author | Christian Urban <christian.urban@kcl.ac.uk> |
Wed, 29 May 2024 13:25:30 +0100 | |
changeset 960 | c7009356ddd8 |
parent 959 | 64ec1884d860 |
child 961 | c0600f8b6427 |
permissions | -rw-r--r-- |
732 | 1 |
// Parser Combinators: Simple Version |
2 |
//==================================== |
|
742 | 3 |
// |
4 |
// Call with |
|
5 |
// |
|
6 |
// amm comb1.sc |
|
732 | 7 |
|
960 | 8 |
|
906 | 9 |
// Note, in the lectures I did not show the type bound |
960 | 10 |
// using is: I => Seq[_], which means that the input |
11 |
// type 'I' needs to be a sequence. |
|
732 | 12 |
|
960 | 13 |
type IsSeq[I] = I => Seq[?] |
941 | 14 |
|
960 | 15 |
abstract class Parser[I: IsSeq, T](using is: IsSeq[I]) { |
16 |
def parse(in: I): Set[(T, I)] |
|
732 | 17 |
|
18 |
def parse_all(in: I) : Set[T] = |
|
960 | 19 |
for ((hd, tl) <- parse(in); |
906 | 20 |
if is(tl).isEmpty) yield hd |
732 | 21 |
} |
22 |
||
954 | 23 |
|
732 | 24 |
// parser combinators |
25 |
||
941 | 26 |
|
799 | 27 |
// alternative parser |
960 | 28 |
class AltParser[I : IsSeq, T](p: => Parser[I, T], |
941 | 29 |
q: => Parser[I, T]) extends Parser[I, T] { |
960 | 30 |
def parse(in: I) = p.parse(in) ++ q.parse(in) |
799 | 31 |
} |
32 |
||
732 | 33 |
// sequence parser |
960 | 34 |
class SeqParser[I: IsSeq, T, S](p: => Parser[I, T], |
941 | 35 |
q: => Parser[I, S]) extends Parser[I, (T, S)] { |
960 | 36 |
def parse(in: I) = |
37 |
for ((hd1, tl1) <- p.parse(in); |
|
732 | 38 |
(hd2, tl2) <- q.parse(tl1)) yield ((hd1, hd2), tl2) |
39 |
} |
|
40 |
||
742 | 41 |
// map parser |
960 | 42 |
class MapParser[I : IsSeq, T, S](p: => Parser[I, T], |
954 | 43 |
f: T => S) extends Parser[I, S] { |
732 | 44 |
def parse(in: I) = for ((hd, tl) <- p.parse(in)) yield (f(hd), tl) |
45 |
} |
|
46 |
||
742 | 47 |
|
48 |
||
732 | 49 |
// an example of an atomic parser for characters |
50 |
case class CharParser(c: Char) extends Parser[String, Char] { |
|
960 | 51 |
def parse(in: String) = |
732 | 52 |
if (in != "" && in.head == c) Set((c, in.tail)) else Set() |
53 |
} |
|
54 |
||
954 | 55 |
val ap = CharParser('a') |
56 |
val bp = CharParser('b') |
|
57 |
||
58 |
val abp = SeqParser(ap, bp) |
|
59 |
MapParser(abp, ab => s"$ab").parse("abc") |
|
799 | 60 |
|
732 | 61 |
// an atomic parser for parsing strings according to a regex |
62 |
import scala.util.matching.Regex |
|
63 |
||
64 |
case class RegexParser(reg: Regex) extends Parser[String, String] { |
|
65 |
def parse(in: String) = reg.findPrefixMatchOf(in) match { |
|
66 |
case None => Set() |
|
960 | 67 |
case Some(m) => Set((m.matched, m.after.toString)) |
732 | 68 |
} |
69 |
} |
|
70 |
||
960 | 71 |
// atomic parsers for numbers and "verbatim" strings |
732 | 72 |
val NumParser = RegexParser("[0-9]+".r) |
73 |
def StrParser(s: String) = RegexParser(Regex.quote(s).r) |
|
74 |
||
960 | 75 |
NumParser.parse("123a123bc") |
906 | 76 |
StrParser("else").parse("elsethen") |
742 | 77 |
|
799 | 78 |
|
732 | 79 |
// NumParserInt transforms a "string integer" into a propper Int |
80 |
// (needs "new" because MapParser is not a case class) |
|
81 |
||
954 | 82 |
val NumParserInt = MapParser(NumParser, (s: String) => s.toInt) |
897 | 83 |
NumParserInt.parse("123abc") |
732 | 84 |
|
960 | 85 |
// the following string interpolation allows us to write |
86 |
// StrParser(_some_string_) more conveniently as |
|
732 | 87 |
// |
960 | 88 |
// p"<_some_string_>" |
732 | 89 |
|
960 | 90 |
extension (sc: StringContext) |
91 |
def p(args: Any*) = StrParser(sc.s(args*)) |
|
897 | 92 |
|
93 |
||
960 | 94 |
(p"else").parse("elsethen") |
799 | 95 |
|
732 | 96 |
// more convenient syntax for parser combinators |
941 | 97 |
extension [I: IsSeq, T](p: Parser[I, T]) { |
