Improvements to Author names parsing.

pensoft · May 19, 2014 · 461c842 · 461c842
1 parent d6652de
commit 461c842
Show file tree

Hide file tree

Showing 3 changed files with 54 additions and 46 deletions.
diff --git a/dbs/tools.js b/dbs/tools.js
@@ -9,8 +9,14 @@ module.exports = {
 		}
 		else {
 			shame.call();
+			if (err === null) {
 			p(('Warning: Request ' + res.req._headers.host + res.req.path), 1);
-			p('  failed with error: ' + ((err === null) ? res.statusCode + ' ' + require('http').STATUS_CODES[res.statusCode] : err), 1);
+				p(('         failed with error: ' + res.statusCode + ' ' + require('http').STATUS_CODES[res.statusCode]), 1);
+			}
+			else{
+				p((err), 1);
+			}
+
 		}
 	},
 
@@ -49,39 +55,28 @@ module.exports = {
 		//returns an array
 		text = text.trim();
 		var firstlast = [
-			/([A-Z]\.[A-Z]\.[A-Z]\.)\s([A-Za-z]+)/,				//A.M.H. Brunsting
-			/([A-Za-z]+\s[A-Z]\.\s[A-Z]\.)\s([A-Za-z]+)/,		//Artur R. M. Serrano
-			/([A-Z]\.\s[A-Z]\.)\s([A-Za-z]+)/,		//J. P. Zaballos
-			/([A-Z]\.\-[A-Z]\.)\s([A-Za-z]+)/,		//S.-A. Bengtson
-			/([A-Za-z]+ [A-Z]\.) ([A-Za-z]+)/,			//Carl H. Lindroth
-			/([A-Za-z]+\s[A-Za-z]+)\s([A-Za-z]+)/,	//Carmen Chavez Ortiz
-			/([A-Za-z]{2,})\s([A-Za-z]{2,})/,		//Lyubomir Penev
-
-			/([A-Z])\s([A-Za-z]+)/,			//M Baehr
-			/([A-Z])\. ([A-Za-z]+)/,		//L. Penev
-			/([A-Za-z])\.\,\s([A-Za-z]+)/,	//L., Penev
-
-			/([A-Z]+)\s([A-Z]+)/,			//BERNHARD KROMP
+			/^([A-Z]\.[A-Z]\.[A-Z]\.)\s+([\S]+)$/,				//A.M.H. Brunsting
+			/^(\S+\s+[A-Z]\.\s+[A-Z]\.)\s+([\S]+)$/,		//Artur R. M. Serrano
+			/^([A-Z]\.\s+[A-Z]\.)\s+([\S]+)$/,		//J. P. Zaballos
+			/^([A-Z]\.\-[A-Z]\.)\s+([\S]+)$/,		//S.-A. Bengtson
+			/^([\S]+\s+[A-Z]\.)\s+([\S]+)$/,			//Carl H. Lindroth
+			/^(\S+\s+\S+)\s+(\S+)$/,	//Carmen Chavez Ortiz
+			/^([\S[^.,]]{2,})\s+([\S]{2,})$/,		//Lyubomir Penev
+			/^([\S^.]\.)\,\s+([\S]+)$/,	//L., Penev
+			/^([\S^.]\.)\s+([\S]+)$/,		//L. Penev
+			/^(\S+)\s+([\S^\.]+)$/,			//M Baehr
 		];
 
-		var lastfirst = [/([A-Za-z]+)\s([A-Z])\./,		//Penev L.
-			/([A-Za-z]+)\,\s([A-Z])\./,		//Penev, L.
-			/([A-Za-z]+)\,\s([A-Za-z]+)/,	//Penev, Lyubomir
-			/([A-Za-z]+)\,\s([A-Z])/,		//Penev, L
-			/([A-Za-z]+)\s\,,\s([A-Z])/,	//Penev ,, L
-
+		var lastfirst = [
 
