scripts/CodeMirror/mode/sas/sas.js (view raw)
1// CodeMirror, copyright (c) by Marijn Haverbeke and others
2// Distributed under an MIT license: https://codemirror.net/LICENSE
3
4
5// SAS mode copyright (c) 2016 Jared Dean, SAS Institute
6// Created by Jared Dean
7
8// TODO
9// indent and de-indent
10// identify macro variables
11
12
13//Definitions
14// comment -- text within * ; or /* */
15// keyword -- SAS language variable
16// variable -- macro variables starts with '&' or variable formats
17// variable-2 -- DATA Step, proc, or macro names
18// string -- text within ' ' or " "
19// operator -- numeric operator + / - * ** le eq ge ... and so on
20// builtin -- proc %macro data run mend
21// atom
22// def
23
24(function(mod) {
25 if (typeof exports == "object" && typeof module == "object") // CommonJS
26 mod(require("../../lib/codemirror"));
27 else if (typeof define == "function" && define.amd) // AMD
28 define(["../../lib/codemirror"], mod);
29 else // Plain browser env
30 mod(CodeMirror);
31})(function(CodeMirror) {
32 "use strict";
33
34 CodeMirror.defineMode("sas", function () {
35 var words = {};
36 var isDoubleOperatorSym = {
37 eq: 'operator',
38 lt: 'operator',
39 le: 'operator',
40 gt: 'operator',
41 ge: 'operator',
42 "in": 'operator',
43 ne: 'operator',
44 or: 'operator'
45 };
46 var isDoubleOperatorChar = /(<=|>=|!=|<>)/;
47 var isSingleOperatorChar = /[=\(:\),{}.*<>+\-\/^\[\]]/;
48
49 // Takes a string of words separated by spaces and adds them as
50 // keys with the value of the first argument 'style'
51 function define(style, string, context) {
52 if (context) {
53 var split = string.split(' ');
54 for (var i = 0; i < split.length; i++) {
55 words[split[i]] = {style: style, state: context};
56 }
57 }
58 }
59 //datastep
60 define('def', 'stack pgm view source debug nesting nolist', ['inDataStep']);
61 define('def', 'if while until for do do; end end; then else cancel', ['inDataStep']);
62 define('def', 'label format _n_ _error_', ['inDataStep']);
63 define('def', 'ALTER BUFNO BUFSIZE CNTLLEV COMPRESS DLDMGACTION ENCRYPT ENCRYPTKEY EXTENDOBSCOUNTER GENMAX GENNUM INDEX LABEL OBSBUF OUTREP PW PWREQ READ REPEMPTY REPLACE REUSE ROLE SORTEDBY SPILL TOBSNO TYPE WRITE FILECLOSE FIRSTOBS IN OBS POINTOBS WHERE WHEREUP IDXNAME IDXWHERE DROP KEEP RENAME', ['inDataStep']);
64 define('def', 'filevar finfo finv fipname fipnamel fipstate first firstobs floor', ['inDataStep']);
65 define('def', 'varfmt varinfmt varlabel varlen varname varnum varray varrayx vartype verify vformat vformatd vformatdx vformatn vformatnx vformatw vformatwx vformatx vinarray vinarrayx vinformat vinformatd vinformatdx vinformatn vinformatnx vinformatw vinformatwx vinformatx vlabel vlabelx vlength vlengthx vname vnamex vnferr vtype vtypex weekday', ['inDataStep']);
66 define('def', 'zipfips zipname zipnamel zipstate', ['inDataStep']);
67 define('def', 'put putc putn', ['inDataStep']);
68 define('builtin', 'data run', ['inDataStep']);
69
70
71 //proc
72 define('def', 'data', ['inProc']);
73
74 // flow control for macros
75 define('def', '%if %end %end; %else %else; %do %do; %then', ['inMacro']);
76
77 //everywhere
78 define('builtin', 'proc run; quit; libname filename %macro %mend option options', ['ALL']);
79
80 define('def', 'footnote title libname ods', ['ALL']);
81 define('def', '%let %put %global %sysfunc %eval ', ['ALL']);
82 // automatic macro variables http://support.sas.com/documentation/cdl/en/mcrolref/61885/HTML/default/viewer.htm#a003167023.htm
83 define('variable', '&sysbuffr &syscc &syscharwidth &syscmd &sysdate &sysdate9 &sysday &sysdevic &sysdmg &sysdsn &sysencoding &sysenv &syserr &syserrortext &sysfilrc &syshostname &sysindex &sysinfo &sysjobid &syslast &syslckrc &syslibrc &syslogapplname &sysmacroname &sysmenv &sysmsg &sysncpu &sysodspath &sysparm &syspbuff &sysprocessid &sysprocessname &sysprocname &sysrc &sysscp &sysscpl &sysscpl &syssite &sysstartid &sysstartname &systcpiphostname &systime &sysuserid &sysver &sysvlong &sysvlong4 &syswarningtext', ['ALL']);
84
85 //footnote[1-9]? title[1-9]?