732 | 98 |
def ||(q : => Parser[I, T]) = new AltParser[I, T](p, q) |
99 |
def ~[S] (q : => Parser[I, S]) = new SeqParser[I, T, S](p, q) |
|
919 | 100 |
def map[S](f: => T => S) = new MapParser[I, T, S](p, f) |
732 | 101 |
} |
102 |
||
960 | 103 |
// simple example of transforming the |
948 | 104 |
// result into capital letters |
897 | 105 |
def toU(s: String) = s.map(_.toUpper) |
106 |
||
960 | 107 |
(p"else").map(toU(_)).parse("elseifthen") |
897 | 108 |
|
732 | 109 |
// these implicits allow us to use an infix notation for |
110 |
// sequences and alternatives; we also can write the usual |
|
111 |
// map for a MapParser |
|
112 |
||
113 |
||
114 |
// with this NumParserInt can now be written more conveniently |
|
115 |
// as: |
|
116 |
||
799 | 117 |
val NumParserInt2 = NumParser.map(_.toInt) |
732 | 118 |
|
954 | 119 |
val x = 1 + 3 |
732 | 120 |
|
121 |
// A parser for palindromes (just returns them as string) |
|
948 | 122 |
// since the parser is recursive it needs to be lazy |
803 | 123 |
lazy val Pal : Parser[String, String] = { |
960 | 124 |
(p"a" ~ Pal ~ p"a").map{ case ((x, y), z) => s"$x$y$z" } || |
125 |
(p"b" ~ Pal ~ p"b").map{ case ((x, y), z) => s"$x$y$z" } || |
|
896 | 126 |
p"a" || p"b" || p"" |
960 | 127 |
} |
732 | 128 |
|
129 |
// examples |
|
954 | 130 |
Pal.parse_all("abacaba") |
131 |
Pal.parse("abacaaba") |
|
732 | 132 |
|
133 |
println("Palindrome: " + Pal.parse_all("abaaaba")) |
|
134 |
||
960 | 135 |
// A parser for wellnested parentheses |
799 | 136 |
// |
137 |
// P ::= ( P ) P | epsilon |
|
138 |
// |
|
139 |
// (transforms '(' -> '{' , ')' -> '}' ) |
|
140 |
lazy val P : Parser[String, String] = { |
|
919 | 141 |
(p"(" ~ P ~ p")" ~ P).map{ case (((_, x), _), y) => "{" + x + "}" + y } || |
799 | 142 |
p"" |
960 | 143 |
} |
732 | 144 |
|
145 |
println(P.parse_all("(((()()))())")) |
|
146 |
println(P.parse_all("(((()()))()))")) |
|
147 |
println(P.parse_all(")(")) |
|
148 |
println(P.parse_all("()")) |
|
149 |
||
150 |
// A parser for arithmetic expressions (Terms and Factors) |
|
898 | 151 |
|
799 | 152 |
lazy val E: Parser[String, Int] = { |
732 | 153 |
(T ~ p"+" ~ E).map{ case ((x, _), z) => x + z } || |
799 | 154 |
(T ~ p"-" ~ E).map{ case ((x, _), z) => x - z } || T } |
155 |
lazy val T: Parser[String, Int] = { |
|
156 |
(F ~ p"*" ~ T).map{ case ((x, _), z) => x * z } || F } |
|
157 |
lazy val F: Parser[String, Int] = { |
|
158 |
(p"(" ~ E ~ p")").map{ case ((_, y), _) => y } || NumParserInt } |
|
956 | 159 |
|
160 |
println(E.parse_all("2*2*2")) |
|
732 | 161 |
println(E.parse_all("1+3+4")) |
162 |
println(E.parse("1+3+4")) |
|
163 |
println(E.parse_all("4*2+3")) |
|
164 |
println(E.parse_all("4*(2+3)")) |
|
954 | 165 |
println(E.parse_all("(4)*(((2+3)))")) |
732 | 166 |
println(E.parse_all("4/2+3")) |
167 |
println(E.parse("1 + 2 * 3")) |
|
168 |
println(E.parse_all("(1+2)+3")) |
|
799 | 169 |
println(E.parse_all("1+2+3")) |
732 | 170 |
|
171 |
||
172 |
// with parser combinators (and other parsing algorithms) |
|
173 |
// no left-recursion is allowed, otherwise the will loop |
|
174 |
||
960 | 175 |
lazy val EL: Parser[String, Int] = |
176 |
((EL ~ p"+" ~ EL).map{ case ((x, y), z) => x + z} || |
|
732 | 177 |
(EL ~ p"*" ~ EL).map{ case ((x, y), z) => x * z} || |
178 |
(p"(" ~ EL ~ p")").map{ case ((x, y), z) => y} || |
|
179 |
NumParserInt) |
|
180 |
||
181 |
// this will run forever: |
|
182 |
//println(EL.parse_all("1+2+3")) |
|
183 |
||
184 |
||
185 |
// non-ambiguous vs ambiguous grammars |
|
186 |
||
187 |
// ambiguous |
|
188 |
lazy val S : Parser[String, String] = |
|
189 |
(p"1" ~ S ~ S).map{ case ((x, y), z) => x + y + z } || p"" |
|
190 |
||
850 | 191 |
//println(time(S.parse("1" * 10))) |