+			/^([\S]+)\,\s+([A-Z]\.)$/,		//Penev, L.
+			/^([\S]+)\,\s+([A-Z]\.\s*[A-Z]\.)$/,		//Vázquez, D. P.
+			/^([^\s.,]+)\,\s*(.+)$/,	//PenevB, Lyubomir
+			/^([^\s,.]+)\,\s+([A-Z])$/,		//Peneve, L
+			/^([^\s.,]+)\s+\,,\s+([A-Z])$/,	//Penevf ,, L
+			/^([^\s]+)\s+([A-Z]\.)$/,		//Penevc L.
 			];
 		var i, n, lst, l;
-		for (i = 0, n = firstlast.length; i < n; i+=1) {
-			if (firstlast[i].test(text)){
-				//p(firstlast[i])
-				//p(text)
-				lst = firstlast[i].exec(text);
-				l = lst.length;
-				return lst.slice(1, l);
-			}
-		}
 		for (i = 0, n = lastfirst.length; i < n; i+=1) {
 			if (lastfirst[i].test(text)){
 				//p(lastfirst[i])
@@ -93,6 +88,15 @@ module.exports = {
 				return first.concat(last);
 			}
 		}
+		for (i = 0, n = firstlast.length; i < n; i+=1) {
+			if (firstlast[i].test(text)){
+				//p(firstlast[i])
+				//p(text)
+				lst = firstlast[i].exec(text);
+				l = lst.length;
+				return lst.slice(1, l);
+			}
+		}
 		return [text];
 	},
 

diff --git a/test/test_tools.js b/test/test_tools.js
@@ -4,22 +4,26 @@ var assert = require("assert");
 describe('Tools', function() {
 	describe('parseAuthorString', function() {
 		var examples = [
-				['M Baehr', ['M', 'Baehr']],
-				['L. Peneva',   ['L', 'Peneva']],
-				['L., Penevb',  ['L', 'Penevb']],
-				['Penevc L.',   ['L', 'Penevc']],
-				['Penevd, L.',  ['L', 'Penevd']],
-				['Peneve, L',   ['L', 'Peneve']],
-				['Penevf ,, L', ['L', 'Penevf']],
-				['Lyubomir PenevA', ['Lyubomir', 'PenevA']],
-				['PenevB, Lyubomir', ['Lyubomir', 'PenevB']],
-				['BERNHARD KROMP', ['BERNHARD', 'KROMP']],
-				['J. P. Zaballos', ['J. P.', 'Zaballos']],
-				['S.-A. Bengtson', ['S.-A.', 'Bengtson']],
-				['Carl H. Lindroth', ['Carl H.', 'Lindroth']],
-				['Carmen Chavez Ortiz', ['Carmen Chavez', 'Ortiz']],
-				['A.M.H. Brunsting', ['A.M.H.', 'Brunsting']],
-				['Artur R. M. Serrano', ['Artur R. M.', 'Serrano']],
+				 ['L., Penevb',  ['L.', 'Penevb']],
+				 ['M Baehr', ['M', 'Baehr']],
+				 ['L. Peneva',   ['L.', 'Peneva']],
+				 ['Lyubomir PenevA', ['Lyubomir', 'PenevA']],
+				 ['BERNHARD KROMP', ['BERNHARD', 'KROMP']],
+				 ['J. P. Zaballos', ['J. P.', 'Zaballos']],
+				 ['S.-A. Bengtson', ['S.-A.', 'Bengtson']],
+				 ['Carl H. Lindroth', ['Carl H.', 'Lindroth']],
+				 ['Carmen Chavez Ortiz', ['Carmen Chavez', 'Ortiz']],
+				 ['A.M.H. Brunsting', ['A.M.H.', 'Brunsting']],
+				 ['Artur R. M. Serrano', ['Artur R. M.', 'Serrano']],
+
+				 ['Vázquez L.',  ['L.', 'Vázquez']],
+				 ['Smith-Ramírez, L.',  ['L.', 'Smith-Ramírez']],
+				 ['Peneve, L',   ['L', 'Peneve']],
+				 ['Penevf ,, L', ['L', 'Penevf']],
+				 ['PenevB, Lyubomir', ['Lyubomir', 'PenevB']],
+				 ['Öckinger,Erik', ['Erik', 'Öckinger']],
+				 ['Vázquez, D. P.', ['D. P.', 'Vázquez']],
+				 ['Andersson, Georg K. S.', ['Georg K. S.', 'Andersson']],
 			];
 
 			examples.forEach(function(e){

diff --git a/tester.js b/tester.js
@@ -13,7 +13,7 @@ module.exports = function (app) {
 		'gnub'  : [],
 		'websrv': [test_websrv]
 	};
-	config.modules2test.filter(function (module) { return config.modules.indexOf(module) > -1; }).forEach(function (module) {
+	config.modules2test.filter(function (module) { return config.modules.indexOf(module) > -1}).forEach(function (module) {
 		tests[module].forEach(function (test) {
 			test.call();
 		});