Merge pull request #768 from OCLC-Developer-Network/factiva

updates to factiva parser
ezpaarse-project · Nov 3, 2023 · adabee1 · adabee1
2 parents 9004e24 + 650f828
commit adabee1
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 3 deletions.
diff --git a/factiva/parser.js b/factiva/parser.js
@@ -14,7 +14,7 @@ module.exports = new Parser(function analyseEC(parsedUrl, ec) {
   let result = {};
   let path   = parsedUrl.pathname;
   // uncomment this line if you need parameters
-  // let param = parsedUrl.query || {};
+  let param = parsedUrl.query || {};
 
   // use console.error for debuging
   // console.error(parsedUrl);
@@ -23,17 +23,36 @@ module.exports = new Parser(function analyseEC(parsedUrl, ec) {
 
   if (/login.asp$/i.test(path)) {
     // https://global.factiva.com:443/en/sess/login.asp?cookie=on&XSID=S00ZWrr1cbyMTZyMTMvODMtM9MuOXmm5DFHY96oYqZlNFFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFB
-    result.rtype    = 'CONNECTION';
+    result.rtype    = 'OTHER';
     result.mime     = 'HTML';
   } else if ((match = /^\/(.*).mp3$/i.exec(path)) !== null) {
     // https://global.factiva.com:443/APNRUW0020181003eea5002ut.mp3?tkn=9lqW0VRwuX1Vy8wElyjuGlQnuO5rd42YIqe8DUknG7ucAXG553GSEh6I2PX_2Fm4YWvkfywGSwJMp5vOyMwklFnkd_2B7dnkjKVzD6fbGhdcubB3qv0WemPaLxIy9M0F9Bz76TIyaKW_2BdoiT73iL3b3V_2F0Xz8lu8o4b9tCXy_2F4Z65tKMBHVCqZl3HT0hg9gkBkmGgStYZk_2F2bUrsbfaz5H3pvO_2FvGVqF8fT3enCbvmynHy0_3D_7C2
     result.rtype    = 'AUDIO';
     result.mime     = 'MISC';
     result.unitid   = match[1];
+  } else if ((match = /^\/SpeachContentHandler.ashx$/i.exec(path)) !== null) {
+    // https://global.factiva.com/SpeachContentHandler.ashx?accessionID=J000000020230905ej950001c&drn=drn:archive.newsarticle.J000000020230905ej950001c&attach=true
+    result.rtype    = 'AUDIO';
+    result.mime     = 'MP3';
+    result.unitid   = param.accessionID;
+  } else if ((match = /^\/redir\/default.aspx$/i.exec(path)) !== null) {
+    // https://global.factiva.com/redir/default.aspx?p=sa&NS=16&AID=9INS008900&an=LBA0000020231018ejai01f5p&drn=drn:archive.newsarticle.LBA0000020231018ejai01f5p&cat=a&ep=asi
+    // https://global.factiva.com/redir/default.aspx?p=sa&an=LBA0000020231018ejai01f5p&drn=drn:archive.newsarticle.LBA0000020231018ejai01f5p&cat=a&ep=ASE
+    result.rtype    = 'ARTICLE';
+    result.mime     = 'HTML';
+    result.unitid   = param.an;
   } else if (/^\/controls\/search\/SearchBuilder/i.test(path)) {
     // https://global.factiva.com:443/controls/search/SearchBuilder0190400ui4sr.ashx
     result.rtype    = 'SEARCH';
     result.mime     = 'HTML';
+  } else if (/^\/Search\/SSResults/i.test(path)) {
+    // https://snapshot.factiva.com/Search/SSResults
+    result.rtype    = 'SEARCH';
+    result.mime     = 'HTML';
+  } else if (/^\/ha\/default\.aspx/i.test(path)) {
+    // https://global.factiva.com/ha/default.aspx?page_driver=searchBuilder_Search#./!?&_suid=169402305530208846596548944554
+    result.rtype    = 'SEARCH';
+    result.mime     = 'HTML';
   }
 
   return result;

