author | Christian Urban <urbanc@in.tum.de> |
Tue, 06 Nov 2018 08:18:53 +0000 | |
changeset 598 | e3ad67cd5123 |
parent 594 | d40d7d7b85bc |
child 599 | 33c4b580092b |
permissions | -rw-r--r-- |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
1 |
import scala.language.implicitConversions |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
2 |
import scala.language.reflectiveCalls |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
3 |
|
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
4 |
/* Note, in the lectures I did not show the type consraint |
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
5 |
* I <% Seq[_] , which means that the input type I can be |
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
6 |
* treated, or seen, as a sequence. */ |
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
7 |
|
461 | 8 |
abstract class Parser[I <% Seq[_], T] { |
9 |
def parse(ts: I): Set[(T, I)] |
|
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
10 |
|
461 | 11 |
def parse_all(ts: I) : Set[T] = |
360
c6c574d2ca0c
update
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
185
diff
changeset
|
12 |
for ((head, tail) <- parse(ts); |
c6c574d2ca0c
update
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
185
diff
changeset
|
13 |
if (tail.isEmpty)) yield head |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
14 |
} |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
15 |
|
461 | 16 |
class SeqParser[I <% Seq[_], T, S](p: => Parser[I, T], |
17 |
q: => Parser[I, S]) extends Parser[I, (T, S)] { |
|
18 |
def parse(sb: I) = |
|
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
19 |
for ((head1, tail1) <- p.parse(sb); |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
20 |
(head2, tail2) <- q.parse(tail1)) yield ((head1, head2), tail2) |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
21 |
} |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
22 |
|
461 | 23 |
class AltParser[I <% Seq[_], T](p: => Parser[I, T], |
24 |
q: => Parser[I, T]) extends Parser[I, T] { |
|
25 |
def parse(sb: I) = p.parse(sb) ++ q.parse(sb) |
|
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
26 |
} |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
27 |
|
461 | 28 |
class FunParser[I <% Seq[_], T, S](p: => Parser[I, T], |
29 |
f: T => S) extends Parser[I, S] { |
|
30 |
def parse(sb: I) = |
|
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
31 |
for ((head, tail) <- p.parse(sb)) yield (f(head), tail) |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
32 |
} |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
33 |
|
183
b17eff695c7f
added new stuff
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
178
diff
changeset
|
34 |
// atomic parsers |
461 | 35 |
case class CharParser(c: Char) extends Parser[String, Char] { |
177
53def1fbf472
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
172
diff
changeset
|
36 |
def parse(sb: String) = |
462 | 37 |
if (sb != "" && sb.head == c) Set((c, sb.tail)) else Set() |
177
53def1fbf472
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
172
diff
changeset
|
38 |
} |
53def1fbf472
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
172
diff
changeset
|
39 |
|
588 | 40 |
import scala.util.matching.Regex |
41 |
case class RegexParser(reg: Regex) extends Parser[String, String] { |
|
42 |
def parse(sb: String) = reg.findPrefixMatchOf(sb) match { |
|
43 |
case None => Set() |
|
44 |
case Some(m) => Set((m.matched, m.after.toString)) |
|
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
45 |
} |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
46 |
} |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
47 |
|
588 | 48 |
val NumParser = RegexParser("[0-9]+".r) |
593 | 49 |
def StringParser(s: String) = RegexParser(Regex.quote(s).r) |
588 | 50 |
|
593 | 51 |
val NumParserInt = NumParser ==> (s => s.toInt) |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
52 |
|
360
c6c574d2ca0c
update
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
185
diff
changeset
|
53 |
// convenience |
462 | 54 |
implicit def string2parser(s: String) = StringParser(s) |
593 | 55 |
implicit def char2parser(c: Char) = CharParser(c) |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
56 |
|
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
57 |
implicit def ParserOps[I<% Seq[_], T](p: Parser[I, T]) = new { |
590 | 58 |
def | (q : => Parser[I, T]) = new AltParser[I, T](p, q) |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
59 |
def ==>[S] (f: => T => S) = new FunParser[I, T, S](p, f) |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
60 |
def ~[S] (q : => Parser[I, S]) = new SeqParser[I, T, S](p, q) |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
61 |
} |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
62 |
|
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
63 |
implicit def StringOps(s: String) = new { |
590 | 64 |
def | (q : => Parser[String, String]) = new AltParser[String, String](s, q) |
65 |
def | (r: String) = new AltParser[String, String](s, r) |
|
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
66 |
def ==>[S] (f: => String => S) = new FunParser[String, String, S](s, f) |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
67 |
def ~[S] (q : => Parser[String, S]) = |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
68 |
new SeqParser[String, String, S](s, q) |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
69 |
def ~ (r: String) = |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
70 |
new SeqParser[String, String, String](s, r) |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
71 |
} |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
72 |
|
588 | 73 |
|
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
74 |
lazy val Pal : Parser[String, String] = |
593 | 75 |
(("a" ~ Pal ~ "a") ==> { case ((x, y), z) => x + y + z } | |
76 |
("b" ~ Pal ~ "b") ==> { case ((x, y), z) => x + y + z } | "a" | "b" | "") |
|
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