86
87 //options statement
88 define('def', 'source2 nosource2 page pageno pagesize', ['ALL']);
89
90 //proc and datastep
91 define('def', '_all_ _character_ _cmd_ _freq_ _i_ _infile_ _last_ _msg_ _null_ _numeric_ _temporary_ _type_ abort abs addr adjrsq airy alpha alter altlog altprint and arcos array arsin as atan attrc attrib attrn authserver autoexec awscontrol awsdef awsmenu awsmenumerge awstitle backward band base betainv between blocksize blshift bnot bor brshift bufno bufsize bxor by byerr byline byte calculated call cards cards4 catcache cbufno cdf ceil center cexist change chisq cinv class cleanup close cnonct cntllev coalesce codegen col collate collin column comamid comaux1 comaux2 comdef compbl compound compress config continue convert cos cosh cpuid create cross crosstab css curobs cv daccdb daccdbsl daccsl daccsyd dacctab dairy datalines datalines4 datejul datepart datetime day dbcslang dbcstype dclose ddfm ddm delete delimiter depdb depdbsl depsl depsyd deptab dequote descending descript design= device dflang dhms dif digamma dim dinfo display distinct dkricond dkrocond dlm dnum do dopen doptname doptnum dread drop dropnote dsname dsnferr echo else emaildlg emailid emailpw emailserver emailsys encrypt end endsas engine eof eov erf erfc error errorcheck errors exist exp fappend fclose fcol fdelete feedback fetch fetchobs fexist fget file fileclose fileexist filefmt filename fileref fmterr fmtsearch fnonct fnote font fontalias fopen foptname foptnum force formatted formchar formdelim formdlim forward fpoint fpos fput fread frewind frlen from fsep fuzz fwrite gaminv gamma getoption getvarc getvarn go goto group gwindow hbar hbound helpenv helploc hms honorappearance hosthelp hostprint hour hpct html hvar ibessel ibr id if index indexc indexw initcmd initstmt inner input inputc inputn inr insert int intck intnx into intrr invaliddata irr is jbessel join juldate keep kentb kurtosis label lag last lbound leave left length levels lgamma lib library libref line linesize link list log log10 log2 logpdf logpmf logsdf lostcard lowcase lrecl ls macro macrogen maps mautosource max maxdec maxr mdy mean measures median memtype merge merror min minute missing missover mlogic mod mode model modify month mopen mort mprint mrecall msglevel msymtabmax mvarsize myy n nest netpv new news nmiss no nobatch nobs nocaps nocardimage nocenter nocharcode nocmdmac nocol nocum nodate nodbcs nodetails nodmr nodms nodmsbatch nodup nodupkey noduplicates noechoauto noequals noerrorabend noexitwindows nofullstimer noicon noimplmac noint nolist noloadlist nomiss nomlogic nomprint nomrecall nomsgcase nomstored nomultenvappl nonotes nonumber noobs noovp nopad nopercent noprint noprintinit normal norow norsasuser nosetinit nosplash nosymbolgen note notes notitle notitles notsorted noverbose noxsync noxwait npv null number numkeys nummousekeys nway obs on open order ordinal otherwise out outer outp= output over ovp p(1 5 10 25 50 75 90 95 99) pad pad2 paired parm parmcards path pathdll pathname pdf peek peekc pfkey pmf point poisson poke position printer probbeta probbnml probchi probf probgam probhypr probit probnegb probnorm probsig probt procleave prt ps pw pwreq qtr quote r ranbin rancau random ranexp rangam range ranks rannor ranpoi rantbl rantri ranuni rcorr read recfm register regr remote remove rename repeat repeated replace resolve retain return reuse reverse rewind right round rsquare rtf rtrace rtraceloc s s2 samploc sasautos sascontrol sasfrscr sasmsg sasmstore sasscript sasuser saving scan sdf second select selection separated seq serror