diff --git a/factiva/test/factiva.2018-10-19.csv b/factiva/test/factiva.2018-10-19.csv
@@ -1,4 +1,4 @@
 out-unitid;out-rtype;out-mime;in-url
-;CONNECTION;HTML;https://global.factiva.com:443/en/sess/login.asp?cookie=on&XSID=S00ZWrr1cbyMTZyMTMvODMtM9MuOXmm5DFHY96oYqZlNFFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFB
+;OTHER;HTML;https://global.factiva.com:443/en/sess/login.asp?cookie=on&XSID=S00ZWrr1cbyMTZyMTMvODMtM9MuOXmm5DFHY96oYqZlNFFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFB
 APNRUW0020181003eea5002ut;AUDIO;MISC;https://global.factiva.com:443/APNRUW0020181003eea5002ut.mp3?tkn=9lqW0VRwuX1Vy8wElyjuGlQnuO5rd42YIqe8DUknG7ucAXG553GSEh6I2PX_2Fm4YWvkfywGSwJMp5vOyMwklFnkd_2B7dnkjKVzD6fbGhdcubB3qv0WemPaLxIy9M0F9Bz76TIyaKW_2BdoiT73iL3b3V_2F0Xz8lu8o4b9tCXy_2F4Z65tKMBHVCqZl3HT0hg9gkBkmGgStYZk_2F2bUrsbfaz5H3pvO_2FvGVqF8fT3enCbvmynHy0_3D_7C2
 ;SEARCH;HTML;https://global.factiva.com:443/controls/search/SearchBuilder0190400ui4sr.ashx
diff --git a/factiva/test/factiva.2023-11-01.csv b/factiva/test/factiva.2023-11-01.csv
@@ -0,0 +1,9 @@
+out-unitid;out-rtype;out-mime;in-url
+LBA0000020231018ejai01f5p;ARTICLE;HTML;https://global.factiva.com/redir/default.aspx?p=sa&NS=16&AID=9INS008900&an=LBA0000020231018ejai01f5p&drn=drn:archive.newsarticle.LBA0000020231018ejai01f5p&cat=a&ep=asi
+LBA0000020231018ejai01f5p;ARTICLE;HTML;https://global.factiva.com/redir/default.aspx?p=sa&an=LBA0000020231018ejai01f5p&drn=drn:archive.newsarticle.LBA0000020231018ejai01f5p&cat=a&ep=ASE
+J000000020230905ej950001c;AUDIO;MP3;https://global.factiva.com/SpeachContentHandler.ashx?accessionID=J000000020230905ej950001c&drn=drn:archive.newsarticle.J000000020230905ej950001c&attach=true
+;SEARCH;HTML;https://snapshot.factiva.com/Search/SSResults
+;SEARCH;HTML;https://global.factiva.com/ha/default.aspx?page_driver=searchBuilder_Search#./!?&_suid=169402305530208846596548944554
+;SEARCH;HTML;https://global.factiva.com:443/controls/search/SearchBuilder0190400ui4sr.ashx
+APNRUW0020181003eea5002ut;AUDIO;MISC;https://global.factiva.com:443/APNRUW0020181003eea5002ut.mp3?tkn=9lqW0VRwuX1Vy8wElyjuGlQnuO5rd42YIqe8DUknG7ucAXG553GSEh6I2PX_2Fm4YWvkfywGSwJMp5vOyMwklFnkd_2B7dnkjKVzD6fbGhdcubB3qv0WemPaLxIy9M0F9Bz76TIyaKW_2BdoiT73iL3b3V_2F0Xz8lu8o4b9tCXy_2F4Z65tKMBHVCqZl3HT0hg9gkBkmGgStYZk_2F2bUrsbfaz5H3pvO_2FvGVqF8fT3enCbvmynHy0_3D_7C2
+;OTHER;HTML;https://global.factiva.com:443/en/sess/login.asp?cookie=on&XSID=S00ZWrr1cbyMTZyMTMvODMtM9MuOXmm5DFHY96oYqZlNFFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFBQUFB