Add support for parsing comments

fletcher · Aug 5, 2024 · df117b0 · df117b0
1 parent 371e872
commit df117b0
Show file tree

Hide file tree

Showing 7 changed files with 4,069 additions and 3,783 deletions.
diff --git a/markdown_lib.c b/markdown_lib.c
@@ -60,6 +60,7 @@ void print_tree(element * elt, int indent) {
             case EMPH:               key = "EMPH"; break;
             case STRONG:             key = "STRONG"; break;
             case DELETE:             key = "DELETE"; break;
+            case COMMENT:            key = "COMMENT"; break;
             case PLAIN:              key = "PLAIN"; break;
             case PARA:               key = "PARA"; break;
             case LISTITEM:           key = "LISTITEM"; break;
@@ -72,8 +73,9 @@ void print_tree(element * elt, int indent) {
             case H5:                 key = "H5"; break;
             case H6:                 key = "H6"; break;
             case BLOCKQUOTE:         key = "BLOCKQUOTE"; break;
+            case COMMENTBLOCK:       key = "COMMENTBLOCK"; break;
             case VERBATIM:           key = "VERBATIM"; break;
-			case CODEBLOCK:			 key = "CODEBLOCK"; break;
+            case CODEBLOCK:			 key = "CODEBLOCK"; break;
             case HTMLBLOCK:          key = "HTMLBLOCK"; break;
             case HRULE:              key = "HRULE"; break;
             case REFERENCE:          key = "REFERENCE"; break;

diff --git a/markdown_lib.h b/markdown_lib.h
@@ -21,6 +21,7 @@ enum markdown_extensions {
 	EXT_MATH						= 1 << 12,
 	EXT_LABELS						= 1 << 13,
 	EXT_DELETE						= 1 << 14,
+	EXT_COMMENT						= 1 << 15,
 };
 
 enum markdown_formats {

diff --git a/markdown_output.c b/markdown_output.c
@@ -316,6 +316,11 @@ static void print_html_element(int ext, GString *out, element *elt, bool obfusca
         print_html_element_list(ext, out, elt->children, obfuscate);
         g_string_append_printf(out, "</del>");
         break;
+	case COMMENT:
+		g_string_append_printf(out, "<!--");
+		print_html_element_list(ext, out, elt->children, obfuscate);
+		g_string_append_printf(out, "-->");
+		break;
     case LIST:
         print_html_element_list(ext, out, elt->children, obfuscate);
         break;
@@ -427,6 +432,13 @@ static void print_html_element(int ext, GString *out, element *elt, bool obfusca
         g_string_append_printf(out, "</blockquote>");
         padded = 0;
         break;
+	case COMMENTBLOCK:
+		pad(out, 2);
+		g_string_append_printf(out, "<!--");
+		print_html_element_list(ext, out, elt->children, obfuscate);
+		g_string_append_printf(out, "-->");
+		padded = 0;
+		break;
     case REFERENCE:
         /* Nonprinting */
         break;

diff --git a/markdown_parser.h b/markdown_parser.h
diff --git a/markdown_parser.leg b/markdown_parser.leg
@@ -127,6 +127,7 @@ MetaDataValue = a:StartList
 
 Block =     BlankLine*
             ( BlockQuote
+            | &{ extension(yy->state, EXT_COMMENT) } CommentBlock
             | Verbatim
             | &{ !extension(yy->state, EXT_COMPATIBILITY) } DefinitionList
             | &{ !extension(yy->state, EXT_COMPATIBILITY) } Glossary
@@ -149,6 +150,7 @@ HeadingSectionBlock =
             BlankLine*
             !Heading
             ( BlockQuote
+            | &{ extension(yy->state, EXT_COMMENT) } CommentBlock
             | Verbatim
             | &{ !extension(yy->state, EXT_COMPATIBILITY) } DefinitionList
             | &{ !extension(yy->state, EXT_COMPATIBILITY) } Glossary
@@ -223,6 +225,23 @@ BlockQuoteRaw =  a:StartList
                      $$->key = RAW;
                  }
 
+CommentBlock = a:CommentBlockRaw
+             {  $$ = mk_element(COMMENTBLOCK);
+                $$->children = a;
+             }
+
+CommentBlockRaw = a:StartList
+                ('<!--'
+				 (!'-->' CommentBlockLine { a = cons($$, a); } )*
+				'-->')+
+                 {   a = cons(mk_str("\033"), a); // add some non-whitespace character that will be stripped later. Required to keep the paragraph even for empty comments.
+                     $$ = mk_str_from_list(a, true);
+                     $$->key = RAW;
+                 }
+
+CommentBlockLine = ( < (!'-->' .)* > )
+				   { $$ = mk_str(yytext); }
+
 NonblankIndentedLine = !BlankLine IndentedLine
 
 FirstVerbatimChunk = a:StartList
@@ -641,6 +660,7 @@ Inline  = &{ check_timeout(yy->state) }
         | Link
         | NoteReference
  #      | InlineNote           # Not used in Markdown/MultiMarkdown
+        | &{ extension(yy->state, EXT_COMMENT) } Comment
         | Code
         | MarkdownHtmlTagOpen
         | RawHtml
@@ -832,6 +852,12 @@ ExplicitLink =  a:StartList l:Label '(' Sp s:Source Spnl t:Title ( &{ !extension
                     free_element(t);
                     free(l);
                 }
+
+Comment     =    '<!--' !Whitespace
+                a:StartList
+                ( !'-->' b:Inline { a = cons(b, a); })+
+                '-->'
+                { $$ = mk_list(COMMENT, a); }
 
 Source  = ( '<' < SourceContents > '>' | < SourceContents > )
           { $$ = mk_str(yytext); }
@@ -1025,6 +1051,7 @@ ExtendedSpecialChar = &{ extension(yy->state, EXT_SMART) } ('.' | '-' | '\'' | '
                     | &{ extension(yy->state, EXT_NOTES) } ( '^' )
                     | &{ extension(yy->state, EXT_MATH) } ( '$' )
                     | &{ extension(yy->state, EXT_DELETE) } ( '~' )
+                    | &{ extension(yy->state, EXT_COMMENT) } ( '<!--' | '-->' )
 
 Smart = &{ extension(yy->state, EXT_SMART) }
         ( Ellipsis | Dash | SingleQuoted | DoubleQuoted | Apostrophe )

diff --git a/markdown_parser_lib.h b/markdown_parser_lib.h
@@ -46,13 +46,15 @@ enum keys { LIST,   /* A generic list of values.  For ordered and bullet lists,
 	EMPH,
 	STRONG,
 	DELETE,
+	COMMENT,
 	PLAIN,
 	PARA,
 	LISTITEM,
 	BULLETLIST,
 	ORDEREDLIST,
 	H1, H2, H3, H4, H5, H6, H7, /* Code assumes that these are in order. */
 	BLOCKQUOTE,
+	COMMENTBLOCK,
 	VERBATIM,
 	CODEBLOCK,
 	HTMLBLOCK,

diff --git a/utility_functions.c b/utility_functions.c
@@ -173,6 +173,7 @@ bool match_inlines(element *l1, element *l2) {
         case EMPH:
         case STRONG:
         case DELETE:
+        case COMMENT:
         case LIST:
         case SINGLEQUOTED:
         case DOUBLEQUOTED: