markdown converter

objs.mk

@@ -26,6 +26,9 @@ src/external_xemil_src_xpointer.o: external/xemil/src/xpointer.c
 OBJS += src/src_action_help.o
 src/src_action_help.o: src/action/help.c
 	$(CC) $(CFLAGS) -c -o $@ src/action/help.c
+OBJS += src/src_action_markdown.o
+src/src_action_markdown.o: src/action/markdown.c
+	$(CC) $(CFLAGS) -c -o $@ src/action/markdown.c
 OBJS += src/src_action_seed.o
 src/src_action_seed.o: src/action/seed.c
 	$(CC) $(CFLAGS) -c -o $@ src/action/seed.c

src/action/markdown.c

@@ -0,0 +1,463 @@
+#include <taiga.h>
+
+static FILE*  out;
+static int    indent	 = 0;
+static int    mode	 = 0;
+static char*  title	 = NULL;
+static int    in_title	 = 0;
+static int    in_section = 0;
+static int    in_pre	 = 0;
+static char** mapping;
+static int    mapping_len;
+
+static void print_indent(void) {
+	int i;
+
+	for(i = 0; i < indent; i++) fprintf(out, "\t");
+}
+
+static char* encode(const char* str, int len) {
+	int   actlen = (len == -1 ? strlen(str) : len);
+	char* r	     = malloc(actlen * 5 + 1);
+	int   n	     = 0;
+	int   i;
+
+	for(i = 0; i < actlen; i++) {
+		if(str[i] == '<') {
+			r[n++] = '&';
+			r[n++] = 'l';
+			r[n++] = 't';
+			r[n++] = ';';
+		} else if(str[i] == '>') {
+			r[n++] = '&';
+			r[n++] = 'g';
+			r[n++] = 't';
+			r[n++] = ';';
+		} else if(str[i] == '&') {
+			r[n++] = '&';
+			r[n++] = 'a';
+			r[n++] = 'm';
+			r[n++] = 'p';
+			r[n++] = ';';
+		} else {
+			r[n++] = str[i];
+		}
+	}
+	r[n] = 0;
+
+	return r;
+}
+
+static int enter_block(MD_BLOCKTYPE type, void* detail, void* user) {
+	if(mode == 0 && type == MD_BLOCK_H && ((MD_BLOCK_H_DETAIL*)detail)->level == 1 && title == NULL) {
+		in_title = 1;
+	}
+
+	if(!mode) return 0;
+
+	if(type == MD_BLOCK_H) {
+		MD_BLOCK_H_DETAIL* det = detail;
+
+		if(det->level <= 2) {
+			if(title != NULL) free(title);
+			title = NULL;
+
+			in_title = 1;
+		} else {
+			print_indent();
+			fprintf(out, "<h%d>\n", det->level);
+			indent++;
+		}
+	} else if(type == MD_BLOCK_P) {
+		print_indent();
+		fprintf(out, "<p>\n");
+		indent++;
+	} else if(type == MD_BLOCK_UL) {
+		print_indent();
+		fprintf(out, "<ul>\n");
+		indent++;
+	} else if(type == MD_BLOCK_OL) {
+		print_indent();
+		fprintf(out, "<ol>\n");
+		indent++;
+	} else if(type == MD_BLOCK_LI) {
+		print_indent();
+		fprintf(out, "<li>\n");
+		indent++;
+	} else if(type == MD_BLOCK_HR) {
+		print_indent();
+		fprintf(out, "<hr>\n");
+		indent++;
+	} else if(type == MD_BLOCK_CODE) {
+		print_indent();
+		fprintf(out, "<pre>");
+		indent++;
+		print_indent();
+		fprintf(out, "<code>");
+		indent++;
+		in_pre = 1;
+	} else if(type == MD_BLOCK_TABLE) {
+		print_indent();
+		fprintf(out, "<table>\n");
+		indent++;
+	} else if(type == MD_BLOCK_TR) {
+		print_indent();
+		fprintf(out, "<tr>\n");
+		indent++;
+	} else if(type == MD_BLOCK_TH) {
+		print_indent();
+		fprintf(out, "<th>\n");
+		indent++;
+	} else if(type == MD_BLOCK_TD) {
+		print_indent();
+		fprintf(out, "<td>\n");
+		indent++;
+	} else if(type == MD_BLOCK_QUOTE) {
+		print_indent();
+		fprintf(out, "<blockquote>\n");
+		indent++;
+	}
+	return 0;
+}
+
+static int leave_block(MD_BLOCKTYPE type, void* detail, void* user) {
+	if(mode == 0 && type == MD_BLOCK_H && ((MD_BLOCK_H_DETAIL*)detail)->level == 1 && in_title) {
+		in_title = 0;
+	}
+
+	if(!mode) return 0;
+
+	if(type == MD_BLOCK_H) {
+		MD_BLOCK_H_DETAIL* det = detail;
+
+		in_title = 0;
+
