1 | /**
|
---|
2 | * @param {string} value
|
---|
3 | * @returns {RegExp}
|
---|
4 | * */
|
---|
5 |
|
---|
6 | /**
|
---|
7 | * @param {RegExp | string } re
|
---|
8 | * @returns {string}
|
---|
9 | */
|
---|
10 | function source(re) {
|
---|
11 | if (!re) return null;
|
---|
12 | if (typeof re === "string") return re;
|
---|
13 |
|
---|
14 | return re.source;
|
---|
15 | }
|
---|
16 |
|
---|
17 | /**
|
---|
18 | * @param {...(RegExp | string) } args
|
---|
19 | * @returns {string}
|
---|
20 | */
|
---|
21 | function concat(...args) {
|
---|
22 | const joined = args.map((x) => source(x)).join("");
|
---|
23 | return joined;
|
---|
24 | }
|
---|
25 |
|
---|
26 | /*
|
---|
27 | Language: LLVM IR
|
---|
28 | Author: Michael Rodler <contact@f0rki.at>
|
---|
29 | Description: language used as intermediate representation in the LLVM compiler framework
|
---|
30 | Website: https://llvm.org/docs/LangRef.html
|
---|
31 | Category: assembler
|
---|
32 | Audit: 2020
|
---|
33 | */
|
---|
34 |
|
---|
35 | /** @type LanguageFn */
|
---|
36 | function llvm(hljs) {
|
---|
37 | const IDENT_RE = /([-a-zA-Z$._][\w$.-]*)/;
|
---|
38 | const TYPE = {
|
---|
39 | className: 'type',
|
---|
40 | begin: /\bi\d+(?=\s|\b)/
|
---|
41 | };
|
---|
42 | const OPERATOR = {
|
---|
43 | className: 'operator',
|
---|
44 | relevance: 0,
|
---|
45 | begin: /=/
|
---|
46 | };
|
---|
47 | const PUNCTUATION = {
|
---|
48 | className: 'punctuation',
|
---|
49 | relevance: 0,
|
---|
50 | begin: /,/
|
---|
51 | };
|
---|
52 | const NUMBER = {
|
---|
53 | className: 'number',
|
---|
54 | variants: [
|
---|
55 | { begin: /0[xX][a-fA-F0-9]+/ },
|
---|
56 | { begin: /-?\d+(?:[.]\d+)?(?:[eE][-+]?\d+(?:[.]\d+)?)?/ }
|
---|
57 | ],
|
---|
58 | relevance: 0
|
---|
59 | };
|
---|
60 | const LABEL = {
|
---|
61 | className: 'symbol',
|
---|
62 | variants: [
|
---|
63 | { begin: /^\s*[a-z]+:/ }, // labels
|
---|
64 | ],
|
---|
65 | relevance: 0
|
---|
66 | };
|
---|
67 | const VARIABLE = {
|
---|
68 | className: 'variable',
|
---|
69 | variants: [
|
---|
70 | { begin: concat(/%/, IDENT_RE) },
|
---|
71 | { begin: /%\d+/ },
|
---|
72 | { begin: /#\d+/ },
|
---|
73 | ]
|
---|
74 | };
|
---|
75 | const FUNCTION = {
|
---|
76 | className: 'title',
|
---|
77 | variants: [
|
---|
78 | { begin: concat(/@/, IDENT_RE) },
|
---|
79 | { begin: /@\d+/ },
|
---|
80 | { begin: concat(/!/, IDENT_RE) },
|
---|
81 | { begin: concat(/!\d+/, IDENT_RE) },
|
---|
82 | // https://llvm.org/docs/LangRef.html#namedmetadatastructure
|
---|
83 | // obviously a single digit can also be used in this fashion
|
---|
84 | { begin: /!\d+/ }
|
---|
85 | ]
|
---|
86 | };
|
---|
87 |
|
---|
88 | return {
|
---|
89 | name: 'LLVM IR',
|
---|
90 | // TODO: split into different categories of keywords
|
---|
91 | keywords:
|
---|
92 | 'begin end true false declare define global ' +
|
---|
93 | 'constant private linker_private internal ' +
|
---|
94 | 'available_externally linkonce linkonce_odr weak ' +
|
---|
