1"""
2 pygments.lexers.minecraft
3 ~~~~~~~~~~~~~~~~~~~~~~~~~
4
5 Lexers for Minecraft related languages.
6
7 SNBT. A data communication format used in Minecraft.
8 wiki: https://minecraft.wiki/w/NBT_format
9
10 MCFunction. The Function file for Minecraft Data packs and Add-ons.
11 official: https://learn.microsoft.com/en-us/minecraft/creator/documents/functionsintroduction
12 wiki: https://minecraft.wiki/w/Function
13
14 MCSchema. A kind of data Schema for Minecraft Add-on Development.
15 official: https://learn.microsoft.com/en-us/minecraft/creator/reference/content/schemasreference/
16 community example: https://www.mcbe-dev.net/addons/data-driven/manifest.html
17
18 :copyright: Copyright 2006-2025 by the Pygments team, see AUTHORS.
19 :license: BSD, see LICENSE for details.
20"""
21
22from pygments.lexer import RegexLexer, default, include, bygroups
23from pygments.token import Comment, Keyword, Literal, Name, Number, Operator, \
24 Punctuation, String, Text, Whitespace
25
26__all__ = ['SNBTLexer', 'MCFunctionLexer', 'MCSchemaLexer']
27
28
29class SNBTLexer(RegexLexer):
30 """Lexer for stringified NBT, a data format used in Minecraft
31 """
32
33 name = "SNBT"
34 url = "https://minecraft.wiki/w/NBT_format"
35 aliases = ["snbt"]
36 filenames = ["*.snbt"]
37 mimetypes = ["text/snbt"]
38 version_added = '2.12'
39
40 tokens = {
41 "root": [
42 # We only look for the open bracket here since square bracket
43 # is only valid in NBT pathing (which is a mcfunction idea).
44 (r"\{", Punctuation, "compound"),
45 (r"[^\{]+", Text),
46 ],
47
48 "whitespace": [
49 (r"\s+", Whitespace),
50 ],
51
52 "operators": [
53 (r"[,:;]", Punctuation),
54 ],
55
56 "literals": [
57 (r"(true|false)", Keyword.Constant),
58 (r"-?\d+[eE]-?\d+", Number.Float),
59 (r"-?\d*\.\d+[fFdD]?", Number.Float),
60 (r"-?\d+[bBsSlLfFdD]?", Number.Integer),
61
62 # Separate states for both types of strings so they don't entangle
63 (r'"', String.Double, "literals.string_double"),
64 (r"'", String.Single, "literals.string_single"),
65 ],
66 "literals.string_double": [
67 (r"\\.", String.Escape),
68 (r'[^\\"\n]+', String.Double),
69 (r'"', String.Double, "#pop"),
70 ],
71 "literals.string_single": [
72 (r"\\.", String.Escape),
73 (r"[^\\'\n]+", String.Single),
74 (r"'", String.Single, "#pop"),
75 ],
76
77 "compound": [
78 # this handles the unquoted snbt keys
79 # note: stringified keys still work
80 (r"[A-Z_a-z]+", Name.Attribute),
81 include("operators"),
82 include("whitespace"),
83 include("literals"),
84 (r"\{", Punctuation, "#push"),
85 (r"\[", Punctuation, "list"),
86 (r"\}", Punctuation, "#pop"),
87 ],
88
89 "list": [
90 (r"[A-Z_a-z]+", Name.Attribute),
91 include("literals"),
92 include("operators"),
93 include("whitespace"),
94 (r"\[", Punctuation, "#push"),
95 (r"\{", Punctuation, "compound"),
96 (r"\]", Punctuation, "#pop"),
97 ],
98 }
99
100
101class MCFunctionLexer(RegexLexer):
102 """Lexer for the mcfunction scripting language used in Minecraft
103 Modelled somewhat after the `GitHub mcfunction grammar <https://github.com/Arcensoth/language-mcfunction>`_.
104 """
105
106 name = "MCFunction"
107 url = "https://minecraft.wiki/w/Commands"
108 aliases = ["mcfunction", "mcf"]
109 filenames = ["*.mcfunction"]
110 mimetypes = ["text/mcfunction"]
111 version_added = '2.12'
112
113 # Used to denotate the start of a block comment, borrowed from Github's mcfunction
114 _block_comment_prefix = "[>!]"
115
116 tokens = {
117 "root": [
118 include("names"),
119 include("comments"),
120 include("literals"),
121 include("whitespace"),
122 include("property"),
123 include("operators"),
124 include("selectors"),
125 ],
126
127 "names": [
128 # The start of a command (either beginning of line OR after the run keyword)
129 # We don't encode a list of keywords since mods, plugins, or even pre-processors
130 # may add new commands, so we have a 'close-enough' regex which catches them.
