lang-detector - npm Package Compare versions

Comparing version 1.0.1 to 1.0.2

index.js

@@ -27,16 +27,7 @@ /**
 
-function getPoints(language, lineOfCode, checkers) {
-	return _.reduce(_.map(checkers, function(checker) {
-		if (checker.pattern.test(lineOfCode)) {
-			return checker.points;
-		}
-		return 0;
-	}), function(memo, num) {
-		return memo + num;
-	}, 0);
-}
-
 /**
  * A checker is an object with the following form:
  *  { pattern: /something/, points: 1 }
+ * or if the pattern only matches code near the top of a given file:
+ *  { pattern: /something/, points: 2, nearTop: true }
  * 
@@ -50,3 +41,3 @@ * Key: Language name.
  * Points scale:
- *  2 = Bonus point:    Almost unique to a given language.
+ *  2 = Bonus points:   Almost unique to a given language.
  *  1 = Regular point:  Not unique to a given language.
@@ -62,3 +53,3 @@ * -1 = Penalty point:  Does not match a given language.
 		// console.log('ayy lmao')
-		{ pattern: /console\.log( )*\(/g, points: 2 },
+		{ pattern: /console\.log( )*\(/, points: 2 },
 		// === operator
@@ -88,3 +79,3 @@ { pattern: /===/g, points: 2 },
 		// #include <whatever.h>
-		{ pattern: /#include (<|")\w+\.h(>|")/, points: 2 },
+		{ pattern: /#include (<|")\w+\.h(>|")/, points: 2, nearTop: true },
 		// pointer
@@ -96,2 +87,4 @@ { pattern: /(\w+)( )*\*( )*\w+/, points: 2 },
 		{ pattern: /(\w+)( )+\w+\[.+\]/, points: 1 },
+		// #define macro
+		{ pattern: /#define( )+.+/, points: 1 },
 		// NULL constant
@@ -117,6 +110,6 @@ { pattern: /NULL/, points: 1 },
 		// #include <whatever.h>
-		{ pattern: /#include( )*(<|")\w+(\.h)?(>|")/, points: 2 },
+		{ pattern: /#include( )*(<|")\w+(\.h)?(>|")/, points: 2, nearTop: true },
 		// using namespace something
-		{ pattern: /using( )*namespace( )*.+( )*;/, points: 2 },
-		// template
+		{ pattern: /using( )*namespace( )*.+( )*;/, points: 2, nearTop: true },
+		// template declaration
 		{ pattern: /template( )*<.*>/, points: 2 },
@@ -129,6 +122,12 @@ // std
 		{ pattern: /(public|protected|private):/, points: 2 },
-		// new Keyword
-		{ pattern: /new \w+(\(.*\))?/, points: 2 },
 		// nullptr
 		{ pattern: /nullptr/, points: 2 },
+		// new Keyword
+		{ pattern: /new \w+(\(.*\))?/, points: 1 },
+		// #define macro
+		{ pattern: /#define( )+.+/, points: 1 },
+		// template usage
+		{ pattern: /\w+<\w+>/, points: 1 },
+		// class keyword
+		{ pattern: /class( )+\w+/, points: 1 },
 		// void keyword
@@ -150,3 +149,3 @@ { pattern: /void/g, points: 1 },
 		// Function definition
-		{ pattern: /def( )+\w+( )*:/, points: 2 },
+		{ pattern: /def( )+\w+\(.*\)( )*:/, points: 2 },
 		// while loop
@@ -165,7 +164,7 @@ { pattern: /while (.+):/, points: 2 },
 		// for loop
-		{ pattern: /for (\w+|\(?\w+,( )*\w+\)?) in (.+):?/, points: 1 },
+		{ pattern: /for (\w+|\(?\w+,( )*\w+\)?) in (.+):/, points: 2 },
 		// Python variable declaration.
-		{ pattern: /\w+( )*=( )*[\w]+/, points: 1 },
+		{ pattern: /\w+( )*=( )*\w+(?!;)(\n|$)/, points: 1 },
 		// import something
-		{ pattern: /import ([[^\.]\w])+/, points: 1 },
+		{ pattern: /import ([[^\.]\w])+/, points: 1, nearTop: true },
 		// print statement/function
@@ -179,3 +178,3 @@ { pattern: /print((( )*\(.+\))|( )+.+)/, points: 1 },
 		// System.out.println() etc.
-		{ pattern: /System\.(in|out)\./, points: 2 },
+		{ pattern: /System\.(in|out)\.\w+/, points: 2 },
 		// Class variable declarations
@@ -198,7 +197,7 @@ { pattern: /(private|protected|public)( )*\w+( )*\w+(( )*=( )*[\w])?/, points: 2 },
 		// new Keyword (Java)
-		{ pattern: /new \w+( )*\(.+\)/, points: 2 },
+		{ pattern: /new [A-Z]\w*( )*\(.+\)/, points: 2 },
 		// C style variable declaration.
 		{ pattern: /(^|\s)(char|long|int|float|double)( )+[\w]+( )*=?/, points: 1 },
 		// extends/implements keywords
-		{ pattern: /(extends|implements)/, points: 1 },
+		{ pattern: /(extends|implements)/, points: 2, nearTop: true },
 		// null keyword
@@ -219,6 +218,7 @@ { pattern: /null/g, points: 1 },
 		// C style include
-		{ pattern: /#include( )*(<|")\w+(\.h)?(>|")/, points: -1 },
+		{ pattern: /#include( )*(<|")\w+(\.h)?(>|")/, points: -1, nearTop: true },
 	],
 
