Initial commit

1 files changed, 96 insertions, 0 deletions

diff --git a/parse-pattern.y b/parse-pattern.y
new file mode 100644
index 0000000..e5acd3c
--- /dev/null
+++ b/parse-pattern.y
@@ -0,0 +1,96 @@
+%{
+
+#include <stdio.h>
+
+#include "ast.h"
+#include "lex-pattern.h"
+
+void yyerror(const char *msg)
+{
+	fprintf(stderr, "%s\n", msg);
+}
+
+%}
+
+%output  "parse-pattern.c"
+%defines "parse-pattern.h"
+
+%union {
+	long  lval;
+	char *sval;
+	struct atom  *atom;
+	struct repeat repeat;
+}
+
+%token END 0 "end of file"
+
+%token T_BADLIT
+%token T_NOMEM
+
+%token T_ALPHABETIC
+%token T_NUMERIC
+%token T_UPPERCASE
+%token T_LOWERCASE
+%token T_CONTROL
+%token T_PUNCTUATION
+%token T_EVERYTHING
+
+%token          T_LPAREN  "("
+%token          T_RPAREN  ")"
+%token          T_COMMA   ","
+%token          T_PERIOD  "."
+%token <lval>   T_INTEGER
+%token <sval>   T_STRING
+
+%type  <atom>   alternation
+%type  <atom>   atom
+%type  <atom>   input
+%type  <atom>   sequence
+%type  <atom>   molecule
+
+%type  <repeat> repeat
+
+%%
+
+start:
+		input END                     { ast = $1; }
+	;
+
+input:
+		molecule
+	|	input molecule                { $$ = mkatom(&(struct atom) { .type = ATOM_SEQUENCE,   .u = { .children = { $1, $2 } } }); }
+	;
+
+molecule:
+		repeat sequence               { $$ = mkatom(&(struct atom) { .type = ATOM_REPETITION, .u = { .repeat = { .counts = $1, .child = $2 } } }); }
+	;
+
+repeat:
+		T_INTEGER T_PERIOD T_INTEGER  { $$ = (struct repeat) { $1, $3 }; }
+	|	T_INTEGER T_PERIOD            { $$ = (struct repeat) { $1, -1 }; }
+	|	T_PERIOD T_INTEGER            { $$ = (struct repeat) {  0, $2 }; }
+	|	T_PERIOD                      { $$ = (struct repeat) {  0, -1 }; }
+	|	T_INTEGER                     { $$ = (struct repeat) { $1, $1 }; }
+	;
+
+sequence:
+		atom
+	|	sequence atom                 { $$ = mkatom(&(struct atom) { .type = ATOM_SEQUENCE, .u = { .children = { $1, $2 } } }); }
+	;
+
+atom:
+		T_ALPHABETIC                  { $$ = mkatom(&(struct atom) { .type = ATOM_ALPHABETIC  }); }
+	|	T_NUMERIC                     { $$ = mkatom(&(struct atom) { .type = ATOM_NUMERIC     }); }
+	|	T_UPPERCASE                   { $$ = mkatom(&(struct atom) { .type = ATOM_UPPERCASE   }); }
+	|	T_LOWERCASE                   { $$ = mkatom(&(struct atom) { .type = ATOM_LOWERCASE   }); }
+	|	T_CONTROL                     { $$ = mkatom(&(struct atom) { .type = ATOM_CONTROL     }); }
+	|	T_PUNCTUATION                 { $$ = mkatom(&(struct atom) { .type = ATOM_PUNCTUATION }); }
+	|	T_EVERYTHING                  { $$ = mkatom(&(struct atom) { .type = ATOM_EVERYTHING  }); }
+	|	T_STRING                      { $$ = mkatom(&(struct atom) { .type = ATOM_LITERAL, .u = { .literal = $1 } }); }
+	|	T_LPAREN alternation T_RPAREN { $$ = $2; }
+	;
+
+alternation:
+		molecule
+	|	alternation T_COMMA molecule  { $$ = mkatom(&(struct atom) { .type = ATOM_ALTERNATION, .u = { .children = { $1, $3 } } }); }
+	;