From 599617843e99807ef118ec57fdf9838e7920ab1d Mon Sep 17 00:00:00 2001
From: david parsons <orc@pell.portland.or.us>
Date: Tue, 22 Mar 2022 17:01:14 -0700
Subject: [PATCH] implement a toc label uniquifier

---
 generate.c  |  9 ++++-----
 markdown.c  |  5 +++--
 markdown.h  |  7 +++++++
 resource.c  |  2 ++
 tests/toc.t | 23 +++++++++++++++++++++++
 toc.c       | 37 +++++++++++++++++++++++++++++++++++--
 6 files changed, 74 insertions(+), 9 deletions(-)

diff --git a/generate.c b/generate.c
index cca1911..1505e7f 100644
--- a/generate.c
+++ b/generate.c
@@ -183,10 +183,9 @@ Qprintf(MMIOT *f, char *fmt, ...)
 /* Qanchor() prints out a suitable-for-id-tag version of a string
  */
 static void
-Qanchor(struct line *p, MMIOT *f)
+Qanchor(char *name, MMIOT *f)
 {
-    mkd_string_to_anchor(T(p->text), S(p->text),
-			 (mkd_sta_function_t)Qchar, f, 1, f);
+    mkd_string_to_anchor(name, strlen(name), (mkd_sta_function_t)Qchar, f, 1, f);
 }
 
 
@@ -1528,14 +1527,14 @@ printheader(Paragraph *pp, MMIOT *f)
 	Qprintf(f, "<h%d", pp->hnumber);
 	if ( is_flag_set(&f->flags, MKD_TOC) ) {
 	    Qstring(" id=\"", f);
-	    Qanchor(pp->text, f);
+	    Qanchor(pp->label, f);
 	    Qchar('"', f);
 	}
 	Qchar('>', f);
     } else {
 	if ( is_flag_set(&f->flags, MKD_TOC) ) {
 	    Qstring("<a name=\"", f);
-	    Qanchor(pp->text, f);
+	    Qanchor(pp->label, f);
 	    Qstring("\"></a>\n", f);
 	}
 	Qprintf(f, "<h%d>", pp->hnumber);
diff --git a/markdown.c b/markdown.c
index 2380869..b0df270 100644
--- a/markdown.c
+++ b/markdown.c
@@ -20,8 +20,6 @@
 
 typedef int (*stfu)(const void*,const void*);
 
-typedef ANCHOR(Paragraph) ParagraphRoot;
-
 static Paragraph *Pp(ParagraphRoot *, Line *, int);
 static Paragraph *compile(Line *, int, MMIOT *);
 
@@ -1358,6 +1356,9 @@ compile(Line *ptr, int toplevel, MMIOT *f)
 	else if ( ishdr(ptr, &hdr_type, &(f->flags) ) ) {
 	    p = Pp(&d, ptr, HDR);
 	    ptr = headerblock(p, hdr_type);
+	    if ( is_flag_set(&(f->flags), MKD_TOC) )
+		p->label = ___mkd_uniquetag(&d, T(p->text->text),
+						S(p->text->text));
 	}
 	else {
 	    /* either markup or an html block element
diff --git a/markdown.h b/markdown.h
index 4efc40c..879444d 100644
--- a/markdown.h
+++ b/markdown.h
@@ -96,6 +96,7 @@ typedef struct paragraph {
     struct paragraph *next;	/* next paragraph */
     struct paragraph *down;	/* recompiled contents of this paragraph */
     struct line *text;		/* all the text in this paragraph */
+    char *label;		/* toc label, uniqued */
     char *ident;		/* %id% tag for QUOTE */
     char *lang;			/* lang attribute for CODE */
     enum { WHITESPACE=0, CODE, QUOTE, MARKUP,
@@ -108,6 +109,8 @@ typedef struct paragraph {
 #define IS_CHECKED		0x02
 } Paragraph;
 
+typedef ANCHOR(Paragraph) ParagraphRoot;
+
 enum { ETX, SETEXT };	/* header types */
 
 /* reference-style links (and images) are stored in an array
@@ -288,6 +291,10 @@ extern void __mkd_trim_line(Line *, int);
 
 extern int  __mkd_io_strget(struct string_stream *);
 
+/* toc uniquifier
+ */
+extern char *___mkd_uniquetag(ParagraphRoot *, char *, int);
+    
 /* utility function to do some operation and exit the current function
  * if it fails
  */
diff --git a/resource.c b/resource.c
index 0b2dd06..256efe4 100644
--- a/resource.c
+++ b/resource.c
@@ -49,6 +49,8 @@ ___mkd_freeParagraph(Paragraph *p)
 	___mkd_freeParagraph(p->down);
     if (p->text)
 	___mkd_freeLines(p->text);
+    if (p->label)
+	free(p->label);
     if (p->ident)
 	free(p->ident);
     if (p->lang)
diff --git a/tests/toc.t b/tests/toc.t
index 21283a7..7c15986 100644
--- a/tests/toc.t
+++ b/tests/toc.t
@@ -91,5 +91,28 @@ try '-T -ftoc,html5anchor' 'html5 multibyte chars' \
 <a name="It’s-an-apostrophe"></a>
 <h1>It’s an apostrophe</h1>'
 
+summary $0
+
+
+# Check that the uniquifier works
+#
+
+title "uniquifying duplicate headers"
+
+
+try '-T -ftoc' 'uniquifying duplicate labels' \
+'# this
+# this' \
+'<ul>
+ <li><a href="#this">this</a></li>
+ <li><a href="#this_1">this</a></li>
+</ul>
+<a name="this"></a>
+<h1>this</h1>
+
+<a name="this_1"></a>
+<h1>this</h1>'
+  
+
 summary $0
 exit $rc
diff --git a/toc.c b/toc.c
index f8e7a15..3822573 100644
--- a/toc.c
+++ b/toc.c
@@ -72,8 +72,7 @@ mkd_toc(Document *p, char **doc)
 			++last_hnumber;
 		    }
 		    Csprintf(&res, "%*s<li><a href=\"#", srcp->hnumber, "");
-		    mkd_string_to_anchor(T(srcp->text->text),
-					 S(srcp->text->text),
+		    mkd_string_to_anchor(srcp->label, strlen(srcp->label),
 					 (mkd_sta_function_t)Csputc,
 					 &res,1,p->ctx);
 		    Csprintf(&res, "\">");
@@ -104,6 +103,40 @@ mkd_toc(Document *p, char **doc)
 }
 
 
+char *
+___mkd_uniquetag(ParagraphRoot *pr, char *name, int length)
+{
+    Paragraph *content;
+    char *label;
+    int suffix;
+    int seq = 0;
+
+    if ( !(pr && name) ) return 0;
+
+    
+    suffix = length;
+    
+    label = calloc(1, suffix+20);
+
+    strcpy(label, name);
+
+restart:
+    for ( content = T(*pr); content; content = content->next ) {
+	
+	if ( content->text && content->typ == HDR ) {
+
+	    if ( content->label && strcmp(label, content->label) == 0 ) {
+		/* collision; bump trailing sequence and try again
+		  */
+		sprintf(label+suffix, "_%d", ++seq);
+		goto restart;
+	    }
+	}
+    }
+    return label;
+}
+
+
 /* write an header index
  */
 int