set setcomm setot sign simple sin sinh siteinfo skewness skip sle sls sortedby sortpgm sortseq sortsize soundex spedis splashlocation split spool sqrt start std stderr stdin stfips stimer stname stnamel stop stopover sub subgroup subpopn substr sum sumwgt symbol symbolgen symget symput sysget sysin sysleave sysmsg sysparm sysprint sysprintfont sysprod sysrc system t table tables tan tanh tapeclose tbufsize terminal test then timepart tinv tnonct to today tol tooldef totper transformout translate trantab tranwrd trigamma trim trimn trunc truncover type unformatted uniform union until upcase update user usericon uss validate value var weight when where while wincharset window work workinit workterm write wsum xsync xwait yearcutoff yes yyq min max', ['inDataStep', 'inProc']);
92 define('operator', 'and not ', ['inDataStep', 'inProc']);
93
94 // Main function
95 function tokenize(stream, state) {
96 // Finally advance the stream
97 var ch = stream.next();
98
99 // BLOCKCOMMENT
100 if (ch === '/' && stream.eat('*')) {
101 state.continueComment = true;
102 return "comment";
103 } else if (state.continueComment === true) { // in comment block
104 //comment ends at the beginning of the line
105 if (ch === '*' && stream.peek() === '/') {
106 stream.next();
107 state.continueComment = false;
108 } else if (stream.skipTo('*')) { //comment is potentially later in line
109 stream.skipTo('*');
110 stream.next();
111 if (stream.eat('/'))
112 state.continueComment = false;
113 } else {
114 stream.skipToEnd();
115 }
116 return "comment";
117 }
118
119 if (ch == "*" && stream.column() == stream.indentation()) {
120 stream.skipToEnd()
121 return "comment"
122 }
123
124 // DoubleOperator match
125 var doubleOperator = ch + stream.peek();
126
127 if ((ch === '"' || ch === "'") && !state.continueString) {
128 state.continueString = ch
129 return "string"
130 } else if (state.continueString) {
131 if (state.continueString == ch) {
132 state.continueString = null;
133 } else if (stream.skipTo(state.continueString)) {
134 // quote found on this line
135 stream.next();
136 state.continueString = null;
137 } else {
138 stream.skipToEnd();
139 }
140 return "string";
141 } else if (state.continueString !== null && stream.eol()) {
142 stream.skipTo(state.continueString) || stream.skipToEnd();
143 return "string";
144 } else if (/[\d\.]/.test(ch)) { //find numbers
145 if (ch === ".")
146 stream.match(/^[0-9]+([eE][\-+]?[0-9]+)?/);
147 else if (ch === "0")
148 stream.match(/^[xX][0-9a-fA-F]+/) || stream.match(/^0[0-7]+/);
149 else
150 stream.match(/^[0-9]*\.?[0-9]*([eE][\-+]?[0-9]+)?/);
151 return "number";
152 } else if (isDoubleOperatorChar.test(ch + stream.peek())) { // TWO SYMBOL TOKENS
153 stream.next();
154 return "operator";
155 } else if (isDoubleOperatorSym.hasOwnProperty(doubleOperator)) {
156 stream.next();
157 if (stream.peek() === ' ')
158 return isDoubleOperatorSym[doubleOperator.toLowerCase()];
159 } else if (isSingleOperatorChar.test(ch)) { // SINGLE SYMBOL TOKENS
160 return "operator";
161 }
162
163 // Matches one whole word -- even if the word is a character
164 var word;
165 if (stream.match(/[%&;\w]+/, false) != null) {
166 word = ch + stream.match(/[%&;\w]+/, true);
167 if (/&/.test(word)) return 'variable'
168 } else {
169 word = ch;
170 }
171 // the word after DATA PROC or MACRO
172 if (state.nextword) {
173 stream.match(/[\w]+/);
174 // match memname.libname
175 if (stream.peek() === '.') stream.skipTo(' ');
176 state.nextword = false;
177 return 'variable-2';
178 }
179
180 word = word.toLowerCase()
181 // Are we in a DATA Step?
