Misc. tweaks and visualization changes, capitalized some rules

3 files changed, 87 insertions, 51 deletions

diff --git a/bpeg.bpeg b/bpeg.bpeg
index 4f55383..f5d490d 100644
--- a/bpeg.bpeg
+++ b/bpeg.bpeg
@@ -1,60 +1,66 @@
 # This is a file defining the BPEG grammar using BPEG syntax
 
-grammar;
-grammar = __ @[main-pattern]extended-pat __ *((__`;__) def) ?(`;__);
-def = @[name]ref __ `= __ @[definition]extended-pat;
+Grammar;
+Grammar = __ @[main-pattern]extended-pat __ *((__`;__) Def) ?(`;__);
+Def = @[name]Ref __ `= __ @[definition]extended-pat;
 
 # This is used for command line arguments:
-string-grammar = *(`\ pat ?`; / .);
+String-grammar = *(`\ pat ?`; / .);
 
-pat = empty / dot / string / char-range / char / escape-range / escape / no / anything-but
-    / upto-and / repeat / after / before / capture / replace / ref / parens;
+pat = Empty / Dot / String / Char-range / Char / Escape-range / Escape / No / Anything-but
+    / Upto-and / Repeat / After / Before / Capture / Replace / Ref / parens;
 
-empty = `/ >(__ (`)/`}));
-dot = `.;
-string = (
-        `" @[s]*(escape / ~`") `"
-      / `' @[s]*(escape / ~`') `'
+Empty = `/ >(__ (`)/`}));
+Dot = `.;
+String = (
+        `" @[s]*(Escape / ~`") `"
+      / `' @[s]*(Escape / ~`') `'
     );
-char-range = `` @[low]. `- @[high].;
-char = `` @[s].;
-escape-range = `\ @[low]escape-sequence `- @[high]escape-sequence;
-escape = `\ @[s]escape-sequence;
+Char-range = `` @[low]. `- @[high].;
+Char = `` @[s].;
+Escape-range = `\ @[low]escape-sequence `- @[high]escape-sequence;
+Escape = `\ @[s]escape-sequence;
 escape-sequence = (
         1-3 `0-7
       / `x 2 (`0-9/`a-f/`A-F)
       /`a/`b/`e/`n/`r/`t/`v / . / \n
     );
-no = `! _ @pat;
-anything-but = `~ ?`~ _ @pat;
-upto-and = `& ?`& _ @pat;
-repeat = (
-           @[min]int _ `- _ @[max]int
-      /    @[min]{=>"0"}    @[max]int _ `-
-      /    @[min]int _ `+   @[max](/)
-      /    @[min]@[max]int
-      / `+ @[min]{=>"1"}    @[max](/)
-      / `* @[min]{=>"0"}    @[max](/)
-      / `? @[min]{=>"0"}    @[max]{=>"1"}
+No = `! _ @pat;
+Anything-but = `~ ?`~ _ @pat;
+Upto-and = `& ?`& _ @pat;
+Repeat = (
+        @[min]int _ `- _ @[max]int
+      /{@[min]{=>"0"}=>} @[max]int _ `-
+      / @[min]int _ `+   @[max](/)
+      / @[min]@[max]int
+      /{@[min]{=>"1"}=>} `+ @[max](/)
+      /{@[min]{=>"0"}=>} `* @[max](/)
+      /{@[min]{=>"0"}=>} `? {@[max]{=>"1"}=>}
     ) _ @[repeat-pat]pat ?( __ `% __ @[sep]pat);
-after = `< _ pat;
-before = `> _ pat;
-capture = `@ ?(_ `[ @[capture-name]ref `]) _ @[capture]pat;
-replace = `{ __ (
-      ?(@[replace-pat]extended-pat __) "=>" ?(__ @[replacement]string)
+After = `< _ pat;
+Before = `> _ pat;
+Capture = `@ ?(_ `[ @[capture-name]Ref `]) _ @[capture]pat;
+Replace = `{ __ (
+      ?(@[replace-pat]extended-pat __) "=>" ?(__ @[replacement]String)
     ) __ `};
-ref = @[name](
+Ref = @[name](
         "^^" / "^" / "__" / "_" / "$$" / "$" /
         (`a-z/`A-Z) *(`a-z/`A-Z/`0-9/`-));
 
 parens = `( __ extended-pat __ `);
 
