1 files changed, 51 insertions(+), 20 deletions(-)

M Src/yyparser.y

@@ 279,6 279,48 @@ static size_t try_parse_class(const char* str, size_t n, const char** out) {
     return 0;
 }
 
+struct escape_seq_kv { char from; const char* to; };
+struct escape_seq_kv escape_seq_map[] = {
+    { 't', "\t" },
+    { 'n', "\n" },
+    { 'r', "\r" },
+    { 'f', "\f" },
+    { 'a', "\a" },
+// Kdevelop doesn't know \e?
+//    { 'e', '\e' },
+    { 'b', "\b" }
+};
+
+static const char* try_parse_escape_seq(char in) {
+    struct escape_seq_kv* cur = escape_seq_map;
+    struct escape_seq_kv* end = escape_seq_map + sizeof(escape_seq_map) / sizeof(*escape_seq_map);
+    for (; cur != end; ++cur) {
+        if (in == cur->from) {
+            return cur->to;
+        }
+    }
+    return NULL;
+}
+
+static size_t try_parse_lit(const char* str, size_t n, Glyph* parsed) {
+    mbstate_t state;
+    memset(&state, '\0', sizeof(state));
+
+    size_t r = mbrtoc32(
+        parsed,
+        str,
+        n,
+        &state
+    );
+    switch (r) {
+        case (size_t)-1: assert(0 && "invalid input");
+        case (size_t)-2: assert(0 && "truncated input");
+        case (size_t)-3: assert(0 && "no surrogate pairs");
+        default: break;
+    }
+    return r;
+}
+
 static int yylex(YYSTYPE* lvalp, struct location* locp, void* ctx, struct slice* src) {
     struct parser_context* pctx = ctx;
     const char** pos = &src->begin;


@@ 301,6 343,14 @@ static int yylex(YYSTYPE* lvalp, struct location* locp, void* ctx, struct slice*
                 *pos += len;
                 return TOKEN_CLASS;
             }
+            const char* seq = try_parse_escape_seq(**pos);
+            if (seq) {
+                size_t seq_len = strlen(seq);
+                size_t r = try_parse_lit(seq, seq_len, &lvalp->glyph);
+                *pos += r;
+                return TOKEN_LIT;
+            }
+
             // rollback, so that we could re-parse it as glyph
             --(*pos);
             break;


@@ 321,29 371,10 @@ static int yylex(YYSTYPE* lvalp, struct location* locp, void* ctx, struct slice*
     }
 
     // TODO: provide escape/control sequences like \w \t \n \f etc
-
-    // this is a lit
-    mbstate_t state;
-    memset(&state, '\0', sizeof(state));
-
-    Glyph glyph;
-
-    size_t r = mbrtoc32(
-        &glyph,
-        *pos,
-        src->end - *pos,
-        &state
-    );
-    switch (r) {
-        case (size_t)-1: assert(0 && "invalid input");
-        case (size_t)-2: assert(0 && "truncated input");
-        case (size_t)-3: assert(0 && "no surrogate pairs");
-        default: break;
-    }
+    size_t r = try_parse_lit(*pos, src->end - *pos, &lvalp->glyph);
     *pos += r;
     locp->pos = *pos;
 
-    lvalp->glyph = glyph;
     return TOKEN_LIT;
 }