|
0
|
1 // CodeMirror, copyright (c) by Marijn Haverbeke and others
|
|
|
2 // Distributed under an MIT license: https://codemirror.net/5/LICENSE
|
|
|
3
|
|
|
4 /*
|
|
|
5 * Pig Latin Mode for CodeMirror 2
|
|
|
6 * @author Prasanth Jayachandran
|
|
|
7 * @link https://github.com/prasanthj/pig-codemirror-2
|
|
|
8 * This implementation is adapted from PL/SQL mode in CodeMirror 2.
|
|
|
9 */
|
|
|
10 (function(mod) {
|
|
|
11 if (typeof exports == "object" && typeof module == "object") // CommonJS
|
|
|
12 mod(require("../../lib/codemirror"));
|
|
|
13 else if (typeof define == "function" && define.amd) // AMD
|
|
|
14 define(["../../lib/codemirror"], mod);
|
|
|
15 else // Plain browser env
|
|
|
16 mod(CodeMirror);
|
|
|
17 })(function(CodeMirror) {
|
|
|
18 "use strict";
|
|
|
19
|
|
|
20 CodeMirror.defineMode("pig", function(_config, parserConfig) {
|
|
|
21 var keywords = parserConfig.keywords,
|
|
|
22 builtins = parserConfig.builtins,
|
|
|
23 types = parserConfig.types,
|
|
|
24 multiLineStrings = parserConfig.multiLineStrings;
|
|
|
25
|
|
|
26 var isOperatorChar = /[*+\-%<>=&?:\/!|]/;
|
|
|
27
|
|
|
28 function chain(stream, state, f) {
|
|
|
29 state.tokenize = f;
|
|
|
30 return f(stream, state);
|
|
|
31 }
|
|
|
32
|
|
|
33 function tokenComment(stream, state) {
|
|
|
34 var isEnd = false;
|
|
|
35 var ch;
|
|
|
36 while(ch = stream.next()) {
|
|
|
37 if(ch == "/" && isEnd) {
|
|
|
38 state.tokenize = tokenBase;
|
|
|
39 break;
|
|
|
40 }
|
|
|
41 isEnd = (ch == "*");
|
|
|
42 }
|
|
|
43 return "comment";
|
|
|
44 }
|
|
|
45
|
|
|
46 function tokenString(quote) {
|
|
|
47 return function(stream, state) {
|
|
|
48 var escaped = false, next, end = false;
|
|
|
49 while((next = stream.next()) != null) {
|
|
|
50 if (next == quote && !escaped) {
|
|
|
51 end = true; break;
|
|
|
52 }
|
|
|
53 escaped = !escaped && next == "\\";
|
|
|
54 }
|
|
|
55 if (end || !(escaped || multiLineStrings))
|
|
|
56 state.tokenize = tokenBase;
|
|
|
57 return "error";
|
|
|
58 };
|
|
|
59 }
|
|
|
60
|
|
|
61
|
|
|
62 function tokenBase(stream, state) {
|
|
|
63 var ch = stream.next();
|
|
|
64
|
|
|
65 // is a start of string?
|
|
|
66 if (ch == '"' || ch == "'")
|
|
|
67 return chain(stream, state, tokenString(ch));
|
|
|
68 // is it one of the special chars
|
|
|
69 else if(/[\[\]{}\(\),;\.]/.test(ch))
|
|
|
70 return null;
|
|
|
71 // is it a number?
|
|
|
72 else if(/\d/.test(ch)) {
|
|
|
73 stream.eatWhile(/[\w\.]/);
|
|
|
74 return "number";
|
|
|
75 }
|
|
|
76 // multi line comment or operator
|
|
|
77 else if (ch == "/") {
|
|
|
78 if (stream.eat("*")) {
|
|
|
79 return chain(stream, state, tokenComment);
|
|
|
80 }
|
|
|
81 else {
|
|
|
82 stream.eatWhile(isOperatorChar);
|
|
|
83 return "operator";
|
|
|
84 }
|
|
|
85 }
|
|
|
86 // single line comment or operator
|
|
|
87 else if (ch=="-") {
|
|
|
88 if(stream.eat("-")){
|
|
|
89 stream.skipToEnd();
|
|
|
90 return "comment";
|
|
|
91 }
|
|
|
92 else {
|
|
|
93 stream.eatWhile(isOperatorChar);
|
|
|
94 return "operator";
|
|
|
95 }
|
|
|
96 }
|
|
|
97 // is it an operator
|
|
|
98 else if (isOperatorChar.test(ch)) {
|
|
|
99 stream.eatWhile(isOperatorChar);
|
|
|
100 return "operator";
|
|
|
101 }
|
|
|
102 else {
|
|
|
103 // get the while word
|
|
|
104 stream.eatWhile(/[\w\$_]/);
|
|
|
105 // is it one of the listed keywords?
|
|
|
106 if (keywords && keywords.propertyIsEnumerable(stream.current().toUpperCase())) {
|
|
|
107 //keywords can be used as variables like flatten(group), group.$0 etc..
|
|
|
108 if (!stream.eat(")") && !stream.eat("."))
