1// CodeMirror, copyright (c) by Marijn Haverbeke and others
2// Distributed under an MIT license: http://codemirror.net/LICENSE
3
4(function(mod) {
5  if (typeof exports == "object" && typeof module == "object") // CommonJS
6    mod(require("../../lib/codemirror"));
7  else if (typeof define == "function" && define.amd) // AMD
8    define(["../../lib/codemirror"], mod);
9  else // Plain browser env
10    mod(CodeMirror);
11})(function(CodeMirror) {
12  "use strict";
13
14  function wordRegexp(words) {
15    return new RegExp("^((" + words.join(")|(") + "))\\b");
16  }
17
18  var wordOperators = wordRegexp(["and", "or", "not", "is", "in"]);
19  var commonKeywords = ["as", "assert", "break", "class", "continue",
20                        "def", "del", "elif", "else", "except", "finally",
21                        "for", "from", "global", "if", "import",
22                        "lambda", "pass", "raise", "return",
23                        "try", "while", "with", "yield"];
24  var commonBuiltins = ["abs", "all", "any", "bin", "bool", "bytearray", "callable", "chr",
25                        "classmethod", "compile", "complex", "delattr", "dict", "dir", "divmod",
26                        "enumerate", "eval", "filter", "float", "format", "frozenset",
27                        "getattr", "globals", "hasattr", "hash", "help", "hex", "id",
28                        "input", "int", "isinstance", "issubclass", "iter", "len",
29                        "list", "locals", "map", "max", "memoryview", "min", "next",
30                        "object", "oct", "open", "ord", "pow", "property", "range",
31                        "repr", "reversed", "round", "set", "setattr", "slice",
32                        "sorted", "staticmethod", "str", "sum", "super", "tuple",
33                        "type", "vars", "zip", "__import__", "NotImplemented",
34                        "Ellipsis", "__debug__"];
35  var py2 = {builtins: ["apply", "basestring", "buffer", "cmp", "coerce", "execfile",
36                        "file", "intern", "long", "raw_input", "reduce", "reload",
37                        "unichr", "unicode", "xrange", "False", "True", "None"],
38             keywords: ["exec", "print"]};
39  var py3 = {builtins: ["ascii", "bytes", "exec", "print"],
40             keywords: ["nonlocal", "False", "True", "None"]};
41
42  CodeMirror.registerHelper("hintWords", "python", commonKeywords.concat(commonBuiltins));
43
44  function top(state) {
45    return state.scopes[state.scopes.length - 1];
46  }
47
48  CodeMirror.defineMode("python", function(conf, parserConf) {
49    var ERRORCLASS = "error";
50
51    var singleOperators = parserConf.singleOperators || new RegExp("^[\\+\\-\\*/%&|\\^~<>!]");
52    var singleDelimiters = parserConf.singleDelimiters || new RegExp("^[\\(\\)\\[\\]\\{\\}@,:`=;\\.]");
53    var doubleOperators = parserConf.doubleOperators || new RegExp("^((==)|(!=)|(<=)|(>=)|(<>)|(<<)|(>>)|(//)|(\\*\\*))");
54    var doubleDelimiters = parserConf.doubleDelimiters || new RegExp("^((\\+=)|(\\-=)|(\\*=)|(%=)|(/=)|(&=)|(\\|=)|(\\^=))");
55    var tripleDelimiters = parserConf.tripleDelimiters || new RegExp("^((//=)|(>>=)|(<<=)|(\\*\\*=))");
56    var identifiers = parserConf.identifiers|| new RegExp("^[_A-Za-z][_A-Za-z0-9]*");
57    var hangingIndent = parserConf.hangingIndent || conf.indentUnit;
58
59    var myKeywords = commonKeywords, myBuiltins = commonBuiltins;
60    if(parserConf.extra_keywords != undefined){
61      myKeywords = myKeywords.concat(parserConf.extra_keywords);
62    }
63    if(parserConf.extra_builtins != undefined){
64      myBuiltins = myBuiltins.concat(parserConf.extra_builtins);
65    }
66    if (parserConf.version && parseInt(parserConf.version, 10) == 3) {
67      myKeywords = myKeywords.concat(py3.keywords);
68      myBuiltins = myBuiltins.concat(py3.builtins);
69      var stringPrefixes = new RegExp("^(([rb]|(br))?('{3}|\"{3}|['\"]))", "i");
70    } else {
71      myKeywords = myKeywords.concat(py2.keywords);
72      myBuiltins = myBuiltins.concat(py2.builtins);
73      var stringPrefixes = new RegExp("^(([rub]|(ur)|(br))?('{3}|\"{3}|['\"]))", "i");
74    }
75    var keywords = wordRegexp(myKeywords);
76    var builtins = wordRegexp(myBuiltins);
77
78    // tokenizers
79    function tokenBase(stream, state) {
80      // Handle scope changes
81      if (stream.sol() && top(state).type == "py") {