+		if(det->level == 2) {
+			char* buf;
+
+			if(in_section) {
+				indent--;
+				print_indent();
+				fprintf(out, "</section>\n");
+			}
+
+			buf = encode(title == NULL ? "" : title, -1);
+
+			print_indent();
+			fprintf(out, "<section title=\"%s\">\n", buf);
+			indent++;
+
+			in_section = 1;
+
+			free(buf);
+		} else if(det->level != 1) {
+			indent--;
+			print_indent();
+			fprintf(out, "</h%d>\n", det->level);
+		}
+	} else if(type == MD_BLOCK_P) {
+		indent--;
+		print_indent();
+		fprintf(out, "</p>\n");
+	} else if(type == MD_BLOCK_UL) {
+		indent--;
+		print_indent();
+		fprintf(out, "</ul>\n");
+	} else if(type == MD_BLOCK_OL) {
+		indent--;
+		print_indent();
+		fprintf(out, "</ol>\n");
+	} else if(type == MD_BLOCK_LI) {
+		indent--;
+		print_indent();
+		fprintf(out, "</li>\n");
+	} else if(type == MD_BLOCK_CODE) {
+		indent--;
+		print_indent();
+		fprintf(out, "</code>\n");
+		indent--;
+		print_indent();
+		fprintf(out, "</pre>\n");
+		in_pre = 0;
+	} else if(type == MD_BLOCK_TABLE) {
+		indent--;
+		print_indent();
+		fprintf(out, "</table>\n");
+	} else if(type == MD_BLOCK_TR) {
+		indent--;
+		print_indent();
+		fprintf(out, "</tr>\n");
+	} else if(type == MD_BLOCK_TH) {
+		indent--;
+		print_indent();
+		fprintf(out, "</th>\n");
+	} else if(type == MD_BLOCK_TD) {
+		indent--;
+		print_indent();
+		fprintf(out, "</td>\n");
+	} else if(type == MD_BLOCK_QUOTE) {
+		indent--;
+		print_indent();
+		fprintf(out, "</blockquote>\n");
+	}
+
+	return 0;
+}
+
+static int enter_span(MD_SPANTYPE type, void* detail, void* user) {
+	if(!mode) return 0;
+
+	if(type == MD_SPAN_EM) {
+		print_indent();
+		fprintf(out, "<em>\n");
+		indent++;
+	} else if(type == MD_SPAN_STRONG) {
+		print_indent();
+		fprintf(out, "<strong>\n");
+		indent++;
+	} else if(type == MD_SPAN_A) {
+		MD_SPAN_A_DETAIL* det  = detail;
+		char*		  href = encode(det->href.text, det->href.size);
+		int		  i;
+
+		for(i = 0; i < mapping_len; i++) {
+			char* key   = u_strdup(mapping[i]);
+			char* value = strchr(key, '=');
+
+			if(value != NULL) {
+				value[0] = 0;
+				value++;
+			}
+
+			if(value != NULL && strcmp(key, href) == 0) {
+				free(href);
+				href = encode(value, -1);
+			}
+
+			free(key);
+		}
+
+		print_indent();
+		fprintf(out, "<a href=\"%s\">\n", href);
+		indent++;
+
+		free(href);
+	} else if(type == MD_SPAN_IMG) {
+		MD_SPAN_IMG_DETAIL* det = detail;
+		char*		    src = encode(det->src.text, det->src.size);
+
+		print_indent();
+		fprintf(out, "<img src=\"%s\" alt=\"", src);
+
+		in_title = 1;
+		if(title != NULL) free(title);
+		title = NULL;
+
+		free(src);
+	} else if(type == MD_SPAN_CODE) {
+		print_indent();
+		fprintf(out, "<code>\n");
+		indent++;
+	} else if(type == MD_SPAN_DEL) {
+		print_indent();
+		fprintf(out, "<del>\n");
+		indent++;
+	} else if(type == MD_SPAN_U) {
+		print_indent();
+		fprintf(out, "<u>\n");
+		indent++;
+	}
+
+	return 0;
+}
+
+static int leave_span(MD_SPANTYPE type, void* detail, void* user) {
+	if(!mode) return 0;
+
+	if(type == MD_SPAN_EM) {
+		indent--;
+		print_indent();
+		fprintf(out, "</em>\n");
+	} else if(type == MD_SPAN_STRONG) {
+		indent--;
+		print_indent();
+		fprintf(out, "</strong>\n");
+	} else if(type == MD_SPAN_A) {
+		indent--;
+		print_indent();
+		fprintf(out, "</a>\n");
+	} else if(type == MD_SPAN_IMG) {
+		char* buf = encode(title, -1);
+
+		fprintf(out, "%s\" />\n", buf);
+		free(buf);
