From a82164505e89dc8257ef87844dfef1476e235a7f Mon Sep 17 00:00:00 2001
From: Bruce Hill <bruce@bruce-hill.com>
Date: Mon, 14 Sep 2020 01:21:49 -0700
Subject: Added nodent support (|)

---
 bpeg.bpeg              |  3 ++-
 compiler.c             | 15 ++++++++++-----
 grammars/builtins.bpeg |  9 ++++++---
 types.h                |  1 +
 utils.c                |  6 +++---
 vm.c                   | 30 ++++++++++++++++++++++++++++++
 6 files changed, 52 insertions(+), 12 deletions(-)

diff --git a/bpeg.bpeg b/bpeg.bpeg
index ae2d3fb..de0c676 100644
--- a/bpeg.bpeg
+++ b/bpeg.bpeg
@@ -8,7 +8,7 @@ String-pattern = *(`\ pat ?`; / .);
 
 pat = suffixed-pat / simple-pat;
 simple-pat = Empty / Upto / Dot / String / Char-range / Char / Escape-range / Escape / No
-    / Repeat / After / Before / Capture / Replace / Ref / parens;
+    / Nodent / Repeat / After / Before / Capture / Replace / Ref / parens;
 suffixed-pat = Eq-pat;
 
 Eq-pat = @[first]simple-pat "==" @[second]pat;
@@ -29,6 +29,7 @@ escape-sequence = (
       /`a/`b/`e/`n/`r/`t/`v / . / \n
     );
 No = `! _ @pat;
+Nodent = `|;
 Upto = 2-3`. ?>(_@pat);
 Repeat = (
         @[min]int _ `- _ @[max]int
diff --git a/compiler.c b/compiler.c
index 267ea27..8279285 100644
--- a/compiler.c
+++ b/compiler.c
@@ -211,6 +211,7 @@ vm_op_t *bpeg_simplepattern(const char *str)
                 str = sep->end;
                 set_range(op, min, max, pat, sep);
             } else {
+                str = pat->end;
                 set_range(op, min, max, pat, NULL);
             }
             break;
@@ -233,6 +234,7 @@ vm_op_t *bpeg_simplepattern(const char *str)
                 str = sep->end;
                 set_range(op, min, max, pat, sep);
             } else {
+                str = pat->end;
                 set_range(op, min, max, pat, NULL);
             }
             break;
@@ -345,8 +347,8 @@ vm_op_t *bpeg_simplepattern(const char *str)
         }
         // Empty choice (/) or {/}
         case '/': {
-            str = after_spaces(str);
-            if (*str == ')' || *str == '}') {
+            const char *next = after_spaces(str);
+            if (*next == ')' || *next == '}') {
                 op->op = VM_EMPTY;
             } else {
                 free(op);
@@ -354,6 +356,10 @@ vm_op_t *bpeg_simplepattern(const char *str)
             }
             break;
         }
+        case '|': {
+            op->op = VM_NODENT;
+            break;
+        }
         default: {
             // Reference
             if (isalpha(c)) {
@@ -373,9 +379,8 @@ vm_op_t *bpeg_simplepattern(const char *str)
 
     // Postfix operators:
   postfix:
-    str = after_spaces(str);
-    if (strncmp(str, "==", 2) == 0) {
-        str += 2;
+    if (strncmp(after_spaces(str), "==", 2) == 0) {
+        str = after_spaces(str)+2;
         vm_op_t *first = op;
         vm_op_t *second = bpeg_simplepattern(str);
         check(second, "Expected pattern after '=='");
diff --git a/grammars/builtins.bpeg b/grammars/builtins.bpeg
index c871408..df03104 100644
--- a/grammars/builtins.bpeg
+++ b/grammars/builtins.bpeg
@@ -2,12 +2,15 @@
 pattern = !(/); # Not defined by default
 replacement = !(/); # Not defined by default
 replace-all = +(...@replacement) ...;
-find-all = {... >matching-line =>} +(matching-line/non-matching-line) ?{!<\n => "\n"};
+find-all = +find-next%\n ?{!<\n => "\n"};
+find-next = matching-line / {..\n =>} find-next;
 only-matches = +{...@pattern=>'@1\n'};
-matching-line = +(..@pattern)..$ ?\n;
-non-matching-line = {..$=>};
+matching-line = +(..@pattern) ..$;
 
 # Helper definitions (commonly used)
+indent = \n |+(\t/' ');
+dedent = >(\n !|);
+indented-block = |` ..$ *(\n|..$);
 crlf = \r\n;
 cr = \r; r = \r;
 anglebraces = `< *(anglebraces / !`>.) `>;
diff --git a/types.h b/types.h
index c8f7a17..36e1aa6 100644
--- a/types.h
+++ b/types.h
@@ -26,6 +26,7 @@ enum VMOpcode {
     VM_REPLACE,
     VM_REF,
     VM_BACKREF,
+    VM_NODENT,
 };
 
 /*
diff --git a/utils.c b/utils.c
index 5078c01..7844934 100644
--- a/utils.c
+++ b/utils.c
@@ -46,9 +46,9 @@ const char *after_name(const char *str)
  */
 int matchchar(const char **str, char c)
 {
-    *str = after_spaces(*str);
-    if (**str == c) {
-        ++(*str);
+    const char *next = after_spaces(*str);
+    if (*next == c) {
+        *str = &next[1];
         return 1;
     } else {
         return 0;
diff --git a/vm.c b/vm.c
index 2a44dd7..9974547 100644
--- a/vm.c
+++ b/vm.c
@@ -30,6 +30,7 @@ static const char *opcode_names[] = {
     [VM_EQUAL] = "EQUAL",
     [VM_REF] = "REF",
     [VM_BACKREF] = "BACKREF",
+    [VM_NODENT] = "NODENT",
 };
 
 const char *opcode_name(enum VMOpcode o)
@@ -337,6 +338,35 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, recursive_ref
         case VM_BACKREF: {
             return match_backref(str, op, (match_t*)op->args.backref);
         }
+        case VM_NODENT: {
+            if (str[-1] == '\0') { // First line
+                match_t *m = calloc(sizeof(match_t), 1);
+                m->start = str;
+                m->end = str;
+                m->op = op;
+                return m;
+            } else if (str[-1] != '\n') {
+                return NULL; // Not at beginning of line
+            }
+            const char *p = &str[-1];
+            while (*p == '\n') --p; // Skip blank lines
+            while (p[-1] && p[-1] != '\n') --p; // Backtrack to start of last (nonblank) line
+            // Count indentation:
+            char denter = *p;
+            int dents = 0;
+            if (denter == ' ' || denter == '\t') {
+                for (; *p == denter; ++p) ++dents;
+            }
+            for (int i = 0; i < dents; i++) {
+                if (str[i] != denter) return NULL;
+            }
+
+            match_t *m = calloc(sizeof(match_t), 1);
+            m->start = str;
+            m->end = &str[dents];
+            m->op = op;
+            return m;
+        }
         default: {
             fprintf(stderr, "Unknown opcode: %d", op->op);
             _exit(1);
-- 
cgit v1.2.3