Merge branch 'ezpaarse-project:master' into micromedex

ben/manifest.json

@@ -0,0 +1,14 @@
+{
+  "longname": "Benjamins",
+  "name": "ben",
+  "describe": "Recognizes the accesses to the platform Benjamins",
+  "contact": "Sean Duffy",
+  "pkb": false,
+  "docurl": "http://analyses.ezpaarse.org/platforms/64ad733df42178dc451bcf6a",
+  "domains": [
+    "www.benjamins.com",
+    "benjamins.com"
+  ],
+  "version": "2023-07-23",
+  "status": "beta"
+}

ben/parser.js

@@ -0,0 +1,47 @@
+#!/usr/bin/env node
+
+'use strict';
+const Parser = require('../.lib/parser.js');
+
+/**
+ * Recognizes the accesses to the platform Benjamins
+ * @param  {Object} parsedUrl an object representing the URL to analyze
+ *                            main attributes: pathname, query, hostname
+ * @param  {Object} ec        an object representing the EC whose URL is being analyzed
+ * @return {Object} the result
+ */
+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 || {};
+
+  // use console.error for debuging
+  // console.error(parsedUrl);
+
+  let match;
+
+  if ((match = /^\/online\/([a-z]+)\/publications\/([0-9]+)$/i.exec(path)) !== null) {
+    // https://benjamins.com/online/etsb/publications/52963
+    result.rtype = 'RECORD';
+    result.mime = 'HTML';
+    result.db_id = match[1];
+    result.unitid = match[2];
+
+  } else if ((match = /^\/online\/([a-z]+)\/articles\/([a-z0-9]+)?(\.[a-z]+)?$/i.exec(path)) !== null) {
+    // https://www.benjamins.com/online/hts/articles/ada1
+    // https://benjamins.com/online/hts/articles/hyb1.fr
+    result.rtype = 'ARTICLE';
+    result.mime = 'HTML';
+    result.db_id = match[1];
+    result.title_id = match[1];
+    result.unitid = match[2];
+  } else if ((match = /^\/online\/([a-z]+)\/search$/i.exec(path)) !== null) {
+    // https://www.benjamins.com/online/hts/search?r=all&q=Adaption
+    result.rtype = 'SEARCH';
+    result.mime = 'HTML';
+    result.db_id = match[1];
+  }
+
+  return result;
+});

ben/test/ben.2023-07-23.csv

@@ -0,0 +1,5 @@
+out-db_id;out-title_id;out-unitid;out-rtype;out-mime;in-url
+hts;hts;ada1;ARTICLE;HTML;https://www.benjamins.com/online/hts/articles/ada1
+hts;hts;hyb1;ARTICLE;HTML;https://benjamins.com/online/hts/articles/hyb1.fr
+hts;;;SEARCH;HTML;https://www.benjamins.com/online/hts/search?r=all&q=Adaption
+etsb;;52963;RECORD;HTML;https://benjamins.com/online/etsb/publications/52963

tair/parser.js

@@ -6,6 +6,7 @@ const Parser = require('../.lib/parser.js');
 module.exports = new Parser(function analyseEC(parsedUrl) {
   let result = {};
   let path   = parsedUrl.pathname;
+  let params = parsedUrl.query || {};
   let match;
 
   if ((match = /^\/(index).jsp$/.exec(path)) !== null) {
@@ -20,12 +21,29 @@ module.exports = new Parser(function analyseEC(parsedUrl) {
     result.mime   = 'HTML';
     result.rtype  = match[1].toUpperCase();
 
-  } else if ((match = /^\/servlets\/(Search)$/.exec(path)) !== null) {
-    // hhttp://www.arabidopsis.org.gate1.inist.fr/servlets/Search?type=general&action=new_search
-    result.rtype  = match[1].toUpperCase();
+  } else if (/^\/servlets\/Search$/.test(path)) {
+    // http://www.arabidopsis.org.gate1.inist.fr/servlets/Search?type=general&action=new_search
+    result.rtype  = 'SEARCH';
+    result.mime   = 'HTML';
+
+  } else if (/^\/servlets\/TairObject$/.test(path)) {
+    // /servlets/TairObject?type=gene_class_symbol&id=1005832620
+    // /servlets/TairObject?type=transposon&id=1866
+    // /servlets/TairObject?type=locus&name=At1g44170
+    const type = params.type || '';
+    const id = params.id || params.name || '';
+
+    result.rtype  = 'RECORD';
     result.mime   = 'HTML';
-    result.unitid = 'servlets/' + match[1];
+    result.unitid = `${type}/${id}`;
+
+  } else if ((match = /^\/download_files(\/.*)$/.exec(path)) !== null) {
+    // /download_files/Genes/Araport11_genome_release/Araport11_blastsets/Araport11_cds_20220914.gz
+    result.rtype  = 'RECORD';
+    result.mime   = 'MISC';
+    result.unitid = match[1];
   }
+
   return result;
 });
 

tair/test/tail.2014-05-19.csv

@@ -1,7 +1,11 @@
 out-unitid;out-mime;out-rtype;in-url
-index;HTML;INDEX;http://www.arabidopsis.org.gate1.inist.fr/index.jsp
-servlets/Search;HTML;SEARCH;http://www.arabidopsis.org.gate1.inist.fr/servlets/Search?type=general&action=new_search
-tools/index;HTML;TOOLS;http://www.arabidopsis.org.gate1.inist.fr/tools/index.jsp
-portals/index;HTML;PORTALS;http://www.arabidopsis.org.gate1.inist.fr/portals/index.jsp
-download/index;HTML;DOWNLOAD;http://www.arabidopsis.org.gate1.inist.fr/download/index.jsp
-submit/index;HTML;SUBMIT;http://www.arabidopsis.org.gate1.inist.fr/submit/index.jsp
+index;HTML;INDEX;http://www.arabidopsis.org/index.jsp
+;HTML;SEARCH;http://www.arabidopsis.org/servlets/Search?type=general&action=new_search
+tools/index;HTML;TOOLS;http://www.arabidopsis.org/tools/index.jsp
+portals/index;HTML;PORTALS;http://www.arabidopsis.org/portals/index.jsp
+download/index;HTML;DOWNLOAD;http://www.arabidopsis.org/download/index.jsp
+submit/index;HTML;SUBMIT;http://www.arabidopsis.org/submit/index.jsp
+gene_class_symbol/1005832620;HTML;RECORD;https://www.arabidopsis.org/servlets/TairObject?type=gene_class_symbol&id=1005832620
+transposon/1866;HTML;RECORD;https://www.arabidopsis.org/servlets/TairObject?type=transposon&id=1866
+locus/At1g44170;HTML;RECORD;https://www.arabidopsis.org/servlets/TairObject?type=locus&name=At1g44170
+/Genes/Araport11_genome_release/Araport11_blastsets/Araport11_cds_20220914.gz;MISC;RECORD;https://www.arabidopsis.org/download_files/Genes/Araport11_genome_release/Araport11_blastsets/Araport11_cds_20220914.gz