Inline html

src/include/mdutil.h

@@ -54,5 +54,7 @@ struct linedata {
 void identifyline(char *line, struct linedata *prev, struct linedata *ret);
 /* prev is almost never used, but sometimes it is. */
 char *realcontent(char *line, struct linedata *data);
+int isgenerictag(char *text);
+/* Identifies open and close tags */
 
 #endif

src/inlines.c

@@ -19,6 +19,7 @@
 #include <string.h>
 
 #include <util.h>
+#include <mdutil.h>
 #include <inlines.h>
 
 static const char *punctuation = "!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~";
@@ -38,12 +39,15 @@ static int writelink(char *data, int i, size_t len, FILE *out);
 static int writeimage(char *data, int i, size_t len, FILE *out);
 static int writeautolink(char *data, int i, size_t len, FILE *out);
 static int writehardbreak(char *data, int i, size_t len, FILE *out);
+static int writerawhtml(char *data, int i, size_t len, FILE *out);
 static int getlinkinfo(char *data, int i, size_t len,
 		int *textstart, int *textend,
 		int *titlestart, int *titleend,
 		int *linkstart, int *linkend);
 static void writeescaped(char *data, size_t len, FILE *out);
 static void writechescape(char c, FILE *out);
+static int inlinehtmlcase(char *data, char *start, char *end, FILE *out);
+/* Returns length of raw html */
 
 void writeline(char *data, FILE *out) {
 	writedata(data, strlen(data), out);
@@ -63,6 +67,8 @@ void writedata(char *data, size_t len, FILE *out) {
 			goto special;
 		if ((newi = writehardbreak(data, i, len, out)) >= 0)
 			goto special;
+		if ((newi = writerawhtml(data, i, len, out)) >= 0)
+			goto special;
 		if (data[i] == '\\') {
 			if (strchr(punctuation, data[i + 1]) == NULL)
 				writechescape('\\', out);
@@ -202,6 +208,48 @@ static int writehardbreak(char *data, int i, size_t len, FILE *out) {
 	return i + codelen;
 }
 
+static int writerawhtml(char *data, int i, size_t len, FILE *out) {
+	int taglen;
+	data += i;
+	taglen = isgenerictag(data);
+	if (taglen > 0) {
+		fwrite(data, 1, taglen, out);
+		return i + taglen;
+	}
+
+	if (memcmp(data, "<!--", 4) == 0) {
+		char *end, *text;
+		int j;
+		text = data + 4;
+		end = strstr(text, "-->");
+		if (end == NULL)
+			goto notcomment;
+		taglen = end - text;
+		if (text[0] == '>' || memcmp(text, "->", 2) == 0)
+			goto notcomment;
+		if (text[taglen - 1] == '-') {
+			goto notcomment;
+		}
+		for (j = 0; j < taglen - 1; ++j) {
+			if (memcmp(text + j, "--", 2) == 0)
+				goto notcomment;
+		}
+
+		fwrite(data, 1, taglen + 7, out);
+
+		return i + taglen + 7;
+	}
+notcomment:
+	if ((taglen = inlinehtmlcase(data, "<?", "?>", out)) > 0)
+		return i + taglen;
+	if ((taglen = inlinehtmlcase(data, "<!", ">", out)) > 0)
+		return i + taglen;
+	if ((taglen = inlinehtmlcase(data, "<![CDATA[", "]]>", out)) > 0)
+		return i + taglen;
+
+	return -1;
+}
+
 static int getlinkinfo(char *data, int i, size_t len,
 		int *textstart, int *textend,
 		int *titlestart, int *titleend,
@@ -355,3 +403,18 @@ static void writechescape(char c, FILE *out) {
 	}
 	fputc(c, out);
 }
+
+static int inlinehtmlcase(char *data, char *start, char *end, FILE *out) {
+	int i;
+	int len;
+	char *endloc;
+	for (i = 0; start[i] != '\0'; ++i)
+		if (data[i] != start[i])
+			return 0;
+	endloc = strstr(data, end);
+	if (endloc == NULL)
+		return 0;
+	len = endloc - data + strlen(end);
+	fwrite(data, 1, len, out);
+	return len;
+}

src/mdutil.c

@@ -127,7 +127,6 @@ notheader:
 
 	if (line[0] == '<') {
 		char *testline;
-		int isopen;
 		testline = line + 1;
 		for (i = 0; i < LEN(concretetags); ++i) {
 			char *aftertag;
@@ -140,12 +139,8 @@ notheader:
 				return;
 			}
 		}
-		if (testline[0] == '/') {
+		if (testline[0] == '/')
 			++testline;
-			isopen = 0;
-		}
-		else
-			isopen = 1;
 		for (i = 0; i < LEN(skeletontags); ++i) {
 			char *aftertag;
 			aftertag = after(skeletontags[i], testline);
@@ -160,73 +155,15 @@ notheader:
 			}
 		}
 
