From a873a190e10227c47c675b8c89e6619659784db9 Mon Sep 17 00:00:00 2001
From: Jim Derry <balthisar@gmail.com>
Date: Wed, 25 Aug 2021 16:03:52 -0400
Subject: [PATCH] Logging for ParseScript(). Improved debug output.

---
 src/lexer.h   |  2 ++
 src/parser.c  |  9 +++++++++
 src/tidylib.c | 56 ++++++++++++++++++++++++++++++++++++++-------------
 3 files changed, 53 insertions(+), 14 deletions(-)

diff --git a/src/lexer.h b/src/lexer.h
index 49fa7fc..83b92ef 100644
--- a/src/lexer.h
+++ b/src/lexer.h
@@ -322,6 +322,8 @@ struct _Node
 
     uint        line;           /**< current line of document */
     uint        column;         /**< current column of document */
+    
+    int         idx;            /**< general purpose register */
 
     Bool        closed;         /**< true if closed by explicit end tag */
     Bool        implicit;       /**< true if inferred */
diff --git a/src/parser.c b/src/parser.c
index f50417d..8a36c7c 100644
--- a/src/parser.c
+++ b/src/parser.c
@@ -5154,6 +5154,12 @@ Node* TY_(ParseRowGroup)( TidyDocImpl* doc, Node *rowgroup, GetTokenMode ARG_UNU
 Node* TY_(ParseScript)( TidyDocImpl* doc, Node *script, GetTokenMode ARG_UNUSED(mode) )
 {
     Node *node = NULL;
+#if defined(ENABLE_DEBUG_LOG)
+    static int depth_parser = 0;
+    static int count_parser = 0;
+#endif
+    
+    DEBUG_LOG_ENTER_WITH_NODE(script);
     
     doc->lexer->parent = script;
     node = TY_(GetToken)(doc, CdataContent);
@@ -5167,10 +5173,12 @@ Node* TY_(ParseScript)( TidyDocImpl* doc, Node *script, GetTokenMode ARG_UNUSED(
     {
         /* handle e.g. a document like "<script>" */
         TY_(Report)(doc, script, NULL, MISSING_ENDTAG_FOR);
+        DEBUG_LOG_EXIT;
         return NULL;
     }
 
     node = TY_(GetToken)(doc, IgnoreWhitespace);
+    DEBUG_LOG_GOT_TOKEN(node);
 
     if (!(node && node->type == EndTag && node->tag &&
         node->tag->id == script->tag->id))
@@ -5184,6 +5192,7 @@ Node* TY_(ParseScript)( TidyDocImpl* doc, Node *script, GetTokenMode ARG_UNUSED(
     {
         TY_(FreeNode)(doc, node);
     }
+    DEBUG_LOG_EXIT;
     return NULL;
 }
 
diff --git a/src/tidylib.c b/src/tidylib.c
index d88cd9f..28ae9be 100644
--- a/src/tidylib.c
+++ b/src/tidylib.c
@@ -2048,24 +2048,52 @@ void dbg_show_node( TidyDocImpl* doc, Node *node, int caller, int indent )
     SPRTF("\n");
 }
 
-/* Tail recursion here with sensible compilers will re-use
-   the stack frame and avoid overflows during debugging.
+/* Make this non-recursive, because we really do want to eliminate
+   recursion that makes us crash, even when debugging.
  */
-void dbg_show_all_nodes_loop( TidyDocImpl* doc, Node *node, int indent )
-{
-    while ( node && (node = node->next) )
-    {
-        dbg_show_node( doc, node, 0, indent );
-        dbg_show_all_nodes_loop( doc, node->content, indent + 1 );
-    }
-}
-
 void dbg_show_all_nodes( TidyDocImpl* doc, Node *node, int indent )
 {
-    dbg_show_node( doc, node, 0, indent );
-    dbg_show_all_nodes_loop( doc, node->content, indent + 1 );
-}
+    Stack *stack = TY_(newStack)(doc, 16);
+    Node *child = NULL;
+    Node *next = NULL;
 
+    dbg_show_node( doc, node, 0, indent++ );
+
+    if ( (child = node->content) )
+    {
+        while ( child )
+        {
+            if ( (next = child->next) )
+            {
+                next->idx = indent;
+            }
+            
+            dbg_show_node( doc, child, 0, indent );
+            
+            if (child->content)
+            {
+                TY_(push)(stack, next);
+                indent++;
+                child = child->content;
+                continue;
+            }
+
+            if (next)
+            {
+                child = next;
+            }
+            else
+            {
+                if ( (child = TY_(pop)(stack)) )
+                {
+                    indent = child->idx;
+                }
+            }
+
+        }
+        TY_(freeStack)(stack);
+    }
+}
 #endif
 
 int         tidyDocCleanAndRepair( TidyDocImpl* doc )