sitemap-generator - npm Package Compare versions

Comparing version 1.1.0 to 1.2.0

package.json

 {
   "name": "sitemap-generator",
-  "version": "1.1.0",
+  "version": "1.2.0",
   "description": "Creates a XML-Sitemap by crawling a given site.",
@@ -22,6 +22,6 @@ "homepage": "https://github.com/lgraubner/node-sitemap-generator",
     "simplecrawler": "^0.5.2",
-    "lodash": "^3.8.0",
-    "xmlbuilder": "^2.6.2",
+    "lodash": "^3.10.0",
+    "xmlbuilder": "^2.6.4",
     "commander": "^2.8.1",
-    "chalk": "^1.0.0"
+    "chalk": "^1.1.0"
   },
@@ -28,0 +28,0 @@ "preferGlobal": true,

README.md

 # Node Sitemap Generator
 Creates a XML-Sitemap by crawling a given site.
 
-## Usage
+## Installation
 
 ```BASH
-npm install -g sitemap-generator
+$ npm install -g sitemap-generator
+```
 
-# sitemap-gen [options] <url>
-sitemap-gen --query example.com
+## Usage
+```BASH
+$ sitemap-gen [options] <url>
 ```
 
-*Important*: Relative Links without leading slash are producing errors.
-
 ### Options
+```BASH
+$ sitemap-gen --help
 
-|command|description
-|:---|:---
-|-q, --query|consider query string
+  Usage: sitemap-gen [options] <url>
+
+  Options:
+
+    -h, --help                output usage information
+    -V, --version             output the version number
+    -q, --query               consider query string
+    -i, --include [ext,ext2]  include fetched links by file extension, comma seperated
+    -e, --exclude [ext,ext2]  exclude fetched links by file extension, comma seperated
+    -o, --output [path]       specify output path
+```
+
+**Important**: Executing the sitemap-generator with sites using HTML `base`-tag along with links *without* leading slashes will probably not work.

sitemap-gen.js

 #!/usr/bin/env node
 
 var Crawler = require("simplecrawler");
-var forIn = require("lodash/object/forIn");
+var _ = require("lodash");
 var fs = require("fs");
@@ -14,2 +14,4 @@ var builder = require("xmlbuilder");
         .option("-q, --query", "consider query string")
+        .option("-i, --include [ext,ext2]", "include fetched links by file extension, comma seperated")
+        .option("-e, --exclude [ext,ext2]", "exclude fetched links by file extension, comma seperated")
         .option("-o, --output [path]", "specify output path")
@@ -34,7 +36,24 @@ .parse(process.argv);
 
-var path = "./";
+var path = ".";
 if (program.output) {
-    path = program.output;
+    path = program.output.replace(/\/+$/, "");
 }
 
+var exclude = ["gif", "jpg", "jpeg", "png", "ico", "bmp", "ogg", "webp", "mp4", "webm", "mp3", "ttf", "woff", "json", "rss", "atom", "gz", "zip", "rar", "7z", "css", "js", "gzip", "exe"];
+
+if (program.include) {
+    exclude = _.difference(exclude, program.include.split(","));
+}
+
+if (program.exclude) {
+    exclude = _.union(exclude, program.exclude.split(","));
+}
+
+var exts = exclude.join("|");
+var regex = new RegExp("\.(" + exts + ")", "i");
+
+c.addFetchCondition(function(parsedURL) {
+    return !parsedURL.path.match(regex);
+});
+
 c.on("fetchcomplete", function(item) {
@@ -58,3 +77,3 @@ chunk.push({
     var xml = builder.create("urlset", { version: "1.0", encoding: "UTF-8" }).att("xmlns", "http://www.sitemaps.org/schemas/sitemap/0.9");
-    forIn(chunk, function(value, key) {
+    _.forIn(chunk, function(value, key) {
         xml.ele("url")
@@ -66,3 +85,3 @@ .ele(value);
 
-    fs.writeFile(path + "sitemap.xml", map, function(err) {
+    fs.writeFile(path + "/sitemap.xml", map, function(err) {
         if (err) {
@@ -77,22 +96,2 @@ return console.log(chalk.red(err));
 
-var image = c.addFetchCondition(function(parsedURL) {
-    return !parsedURL.path.match(/\.(gif|jpg|jpeg|png|ico|bmp)/i);
-});
-
-var media = c.addFetchCondition(function(parsedURL) {
-    return !parsedURL.path.match(/\.(ogg|webp|mp4|webm|mp3)/i);
-});
-
-var font = c.addFetchCondition(function(parsedURL) {
-    return !parsedURL.path.match(/\.(ttf|woff)$/i);
-});
-
-var data = c.addFetchCondition(function(parsedURL) {
-    return !parsedURL.path.match(/\.(json|rss|atom|gz|zip|rar|7z)/i);
-});
-
-var misc = c.addFetchCondition(function(parsedURL) {
-    return !parsedURL.path.match(/\.(css|js|gzip|exe)/i);
-});
-
 c.start();

No alert changes

Improved metrics

Total package byte prevSize

5885

increased by11.8%

Number of lines in readme file

32

increased by60%

Worsened metrics

Lines of code

71

decreased by-1.39%

Dependency changes

• Updatedchalk@^1.1.0

• Updatedlodash@^3.10.0

• Updatedxmlbuilder@^2.6.4