-		/* < || </ */
+		if (isgenerictag(line)) {
-
-		if (!isalpha(testline[0]))
-			goto nothtml;
-		++testline;
-		while (isalnum(testline[0]) || testline[0] == '-')
-			++testline;
-
-		/* <[tag name] || </[tag name] */
-
-		for (;;) {
-			char *newtestline;
-			newtestline = truncate(testline);
-
-			if (!isalpha(newtestline[0]) &&
-					strchr("_:", newtestline[0]) == NULL)
-				break;
-			++newtestline;
-			while (isalnum(newtestline[0]) ||
-					strchr("_.:-", newtestline[0]) != NULL)
-				++newtestline;
-			/* Swallow attribute name */
-
-			newtestline = truncate(newtestline);
-			if (newtestline[0] == '=') {
-				char start;
-				++newtestline;
-				newtestline = truncate(newtestline);
-				start = newtestline[0];
-				switch (start) {
-				case '\'': case '"':
-					while (newtestline[0] != start &&
-							newtestline[0] != '\0')
-						++newtestline;
-					if (newtestline[0] == '\0')
-						goto nothtml;
-					break;
-				/* Swallow single/double quoted attribute
-				 * value */
-				default:
-					while (strchr("\"'=<>`", newtestline[0])
-							== NULL &&
-							newtestline[0] != '\0')
-						++newtestline;
-					if (newtestline[0] == '\0')
-						goto nothtml;
-					break;
-				/* Swallow unquoted attribute value */
-				}
-			}
-			/* Swallow attribute value */
-
-			testline = newtestline;
-		}
-		
-		/* <[tag name][attribute]* || </tag name][attribute]* */
-		if (isopen && testline[0] == '/')
-			++testline;
-		if (testline[0] == '>') {
 			ret->type = GENERICTAG;
 			ret->data.isfirst = 1;
 			return;
 		}
 	}
-nothtml:
 
 	ret->type = PLAIN;
+	ret->data.isfirst = 1;
 	return;
 }
 
@@ -246,6 +183,82 @@ char *realcontent(char *line, struct linedata *data) {
 	return NULL;
 }
 
+int isgenerictag(char *text) {
+	int isopen;
+	char *initialtext;
+	initialtext = text;
+	if (text[0] != '<')
+		return 0;
+	++text;
+	if (text[0] == '/') {
+		++text;
+		isopen = 0;
+	}
+	else
+		isopen = 1;
+
+	/* < || </ */
+
+	if (!isalpha(text[0]))
+		return 0;
+	while (isalnum(text[0]) || text[0] == '-')
+		++text;
+
+	/* <[tag name] || </[tag name] */
+
+	for (;;) {
+		char *newtext;
+		newtext = truncate(text);
+
+		if (!isalpha(newtext[0]) &&
+				strchr("_:", newtext[0]) == NULL)
+			break;
+		++newtext;
+		while (isalnum(newtext[0]) ||
+				strchr("_.:-", newtext[0]) != NULL)
+			++newtext;
+		/* Swallow attribute name */
+
+		newtext = truncate(newtext);
+		if (newtext[0] == '=') {
+			char start;
+			++newtext;
+			newtext = truncate(newtext);
+			start = newtext[0];
+			switch (start) {
+			case '\'': case '"':
+				while (newtext[0] != start &&
+						newtext[0] != '\0')
+					++newtext;
+				if (newtext[0] == '\0')
+					return 0;
+				break;
+			/* Swallow single/double quoted attribute
+			 * value */
+			default:
+				while (strchr("\"'=<>`", newtext[0])
+						== NULL &&
+						newtext[0] != '\0')
+					++newtext;
+				if (newtext[0] == '\0')
+					return 0;
+				break;
+			/* Swallow unquoted attribute value */
+			}
+		}
+		/* Swallow attribute value */
+
+		text = newtext;
+	}
+	
+	/* <[tag name][attribute]* || </tag name][attribute]* */
+	if (isopen && text[0] == '/')
+		++text;
+	if (text[0] == '>')
+		return text - initialtext + 1;
+	return 0;
+}
+
 static char *truncate(char *str) {
 	while (isspace(str[0]))
 		++str;