1 files changed, 47 insertions(+), 21 deletions(-)

M Src/yyparser.y

@@ 15,10 15,11 @@
 #include <slice.h>
 #include <ast.h>
 #include <parser_context.h>
+#include <stdio.h>
 
 struct location {
-    Byte* pos;
-    Byte* end;
+    const Byte* pos;
+    const Byte* end;
 };
 
 #define YYLLOC_DEFAULT(Cur, Rhs, N)                                          \


@@ 53,12 54,13 @@ struct location {
     struct set_piece set_elem;
     Glyph            glyph;
     char             repeat;
+    const char*      class_;
 }
 
 %token          TOKEN_EOF
 %token <glyph>  TOKEN_LIT
 %token <repeat> TOKEN_REPEAT
-%token <class>  TOKEN_CLASS
+%token <class_> TOKEN_CLASS
 
 %type  <node>     regex seq expr plain_expr repeatable_expr repeated_expr or atom
 %type  <set>      set set_expr_list


@@ 240,29 242,44 @@ set_expr_list: set_expr
              }
              ;
 
-class: '\\' class_
+class: TOKEN_CLASS
+     {
+         fprintf(stderr, "Got class: %s\n", $1);
+     }
      ;
 
-class_: 'a'
-      | 'b'
-      | 'B'
-      | 'e'
-      | 'f'
-      | 'v'
-      | 'd'
-      | 'D'
-      | 's'
-      | 'S'
-      | 'w'
-      | 'W'
-      ;
-
 
 %%
 
 #include <stdio.h>
 
-int yylex(YYSTYPE* lvalp, struct location* locp, void* ctx, struct slice* src) {
+const char* known_classes[] = {
+    "e",
+    "v",
+    "d",
+    "D",
+    "s",
+    "S",
+    "w",
+    "W",
+    "N"
+};
+
+static size_t try_parse_class(const char* str, size_t n, const char** out) {
+    const char** cur = known_classes;
+    const char** end = known_classes + sizeof(known_classes) / sizeof(*known_classes);
+    size_t       cur_len;
+    for (; cur != end; ++cur) {
+        cur_len = strlen(*cur);
+        if (cur_len <= n && strncmp(*cur, str, cur_len) == 0) {
+            *out = *cur;
+            return cur_len;
+        }
+    }
+    return 0;
+}
+
+static int yylex(YYSTYPE* lvalp, struct location* locp, void* ctx, struct slice* src) {
     struct parser_context* pctx = ctx;
     const char** pos = &src->begin;
 


@@ 272,15 289,22 @@ int yylex(YYSTYPE* lvalp, struct location* locp, void* ctx, struct slice* src) {
         return YYEOF;
     }
 
-    int escaped = 0;
     switch (**pos) {
         case '\\':
-            escaped = 1;
+        {
             ++(*pos);
             if (*pos == src->end) {
                 return YYEOF;
             }
+            size_t len = try_parse_class(*pos, src->end - *pos, &lvalp->class_);
+            if (len) {
+                *pos += len;
+                return TOKEN_CLASS;
+            }
+            // rollback, so that we could re-parse it as glyph
+            --(*pos);
             break;
+        }
         case '|': ++(*pos); return '|';
         case '+':
         case '*':


@@ 296,6 320,8 @@ int yylex(YYSTYPE* lvalp, struct location* locp, void* ctx, struct slice* src) {
         default: break;
     }
 
+    // TODO: provide escape/control sequences like \w \t \n \f etc
+
     // this is a lit
     mbstate_t state;
     memset(&state, '\0', sizeof(state));