+
+		in_title = 0;
+	} else if(type == MD_SPAN_CODE) {
+		indent--;
+		print_indent();
+		fprintf(out, "</code>\n");
+	} else if(type == MD_SPAN_DEL) {
+		indent--;
+		print_indent();
+		fprintf(out, "</del>\n");
+	} else if(type == MD_SPAN_U) {
+		indent--;
+		print_indent();
+		fprintf(out, "</u>\n");
+	}
+
+	return 0;
+}
+
+static int text(MD_TEXTTYPE type, const MD_CHAR* text, MD_SIZE size, void* user) {
+	if(in_title && (type == MD_TEXT_NORMAL || type == MD_TEXT_ENTITY)) {
+		if(title == NULL) {
+			title = malloc(size + 1);
+
+			memcpy(title, text, size);
+			title[size] = 0;
+		} else {
+			char* old = title;
+			char* buf = malloc(size + 1);
+
+			memcpy(buf, text, size);
+			buf[size] = 0;
+
+			title = u_strvacat(old, " ", buf, NULL);
+			free(old);
+
+			free(buf);
+		}
+	}
+
+	if(!mode) return 0;
+
+	if(in_title) return 0;
+
+	if(type == MD_TEXT_BR) {
+		print_indent();
+		fprintf(out, "<br />\n");
+	} else if(type == MD_TEXT_NORMAL || type == MD_TEXT_ENTITY || type == MD_TEXT_CODE) {
+		char* buf;
+
+		print_indent();
+
+		buf = encode(text, size);
+		fwrite(buf, 1, strlen(buf), out);
+		free(buf);
+
+		fprintf(out, "\n");
+	}
+
+	return 0;
+}
+
+int action_markdown(int argc, char** argv) {
+	FILE*	  f;
+	char*	  buffer;
+	int	  size;
+	MD_PARSER parser;
+	char*	  argin	 = NULL;
+	char*	  argout = NULL;
+	int	  i;
+
+	memset(&parser, 0, sizeof(parser));
+	parser.flags	   = MD_FLAG_COLLAPSEWHITESPACE | MD_FLAG_TABLES | MD_FLAG_STRIKETHROUGH | MD_FLAG_UNDERLINE;
+	parser.enter_block = enter_block;
+	parser.leave_block = leave_block;
+	parser.enter_span  = enter_span;
+	parser.leave_span  = leave_span;
+	parser.text	   = text;
+
+	mapping	    = NULL;
+	mapping_len = 0;
+	for(i = 0; i < argc; i++) {
+		if(argin == NULL) {
+			argin = argv[i];
+		} else if(argout == NULL) {
+			argout = argv[i];
+		} else if(mapping == NULL) {
+			mapping	    = &argv[i];
+			mapping_len = argc - i;
+		}
+	}
+
+	if(argin == NULL || argout == NULL) {
+		fprintf(stderr, "Usage: markdown input output [linksrc=linkdest ...]\n");
+		return 1;
+	}
+
+	if((f = fopen(argin, "r")) == NULL) {
+		fprintf(stderr, "Failed to open input\n");
+		return 1;
+	}
+
+	if((out = fopen(argout, "w")) == NULL) {
+		fprintf(stderr, "Failed to open output\n");
+		return 1;
+	}
+
+	fseek(f, 0, SEEK_END);
+	size = ftell(f);
+	fseek(f, 0, SEEK_SET);
+
+	buffer = malloc(size);
+	fread(buffer, 1, size, f);
+	fclose(f);
+
+	fprintf(out, "<document>\n");
+	indent++;
+
+	in_title   = 0;
+	in_section = 0;
+	in_pre	   = 0;
+	mode	   = 0;
+	if(md_parse(buffer, size, &parser, NULL)) {
+		free(buffer);
+		fclose(out);
+		remove(argout);
+		fprintf(stderr, "Failed to parse Markdown\n");
+		return 1;
+	}
+
+	fprintf(out, "	<header>\n");
+	indent++;
+	if(title != NULL) fprintf(out, "		<title>%s</title>\n", title);
+	fprintf(out, "	</header>\n");
+	indent--;
+
+	fprintf(out, "	<body>\n");
+	indent++;
+
+	in_title   = 0;
+	in_section = 0;
+	in_pre	   = 0;
+	mode	   = 1;
+	if(md_parse(buffer, size, &parser, NULL)) {
+		free(buffer);
+		fclose(out);
+		remove(argout);
+		fprintf(stderr, "Failed to parse Markdown\n");
+		return 1;
+	}
+
+	if(in_section) {
+		indent--;
+		print_indent();
+		fprintf(out, "</section>\n");
+	}
+
+	fprintf(out, "	</body>\n");
+	indent--;
+	fprintf(out, "</document>\n");
+	indent--;
+
+	if(title != NULL) free(title);
+
+	free(buffer);
+
+	fclose(out);
+
+	return 0;
+}

