Added tagged definitions: `foo:: blah`

5 files changed, 21 insertions, 5 deletions

diff --git a/Lua/lbp.c b/Lua/lbp.c
index a8b573e..7306c86 100644
--- a/Lua/lbp.c
+++ b/Lua/lbp.c
@@ -72,7 +72,11 @@ static void push_matchstring(lua_State *L, match_t *m)
 
 static void set_capture_fields(lua_State *L, match_t *m, int *n, const char *start)
 {
-    if (m->pat->type == BP_CAPTURE) {
+    if (m->pat->type == BP_TAGGED) {
+        lua_pushlstring(L, m->pat->args.capture.name, m->pat->args.capture.namelen);
+        lua_setfield(L, -2, "__tag");
+        set_capture_fields(L, m->children[0], n, start);
+    } else if (m->pat->type == BP_CAPTURE) {
         if (m->pat->args.capture.namelen > 0) {
             lua_pushlstring(L, m->pat->args.capture.name, m->pat->args.capture.namelen);
             push_match(L, m->children[0], start);
diff --git a/json.c b/json.c
index 231ccce..297aa30 100644
--- a/json.c
+++ b/json.c
@@ -30,6 +30,9 @@ static int _json_match(const char *text, match_t *m, int comma, bool verbose)
     }
     printf("\",\"range\":[%ld,%ld]", m->start - text, m->end - text);
 
+    if (m->pat->type == BP_TAGGED)
+        printf(",\"tag\":\"%.*s\"", (int)m->pat->args.capture.namelen, m->pat->args.capture.name);
+
     if (m->children && (verbose || m->pat->type != BP_REF)) {
         printf(",\"children\":[");
         for (int i = 0; m->children && m->children[i]; i++)
diff --git a/match.c b/match.c
index fcea58b..2252fc3 100644
--- a/match.c
+++ b/match.c
@@ -256,7 +256,7 @@ static pat_t *get_prerequisite(match_ctx_t *ctx, pat_t *pat)
             if (p->args.repetitions.min == 0)
                 return p;
             p = p->args.repetitions.repeat_pat; break;
-        case BP_CAPTURE:
+        case BP_CAPTURE: case BP_TAGGED:
             p = p->args.capture.capture_pat; break;
         case BP_CHAIN: {
             pat_t *f = p->args.multiple.first;
@@ -538,7 +538,7 @@ static match_t *match(match_ctx_t *ctx, const char *str, pat_t *pat)
         match_t *after = match(ctx, str, pat->args.pat);
         return after ? new_match(pat, str, str, MATCHES(after)) : NULL;
     }
-    case BP_CAPTURE: {
+    case BP_CAPTURE: case BP_TAGGED: {
         match_t *p = match(ctx, str, pat->args.pat);
         return p ? new_match(pat, str, p->end, MATCHES(p)) : NULL;
     }
diff --git a/pattern.c b/pattern.c
index 78cf6e6..d0e364f 100644
--- a/pattern.c
+++ b/pattern.c
@@ -210,6 +210,7 @@ static pat_t *_bp_definition(const char *start, const char *end)
     const char *str = after_name(start, end);
     size_t namelen = (size_t)(str - start);
     if (!matchchar(&str, ':', false, end)) return NULL;
+    bool is_tagged = matchchar(&str, ':', false, end);
     pat_t *def = bp_pattern_nl(str, end, false);
     if (!def) parse_err(str, end, "Could not parse this definition.");
     str = def->end;
@@ -217,6 +218,13 @@ static pat_t *_bp_definition(const char *start, const char *end)
     pat_t *ret = new_pat(BP_DEFINITIONS, start, str, 0, -1);
     ret->args.def.name = start;
     ret->args.def.namelen = namelen;
+    if (is_tagged) { // `id:: foo` means define a rule named `id` that gives captures an `id` tag
+        pat_t *capture = new_pat(BP_TAGGED, def->start, def->end, def->min_matchlen, def->max_matchlen);
+        capture->args.capture.capture_pat = def;
+        capture->args.capture.name = start;
+        capture->args.capture.namelen = namelen;
+        def = capture;
+    }
     ret->args.def.meaning = def;
     ret->args.def.next_def = _bp_definition(after_spaces(str, true, end), end);
     if (ret->args.def.next_def)
@@ -679,7 +687,7 @@ void delete_pat(pat_t **at_pat, bool recursive)
         case BP_REPLACE:
             delete_pat(&pat->args.replace.pat, true);
             break;
-        case BP_CAPTURE:
+        case BP_CAPTURE: case BP_TAGGED:
             delete_pat(&pat->args.capture.capture_pat, true);
             break;
         case BP_NOT: case BP_AFTER: case BP_BEFORE:
diff --git a/pattern.h b/pattern.h
index 218a997..8dc40d2 100644
--- a/pattern.h
+++ b/pattern.h
@@ -35,7 +35,8 @@ enum pattype_e {
     BP_END_OF_LINE   = 23,
     BP_WORD_BOUNDARY = 24,
     BP_DEFINITIONS   = 25,
-    BP_LEFTRECURSION = 26,
+    BP_TAGGED        = 26,
+    BP_LEFTRECURSION = 27,
 };
 
 //