Merge branch 'refs/heads/ninja-internal-master' into copy-paste-bugfix

author: Ananya Sen 2012-07-17 14:25:53 -0700
committer: Ananya Sen 2012-07-17 14:25:53 -0700
commit: 09457e39532e9b35afb9c635266755bfcb9b488a (patch)
tree: 665dcbd41377b98dcab15ff995e9434d9db1a462 /imports/codemirror/mode/pig
parent: f2dbca782bbaca3bed96dff808693693ba083ea9 (diff)
parent: 4f737b24c19ddc02d20f9783b8b080fc6ef11142 (diff)
download: ninja-09457e39532e9b35afb9c635266755bfcb9b488a.tar.gz
2 files changed, 214 insertions, 0 deletions
diff --git a/imports/codemirror/mode/pig/index.html b/imports/codemirror/mode/pig/index.html
new file mode 100644
index 00000000..9cd69c4f
--- /dev/null
+++ b/imports/codemirror/mode/pig/index.html
@@ -0,0 +1,42 @@
+<!doctype html>
+<html>
+  <head>
+    <title>CodeMirror: Pig Latin mode</title>
+    <link rel="stylesheet" href="../../lib/codemirror.css">
+    <script src="../../lib/codemirror.js"></script>
+    <script src="pig.js"></script>
+    <link rel="stylesheet" href="../../doc/docs.css">
+    <style>.CodeMirror {border: 2px inset #dee;}</style>
+  </head>
+  <body>
+    <h1>CodeMirror: Pig Latin mode</h1>
+<form><textarea id="code" name="code">
+-- Apache Pig (Pig Latin Language) Demo
+/* 
+This is a multiline comment.
+*/
+a = LOAD "\path\to\input" USING PigStorage('\t') AS (x:long, y:chararray, z:bytearray);
+b = GROUP a BY (x,y,3+4);
+c = FOREACH b GENERATE flatten(group) as (x,y), SUM(group.$2) as z;
+STORE c INTO "\path\to\output";
+--
+</textarea></form>
+    <script>
+      var editor = CodeMirror.fromTextArea(document.getElementById("code"), {
+        lineNumbers: true,
+        matchBrackets: true,
+        indentUnit: 4,
+        mode: "text/x-pig"
+      });
+    </script>
+    <p>
+        Simple mode that handles Pig Latin language.
+    </p>
+    <p><strong>MIME type defined:</strong> <code>text/x-pig</code>
+    (PIG code)
+</html>
diff --git a/imports/codemirror/mode/pig/pig.js b/imports/codemirror/mode/pig/pig.js
new file mode 100644
index 00000000..b7399b43
--- /dev/null
+++ b/imports/codemirror/mode/pig/pig.js
@@ -0,0 +1,172 @@
+/*
+ *      Pig Latin Mode for CodeMirror 2 
+ *      @author Prasanth Jayachandran
+ *      @link   https://github.com/prasanthj/pig-codemirror-2
+ *  This implementation is adapted from PL/SQL mode in CodeMirror 2.
+*/
+CodeMirror.defineMode("pig", function(config, parserConfig) {
+        var indentUnit = config.indentUnit,
+                keywords = parserConfig.keywords,
+                builtins = parserConfig.builtins,
+                types = parserConfig.types,
+                multiLineStrings = parserConfig.multiLineStrings;
+        
+        var isOperatorChar = /[*+\-%<>=&?:\/!|]/;
+        
+        function chain(stream, state, f) {
+                state.tokenize = f;
+                return f(stream, state);
+        }
+        
+        var type;
+        function ret(tp, style) {
+                type = tp;
+                return style;
+        }
+        
+        function tokenComment(stream, state) {
+                var isEnd = false;
+                var ch;
+                while(ch = stream.next()) {
+                        if(ch == "/" && isEnd) {
+                                state.tokenize = tokenBase;
+                                break;
+                        }
+                        isEnd = (ch == "*");
+                }
+                return ret("comment", "comment");
+        }
+        
+        function tokenString(quote) {
+                return function(stream, state) {
+                        var escaped = false, next, end = false;
+                        while((next = stream.next()) != null) {
+                                if (next == quote && !escaped) {
+                                        end = true; break;
+                                }
+                                escaped = !escaped && next == "\\";
+                        }
+                        if (end || !(escaped || multiLineStrings))
+                                state.tokenize = tokenBase;
+                        return ret("string", "error");
+                };
+        }
+        
+        function tokenBase(stream, state) {
+                var ch = stream.next();
+                
+                // is a start of string?