src/default.c

@@ -1,17 +1,15 @@
 #include <taiga.h>
 
-static void print(FILE* out, const char* txt, int indent) {
+static void print(FILE* out, const char* txt, int pre, int indent) {
 	int i;
 
 	for(i = 0;; i++) {
 		if(i == 0 || txt[i] == '\n' || txt[i] == 0) {
 			int j;
-			if(i > 0)
-				fprintf(out, "\n");
-			if(txt[i] == 0)
-				break;
-			for(j = 0; j < indent; j++)
-				fprintf(out, "\t");
+			if(pre ? (txt[i] == '\n') : (i > 0)) fprintf(out, "\n");
+			if(txt[i] == 0) break;
+			if(!pre)
+				for(j = 0; j < indent; j++) fprintf(out, "\t");
 		}
 
 		if(txt[i] != '\n') fprintf(out, "%c", txt[i]);
@@ -70,7 +68,7 @@ void default_head(FILE* out, const char* top, xl_node_t* element, int indent) {
 		sprintf(tag, "<%s%s>", element->name, text);
 	}
 
-	if(tag[0] != 0) print(out, tag, indent);
+	if(tag[0] != 0) print(out, tag, 0, indent);
 
 	child = element->first_child;
 	while(child != NULL) {
@@ -86,11 +84,10 @@ void default_head(FILE* out, const char* top, xl_node_t* element, int indent) {
 		child = child->next;
 	}
 
