Commit | Line | Data |
---|---|---|
53c2ea70 JFI |
1 | package mal |
2 | ||
3 | import kotlin.text.Regex | |
4 | ||
5 | val TOKEN_REGEX = Regex("[\\s,]*(~@|[\\[\\]{}()'`~^@]|\"(?:\\\\.|[^\\\\\"])*\"|;.*|[^\\s\\[\\]{}('\"`,;)]*)") | |
6 | val ATOM_REGEX = Regex("(^-?[0-9]+$)|(^nil$)|(^true$)|(^false$)|^\"(.*)\"$|:(.*)|(^[^\"]*$)") | |
7 | ||
8 | class Reader(sequence: Sequence<String>) { | |
9 | val tokens = sequence.iterator() | |
10 | var current = advance() | |
11 | ||
12 | fun next(): String? { | |
13 | var result = current | |
14 | current = advance() | |
15 | return result | |
16 | } | |
17 | ||
18 | fun peek(): String? = current | |
19 | ||
20 | private fun advance(): String? = if (tokens.hasNext()) tokens.next() else null | |
21 | } | |
22 | ||
23 | fun read_str(input: String?): MalType { | |
24 | val tokens = tokenizer(input) ?: return NIL | |
25 | return read_form(Reader(tokens)) | |
26 | } | |
27 | ||
28 | fun tokenizer(input: String?): Sequence<String>? { | |
29 | if (input == null) return null | |
30 | ||
31 | return TOKEN_REGEX.findAll(input) | |
32 | .map({ it -> it.groups[1]?.value as String }) | |
33 | .filter({ it != "" && !it.startsWith(";")}) | |
34 | } | |
35 | ||
36 | fun read_form(reader: Reader): MalType = | |
37 | when (reader.peek()) { | |
38 | null -> throw MalContinue() | |
940bb6ff JFI |
39 | "(" -> read_list(reader) |
40 | ")" -> throw MalReaderException("expected form, got ')'") | |
41 | "[" -> read_vector(reader) | |
42 | "]" -> throw MalReaderException("expected form, got ']'") | |
43 | "{" -> read_hashmap(reader) | |
44 | "}" -> throw MalReaderException("expected form, got '}'") | |
45 | "'" -> read_shorthand(reader, "quote") | |
46 | "`" -> read_shorthand(reader, "quasiquote") | |
47 | "~" -> read_shorthand(reader, "unquote") | |
48 | "~@" -> read_shorthand(reader, "splice-unquote") | |
2ec7fa0b JFI |
49 | "^" -> { |
50 | reader.next() | |
51 | ||
52 | val meta = read_form(reader) | |
53 | val obj = read_form(reader) | |
54 | ||
55 | val list = MalList() | |
56 | list.conj_BANG(MalSymbol("with-meta")) | |
57 | list.conj_BANG(obj) | |
58 | list.conj_BANG(meta) | |
59 | ||
60 | list | |
61 | } | |
53c2ea70 JFI |
62 | else -> read_atom(reader) |
63 | } | |
64 | ||
65 | fun read_list(reader: Reader): MalType = read_sequence(reader, MalList(), ")") | |
66 | fun read_vector(reader: Reader): MalType = read_sequence(reader, MalVector(), "]") | |
67 | ||
68 | private fun read_sequence(reader: Reader, sequence: IMutableSeq, end: String): MalType { | |
69 | reader.next() | |
70 | ||
71 | do { | |
72 | val form = when (reader.peek()) { | |
73 | null -> throw MalReaderException("expected '$end', got EOF") | |
74 | end -> { reader.next(); null } | |
75 | else -> read_form(reader) | |
76 | } | |
77 | ||
78 | if (form != null) { | |
79 | sequence.conj_BANG(form) | |
80 | } | |
81 | } while (form != null) | |
82 | ||
83 | return sequence | |
84 | } | |
85 | ||
86 | fun read_hashmap(reader: Reader): MalType { | |
87 | reader.next() | |
88 | val hashMap = MalHashMap() | |
89 | ||
90 | do { | |
91 | var value : MalType? = null; | |
92 | val key = when (reader.peek()) { | |
93 | null -> throw MalReaderException("expected '}', got EOF") | |
94 | "}" -> { reader.next(); null } | |
95 | else -> { | |
96 | var key = read_form(reader) | |
97 | if (key !is MalString) { | |
98 | throw MalReaderException("hash-map keys must be strings or keywords") | |
99 | } | |
100 | value = when (reader.peek()) { | |
101 | null -> throw MalReaderException("expected form, got EOF") | |
102 | else -> read_form(reader) | |
103 | } | |
104 | key | |
105 | } | |
106 | } | |
107 | ||
108 | if (key != null) { | |
109 | hashMap.assoc_BANG(key as MalString, value as MalType) | |
110 | } | |
111 | } while (key != null) | |
112 | ||
113 | return hashMap | |
114 | } | |
115 | ||
940bb6ff JFI |
116 | fun read_shorthand(reader: Reader, symbol: String): MalType { |
117 | reader.next() | |
118 | ||
119 | val list = MalList() | |
120 | list.conj_BANG(MalSymbol(symbol)) | |
121 | list.conj_BANG(read_form(reader)) | |
122 | ||
123 | return list | |
124 | } | |
125 | ||
53c2ea70 JFI |
126 | fun read_atom(reader: Reader): MalType { |
127 | val next = reader.next() ?: throw MalReaderException("Unexpected null token") | |
128 | val groups = ATOM_REGEX.find(next)?.groups ?: throw MalReaderException("Unrecognized token: " + next) | |
129 | ||
130 | return if (groups[1]?.value != null) { | |
131 | MalInteger(Integer.valueOf(groups[1]?.value)) | |
132 | } else if (groups[2]?.value != null) { | |
133 | NIL | |
134 | } else if (groups[3]?.value != null) { | |
135 | TRUE | |
136 | } else if (groups[4]?.value != null) { | |
137 | FALSE | |
138 | } else if (groups[5]?.value != null) { | |
139 | MalString((groups[5]?.value as String).replace("\\n", "\n").replace("\\\"", "\"")) | |
140 | } else if (groups[6]?.value != null) { | |
141 | MalKeyword(groups[6]?.value as String) | |
142 | } else if (groups[7]?.value != null) { | |
143 | MalSymbol(groups[7]?.value as String) | |
144 | } else { | |
145 | throw MalReaderException("Unrecognized token: " + next) | |
146 | } | |
147 | } |