+                if (ch == '"' || ch == "'")
+                        return chain(stream, state, tokenString(ch));
+                // is it one of the special chars
+                else if(/[\[\]{}\(\),;\.]/.test(ch))
+                        return ret(ch);
+                // is it a number?
+                else if(/\d/.test(ch)) {
+                        stream.eatWhile(/[\w\.]/);
+                        return ret("number", "number");
+                }
+                // multi line comment or operator
+                else if (ch == "/") {
+                        if (stream.eat("*")) {
+                                return chain(stream, state, tokenComment);
+                        }
+                        else {
+                                stream.eatWhile(isOperatorChar);
+                                return ret("operator", "operator");
+                        }
+                }
+                // single line comment or operator
+                else if (ch=="-") {
+                        if(stream.eat("-")){
+                                stream.skipToEnd();
+                                return ret("comment", "comment");
+                        }
+                        else {
+                                stream.eatWhile(isOperatorChar);
+                                return ret("operator", "operator");
+                        }
+                }
+                // is it an operator
+                else if (isOperatorChar.test(ch)) {
+                        stream.eatWhile(isOperatorChar);
+                        return ret("operator", "operator");
+                }
+                else {
+                        // get the while word
+                        stream.eatWhile(/[\w\$_]/);
+                        // is it one of the listed keywords?
+                        if (keywords && keywords.propertyIsEnumerable(stream.current().toUpperCase())) {
+                                if (stream.eat(")") || stream.eat(".")) {
+                                        //keywords can be used as variables like flatten(group), group.$0 etc..
+                                }
+                                else {
+                                        return ("keyword", "keyword");
+                                }
+                        }
+                        // is it one of the builtin functions?
+                        if (builtins && builtins.propertyIsEnumerable(stream.current().toUpperCase()))
+                        {
+                                return ("keyword", "variable-2")
+                        }
+                        // is it one of the listed types?
+                        if (types && types.propertyIsEnumerable(stream.current().toUpperCase()))
+                                return ("keyword", "variable-3")
+                        // default is a 'word'
+                        return ret("word", "pig-word");
+                }
+        }
+        
+        // Interface
+        return {
+                startState: function(basecolumn) {
+                        return {
+                                tokenize: tokenBase,
+                                startOfLine: true
+                        };
+                },
+                
+                token: function(stream, state) {
+                        if(stream.eatSpace()) return null;
+                        var style = state.tokenize(stream, state);
+                        return style;
+                }
+        };
+});
+(function() {
+        function keywords(str) {
+                var obj = {}, words = str.split(" ");