-	if(end[0] != 0)
-		print(out, end, indent);
+	if(end[0] != 0) print(out, end, 0, indent);
 }
 
-void default_body(FILE* out, const char* top, xl_node_t* element, int spec, int indent) {
+void default_body(FILE* out, const char* top, xl_node_t* element, int spec, int pre, int indent) {
 	xl_node_t* child;
 	char	   tag[2048]; /* enough for most cases... :) */
 	char	   end[128];
@@ -149,11 +146,14 @@ void default_body(FILE* out, const char* top, xl_node_t* element, int spec, int
 		add = 3;
 	} else if(strcmp(element->name, "p") == 0 ||	      /**/
 		  strcmp(element->name, "code") == 0 ||	      /**/
+		  strcmp(element->name, "pre") == 0 ||	      /**/
 		  strcmp(element->name, "blockquote") == 0 || /**/
 		  strcmp(element->name, "li") == 0 ||	      /**/
 		  strcmp(element->name, "dt") == 0 ||	      /**/
 		  strcmp(element->name, "dd") == 0 ||	      /**/
 		  strcmp(element->name, "strong") == 0 ||     /**/
+		  strcmp(element->name, "del") == 0 ||	      /**/
+		  strcmp(element->name, "u") == 0 ||	      /**/
 		  strcmp(element->name, "sub") == 0 ||	      /**/
 		  strcmp(element->name, "sup") == 0 ||	      /**/
 		  strcmp(element->name, "tr") == 0 ||	      /**/
@@ -166,6 +166,8 @@ void default_body(FILE* out, const char* top, xl_node_t* element, int spec, int
 		  strcmp(element->name, "h6") == 0) {
 		accept_attr(top, text, 0, element, "id", "class", NULL);
 
+		if(strcmp(element->name, "pre") == 0) pre = 1;
+
 		sprintf(tag, "<%s%s>", element->name, text);
 
 		sprintf(end, "</%s>", element->name);
@@ -231,20 +233,21 @@ void default_body(FILE* out, const char* top, xl_node_t* element, int spec, int
 		sprintf(tag, "<%s%s>", element->name, text);
 	}
 
-	if(tag[0] != 0) print(out, tag, indent);
+	if(tag[0] != 0) print(out, tag, pre, indent);
 
 	child = element->first_child;
 	while(child != NULL) {
 		if(child->type == XL_NODE_NODE && child->name != NULL) {
-			default_body(out, top, child, spec, indent + 1 + add);
+			default_body(out, top, child, spec, pre, indent + 1 + add);
 		} else if(child->type == XL_NODE_TEXT && child->text != NULL) {
-			print(out, child->text, indent + 1 + add);
+			print(out, child->text, pre, pre ? 0 : (indent + 1 + add));
 		}
 
 		child = child->next;
 	}
 
-	if(end[0] != 0) print(out, end, indent);
+	if(end[0] != 0) print(out, end, pre, indent);
+	if(strcmp(end, "</pre>") == 0) fprintf(out, "\n");
 }
 
 void default_nav(FILE* out, const char* top, xl_node_t* element, int indent) {

src/main.c

@@ -7,7 +7,7 @@ int main(int argc, char** argv) {
 	char** target_argv = malloc(sizeof(*target_argv));
 	target_argv[0]	   = NULL;
 
-	printf("Taiga, static website generator - run `%s help' for help\n\n", argv[0]);
+	fprintf(stderr, "Taiga, static website generator - run `%s help' for help\n\n", argv[0]);
 
 	for(i = 1; i < argc; i++) {
 		if(argv[i][0] == '-') {
@@ -30,6 +30,8 @@ int main(int argc, char** argv) {
 		return action_help(target_argc, target_argv);
 	} else if(strcmp(action, "seed") == 0) {
 		return action_seed(target_argc, target_argv);
+	} else if(strcmp(action, "markdown") == 0) {
+		return action_markdown(target_argc, target_argv);
 	}
 
 	fprintf(stderr, "Unknown action\n");

src/process.c

@@ -46,8 +46,7 @@ static int parse_file(const char* top, const char* full) {
 				node = node->next;
 			}
 