-chain = +@pat % (__);
-otherwise = +@(chain/pat) % (__`/__);
-extended-pat = otherwise / chain / pat;
+Chain = +@pat % (__);
+Otherwise = +@(Chain/pat) % (__`/__);
+extended-pat = Otherwise / Chain / pat;
 
-_ = *(`  / \t);
+# Special-symbol rules:
+_  = *(`  / \t);
 __ = *(`  / \t / \r / \n / comment);
+$$ = !$.;
+$  = !.;
+^^ = !<$.;
+^  = !<.;
+
 hash-comment = `# *.;
 
 # Note: comments are undefined by default in regular BPEG
diff --git a/bpeg.c b/bpeg.c
index 8a8c1f9..6607527 100644
--- a/bpeg.c
+++ b/bpeg.c
@@ -52,10 +52,11 @@ static match_t *free_match(match_t *m)
  */
 static match_t *match(const char *str, vm_op_t *op)
 {
-  tailcall:
+  //tailcall:
     switch (op->op) {
         case VM_EMPTY: {
             match_t *m = calloc(sizeof(match_t), 1);
+            m->op = op;
             m->start = str;
             m->end = str;
             return m;
@@ -64,6 +65,7 @@ static match_t *match(const char *str, vm_op_t *op)
             if (!*str || (!op->multiline && *str == '\n'))
                 return NULL;
             match_t *m = calloc(sizeof(match_t), 1);
+            m->op = op;
             m->start = str;
             m->end = str+1;
             return m;
@@ -72,6 +74,7 @@ static match_t *match(const char *str, vm_op_t *op)
             if (strncmp(str, op->args.s, op->len) != 0)
                 return NULL;
             match_t *m = calloc(sizeof(match_t), 1);
+            m->op = op;
             m->start = str;
             m->end = str + op->len;
             return m;
@@ -80,6 +83,7 @@ static match_t *match(const char *str, vm_op_t *op)
             if (*str < op->args.range.low || *str > op->args.range.high)
                 return NULL;
             match_t *m = calloc(sizeof(match_t), 1);
+            m->op = op;
             m->start = str;
             m->end = str + 1;
             return m;
@@ -94,6 +98,7 @@ static match_t *match(const char *str, vm_op_t *op)
                 return NULL;
             }
             m = calloc(sizeof(match_t), 1);
+            m->op = op;
             m->start = str;
             if (op->op == VM_ANYTHING_BUT) ++str;
             m->end = str;
@@ -102,6 +107,7 @@ static match_t *match(const char *str, vm_op_t *op)
         case VM_UPTO_AND: {
             match_t *m = calloc(sizeof(match_t), 1);
             m->start = str;
+            m->op = op;
             match_t *p = NULL;
             for (const char *prev = NULL; p == NULL && prev < str; ) {
                 prev = str;
@@ -121,6 +127,7 @@ static match_t *match(const char *str, vm_op_t *op)
             match_t *m = calloc(sizeof(match_t), 1);
             m->start = str;
             m->end = str;
+            m->op = op;
             if (op->args.repetitions.max == 0) return m;
 
