3 files changed, 67 insertions, 4 deletions

diff --git a/src/buffer.c b/src/buffer.c
index 2ca0fd1..af77b70 100644
--- a/src/buffer.c
+++ b/src/buffer.c
@@ -62,6 +62,13 @@ hoedown_buffer_init(
 	buf->buffer_free = buffer_free;
 }
 
+void
+hoedown_buffer_uninit(hoedown_buffer *buf)
+{
+	assert(buf && buf->unit);
+	buf->data_free(buf->data);
+}
+
 hoedown_buffer *
 hoedown_buffer_new(size_t unit)
 {
@@ -74,6 +81,7 @@ void
 hoedown_buffer_free(hoedown_buffer *buf)
 {
 	if (!buf) return;
+	assert(buf && buf->unit);
 
 	buf->data_free(buf->data);
 
@@ -138,6 +146,19 @@ hoedown_buffer_putc(hoedown_buffer *buf, uint8_t c)
 	buf->size += 1;
 }
 
+int
+hoedown_buffer_putf(hoedown_buffer *buf, FILE *file)
+{
+	assert(buf && buf->unit);
+
+	while (!(feof(file) || ferror(file))) {
+		hoedown_buffer_grow(buf, buf->size + buf->unit);
+		buf->size += fread(buf->data + buf->size, 1, buf->unit, file);
+	}
+
+	return ferror(file);
+}
+
 void
 hoedown_buffer_set(hoedown_buffer *buf, const uint8_t *data, size_t size)
 {
@@ -174,8 +195,6 @@ hoedown_buffer_prefix(const hoedown_buffer *buf, const char *prefix)
 {
 	size_t i;
 
-	assert(buf && buf->unit);
-
 	for (i = 0; i < buf->size; ++i) {
 		if (prefix[i] == 0)
 			return 0;
@@ -253,3 +272,37 @@ hoedown_buffer_printf(hoedown_buffer *buf, const char *fmt, ...)
 
 	buf->size += n;
 }
+
+void hoedown_buffer_put_utf8(hoedown_buffer *buf, unsigned int c) {
+	unsigned char unichar[4];
+
+	assert(buf && buf->unit);
+
+	if (c < 0x80) {
+		hoedown_buffer_putc(buf, c);
+	}
+	else if (c < 0x800) {
+		unichar[0] = 192 + (c / 64);
+		unichar[1] = 128 + (c % 64);
+		hoedown_buffer_put(buf, unichar, 2);
+	}
+	else if (c - 0xd800u < 0x800) {
+		HOEDOWN_BUFPUTSL(buf, "\xef\xbf\xbd");
+	}
+	else if (c < 0x10000) {
+		unichar[0] = 224 + (c / 4096);
+		unichar[1] = 128 + (c / 64) % 64;
+		unichar[2] = 128 + (c % 64);
+		hoedown_buffer_put(buf, unichar, 3);
+	}
+	else if (c < 0x110000) {
+		unichar[0] = 240 + (c / 262144);
+		unichar[1] = 128 + (c / 4096) % 64;
+		unichar[2] = 128 + (c / 64) % 64;
+		unichar[3] = 128 + (c % 64);
+		hoedown_buffer_put(buf, unichar, 4);
+	}
+	else {
+		HOEDOWN_BUFPUTSL(buf, "\xef\xbf\xbd");
+	}
+}
diff --git a/src/buffer.h b/src/buffer.h
index fcda83c..d7703f8 100644
--- a/src/buffer.h
+++ b/src/buffer.h
@@ -59,6 +59,9 @@ void hoedown_buffer_init(
 	hoedown_free_callback buffer_free
 );
 
+/* hoedown_buffer_uninit: uninitialize an existing buffer */
+void hoedown_buffer_uninit(hoedown_buffer *buf);
+
 /* hoedown_buffer_new: allocate a new buffer */
 hoedown_buffer *hoedown_buffer_new(size_t unit) __attribute__ ((malloc));
 
@@ -77,6 +80,9 @@ void hoedown_buffer_puts(hoedown_buffer *buf, const char *str);
 /* hoedown_buffer_putc: append a single char to a buffer */
 void hoedown_buffer_putc(hoedown_buffer *buf, uint8_t c);
 
+/* hoedown_buffer_putf: read from a file and append to a buffer, until EOF or error */
+int hoedown_buffer_putf(hoedown_buffer *buf, FILE* file);
+
 /* hoedown_buffer_set: replace the buffer's contents with raw data */
 void hoedown_buffer_set(hoedown_buffer *buf, const uint8_t *data, size_t size);
 
@@ -101,6 +107,9 @@ const char *hoedown_buffer_cstr(hoedown_buffer *buf);
 /* hoedown_buffer_printf: formatted printing to a buffer */
 void hoedown_buffer_printf(hoedown_buffer *buf, const char *fmt, ...) __attribute__ ((format (printf, 2, 3)));
 
+/* hoedown_buffer_put_utf8: put a Unicode character encoded as UTF-8 */
+void hoedown_buffer_put_utf8(hoedown_buffer *buf, unsigned int codepoint);
+
 /* hoedown_buffer_free: free the buffer */
 void hoedown_buffer_free(hoedown_buffer *buf);
 
diff --git a/src/document.c b/src/document.c
index 17feb24..27c17b2 100644
--- a/src/document.c
+++ b/src/document.c
@@ -459,7 +459,7 @@ tag_length(uint8_t *data, size_t size, hoedown_autolink_type *autolink)
 static void
 parse_inline(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t size)
 {
-	size_t i = 0, end = 0;
+	size_t i = 0, end = 0, consumed = 0;
 	hoedown_buffer work = { 0, 0, 0, 0, NULL, NULL, NULL };
 	uint8_t *active_char = doc->active_char;
 
@@ -483,12 +483,13 @@ parse_inline(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t si
 		if (end >= size) break;
 		i = end;
 
-		end = markdown_char_ptrs[ (int)active_char[data[end]] ](ob, doc, data + i, i, size - i);
+		end = markdown_char_ptrs[ (int)active_char[data[end]] ](ob, doc, data + i, i - consumed, size - i);
 		if (!end) /* no action from the callback */
 			end = i + 1;
 		else {
 			i += end;
 			end = i;
+			consumed = i;
 		}
 	}
 }