95 | 'weak_odr appending dllimport dllexport common ' +
|
---|
96 | 'default hidden protected extern_weak external ' +
|
---|
97 | 'thread_local zeroinitializer undef null to tail ' +
|
---|
98 | 'target triple datalayout volatile nuw nsw nnan ' +
|
---|
99 | 'ninf nsz arcp fast exact inbounds align ' +
|
---|
100 | 'addrspace section alias module asm sideeffect ' +
|
---|
101 | 'gc dbg linker_private_weak attributes blockaddress ' +
|
---|
102 | 'initialexec localdynamic localexec prefix unnamed_addr ' +
|
---|
103 | 'ccc fastcc coldcc x86_stdcallcc x86_fastcallcc ' +
|
---|
104 | 'arm_apcscc arm_aapcscc arm_aapcs_vfpcc ptx_device ' +
|
---|
105 | 'ptx_kernel intel_ocl_bicc msp430_intrcc spir_func ' +
|
---|
106 | 'spir_kernel x86_64_sysvcc x86_64_win64cc x86_thiscallcc ' +
|
---|
107 | 'cc c signext zeroext inreg sret nounwind ' +
|
---|
108 | 'noreturn noalias nocapture byval nest readnone ' +
|
---|
109 | 'readonly inlinehint noinline alwaysinline optsize ssp ' +
|
---|
110 | 'sspreq noredzone noimplicitfloat naked builtin cold ' +
|
---|
111 | 'nobuiltin noduplicate nonlazybind optnone returns_twice ' +
|
---|
112 | 'sanitize_address sanitize_memory sanitize_thread sspstrong ' +
|
---|
113 | 'uwtable returned type opaque eq ne slt sgt ' +
|
---|
114 | 'sle sge ult ugt ule uge oeq one olt ogt ' +
|
---|
115 | 'ole oge ord uno ueq une x acq_rel acquire ' +
|
---|
116 | 'alignstack atomic catch cleanup filter inteldialect ' +
|
---|
117 | 'max min monotonic nand personality release seq_cst ' +
|
---|
118 | 'singlethread umax umin unordered xchg add fadd ' +
|
---|
119 | 'sub fsub mul fmul udiv sdiv fdiv urem srem ' +
|
---|
120 | 'frem shl lshr ashr and or xor icmp fcmp ' +
|
---|
121 | 'phi call trunc zext sext fptrunc fpext uitofp ' +
|
---|
122 | 'sitofp fptoui fptosi inttoptr ptrtoint bitcast ' +
|
---|
123 | 'addrspacecast select va_arg ret br switch invoke ' +
|
---|
124 | 'unwind unreachable indirectbr landingpad resume ' +
|
---|
125 | 'malloc alloca free load store getelementptr ' +
|
---|
126 | 'extractelement insertelement shufflevector getresult ' +
|
---|
127 | 'extractvalue insertvalue atomicrmw cmpxchg fence ' +
|
---|
128 | 'argmemonly double',
|
---|
129 | contains: [
|
---|
130 | TYPE,
|
---|
131 | // this matches "empty comments"...
|
---|
132 | // ...because it's far more likely this is a statement terminator in
|
---|
133 | // another language than an actual comment
|
---|
134 | hljs.COMMENT(/;\s*$/, null, { relevance: 0 }),
|
---|
135 | hljs.COMMENT(/;/, /$/),
|
---|
136 | hljs.QUOTE_STRING_MODE,
|
---|
137 | {
|
---|
138 | className: 'string',
|
---|
139 | variants: [
|
---|
140 | // Double-quoted string
|
---|
141 | { begin: /"/, end: /[^\\]"/ },
|
---|
142 | ]
|
---|
143 | },
|
---|
144 | FUNCTION,
|
---|
145 | PUNCTUATION,
|
---|
146 | OPERATOR,
|
---|
147 | VARIABLE,
|
---|
148 | LABEL,
|
---|
149 | NUMBER
|
---|
150 | ]
|
---|
151 | };
|
---|
152 | }
|
---|
153 |
|
---|
154 | module.exports = llvm;
|
---|