82        var scopeOffset = top(state).offset;
83        if (stream.eatSpace()) {
84          var lineOffset = stream.indentation();
85          if (lineOffset > scopeOffset)
86            pushScope(stream, state, "py");
87          else if (lineOffset < scopeOffset && dedent(stream, state))
88            state.errorToken = true;
89          return null;
90        } else {
91          var style = tokenBaseInner(stream, state);
92          if (scopeOffset > 0 && dedent(stream, state))
93            style += " " + ERRORCLASS;
94          return style;
95        }
96      }
97      return tokenBaseInner(stream, state);
98    }
99
100    function tokenBaseInner(stream, state) {
101      if (stream.eatSpace()) return null;
102
103      var ch = stream.peek();
104
105      // Handle Comments
106      if (ch == "#") {
107        stream.skipToEnd();
108        return "comment";
109      }
110
111      // Handle Number Literals
112      if (stream.match(/^[0-9\.]/, false)) {
113        var floatLiteral = false;
114        // Floats
115        if (stream.match(/^\d*\.\d+(e[\+\-]?\d+)?/i)) { floatLiteral = true; }
116        if (stream.match(/^\d+\.\d*/)) { floatLiteral = true; }
117        if (stream.match(/^\.\d+/)) { floatLiteral = true; }
118        if (floatLiteral) {
119          // Float literals may be "imaginary"
120          stream.eat(/J/i);
121          return "number";
122        }
123        // Integers
124        var intLiteral = false;
125        // Hex
126        if (stream.match(/^0x[0-9a-f]+/i)) intLiteral = true;
127        // Binary
128        if (stream.match(/^0b[01]+/i)) intLiteral = true;
129        // Octal
130        if (stream.match(/^0o[0-7]+/i)) intLiteral = true;
131        // Decimal
132        if (stream.match(/^[1-9]\d*(e[\+\-]?\d+)?/)) {
133          // Decimal literals may be "imaginary"
134          stream.eat(/J/i);
135          // TODO - Can you have imaginary longs?
136          intLiteral = true;
137        }
138        // Zero by itself with no other piece of number.
139        if (stream.match(/^0(?![\dx])/i)) intLiteral = true;
140        if (intLiteral) {
141          // Integer literals may be "long"
142          stream.eat(/L/i);
143          return "number";
144        }
145      }
146
147      // Handle Strings
148      if (stream.match(stringPrefixes)) {
149        state.tokenize = tokenStringFactory(stream.current());
150        return state.tokenize(stream, state);
151      }
152
153      // Handle operators and Delimiters
154      if (stream.match(tripleDelimiters) || stream.match(doubleDelimiters))
155        return null;
156
157      if (stream.match(doubleOperators)
158          || stream.match(singleOperators)
159          || stream.match(wordOperators))
160        return "operator";
161
162      if (stream.match(singleDelimiters))
163        return null;
164
165      if (stream.match(keywords))
166        return "keyword";
167
168      if (stream.match(builtins))
169        return "builtin";
170
171      if (stream.match(/^(self|cls)\b/))
172        return "variable-2";
173
174      if (stream.match(identifiers)) {
175        if (state.lastToken == "def" || state.lastToken == "class")
176          return "def";
177        return "variable";
178      }
179
180      // Handle non-detected items
181      stream.next();
182      return ERRORCLASS;
183    }
184
185    function tokenStringFactory(delimiter) {
186      while ("rub".indexOf(delimiter.charAt(0).toLowerCase()) >= 0)
187        delimiter = delimiter.substr(1);
188
189      var singleline = delimiter.length == 1;
190      var OUTCLASS = "string";
191
192      function tokenString(stream, state) {
193        while (!stream.eol()) {
194          stream.eatWhile(/[^'"\\]/);
195          if (stream.eat("\\")) {
196            stream.next();
197            if (singleline && stream.eol())
198              return OUTCLASS;
199          } else if (stream.match(delimiter)) {
200            state.tokenize = tokenBase;
201            return OUTCLASS;
202          } else {
203            stream.eat(/['"]/);
204          }
205        }
206        if (singleline) {
207          if (parserConf.singleLineStringErrors)
208            return ERRORCLASS;
209          else
210            state.tokenize = tokenBase;
211        }
212        return OUTCLASS;
213      }
214      tokenString.isString = true;
215      return tokenString;
216    }
217