             match_t **dest = &m->child;
@@ -171,6 +178,7 @@ static match_t *match(const char *str, vm_op_t *op)
             match_t *m = calloc(sizeof(match_t), 1);
             m->start = str;
             m->end = str;
+            m->op = op;
             return m;
         }
         case VM_BEFORE: {
@@ -180,6 +188,7 @@ static match_t *match(const char *str, vm_op_t *op)
             match_t *m = calloc(sizeof(match_t), 1);
             m->start = str;
             m->end = str;
+            m->op = op;
             return m;
         }
         case VM_CAPTURE: {
@@ -188,6 +197,7 @@ static match_t *match(const char *str, vm_op_t *op)
             match_t *m = calloc(sizeof(match_t), 1);
             m->start = str;
             m->end = p->end;
+            m->op = op;
             m->child = p;
             m->is_capture = 1;
             if (op->args.capture.name)
@@ -210,6 +220,7 @@ static match_t *match(const char *str, vm_op_t *op)
             match_t *m = calloc(sizeof(match_t), 1);
             m->start = str;
             m->end = m2->end;
+            m->op = op;
             m->child = m1;
             m1->nextsibling = m2;
             return m;
@@ -217,6 +228,7 @@ static match_t *match(const char *str, vm_op_t *op)
         case VM_REPLACE: {
             match_t *m = calloc(sizeof(match_t), 1);
             m->start = str;
+            m->op = op;
             if (op->args.replace.replace_pat) {
                 match_t *p = match(str, op->args.replace.replace_pat);
                 if (p == NULL) return NULL;
@@ -234,8 +246,20 @@ static match_t *match(const char *str, vm_op_t *op)
             for (int i = ndefs-1; i >= 0; i--) {
                 if (streq(defs[i].name, op->args.s)) {
                     // Bingo!
+                    /*
                     op = defs[i].op;
                     goto tailcall;
+                    */
+                    match_t *p = match(str, defs[i].op);
+                    if (p == NULL) return NULL;
+                    match_t *m = calloc(sizeof(match_t), 1);
+                    m->start = p->start;
+                    m->end = p->end;
+                    m->op = op;
+                    m->child = p;
+                    m->name_or_replacement = defs[i].name;
+                    m->is_ref = 1;
+                    return m;
                 }
             }
             check(0, "Unknown identifier: '%s'", op->args.s);
@@ -905,8 +929,11 @@ static void print_match(match_t *m, const char *color)
             }
         }
     } else {
-        if (m->is_capture && m->name_or_replacement)
-          printf("\033[0;2;33m[%s:", m->name_or_replacement);
+        const char *name = m->name_or_replacement;
+        if (verbose && m->is_ref && name && isupper(name[0]))
+            printf("\033[0;2;35m{%s:", name);
+        //if (m->is_capture && name)
+        //    printf("\033[0;2;33m[%s:", name);
         const char *prev = m->start;
         for (match_t *child = m->child; child; child = child->nextsibling) {
             if (child->start > prev)
@@ -916,8 +943,10 @@ static void print_match(match_t *m, const char *color)
         }
         if (m->end > prev)
             printf("%s%.*s", color, (int)(m->end - prev), prev);
-        if (m->is_capture && m->name_or_replacement)
-          printf("\033[0;2;33m]");
+        if (verbose && m->is_ref && name && isupper(name[0]))
+            printf("\033[0;2;35m}");
+        //if (m->is_capture && name)
+        //    printf("\033[0;2;33m]");
     }
 }
 
diff --git a/bpeg.h b/bpeg.h
index 7836250..459bd65 100644
--- a/bpeg.h
+++ b/bpeg.h
@@ -19,16 +19,6 @@ const char *usage = (
     "  -s --slow\t run in slow mode for debugging\n"
     "  -r --replace <replacement>   replace the input pattern with the given replacement\n"
     "  -g --grammar <grammar file>  use the specified file as a grammar\n");
-/*
- * Pattern matching result object
- */
-typedef struct match_s {
-    // Where the match starts and ends (end is after the last character)
-    const char *start, *end;
-    unsigned int is_capture:1, is_replacement:1;
-    const char *name_or_replacement;
-    struct match_s *child, *nextsibling;
-} match_t;
 
 /*
  * BPEG virtual machine opcodes
@@ -85,6 +75,17 @@ typedef struct vm_op_s {
     } args;
 } vm_op_t;
 
+/*
+ * Pattern matching result object
+ */
+typedef struct match_s {
+    // Where the match starts and ends (end is after the last character)
+    const char *start, *end;
+    unsigned int is_capture:1, is_replacement:1, is_ref:1;
+    const char *name_or_replacement;
+    struct match_s *child, *nextsibling;
+    vm_op_t *op;
+} match_t;
 
 static inline const char *after_spaces(const char *str);
 static match_t *free_match(match_t *m);