|
|
|
109 return "keyword";
|
|
|
110 }
|
|
|
111 // is it one of the builtin functions?
|
|
|
112 if (builtins && builtins.propertyIsEnumerable(stream.current().toUpperCase()))
|
|
|
113 return "variable-2";
|
|
|
114 // is it one of the listed types?
|
|
|
115 if (types && types.propertyIsEnumerable(stream.current().toUpperCase()))
|
|
|
116 return "variable-3";
|
|
|
117 // default is a 'variable'
|
|
|
118 return "variable";
|
|
|
119 }
|
|
|
120 }
|
|
|
121
|
|
|
122 // Interface
|
|
|
123 return {
|
|
|
124 startState: function() {
|
|
|
125 return {
|
|
|
126 tokenize: tokenBase,
|
|
|
127 startOfLine: true
|
|
|
128 };
|
|
|
129 },
|
|
|
130
|
|
|
131 token: function(stream, state) {
|
|
|
132 if(stream.eatSpace()) return null;
|
|
|
133 var style = state.tokenize(stream, state);
|
|
|
134 return style;
|
|
|
135 }
|
|
|
136 };
|
|
|
137 });
|
|
|
138
|
|
|
139 (function() {
|
|
|
140 function keywords(str) {
|
|
|
141 var obj = {}, words = str.split(" ");
|
|
|
142 for (var i = 0; i < words.length; ++i) obj[words[i]] = true;
|
|
|
143 return obj;
|
|
|
144 }
|
|
|
145
|
|
|
146 // builtin funcs taken from trunk revision 1303237
|
|
|
147 var pBuiltins = "ABS ACOS ARITY ASIN ATAN AVG BAGSIZE BINSTORAGE BLOOM BUILDBLOOM CBRT CEIL "
|
|
|
148 + "CONCAT COR COS COSH COUNT COUNT_STAR COV CONSTANTSIZE CUBEDIMENSIONS DIFF DISTINCT DOUBLEABS "
|
|
|
149 + "DOUBLEAVG DOUBLEBASE DOUBLEMAX DOUBLEMIN DOUBLEROUND DOUBLESUM EXP FLOOR FLOATABS FLOATAVG "
|
|
|
150 + "FLOATMAX FLOATMIN FLOATROUND FLOATSUM GENERICINVOKER INDEXOF INTABS INTAVG INTMAX INTMIN "
|
|
|
151 + "INTSUM INVOKEFORDOUBLE INVOKEFORFLOAT INVOKEFORINT INVOKEFORLONG INVOKEFORSTRING INVOKER "
|
|
|
152 + "ISEMPTY JSONLOADER JSONMETADATA JSONSTORAGE LAST_INDEX_OF LCFIRST LOG LOG10 LOWER LONGABS "
|
|
|
153 + "LONGAVG LONGMAX LONGMIN LONGSUM MAX MIN MAPSIZE MONITOREDUDF NONDETERMINISTIC OUTPUTSCHEMA "
|
|
|
154 + "PIGSTORAGE PIGSTREAMING RANDOM REGEX_EXTRACT REGEX_EXTRACT_ALL REPLACE ROUND SIN SINH SIZE "
|
|
|
155 + "SQRT STRSPLIT SUBSTRING SUM STRINGCONCAT STRINGMAX STRINGMIN STRINGSIZE TAN TANH TOBAG "
|
|
|
156 + "TOKENIZE TOMAP TOP TOTUPLE TRIM TEXTLOADER TUPLESIZE UCFIRST UPPER UTF8STORAGECONVERTER ";
|
|
|
157
|
|
|
158 // taken from QueryLexer.g
|
|
|
159 var pKeywords = "VOID IMPORT RETURNS DEFINE LOAD FILTER FOREACH ORDER CUBE DISTINCT COGROUP "
|
|
|
160 + "JOIN CROSS UNION SPLIT INTO IF OTHERWISE ALL AS BY USING INNER OUTER ONSCHEMA PARALLEL "
|
|
|
161 + "PARTITION GROUP AND OR NOT GENERATE FLATTEN ASC DESC IS STREAM THROUGH STORE MAPREDUCE "
|
|
|
162 + "SHIP CACHE INPUT OUTPUT STDERROR STDIN STDOUT LIMIT SAMPLE LEFT RIGHT FULL EQ GT LT GTE LTE "
|
|
|
163 + "NEQ MATCHES TRUE FALSE DUMP";
|
|
|
164
|
|
|
165 // data types
|
|
|
166 var pTypes = "BOOLEAN INT LONG FLOAT DOUBLE CHARARRAY BYTEARRAY BAG TUPLE MAP ";
|
|
|
167
|
|
|
168 CodeMirror.defineMIME("text/x-pig", {
|
|
|
169 name: "pig",
|
|
|
170 builtins: keywords(pBuiltins),
|
|
|
171 keywords: keywords(pKeywords),
|
|
|
172 types: keywords(pTypes)
|
|
|
173 });
|
|
|
174
|
|
|
175 CodeMirror.registerHelper("hintWords", "pig", (pBuiltins + pTypes + pKeywords).split(" "));
|
|
|
176 }());
|
|
|
177
|
|
|
178 });
|