131 (r"^(\s*)([a-z_]+)", bygroups(Whitespace, Name.Builtin)),
132 (r"(?<=run)\s+[a-z_]+", Name.Builtin),
133
134 # UUID
135 (r"\b[0-9a-fA-F]+(?:-[0-9a-fA-F]+){4}\b", Name.Variable),
136 include("resource-name"),
137 # normal command names and scoreboards
138 # there's no way to know the differences unfortuntely
139 (r"[A-Za-z_][\w.#%$]+", Keyword.Constant),
140 (r"[#%$][\w.#%$]+", Name.Variable.Magic),
141 ],
142
143 "resource-name": [
144 # resource names have to be lowercase
145 (r"#?[a-z_][a-z_.-]*:[a-z0-9_./-]+", Name.Function),
146 # similar to above except optional `:``
147 # a `/` must be present "somewhere"
148 (r"#?[a-z0-9_\.\-]+\/[a-z0-9_\.\-\/]+", Name.Function),
149 ],
150
151 "whitespace": [
152 (r"\s+", Whitespace),
153 ],
154
155 "comments": [
156 (rf"^\s*(#{_block_comment_prefix})", Comment.Multiline,
157 ("comments.block", "comments.block.emphasized")),
158 (r"#.*$", Comment.Single),
159 ],
160 "comments.block": [
161 (rf"^\s*#{_block_comment_prefix}", Comment.Multiline,
162 "comments.block.emphasized"),
163 (r"^\s*#", Comment.Multiline, "comments.block.normal"),
164 default("#pop"),
165 ],
166 "comments.block.normal": [
167 include("comments.block.special"),
168 (r"\S+", Comment.Multiline),
169 (r"\n", Text, "#pop"),
170 include("whitespace"),
171 ],
172 "comments.block.emphasized": [
173 include("comments.block.special"),
174 (r"\S+", String.Doc),
175 (r"\n", Text, "#pop"),
176 include("whitespace"),
177 ],
178 "comments.block.special": [
179 # Params
180 (r"@\S+", Name.Decorator),
181
182 include("resource-name"),
183
184 # Scoreboard player names
185 (r"[#%$][\w.#%$]+", Name.Variable.Magic),
186 ],
187
188 "operators": [
189 (r"[\-~%^?!+*<>\\/|&=.]", Operator),
190 ],
191
192 "literals": [
193 (r"\.\.", Literal),
194 (r"(true|false)", Keyword.Pseudo),
195
196 # these are like unquoted strings and appear in many places
197 (r"[A-Za-z_]+", Name.Variable.Class),
198
199 (r"[0-7]b", Number.Byte),
200 (r"[+-]?\d*\.?\d+([eE]?[+-]?\d+)?[df]?\b", Number.Float),
201 (r"[+-]?\d+\b", Number.Integer),
202 (r'"', String.Double, "literals.string-double"),
203 (r"'", String.Single, "literals.string-single"),
204 ],
205 "literals.string-double": [
206 (r"\\.", String.Escape),
207 (r'[^\\"\n]+', String.Double),
208 (r'"', String.Double, "#pop"),
209 ],
210 "literals.string-single": [
211 (r"\\.", String.Escape),
212 (r"[^\\'\n]+", String.Single),
213 (r"'", String.Single, "#pop"),
214 ],
215
216 "selectors": [
217 (r"@[a-z]", Name.Variable),
218 ],
219
220
221 ## Generic Property Container
222 # There are several, differing instances where the language accepts
223 # specific contained keys or contained key, value pairings.
224 #
225 # Property Maps:
226 # - Starts with either `[` or `{`
227 # - Key separated by `:` or `=`
228 # - Deliminated by `,`
229 #
230 # Property Lists:
231 # - Starts with `[`
232 # - Deliminated by `,`
233 #
234 # For simplicity, these patterns match a generic, nestable structure
235 # which follow a key, value pattern. For normal lists, there's only keys.
236 # This allow some "illegal" structures, but we'll accept those for
237 # sake of simplicity
238 #
239 # Examples:
240 # - `[facing=up, powered=true]` (blockstate)
241 # - `[name="hello world", nbt={key: 1b}]` (selector + nbt)
242 # - `[{"text": "value"}, "literal"]` (json)
243 ##
244 "property": [
245 # This state gets included in root and also several substates
246 # We do this to shortcut the starting of new properties
247 # within other properties. Lists can have sublists and compounds
248 # and values can start a new property (see the `difficult_1.txt`
249 # snippet).