 	'HTML': [
+		{ pattern: /<!DOCTYPE (html|HTML PUBLIC .+)>/, points: 2, nearTop: true },
 		// Tags
@@ -232,3 +232,5 @@ { pattern: /<[a-z0-9]+(( )*[\w]+=('|").+('|")( )*)?>.*<\/[a-z0-9]+>/g, points: 2 },
 		// Properties
-		{ pattern: /[a-z\-]+:.+;/, points: 2 },
+		{ pattern: /[a-z\-]+:(?!:).+;/, points: 2 },
+		// <style> tag from HTML (This is a rare case where a lot of penalty points are needed.)
+		{ pattern: /<(\/)?style>/, points: -50 },
 	],
@@ -238,3 +240,3 @@
 		// require/include
-		{ pattern: /(require|include)( )*'\w+(\.rb)?'/, points: 2 },
+		{ pattern: /(require|include)( )*'\w+(\.rb)?'/, points: 2, nearTop: true },
 		// Function definition
@@ -247,3 +249,3 @@ { pattern: /def( )+\w+( )*(\(.+\))?( )*\n/, points: 2 },
 		// puts (Ruby print)
-		{ pattern: /puts( )*("|').+("|')/, points: 2 },
+		{ pattern: /puts( )+("|').+("|')/, points: 2 },
 		// Inheriting class
@@ -268,2 +270,6 @@ { pattern: /class [A-Z]\w*( )*<( )*([A-Z]\w*(::)?)+/, points: 2 },
 	'Go': [
+		// package something
+		{ pattern: /package( )+[a-z]+\n/, points: 2, nearTop: true },
+		// import
+		{ pattern: /(import( )*\(( )*\n)|(import( )+"[a-z0-9\/\.]+")/, points: 2, nearTop: true },
 		// error check
@@ -276,3 +282,3 @@ { pattern: /if.+err( )*!=( )*nil.+{/, points: 2 },
 		// variable initialisation
-		{ pattern: /\w+( )*:=( )*.+[;\n]/, points: 2 },
+		{ pattern: /\w+( )*:=( )*.+[^;\n]/, points: 2 },
 		// if/else if
@@ -293,2 +299,13 @@ { pattern: /(} else )?if.+{/, points: 2 },
 
+function getPoints(language, lineOfCode, checkers) {
+	return _.reduce(_.map(checkers, function(checker) {
+		if (checker.pattern.test(lineOfCode)) {
+			return checker.points;
+		}
+		return 0;
+	}), function(memo, num) {
+		return memo + num;
+	}, 0);
+}
+
 function detectLang(snippet, options) {
@@ -305,5 +322,12 @@ var opts = _.defaults(options || {}, {
 
-	if (opts.heuristic && linesOfCode.length > 1000) {
+	function nearTop(index) {
+		if (linesOfCode.length <= 10) {
+			return true;
+		}
+		return index < linesOfCode.length / 10;
+	}
+
+	if (opts.heuristic && linesOfCode.length >= 500) {
 		linesOfCode = linesOfCode.filter(function(lineOfCode, index) {
-			if (index <= linesOfCode.length / 10) {
+			if (nearTop(index)) {
 				return true;
@@ -327,7 +351,15 @@ }
 
-		var points = _.reduce(_.map(linesOfCode, function(lineOfCode) {
-			return getPoints(language, lineOfCode, checkers);
-		}), function(memo, num) {
+		var pointsList = linesOfCode.map(function(lineOfCode, index) {
+			if (!nearTop(index)) {
+				return getPoints(language, lineOfCode, _.reject(checkers, function(checker) {
+					return checker.nearTop;
+				}));
+			} else {
+				return getPoints(language, lineOfCode, checkers);
+			}
+		});
+
+		var points = _.reduce(pointsList, function(memo, num) {
 			return memo + num;
-		}, 0);
+		});
 
@@ -334,0 +366,0 @@ return { language: language, points: points };

package.json

 {
   "name": "lang-detector",
-  "version": "1.0.1",
+  "version": "1.0.2",
   "description": "A library for detecting the programming language of a code snippet.",
@@ -5,0 +5,0 @@ "main": "index.js",

No alert changes

Improved metrics

Total package byte prevSize

482391

increased by0.52%

Lines of code

10245

increased by0.61%

No dependency changes