrevNext();
+    $('.overlay article').html(iframe);
+};
+
+/**
+ * Hide the overlay and clean up.
+ */
+SiteDiff.destroyOverlay = function () {
+    $('.overlay article').empty();
+    SiteDiff.removeKeyHandlers();
+    $('.overlay').fadeOut(300, SiteDiff.scrollToButton);
+};
+
+/**
+ * Display the previous diff.
+ */
+SiteDiff.prevDiff = function () {
+    if (SiteDiff.currentDiff > 0) {
+        SiteDiff.currentDiff--;
+        SiteDiff.showDiff();
+    }
+};
+
+/**
+ * Display the next diff.
+ */
+SiteDiff.nextDiff = function () {
+    if (SiteDiff.currentDiff < SiteDiff.diffs.length - 1) {
+        SiteDiff.currentDiff++;
+        SiteDiff.showDiff();
+    }
+};
+
+/**
+ * Enable or disable prev and next buttons based on current diff.
+ */
+SiteDiff.setPrevNext = function () {
+    if (SiteDiff.currentDiff <= 0) {
+        // set prev disabled
+        $('.overlay header .prev').addClass('disabled');
+    }
+    else {
+        $('.overlay header .prev.disabled').removeClass('disabled');
+    }
+    if (SiteDiff.currentDiff >= SiteDiff.diffs.length - 1) {
+        // set next disabled
+        $('.overlay header .next').addClass('disabled');
+    }
+    else {
+        $('.overlay header .next.disabled').removeClass('disabled');
+    }
+};
+
+/**
+ * Scroll to the button associated with the current diff.
+ */
+SiteDiff.scrollToButton = function () {
+    var $diffButton = SiteDiff.diffs[SiteDiff.currentDiff].element;
+    if (! SiteDiff.isElementVisible($diffButton)) {
+        SiteDiff.scrollToElement($diffButton);
+    }
+};
+
+/**
+ * Check if an element is at least partly visible.
+ * @param element
+ */
+SiteDiff.isElementVisible = function (element) {
+    var topVisible = $(window).scrollTop();
+    var bottomVisible = topVisible + $(window).height();
+    var elemTop = $(element).offset().top;
+    var elemBottom = elemTop + $(element).height();
+    return ((elemBottom <= bottomVisible) && (elemTop >= topVisible));
+};
+
 /**
  * Jumps to the first diff on the page.
  */