250 (r"\{", Punctuation, ("property.curly", "property.key")),
251 (r"\[", Punctuation, ("property.square", "property.key")),
252 ],
253 "property.curly": [
254 include("whitespace"),
255 include("property"),
256 (r"\}", Punctuation, "#pop"),
257 ],
258 "property.square": [
259 include("whitespace"),
260 include("property"),
261 (r"\]", Punctuation, "#pop"),
262
263 # lists can have sequences of items
264 (r",", Punctuation),
265 ],
266 "property.key": [
267 include("whitespace"),
268
269 # resource names (for advancements)
270 # can omit `:` to default `minecraft:`
271 # must check if there is a future equals sign if `:` is in the name
272 (r"#?[a-z_][a-z_\.\-]*\:[a-z0-9_\.\-/]+(?=\s*\=)", Name.Attribute, "property.delimiter"),
273 (r"#?[a-z_][a-z0-9_\.\-/]+", Name.Attribute, "property.delimiter"),
274
275 # unquoted NBT key
276 (r"[A-Za-z_\-\+]+", Name.Attribute, "property.delimiter"),
277
278 # quoted JSON or NBT key
279 (r'"', Name.Attribute, "property.delimiter", "literals.string-double"),
280 (r"'", Name.Attribute, "property.delimiter", "literals.string-single"),
281
282 # index for a list
283 (r"-?\d+", Number.Integer, "property.delimiter"),
284
285 default("#pop"),
286 ],
287 "property.key.string-double": [
288 (r"\\.", String.Escape),
289 (r'[^\\"\n]+', Name.Attribute),
290 (r'"', Name.Attribute, "#pop"),
291 ],
292 "property.key.string-single": [
293 (r"\\.", String.Escape),
294 (r"[^\\'\n]+", Name.Attribute),
295 (r"'", Name.Attribute, "#pop"),
296 ],
297 "property.delimiter": [
298 include("whitespace"),
299
300 (r"[:=]!?", Punctuation, "property.value"),
301 (r",", Punctuation),
302
303 default("#pop"),
304 ],
305 "property.value": [
306 include("whitespace"),
307
308 # unquoted resource names are valid literals here
309 (r"#?[a-z_][a-z_\.\-]*\:[a-z0-9_\.\-/]+", Name.Tag),
310 (r"#?[a-z_][a-z0-9_\.\-/]+", Name.Tag),
311
312 include("literals"),
313 include("property"),
314
315 default("#pop"),
316 ],
317 }
318
319
320class MCSchemaLexer(RegexLexer):
321 """Lexer for Minecraft Add-ons data Schemas, an interface structure standard used in Minecraft
322 """
323
324 name = 'MCSchema'
325 url = 'https://learn.microsoft.com/en-us/minecraft/creator/reference/content/schemasreference/'
326 aliases = ['mcschema']
327 filenames = ['*.mcschema']
328 mimetypes = ['text/mcschema']
329 version_added = '2.14'
330
331 tokens = {
332 'commentsandwhitespace': [
333 (r'\s+', Whitespace),
334 (r'//.*?$', Comment.Single),
335 (r'/\*.*?\*/', Comment.Multiline)
336 ],
337 'slashstartsregex': [
338 include('commentsandwhitespace'),
339 (r'/(\\.|[^[/\\\n]|\[(\\.|[^\]\\\n])*])+/'
340 r'([gimuysd]+\b|\B)', String.Regex, '#pop'),
341 (r'(?=/)', Text, ('#pop', 'badregex')),
342 default('#pop')
343 ],
344 'badregex': [
345 (r'\n', Whitespace, '#pop')
346 ],
347 'singlestring': [
348 (r'\\.', String.Escape),
349 (r"'", String.Single, '#pop'),
350 (r"[^\\']+", String.Single),
351 ],
352 'doublestring': [
353 (r'\\.', String.Escape),
354 (r'"', String.Double, '#pop'),
355 (r'[^\\"]+', String.Double),
356 ],
357 'root': [
358 (r'^(?=\s|/|<!--)', Text, 'slashstartsregex'),
359 include('commentsandwhitespace'),
360
361 # keywords for optional word and field types
362 (r'(?<=: )opt', Operator.Word),
363 (r'(?<=\s)[\w-]*(?=(\s+"|\n))', Keyword.Declaration),
364
365 # numeric literals
366 (r'0[bB][01]+', Number.Bin),
367 (r'0[oO]?[0-7]+', Number.Oct),
368 (r'0[xX][0-9a-fA-F]+', Number.Hex),
369 (r'\d+', Number.Integer),
370 (r'(\.\d+|\d+\.\d*|\d+)([eE][-+]?\d+)?', Number.Float),
371
372 # possible punctuations
373 (r'\.\.\.|=>', Punctuation),
374 (r'\+\+|--|~|\?\?=?|\?|:|\\(?=\n)|'
375 r'(<<|>>>?|==?|!=?|(?:\*\*|\|\||&&|[-<>+*%&|^/]))=?', Operator, 'slashstartsregex'),
376 (r'[{(\[;,]', Punctuation, 'slashstartsregex'),
377 (r'[})\].]', Punctuation),
378
379 # strings
380 (r"'", String.Single, 'singlestring'),
381 (r'"', String.Double, 'doublestring'),
382
383 # title line
384 (r'[\w-]*?(?=:\{?\n)', String.Symbol),
385 # title line with a version code, formatted
386 # `major.minor.patch-prerelease+buildmeta`
387 (r'([\w-]*?)(:)(\d+)(?:(\.)(\d+)(?:(\.)(\d+)(?:(\-)((?:[^\W_]|-)*(?:\.(?:[^\W_]|-)*)*))?(?:(\+)((?:[^\W_]|-)+(?:\.(?:[^\W_]|-)+)*))?)?)?(?=:\{?\n)', bygroups(String.Symbol, Operator, Number.Integer, Operator, Number.Integer, Operator, Number.Integer, Operator, String, Operator, String)),
388
389 (r'.*\n', Text),
390 ]
391 }