77 |
|
586 | 78 |
Pal.parse_all("abaaaba") |
593 | 79 |
Pal.parse("abaaaba") |
586 | 80 |
|
531 | 81 |
println("Palindrome: " + Pal.parse_all("abaaaba")) |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
82 |
|
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
83 |
// well-nested parenthesis parser |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
84 |
lazy val P : Parser[String, String] = |
593 | 85 |
"(" ~ P ~ ")" ~ P ==> { case (((_, x), _), y) => "{" + x + "}" + y } | "" |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
86 |
|
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
87 |
P.parse_all("(((()()))())") |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
88 |
P.parse_all("(((()()))()))") |
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
89 |
P.parse_all(")(") |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
90 |
P.parse_all("()") |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
91 |
|
360
c6c574d2ca0c
update
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
185
diff
changeset
|
92 |
// arithmetic expressions |
586 | 93 |
|
367
04127a5aad23
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
366
diff
changeset
|
94 |
lazy val E: Parser[String, Int] = |
593 | 95 |
(T ~ "+" ~ E) ==> { case ((x, y), z) => x + z } | |
96 |
(T ~ "-" ~ E) ==> { case ((x, y), z) => x - z } | T |
|
470
8df654e9eb4e
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
467
diff
changeset
|
97 |
lazy val T: Parser[String, Int] = |
593 | 98 |
(F ~ "*" ~ T) ==> { case ((x, y), z) => x * z } | F |
367
04127a5aad23
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
366
diff
changeset
|
99 |
lazy val F: Parser[String, Int] = |
593 | 100 |
("(" ~ E ~ ")") ==> { case ((x, y), z) => y } | NumParserInt |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
101 |
|
586 | 102 |
|
593 | 103 |
println(E.parse_all("4*2+3")) |
104 |
println(E.parse_all("4*(2+3)")) |
|
594 | 105 |
println(E.parse_all("(4)*((2+3))")) |
586 | 106 |
println(E.parse_all("4/2+3")) |
367
04127a5aad23
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
366
diff
changeset
|
107 |
println(E.parse("1 + 2 * 3")) |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
108 |
println(E.parse_all("(1+2)+3")) |
531 | 109 |
println(E.parse_all("1+2+3")) |
110 |
||
586 | 111 |
|
112 |
||
531 | 113 |
// no left-recursion allowed, otherwise will loop |
114 |
lazy val EL: Parser[String, Int] = |
|
593 | 115 |
(EL ~ "+" ~ EL ==> { case ((x, y), z) => x + z} | |
116 |
EL ~ "*" ~ EL ==> { case ((x, y), z) => x * z} | |
|
117 |
"(" ~ EL ~ ")" ==> { case ((x, y), z) => y} | |
|
118 |
NumParserInt) |
|
531 | 119 |
|
593 | 120 |
//println(EL.parse_all("1+2+3")) |
531 | 121 |
|
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
122 |
|
462 | 123 |
|
124 |
||
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
125 |
// non-ambiguous vs ambiguous grammars |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
126 |
lazy val S : Parser[String, String] = |
593 | 127 |
("1" ~ S ~ S) ==> { case ((x, y), z) => x + y + z } | "" |
172
47b5c91eff47
added
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
diff
changeset
|
128 |
|
531 | 129 |
S.parse("1" * 15) |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
130 |
|
366
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
131 |
lazy val U : Parser[String, String] = |
593 | 132 |
("1" ~ U) ==> { case (x, y) => x + y } | "" |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
133 |
|
531 | 134 |
U.parse("1" * 15) |
135 |
||
366
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
136 |
U.parse("11") |
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
137 |
U.parse("11111") |
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
138 |
U.parse("11011") |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
139 |
|
531 | 140 |
U.parse_all("1" * 100) |
366
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
141 |
U.parse_all("1" * 100 + "0") |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
142 |
|
366
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
143 |
lazy val UCount : Parser[String, Int] = |
593 | 144 |
("1" ~ UCount) ==> { case (x, y) => y + 1 } | |
598 | 145 |
"" ==> { x => 0 } |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
146 |
|
366
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
147 |
UCount.parse("11111") |
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
148 |
UCount.parse_all("11111") |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
149 |
|
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
150 |
|
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
151 |
|
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
152 |
|
366
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
153 |
// Single Character parser |
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
154 |
lazy val One : Parser[String, String] = "1" |
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
155 |
lazy val Two : Parser[String, String] = "2" |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
156 |
|
366
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
157 |
One.parse("1") |
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
158 |
One.parse("111") |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
159 |
|
366
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
160 |
(One ~ One).parse("111") |
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
161 |
(One ~ One ~ One).parse("111") |
5a83336a9690
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
362
diff
changeset
|
162 |
(One ~ One ~ One ~ One).parse("1111") |
362
57ea439feaff
updated
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
360
diff
changeset
|
163 |
|
593 | 164 |
(One | Two).parse("111") |
467
b5ec11e89768
fixed bug
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
462
diff
changeset
|
165 |
|
b5ec11e89768
fixed bug
Christian Urban <christian dot urban at kcl dot ac dot uk>
parents:
462
diff
changeset
|
166 |
|
531 | 167 |