182 if (state.inDataStep) {
183 if (word === 'run;' || stream.match(/run\s;/)) {
184 state.inDataStep = false;
185 return 'builtin';
186 }
187 // variable formats
188 if ((word) && stream.next() === '.') {
189 //either a format or libname.memname
190 if (/\w/.test(stream.peek())) return 'variable-2';
191 else return 'variable';
192 }
193 // do we have a DATA Step keyword
194 if (word && words.hasOwnProperty(word) &&
195 (words[word].state.indexOf("inDataStep") !== -1 ||
196 words[word].state.indexOf("ALL") !== -1)) {
197 //backup to the start of the word
198 if (stream.start < stream.pos)
199 stream.backUp(stream.pos - stream.start);
200 //advance the length of the word and return
201 for (var i = 0; i < word.length; ++i) stream.next();
202 return words[word].style;
203 }
204 }
205 // Are we in an Proc statement?
206 if (state.inProc) {
207 if (word === 'run;' || word === 'quit;') {
208 state.inProc = false;
209 return 'builtin';
210 }
211 // do we have a proc keyword
212 if (word && words.hasOwnProperty(word) &&
213 (words[word].state.indexOf("inProc") !== -1 ||
214 words[word].state.indexOf("ALL") !== -1)) {
215 stream.match(/[\w]+/);
216 return words[word].style;
217 }
218 }
219 // Are we in a Macro statement?
220 if (state.inMacro) {
221 if (word === '%mend') {
222 if (stream.peek() === ';') stream.next();
223 state.inMacro = false;
224 return 'builtin';
225 }
226 if (word && words.hasOwnProperty(word) &&
227 (words[word].state.indexOf("inMacro") !== -1 ||
228 words[word].state.indexOf("ALL") !== -1)) {
229 stream.match(/[\w]+/);
230 return words[word].style;
231 }
232
233 return 'atom';
234 }
235 // Do we have Keywords specific words?
236 if (word && words.hasOwnProperty(word)) {
237 // Negates the initial next()
238 stream.backUp(1);
239 // Actually move the stream
240 stream.match(/[\w]+/);
241 if (word === 'data' && /=/.test(stream.peek()) === false) {
242 state.inDataStep = true;
243 state.nextword = true;
244 return 'builtin';
245 }
246 if (word === 'proc') {
247 state.inProc = true;
248 state.nextword = true;
249 return 'builtin';
250 }
251 if (word === '%macro') {
252 state.inMacro = true;
253 state.nextword = true;
254 return 'builtin';
255 }
256 if (/title[1-9]/.test(word)) return 'def';
257
258 if (word === 'footnote') {
259 stream.eat(/[1-9]/);
260 return 'def';
261 }
262
263 // Returns their value as state in the prior define methods
264 if (state.inDataStep === true && words[word].state.indexOf("inDataStep") !== -1)
265 return words[word].style;
266 if (state.inProc === true && words[word].state.indexOf("inProc") !== -1)
267 return words[word].style;
268 if (state.inMacro === true && words[word].state.indexOf("inMacro") !== -1)
269 return words[word].style;
270 if (words[word].state.indexOf("ALL") !== -1)
271 return words[word].style;
272 return null;
273 }
274 // Unrecognized syntax
275 return null;
276 }
277
278 return {
279 startState: function () {
280 return {
281 inDataStep: false,
282 inProc: false,
283 inMacro: false,
284 nextword: false,
285 continueString: null,
286 continueComment: false
287 };
288 },
289 token: function (stream, state) {
290 // Strip the spaces, but regex will account for them either way
291 if (stream.eatSpace()) return null;
292 // Go through the main process
293 return tokenize(stream, state);
294 },
295
296 blockCommentStart: "/*",
297 blockCommentEnd: "*/"
298 };
299
300 });
301
302 CodeMirror.defineMIME("text/x-sas", "sas");
303});