From 98df52bf69b6f2d838264902de21369013602102 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Paul=20H=C3=A4nsch?= <paul@plutz.net>
Date: Fri, 30 Aug 2024 19:18:04 +0200
Subject: [PATCH] modify regexes for use in mawk

---
 markdown.awk | 30 +++++++++++++++++++-----------
 1 file changed, 19 insertions(+), 11 deletions(-)

diff --git a/markdown.awk b/markdown.awk
index fc2f203..6e3febe 100755
--- a/markdown.awk
+++ b/markdown.awk
@@ -701,23 +701,25 @@ function _block( block, LOCAL, st, len, text, title, attrib, href, guard, code,
       ret = ret headline(2, text, 0) ; block = substr( block, len + 1);
       continue;
 
-    # Nth Order Heading H1 H2 H3 H4 H5 H6 + Attrib
-    } else if ( match( block, /^(##?#?#?#?#?)[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*([ \t]*\{([a-zA-Z \t-]*)\})(\n|$)/ ) ) {
+    # # Nth Order Heading H1 H2 H3 H4 H5 H6 + Attrib
+    # } else if ( match( block, /^(##?#?#?#?#?)[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*[ \t]*\{[a-zA-Z \t-]*\}(\n|$)/ ) ) {
+    } else if ( match( block, /^##?#?#?#?#?[^#\n]([^\n#]|#[^\t\n# ]|#[\t ]+[^\t\n ])+#*[\t ]*\{[\ta-zA-Z -]*\}(\n|$)/ ) ) {
       len = RLENGTH; text = attrib = substr(block, 1, len);
       match(block, /^##?#?#?#?#?[^#]/); n = RLENGTH - 1;
-
       # sub(/^(##?#?#?#?#?)[ \t]*/, "", text);  # not working in mawk
       text = substr(text, n + 1); sub(/^[ \t]*/, "", text);
       sub(/[ \t]*#*([ \t]*\{([a-zA-Z \t-]*)\})(\n.*)?$/, "", text);
-      sub(/^(##?#?#?#?#?)[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*[ \t]*\{/, "", attrib);
+
+      sub(/^##?#?#?#?#?[^#\n]([^\n#]|#[^\t\n# ]|#[\t ]+[^\t\n ])+#*[\t ]*\{/, "", attrib);
       sub(/\}(\n.*)?$/, "", attrib);
       gsub(/[^a-zA-Z0-9_-]+/, " ", attrib); gsub(/(^ | $)/, "", attrib);
 
-      ret = ret headline( n, text, attrib ) ; block = substr( block, len + 1);
+      ret = ret headline( n, text, attrib ); block = substr( block, len + 1);
       continue;
 
     # Nth Order Heading H1 H2 H3 H4 H5 H6
-    } else if ( match( block, /^(##?#?#?#?#?)[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*(\n|$)/ ) ) {
+    # } else if ( match( block, /^(##?#?#?#?#?)[ \t]*(([^ \t\n]+|[ \t]+[^ \t\n#]|[ \t]+#+[ \t]*[^ \t\n#])+)[ \t]*#*(\n|$)/ ) ) {
+    } else if ( match( block, /^##?#?#?#?#?[^#\n]([^\n#]|#[^\t\n# ]|#[\t ]+[^\t\n ])+#*(\n|$)/ ) ) {
       len = RLENGTH; text = substr(block, 1, len);
       match(block, /^##?#?#?#?#?[^#]/); n = RLENGTH - 1;
       # sub(/^(##?#?#?#?#?)[ \t]+/, "", text);  # not working in mawk
@@ -854,10 +856,12 @@ function _startlist(block, type, mark, exclude, LOCAL, st, len, list, indent, it
                                    "|[^\n \t][^\n]+(\n|$))*" ) ) {
     st = RSTART; len = RLENGTH; list = substr( block, st, len);
 
-    sub("^\n", "", list); match(list, "^ ? ? ?"); indent = RLENGTH;
-    it = ""; while ( indent > 0 ) { it = it " ?"; indent--; }
+    sub("^\n", "", list); match(list, "^(   |  | |)"); indent = RLENGTH;
     # gsub( "(^|\n) {0," indent "}", "\n", list); sub("^\n", "", list);
-    gsub( "(^|\n)" it, "\n", list); sub("^\n", "", list);
+    # emulate greedy range matcher for mawk
+    it = "("; while ( indent > 0 ) { for (k = indent; k > 0; k--) { it = it " "; } it = it "|"; indent--; }
+    sub(/\|$/, ")?", it); sub(/^\($/, "", it);
+    gsub( "(^|\n)" it, "\n", list ); sub("^\n", "", list);
 
     text = substr(block, 1, st - 1); block = substr(block, st + len);
     if (match(text, /\n[[:space:]]*\n/)) return 0;
@@ -875,7 +879,6 @@ function _list (block, mark, p, LOCAL, len, st, text, indent, it, task) {
   if ( match(block, "^([ \t]*\n)*$")) return;
 
   match(block, "^" mark "[ \t]"); indent = RLENGTH;
-  it = ""; while ( indent > 0 ) { it = it " ?"; indent--; }
 
   sub("^" mark "[ \t]", "", block);
 
@@ -886,6 +889,9 @@ function _list (block, mark, p, LOCAL, len, st, text, indent, it, task) {
   text = substr(block, 1, st); block = substr(block, st + 1);
 
   # gsub("\n {0," indent "}", "\n", text);
+  # emulate greedy range matcher for mawk
+  it = "("; while ( indent > 0 ) { for (k = indent; k > 0; k--) { it = it " "; } it = it "|"; indent--; }
+  sub(/\|$/, ")?", it); sub(/^\($/, "", it);
   gsub("\n" it, "\n", text);
 
   task = match( text, /^\[ \]/   ) ? "<li class=\"task pending\"><input type=checkbox disabled>"      : \
@@ -915,9 +921,11 @@ function _dlist (block, LOCAL, len, st, text, indent, it, p) {
     len = RLENGTH; text = substr(block, 1, len);
     sub( "^([ \t]*\n)*", "", text);
     match(text, "^ ? ? ?:(\t| +)"); indent = RLENGTH;
-    it = ""; while ( indent > 0 ) { it = it " ?"; indent--; }
     sub( "^ ? ? ?:(\t| +)", "", text);
     # gsub( "(^|\n) {0," indent "}", "\n", text );
+    # emulate greedy range matcher for mawk
+    it = "("; while ( indent > 0 ) { for (k = indent; k > 0; k--) { it = it " "; } it = it "|"; indent--; }
+    sub(/\|$/, ")?", it); sub(/^\($/, "", it);
     gsub( "(^|\n)" it, "\n", text );
 
     text = _nblock(text);
-- 
2.39.2