v 2.7.1 Fix an issue that '#' is not included in the 'onlyUri' property

Author: patternhelloworld

dist/extract-normalize-urls.bundle.js

@@ -1,6 +1,6 @@
 {
  "name": "extract-normalize-urls",
- "version": "2.6.0",
+ "version": "2.7.0",
  "description": "Extract and normalize urls, fuzzy urls, urls without protocols, uris in natural language texts.",
  "main": "src/entry.js",
  "scripts": {

package.json

@@ -354,8 +354,8 @@ <h3>TextArea (recommended)</h3>
  }
 
  start();
- console.log('final_3_2 : ' + JSON.stringify(Pattern.UrlArea.parseUrl("http://abc.def:9094/park/noon"), null, 2));
- console.log('final_3_3 : ' + JSON.stringify(Pattern.UrlArea.normalizeUrl("://abc.jppp:9091 /park/noon"), null, 2));
+ console.log('final_3_2 : ' + JSON.stringify(Pattern.UrlArea.parseUrl("http://abc.def:9094#park/noon"), null, 2));
+ console.log('final_3_3 : ' + JSON.stringify(Pattern.UrlArea.normalizeUrl("://abc.jppp:9091#park/noon"), null, 2));
  //console.log('final_5 : ' + JSON.stringify(Pattern.TextArea.extractCertainUris(textStr1, [['{number}', 'kak'], ['nice','guy'],['abc', '{number}']], true), null, 2));
  // console.log('final_3 : ' + JSON.stringify(Pattern.XmlArea.extractAllUrls(xmlStr), null, 2));
 /*

public/index.html

@@ -550,7 +550,7 @@ const Url = {
 
 
  // URI
- let rx4 = new RegExp('\\/(?:.|[\\n\\r\\t\\s])*$', 'gi');
+ let rx4 = new RegExp('[#/](?:.|[\\n\\r\\t\\s])*$', 'gi');
  let match4 = {};
  while ((match4 = rx4.exec(modified_url)) !== null) {
  obj['onlyUri'] = match4[0];
@@ -833,14 +833,14 @@ const Url = {
  }
 
  // 5. Separate uri
- let rx2 = new RegExp('\\/(?:.|[\\n\\r\\t\\s])*$', 'gi');
+ let rx2 = new RegExp('[#/](?:.|[\\n\\r\\t\\s])*$', 'gi');
  let match2 = {};
  while ((match2 = rx2.exec(url)) !== null) {
  obj['onlyUri'] = match2[0];
  }
  url = url.replace(rx2, '');
 
- if (obj['onlyUri'] === "/") {
+ if (/^[#/]+$/.test(obj['onlyUri'])) {
  obj['onlyUri'] = null;
  }
 
@@ -930,7 +930,7 @@ const Url = {
 
  } else if (obj['onlyUri'] !== null && obj['onlyParams'] === null) {
 
- let rm_part_matches = new RegExp('\\/([^/\\n\\r\\t\\s]+?)(' + Pattern.Ancestors.no_lang_char_num + '+)$', 'gi').exec(obj['url']);
+ let rm_part_matches = new RegExp('[#/]([^/\\n\\r\\t\\s]+?)(' + Pattern.Ancestors.no_lang_char_num + '+)$', 'gi').exec(obj['url']);
 
  //console.log(obj['url'] + ' : ' + rm_part_matches[1]);