192 |
//println(time(S.parse_all("1" * 10))) |
|
732 | 193 |
|
194 |
// non-ambiguous |
|
195 |
lazy val U : Parser[String, String] = |
|
196 |
(p"1" ~ U).map{ case (x, y) => x + y } || p"" |
|
197 |
||
850 | 198 |
//println(time(U.parse("1" * 10))) |
199 |
//println(time(U.parse_all("1" * 10))) |
|
732 | 200 |
println(U.parse("1" * 25)) |
201 |
||
202 |
U.parse("11") |
|
203 |
U.parse("11111") |
|
204 |
U.parse("11011") |
|
205 |
||
206 |
U.parse_all("1" * 100) |
|
207 |
U.parse_all("1" * 100 + "0") |
|
208 |
||
209 |
// you can see the difference in second example |
|
210 |
//S.parse_all("1" * 100) // succeeds |
|
211 |
//S.parse_all("1" * 100 + "0") // fails |
|
212 |
||
213 |
||
214 |
// A variant which counts how many 1s are parsed |
|
215 |
lazy val UCount : Parser[String, Int] = |
|
799 | 216 |
(p"1" ~ UCount).map{ case (_, y) => y + 1 } || p"".map{ _ => 0 } |
732 | 217 |
|
218 |
println(UCount.parse("11111")) |
|
219 |
println(UCount.parse_all("11111")) |
|
220 |
||
221 |
// Two single character parsers |
|
222 |
lazy val One : Parser[String, String] = p"a" |
|
223 |
lazy val Two : Parser[String, String] = p"b" |
|
224 |
||
225 |
One.parse("a") |
|
226 |
One.parse("aaa") |
|
227 |
||
228 |
// note how the pairs nest to the left with sequence parsers |
|
229 |
(One ~ One).parse("aaa") |
|
230 |
(One ~ One ~ One).parse("aaa") |
|
231 |
(One ~ One ~ One ~ One).parse("aaaa") |
|
232 |
||
233 |
(One || Two).parse("aaa") |
|
234 |
||
235 |
||
236 |
||
960 | 237 |
// a problem with the arithmetic expression parser: it |
742 | 238 |
// gets very slow with deeply nested parentheses |
732 | 239 |
|
919 | 240 |
println("A runtime problem") |
732 | 241 |
println(E.parse("1")) |
242 |
println(E.parse("(1)")) |
|
243 |
println(E.parse("((1))")) |
|
906 | 244 |
println(E.parse("(((1)))")) |
245 |
println(E.parse("((((1))))")) |
|
919 | 246 |
println(E.parse("((((((1))))))")) |
247 |
println(E.parse("(((((((1)))))))")) |
|
906 | 248 |
//println(E.parse("((((((((1))))))))")) |
828 | 249 |
|
250 |
||
919 | 251 |
// faster because of merge in the +/- case |
906 | 252 |
lazy val E2: Parser[String, Int] = { |
919 | 253 |
(T2 ~ (p"+" || p"-") ~ E2).map[Int]{ case ((x, y), z) => if (y == "+") x + z else x - z} || T2 } |
906 | 254 |
lazy val T2: Parser[String, Int] = { |
919 | 255 |
(F2 ~ p"*" ~ T2).map[Int]{ case ((x, _), z) => x * z } || F2 } |
906 | 256 |
lazy val F2: Parser[String, Int] = { |
919 | 257 |
(p"(" ~ E2 ~ p")").map[Int]{ case ((_, y), _) => y } || NumParserInt } |
828 | 258 |
|
259 |
||
919 | 260 |
println("mitigated by merging clauses") |
906 | 261 |
println(E2.parse("1")) |
262 |
println(E2.parse("(1)")) |
|
263 |
println(E2.parse("((1))")) |
|
264 |
println(E2.parse("(((1)))")) |
|
265 |
println(E2.parse("((((1))))")) |
|
919 | 266 |
println(E2.parse("((((((1))))))")) |
267 |
println(E2.parse("(((((((1)))))))")) |
|
937 | 268 |
println(E2.parse("((((((((1))))))))")) |
959
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
269 |
|
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
270 |
|
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
271 |
|
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
272 |
|
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
273 |
|
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
274 |
/* |
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
275 |
Try |
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
276 |
|
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
277 |
6 / 2 * (2+1) |
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
278 |
|
64ec1884d860
updated and added pascal.while file
Christian Urban <christian.urban@kcl.ac.uk>
parents:
956
diff
changeset
|
279 |
*/ |