+                for (var i = 0; i < words.length; ++i) obj[words[i]] = true;
+                return obj;
+        }
+        // builtin funcs taken from trunk revision 1303237
+        var pBuiltins = "ABS ACOS ARITY ASIN ATAN AVG BAGSIZE BINSTORAGE BLOOM BUILDBLOOM CBRT CEIL " 
+        + "CONCAT COR COS COSH COUNT COUNT_STAR COV CONSTANTSIZE CUBEDIMENSIONS DIFF DISTINCT DOUBLEABS "
+        + "DOUBLEAVG DOUBLEBASE DOUBLEMAX DOUBLEMIN DOUBLEROUND DOUBLESUM EXP FLOOR FLOATABS FLOATAVG "
+        + "FLOATMAX FLOATMIN FLOATROUND FLOATSUM GENERICINVOKER INDEXOF INTABS INTAVG INTMAX INTMIN "
+        + "INTSUM INVOKEFORDOUBLE INVOKEFORFLOAT INVOKEFORINT INVOKEFORLONG INVOKEFORSTRING INVOKER "
+        + "ISEMPTY JSONLOADER JSONMETADATA JSONSTORAGE LAST_INDEX_OF LCFIRST LOG LOG10 LOWER LONGABS "
+        + "LONGAVG LONGMAX LONGMIN LONGSUM MAX MIN MAPSIZE MONITOREDUDF NONDETERMINISTIC OUTPUTSCHEMA  "
+        + "PIGSTORAGE PIGSTREAMING RANDOM REGEX_EXTRACT REGEX_EXTRACT_ALL REPLACE ROUND SIN SINH SIZE "
+        + "SQRT STRSPLIT SUBSTRING SUM STRINGCONCAT STRINGMAX STRINGMIN STRINGSIZE TAN TANH TOBAG "
+        + "TOKENIZE TOMAP TOP TOTUPLE TRIM TEXTLOADER TUPLESIZE UCFIRST UPPER UTF8STORAGECONVERTER "; 
+        
+        // taken from QueryLexer.g
+        var pKeywords = "VOID IMPORT RETURNS DEFINE LOAD FILTER FOREACH ORDER CUBE DISTINCT COGROUP "
+        + "JOIN CROSS UNION SPLIT INTO IF OTHERWISE ALL AS BY USING INNER OUTER ONSCHEMA PARALLEL "
+        + "PARTITION GROUP AND OR NOT GENERATE FLATTEN ASC DESC IS STREAM THROUGH STORE MAPREDUCE "
+        + "SHIP CACHE INPUT OUTPUT STDERROR STDIN STDOUT LIMIT SAMPLE LEFT RIGHT FULL EQ GT LT GTE LTE " 
+        + "NEQ MATCHES TRUE FALSE "; 
+        
+        // data types
+        var pTypes = "BOOLEAN INT LONG FLOAT DOUBLE CHARARRAY BYTEARRAY BAG TUPLE MAP "
+        
+        CodeMirror.defineMIME("text/x-pig", {
+         name: "pig",
+         builtins: keywords(pBuiltins),
+         keywords: keywords(pKeywords),
+         types: keywords(pTypes)
+         });
+}());
author	Ananya Sen	2012-07-17 14:25:53 -0700
committer	Ananya Sen	2012-07-17 14:25:53 -0700
commit	09457e39532e9b35afb9c635266755bfcb9b488a (patch)
tree	665dcbd41377b98dcab15ff995e9434d9db1a462 /imports/codemirror/mode/pig
parent	f2dbca782bbaca3bed96dff808693693ba083ea9 (diff)
parent	4f737b24c19ddc02d20f9783b8b080fc6ef11142 (diff)
download	ninja-09457e39532e9b35afb9c635266755bfcb9b488a.tar.gz

diff --git a/imports/codemirror/mode/pig/index.html b/imports/codemirror/mode/pig/index.html new file mode 100644 index 00000000..9cd69c4f --- /dev/null +++ b/imports/codemirror/mode/pig/index.html
@@ -0,0 +1,42 @@
	1	<!doctype html>
	2	<html>
	3	<head>
	4	<title>CodeMirror: Pig Latin mode</title>
	5	<link rel="stylesheet" href="../../lib/codemirror.css">
	6	<script src="../../lib/codemirror.js"></script>
	7	<script src="pig.js"></script>
	8	<link rel="stylesheet" href="../../doc/docs.css">
	9	<style>.CodeMirror {border: 2px inset #dee;}</style>
	10	</head>
	11	<body>
	12	<h1>CodeMirror: Pig Latin mode</h1>
	13
	14	<form><textarea id="code" name="code">
	15	-- Apache Pig (Pig Latin Language) Demo
	16	/*
	17	This is a multiline comment.
	18	*/
	19	a = LOAD "\path\to\input" USING PigStorage('\t') AS (x:long, y:chararray, z:bytearray);
	20	b = GROUP a BY (x,y,3+4);
	21	c = FOREACH b GENERATE flatten(group) as (x,y), SUM(group.$2) as z;
	22	STORE c INTO "\path\to\output";
	23
	24	--
	25	</textarea></form>
	26
	27	<script>
	28	var editor = CodeMirror.fromTextArea(document.getElementById("code"), {
	29	lineNumbers: true,
	30	matchBrackets: true,
	31	indentUnit: 4,
	32	mode: "text/x-pig"
	33	});