218    function pushScope(stream, state, type) {
219      var offset = 0, align = null;
220      if (type == "py") {
221        while (top(state).type != "py")
222          state.scopes.pop();
223      }
224      offset = top(state).offset + (type == "py" ? conf.indentUnit : hangingIndent);
225      if (type != "py" && !stream.match(/^(\s|#.*)*$/, false))
226        align = stream.column() + 1;
227      state.scopes.push({offset: offset, type: type, align: align});
228    }
229
230    function dedent(stream, state) {
231      var indented = stream.indentation();
232      while (top(state).offset > indented) {
233        if (top(state).type != "py") return true;
234        state.scopes.pop();
235      }
236      return top(state).offset != indented;
237    }
238
239    function tokenLexer(stream, state) {
240      var style = state.tokenize(stream, state);
241      var current = stream.current();
242
243      // Handle '.' connected identifiers
244      if (current == ".") {
245        style = stream.match(identifiers, false) ? null : ERRORCLASS;
246        if (style == null && state.lastStyle == "meta") {
247          // Apply 'meta' style to '.' connected identifiers when
248          // appropriate.
249          style = "meta";
250        }
251        return style;
252      }
253
254      // Handle decorators
255      if (current == "@")
256        return stream.match(identifiers, false) ? "meta" : ERRORCLASS;
257
258      if ((style == "variable" || style == "builtin")
259          && state.lastStyle == "meta")
260        style = "meta";
261
262      // Handle scope changes.
263      if (current == "pass" || current == "return")
264        state.dedent += 1;
265
266      if (current == "lambda") state.lambda = true;
267      if (current == ":" && !state.lambda && top(state).type == "py")
268        pushScope(stream, state, "py");
269
270      var delimiter_index = current.length == 1 ? "[({".indexOf(current) : -1;
271      if (delimiter_index != -1)
272        pushScope(stream, state, "])}".slice(delimiter_index, delimiter_index+1));
273
274      delimiter_index = "])}".indexOf(current);
275      if (delimiter_index != -1) {
276        if (top(state).type == current) state.scopes.pop();
277        else return ERRORCLASS;
278      }
279      if (state.dedent > 0 && stream.eol() && top(state).type == "py") {
280        if (state.scopes.length > 1) state.scopes.pop();
281        state.dedent -= 1;
282      }
283
284      return style;
285    }
286
287    var external = {
288      startState: function(basecolumn) {
289        return {
290          tokenize: tokenBase,
291          scopes: [{offset: basecolumn || 0, type: "py", align: null}],
292          lastStyle: null,
293          lastToken: null,
294          lambda: false,
295          dedent: 0
296        };
297      },
298
299      token: function(stream, state) {
300        var addErr = state.errorToken;
301        if (addErr) state.errorToken = false;
302        var style = tokenLexer(stream, state);
303
304        state.lastStyle = style;
305
306        var current = stream.current();
307        if (current && style)
308          state.lastToken = current;
309
310        if (stream.eol() && state.lambda)
311          state.lambda = false;
312        return addErr ? style + " " + ERRORCLASS : style;
313      },
314
315      indent: function(state, textAfter) {
316        if (state.tokenize != tokenBase)
317          return state.tokenize.isString ? CodeMirror.Pass : 0;
318
319        var scope = top(state);
320        var closing = textAfter && textAfter.charAt(0) == scope.type;
321        if (scope.align != null)
322          return scope.align - (closing ? 1 : 0);
323        else if (closing && state.scopes.length > 1)
324          return state.scopes[state.scopes.length - 2].offset;
325        else
326          return scope.offset;
327      },
328
329      lineComment: "#",
330      fold: "indent"
331    };
332    return external;
333  });
334
335  CodeMirror.defineMIME("text/x-python", "python");
336
337  var words = function(str) { return str.split(" "); };
338
339  CodeMirror.defineMIME("text/x-cython", {
340    name: "python",
341    extra_keywords: words("by cdef cimport cpdef ctypedef enum except"+
342                          "extern gil include nogil property public"+
343                          "readonly struct union DEF IF ELIF ELSE")
344  });
345
346});
347