-			if(title == NULL)
-				title = u_strdup("Untitled");
+			if(title == NULL) title = u_strdup("Untitled");
 
 			fprintf(f, "<!DOCTYPE html PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\n");
 			fprintf(f, "<html>\n");

src/skin/classic.c

@@ -323,11 +323,13 @@ void classic_body(FILE* out, const char* top, const char* title, xl_node_t* body
 	fprintf(out, "				<td valign=\"top\">\n");
 	fprintf(out, "					<div id=\"content\">\n");
 
-	child = body->first_child;
-	while(child != NULL) {
-		default_body(out, top, child, 0, 6);
+	if(body != NULL) {
+		child = body->first_child;
+		while(child != NULL) {
+			default_body(out, top, child, 0, 0, 6);
 
-		child = child->next;
+			child = child->next;
+		}
 	}
 
 	fprintf(out, "					</div>\n");
@@ -336,7 +338,7 @@ void classic_body(FILE* out, const char* top, const char* title, xl_node_t* body
 	if((nodes = xl_get_path(skinconf->root, "footer")) != NULL) {
 		child = nodes[0]->first_child;
 		while(child != NULL) {
-			default_body(out, top, child, 1, 6);
+			default_body(out, top, child, 1, 0, 6);
 
 			child = child->next;
 		}

src/skin/simple.c

@@ -334,11 +334,13 @@ void simple_body(FILE* out, const char* top, const char* title, xl_node_t* body)
 	fprintf(out, "				</td>\n");
 	fprintf(out, "				<td id=\"content\" valign=\"top\">\n");
 
-	child = body->first_child;
-	while(child != NULL) {
-		default_body(out, top, child, 0, 5);
+	if(body != NULL) {
+		child = body->first_child;
+		while(child != NULL) {
+			default_body(out, top, child, 0, 0, 5);
 
-		child = child->next;
+			child = child->next;
+		}
 	}
 
 	fprintf(out, "					<div id=\"footer\">\n");
@@ -346,7 +348,7 @@ void simple_body(FILE* out, const char* top, const char* title, xl_node_t* body)
 	if((nodes = xl_get_path(skinconf->root, "footer")) != NULL) {
 		child = nodes[0]->first_child;
 		while(child != NULL) {
-			default_body(out, top, child, 1, 6);
+			default_body(out, top, child, 1, 0, 6);
 
 			child = child->next;
 		}

src/taiga.h

@@ -62,6 +62,9 @@ int action_help(int argc, char** argv);
 /* action/seed.c */
 int action_seed(int argc, char** argv);
 
+/* action/markdown.c */
+int action_markdown(int argc, char** argv);
+
 /* util.c */
 char* u_strvacat(const char* str, ...);
 char* u_strdup(const char* str);
@@ -78,7 +81,7 @@ int process(const char* top, const char* out, const char* full);
 
 /* default.c */
 void default_head(FILE* out, const char* top, xl_node_t* element, int indent);
-void default_body(FILE* out, const char* top, xl_node_t* element, int spec, int indent); /* pass non-zero to spec if you want links to be relative from top */
+void default_body(FILE* out, const char* top, xl_node_t* element, int spec, int pre, int indent); /* pass non-zero to spec if you want links to be relative from top */
 void default_nav(FILE* out, const char* top, xl_node_t* element, int indent);
 
 /*** skins ***/

src/util.c

@@ -131,7 +131,7 @@ char* u_section_id(xl_node_t* node) {
 
 		id = malloc(64);
 
-		sprintf(id, "%d-%d", unique, crc32(name, strlen(name)));
+		sprintf(id, "section%d-%d", unique, crc32(name, strlen(name)));
 	}
 
 	return id;