	34	</script>
	35
	36	<p>
	37	Simple mode that handles Pig Latin language.
	38	</p>
	39
	40	<p><strong>MIME type defined:</strong> <code>text/x-pig</code>
	41	(PIG code)
	42	</html>


diff --git a/imports/codemirror/mode/pig/pig.js b/imports/codemirror/mode/pig/pig.js new file mode 100644 index 00000000..b7399b43 --- /dev/null +++ b/imports/codemirror/mode/pig/pig.js
@@ -0,0 +1,172 @@
	1	/*
	2	* Pig Latin Mode for CodeMirror 2
	3	* @author Prasanth Jayachandran
	4	* @link https://github.com/prasanthj/pig-codemirror-2
	5	* This implementation is adapted from PL/SQL mode in CodeMirror 2.
	6	*/
	7	CodeMirror.defineMode("pig", function(config, parserConfig) {
	8	var indentUnit = config.indentUnit,
	9	keywords = parserConfig.keywords,
	10	builtins = parserConfig.builtins,
	11	types = parserConfig.types,
	12	multiLineStrings = parserConfig.multiLineStrings;
	13
	14	var isOperatorChar = /[*+\-%<>=&?:\/!\|]/;
	15
	16	function chain(stream, state, f) {
	17	state.tokenize = f;
	18	return f(stream, state);
	19	}
	20
	21	var type;
	22	function ret(tp, style) {
	23	type = tp;
	24	return style;
	25	}
	26
	27	function tokenComment(stream, state) {
	28	var isEnd = false;
	29	var ch;
	30	while(ch = stream.next()) {
	31	if(ch == "/" && isEnd) {
	32	state.tokenize = tokenBase;
	33	break;
	34	}
	35	isEnd = (ch == "*");
	36	}
	37	return ret("comment", "comment");
	38	}
	39
	40	function tokenString(quote) {
	41	return function(stream, state) {
	42	var escaped = false, next, end = false;
	43	while((next = stream.next()) != null) {
	44	if (next == quote && !escaped) {
	45	end = true; break;
	46	}
	47	escaped = !escaped && next == "\\";
	48	}
	49	if (end \|\| !(escaped \|\| multiLineStrings))
	50	state.tokenize = tokenBase;
	51	return ret("string", "error");
	52	};
	53	}
	54
	55	function tokenBase(stream, state) {
	56	var ch = stream.next();
	57
	58	// is a start of string?
	59	if (ch == '"' \|\| ch == "'")
	60	return chain(stream, state, tokenString(ch));
	61	// is it one of the special chars
	62	else if(/[\[\]{}\(\),;\.]/.test(ch))
	63	return ret(ch);
	64	// is it a number?
	65	else if(/\d/.test(ch)) {
	66	stream.eatWhile(/[\w\.]/);
	67	return ret("number", "number");
	68	}
	69	// multi line comment or operator
	70	else if (ch == "/") {
	71	if (stream.eat("*")) {
	72	return chain(stream, state, tokenComment);
	73	}
	74	else {
	75	stream.eatWhile(isOperatorChar);
	76	return ret("operator", "operator");
	77	}
	78	}
	79	// single line comment or operator
	80	else if (ch=="-") {
	81	if(stream.eat("-")){
	82	stream.skipToEnd();
	83	return ret("comment", "comment");
	84	}
	85	else {
	86	stream.eatWhile(isOperatorChar);
	87	return ret("operator", "operator");
	88	}
	89	}
	90	// is it an operator
	91	else if (isOperatorChar.test(ch)) {
	92	stream.eatWhile(isOperatorChar);
	93	return ret("operator", "operator");
	94	}
	95	else {
	96	// get the while word
	97	stream.eatWhile(/[\w\$_]/);
	98	// is it one of the listed keywords?
	99	if (keywords && keywords.propertyIsEnumerable(stream.current().toUpperCase())) {
	100	if (stream.eat(")") \|\| stream.eat(".")) {
	101	//keywords can be used as variables like flatten(group), group.$0 etc..
	102	}
	103	else {
	104	return ("keyword", "keyword");
	105	}
	106	}
	107	// is it one of the builtin functions?
	108	if (builtins && builtins.propertyIsEnumerable(stream.current().toUpperCase()))
	109	{
	110	return ("keyword", "variable-2")
	111	}
	112	// is it one of the listed types?
	113	if (types && types.propertyIsEnumerable(stream.current().toUpperCase()))
	114	return ("keyword", "variable-3")
	115	// default is a 'word'
	116	return ret("word", "pig-word");
	117	}
	118	}
	119
	120	// Interface
	121	return {
	122	startState: function(basecolumn) {
	123	return {
	124	tokenize: tokenBase,
	125	startOfLine: true
	126	};
	127	},
	128
	129	token: function(stream, state) {
	130	if(stream.eatSpace()) return null;
	131	var style = state.tokenize(stream, state);
	132	return style;
	133	}
	134	};
	135	});
	136
	137	(function() {
	138	function keywords(str) {
	139	var obj = {}, words = str.split(" ");
	140	for (var i = 0; i < words.length; ++i) obj[words[i]] = true;
	141	return obj;
	142	}
	143
	144	// builtin funcs taken from trunk revision 1303237
	145	var pBuiltins = "ABS ACOS ARITY ASIN ATAN AVG BAGSIZE BINSTORAGE BLOOM BUILDBLOOM CBRT CEIL "
	146	+ "CONCAT COR COS COSH COUNT COUNT_STAR COV CONSTANTSIZE CUBEDIMENSIONS DIFF DISTINCT DOUBLEABS "
	147	+ "DOUBLEAVG DOUBLEBASE DOUBLEMAX DOUBLEMIN DOUBLEROUND DOUBLESUM EXP FLOOR FLOATABS FLOATAVG "
	148	+ "FLOATMAX FLOATMIN FLOATROUND FLOATSUM GENERICINVOKER INDEXOF INTABS INTAVG INTMAX INTMIN "
	149	+ "INTSUM INVOKEFORDOUBLE INVOKEFORFLOAT INVOKEFORINT INVOKEFORLONG INVOKEFORSTRING INVOKER "
	150	+ "ISEMPTY JSONLOADER JSONMETADATA JSONSTORAGE LAST_INDEX_OF LCFIRST LOG LOG10 LOWER LONGABS "
	151	+ "LONGAVG LONGMAX LONGMIN LONGSUM MAX MIN MAPSIZE MONITOREDUDF NONDETERMINISTIC OUTPUTSCHEMA "
	152	+ "PIGSTORAGE PIGSTREAMING RANDOM REGEX_EXTRACT REGEX_EXTRACT_ALL REPLACE ROUND SIN SINH SIZE "
	153	+ "SQRT STRSPLIT SUBSTRING SUM STRINGCONCAT STRINGMAX STRINGMIN STRINGSIZE TAN TANH TOBAG "
	154	+ "TOKENIZE TOMAP TOP TOTUPLE TRIM TEXTLOADER TUPLESIZE UCFIRST UPPER UTF8STORAGECONVERTER ";
	155
	156	// taken from QueryLexer.g
	157	var pKeywords = "VOID IMPORT RETURNS DEFINE LOAD FILTER FOREACH ORDER CUBE DISTINCT COGROUP "
	158	+ "JOIN CROSS UNION SPLIT INTO IF OTHERWISE ALL AS BY USING INNER OUTER ONSCHEMA PARALLEL "
	159	+ "PARTITION GROUP AND OR NOT GENERATE FLATTEN ASC DESC IS STREAM THROUGH STORE MAPREDUCE "
	160	+ "SHIP CACHE INPUT OUTPUT STDERROR STDIN STDOUT LIMIT SAMPLE LEFT RIGHT FULL EQ GT LT GTE LTE "
	161	+ "NEQ MATCHES TRUE FALSE ";
	162
	163	// data types
	164	var pTypes = "BOOLEAN INT LONG FLOAT DOUBLE CHARARRAY BYTEARRAY BAG TUPLE MAP "
	165
	166	CodeMirror.defineMIME("text/x-pig", {
	167	name: "pig",
	168	builtins: keywords(pBuiltins),
	169	keywords: keywords(pKeywords),
	170	types: keywords(pTypes)
	171	});
	172	}());