Conent optimizations

README.md

@@ -29,11 +29,18 @@ search:
 
 - **path** - file path. By default is `search.xml` . If the file extension is `.json`, the output format will be JSON. Otherwise XML format file will be exported.
 - **field** - the search scope you want to search, you can chose:
-  * **post** (Default) - will only covers all the posts of your blog.
-  * **page** - will only covers all the pages of your blog.
-  * **all** - will covers all the posts and pages of your blog.
-- **content** - whether contains the whole content of each article. If `false`, the generated results only cover title and other meta info without mainbody. By default is `true`.
+  * `post` (Default) - will only covers all the posts of your blog.
+  * `page` - will only covers all the pages of your blog.
+  * `all` - will covers all the posts and pages of your blog.
+- **content** - whether contains the whole content of each article.
+  * `true` (Default) - generated results use the mainbody.
+  * `rendered` - generated results use the rendered mainbody if available. (json only)
+  * `excerpt` - generated results use the excerpts as content if available.
+  * `raw` - generated results use the raw mainbody if available. (json only, also contains the front-matter)
+  * `false` - generated results only cover title and other meta info without mainbody.
 - **template** (Optional) - path to a custom XML template
+- **strip_html** (Optional) - when `true` all HTML tags will be removed from the content.
+- **permalinks** (Optional) - when `true` the tags and categories in json output will also contain the permalinks.
 
 ## Exclude indexing
 

lib/json_generator.js

@@ -6,6 +6,11 @@ module.exports = function(locals){
     var searchConfig = config.search;
     var searchfield = searchConfig.field;
     var content = searchConfig.content;
+    if (content == undefined) content = true;
+    var stripHtml = searchConfig.strip_html;
+    if (stripHtml == undefined) stripHtml = false;
+    var permalinks = searchConfig.permalinks;
+    if (permalinks == undefined) permalinks = false;
 
     var posts, pages;
 
@@ -23,67 +28,81 @@ module.exports = function(locals){
         posts = locals.posts.sort('-date');
     }
 
-    var res = new Array() 
+    var res = new Array()
     var index = 0
-    
-    if(posts){     
+
+    if(posts){
         posts.each(function(post) {
             if (post.indexing != undefined && !post.indexing) return;
-            var temp_post = new Object() 
+            var temp_post = new Object()
             temp_post.title = post.title || 'No Title'
-            if (post.path) { 
-                temp_post.url = config.root + post.path 
-            } 
-            if (content != false && post._content) { 
-                temp_post.content = post._content 
-            } 
-            if (post.tags && post.tags.length > 0) { 
-                var tags = [];
+            if (post.path) {
+                temp_post.url = config.root + post.path
+            }
+            if (content != false) {
+                if (content == 'rendered' && post.content) {
+                    temp_post.content = post.content
+                } else if (content == 'excerpt' && post.excerpt) {
+                    temp_post.content = post.excerpt
+                } else if (content == 'raw' && post.raw) {
+                    temp_post.content = post.raw
+                } else if (post._content) {
+                    temp_post.content = post._content
+                }
+                if (stripHtml && temp_post.content) {
+                    temp_post.content = temp_post.content.replace(/<[^>]+>/g, '')
+                }
+            }
+            temp_post.tags = [];
+            if (post.tags && post.tags.length > 0) {
                 post.tags.forEach(function (tag) {
-                    tags.push(tag.name);
-                }); 
-                temp_post.tags = tags 
-            } 
-            if (post.categories && post.categories.length > 0) { 
-                var categories = [];
+                    if (permalinks) {
+                        temp_post.tags.push([ tag.name, tag.permalink ]);
+                    } else {
+                        temp_post.tags.push(tag.name);
+                    }
+                });
+            }
+            temp_post.categories = [];
+            if (post.categories && post.categories.length > 0) {
                 post.categories.forEach(function (cate) {
-                    categories.push(cate.name);
-                }); 
-                temp_post.categories = categories 
-            } 
-            res[index] = temp_post;  
-            index += 1; 
-	    }); 
-    } 
-    if(pages){ 
+                    if (permalinks) {
+                        temp_post.categories.push([ cate.name, cate.permalink ]);
+                    } else {
+                        temp_post.categories.push(cate.name);
+                    }
+                });
+            }
+            res[index] = temp_post;
+            index += 1;
+	    });
+    }
+    if(pages){
         pages.each(function(page){
             if (page.indexing != undefined && !page.indexing) return;
-            var temp_page = new Object() 
-            temp_post.title = post.title || 'No Title'
-            if (page.path) { 
-                temp_page.url = config.root + page.path 
-            } 
-            if (content != false && page._content) { 
-                temp_page.content = page._content 
-            } 
-            if (page.tags && page.tags.length > 0) { 
-                var tags = new Array() 
-                var tag_index = 0 
-                page.tags.each(function (tag) {
-                    tags[tag_index] = tag.name; 
-                }); 
-                temp_page.tags = tags 
-            } 
-            if (page.categories && page.categories.length > 0) {
-                temp_page.categories = []
-                (page.categories.each || page.categories.forEach)(function (item) {
-                    temp_page.categories.push(item);
-                });
-            } 
-            res[index] = temp_page;  
-            index += 1; 
-        }); 
-    } 
+            var temp_page = new Object()
+            temp_page.title = page.title || 'No Title'
+            if (page.path) {
+                temp_page.url = config.root + page.path
+            }
+            if (content != false) {
+                if (content == 'rendered' && page.content) {
+                    temp_page.content = page.content
+                } else if (content == 'excerpt' && page.excerpt) {
+                    temp_page.content = page.excerpt
+                } else if (content == 'raw' && page.raw) {
+                    temp_page.content = page.raw
+                } else if (page._content) {
+                    temp_page.content = page._content
+                }
+                if (stripHtml && temp_page.content) {
+                    temp_page.content = temp_page.content.replace(/<[^>]+>/g, '')
+                }
+            }
+            res[index] = temp_page;
+            index += 1;
+        });
+    }
 
 
     var json = JSON.stringify(res);

lib/xml_generator.js

@@ -22,7 +22,9 @@ module.exports = function(locals){
   var template = searchTmpl;
   var searchfield = searchConfig.field;
   var content = searchConfig.content;
-  if (content == undefined) content=true;
+  if (content == undefined) content = true;
+  var stripHtml = searchConfig.strip_html;
+  if (stripHtml == undefined) stripHtml = false;
 
   var posts, pages;
 
@@ -40,6 +42,32 @@ module.exports = function(locals){
     posts = locals.posts.sort('-date');
   }
 
+  if (content == 'excerpt') {
+    posts.data.forEach(function (p) {
+      if (p.excerpt) {
+        p.content = p.excerpt;
+      }
+    });
+    pages.data.forEach(function (p) {
+      if (p.excerpt) {
+        p.content = p.excerpt;
+      }
+    });
+  }
+
+  if (stripHtml) {
+    posts.data.forEach(function (p) {
+      if (p.content) {
+        p.content = p.content.replace(/<[^>]+>/g, '');
+      }
+    });
+    pages.data.forEach(function (p) {
+      if (p.content) {
+        p.content = p.content.replace(/<[^>]+>/g, '');
+      }
+    });
+  }
+
   var rootURL;
   if (config.root == null){
     rootURL = "/";