cmark_chunk cmark_clean_title(cmark_chunk *title)
{
cmark_strbuf buf = GH_BUF_INIT;
unsigned char first, last;
if (title->len == 0) {
cmark_chunk result = CMARK_CHUNK_EMPTY;
return result;
}
first = title->data[0];
last = title->data[title->len - 1];
// remove surrounding quotes if any:
if ((first == '\'' && last == '\'') ||
(first == '(' && last == ')') ||
(first == '"' && last == '"')) {
houdini_unescape_html_f(&buf, title->data + 1, title->len - 2);
} else {
houdini_unescape_html_f(&buf, title->data, title->len);
}
cmark_strbuf_unescape(&buf);
return cmark_chunk_buf_detach(&buf);
}
// Clean a URL: remove surrounding whitespace and surrounding <>,
// and remove \ that escape punctuation.
unsigned char *cmark_clean_url(cmark_chunk *url)
{
cmark_strbuf buf = GH_BUF_INIT;
cmark_chunk_trim(url);
if (url->len == 0)
return NULL;
if (url->data[0] == '<' && url->data[url->len - 1] == '>') {
houdini_unescape_html_f(&buf, url->data + 1, url->len - 2);
} else {
houdini_unescape_html_f(&buf, url->data, url->len);
}
cmark_strbuf_unescape(&buf);
return cmark_strbuf_detach(&buf);
}
// Clean a URL: remove surrounding whitespace and surrounding <>,
// and remove \ that escape punctuation.
cmark_chunk cmark_clean_url(cmark_mem *mem, cmark_chunk *url) {
cmark_strbuf buf = CMARK_BUF_INIT(mem);
cmark_chunk_trim(url);
if (url->len == 0) {
cmark_chunk result = CMARK_CHUNK_EMPTY;
return result;
}
if (url->data[0] == '<' && url->data[url->len - 1] == '>') {
houdini_unescape_html_f(&buf, url->data + 1, url->len - 2);
} else {
houdini_unescape_html_f(&buf, url->data, url->len);
}
cmark_strbuf_unescape(&buf);
return cmark_chunk_buf_detach(&buf);
}
unsigned char *cmark_clean_title(cmark_chunk *title)
{
cmark_strbuf buf = GH_BUF_INIT;
unsigned char first, last;
if (title->len == 0)
return NULL;
first = title->data[0];
last = title->data[title->len - 1];
// remove surrounding quotes if any:
if ((first == '\'' && last == '\'') ||
(first == '(' && last == ')') ||
(first == '"' && last == '"')) {
houdini_unescape_html_f(&buf, title->data + 1, title->len - 2);
} else {
houdini_unescape_html_f(&buf, title->data, title->len);
}
cmark_strbuf_unescape(&buf);
return buf.size == 0 ? NULL : cmark_strbuf_detach(&buf);
}
static unsigned char *cmark_clean_autolink(cmark_chunk *url, int is_email)
{
cmark_strbuf buf = GH_BUF_INIT;
cmark_chunk_trim(url);
if (url->len == 0)
return NULL;
if (is_email)
cmark_strbuf_puts(&buf, "mailto:");
houdini_unescape_html_f(&buf, url->data, url->len);
return cmark_strbuf_detach(&buf);
}
static cmark_chunk cmark_clean_autolink(cmark_mem *mem, cmark_chunk *url,
int is_email) {
cmark_strbuf buf = CMARK_BUF_INIT(mem);
cmark_chunk_trim(url);
if (url->len == 0) {
cmark_chunk result = CMARK_CHUNK_EMPTY;
return result;
}
if (is_email)
cmark_strbuf_puts(&buf, "mailto:");
houdini_unescape_html_f(&buf, url->data, url->len);
return cmark_chunk_buf_detach(&buf);
}
static cmark_node*
finalize(cmark_parser *parser, cmark_node* b)
{
int pos;
cmark_node* item;
cmark_node* subitem;
cmark_node* parent;
parent = b->parent;
assert(b->open); // shouldn't call finalize on closed blocks
b->open = false;
if (parser->curline->size == 0) {
// end of input - line number has not been incremented
b->end_line = parser->line_number;
b->end_column = parser->last_line_length;
} else if (b->type == NODE_DOCUMENT ||
(b->type == NODE_CODE_BLOCK && b->as.code.fenced) ||
(b->type == NODE_HEADER && b->as.header.setext)) {
b->end_line = parser->line_number;
b->end_column = parser->curline->size;
if (b->end_column && parser->curline->ptr[b->end_column - 1] == '\n')
b->end_column--;
if (b->end_column && parser->curline->ptr[b->end_column - 1] == '\r')
b->end_column--;
} else {
b->end_line = parser->line_number - 1;
b->end_column = parser->last_line_length;
}
switch (b->type) {
case NODE_PARAGRAPH:
while (cmark_strbuf_at(&b->string_content, 0) == '[' &&
(pos = cmark_parse_reference_inline(&b->string_content, parser->refmap))) {
cmark_strbuf_drop(&b->string_content, pos);
}
if (is_blank(&b->string_content, 0)) {
// remove blank node (former reference def)
cmark_node_free(b);
}
break;
case NODE_CODE_BLOCK:
if (!b->as.code.fenced) { // indented code
remove_trailing_blank_lines(&b->string_content);
cmark_strbuf_putc(&b->string_content, '\n');
} else {
// first line of contents becomes info
for (pos = 0; pos < b->string_content.size; ++pos) {
if (b->string_content.ptr[pos] == '\r' ||
b->string_content.ptr[pos] == '\n')
break;
}
assert(pos < b->string_content.size);
cmark_strbuf tmp = GH_BUF_INIT;
houdini_unescape_html_f(
&tmp,
b->string_content.ptr,
pos
);
cmark_strbuf_trim(&tmp);
cmark_strbuf_unescape(&tmp);
b->as.code.info = cmark_chunk_buf_detach(&tmp);
if (b->string_content.ptr[pos] == '\r')
pos += 1;
if (b->string_content.ptr[pos] == '\n')
pos += 1;
cmark_strbuf_drop(&b->string_content, pos);
}
b->as.code.literal = cmark_chunk_buf_detach(&b->string_content);
break;
case NODE_HTML:
b->as.literal = cmark_chunk_buf_detach(&b->string_content);
break;
case NODE_LIST: // determine tight/loose status
b->as.list.tight = true; // tight by default
item = b->first_child;
while (item) {
// check for non-final non-empty list item ending with blank line:
if (item->last_line_blank && item->next) {
b->as.list.tight = false;
break;
}
// recurse into children of list item, to see if there are
// spaces between them:
subitem = item->first_child;
while (subitem) {
if (ends_with_blank_line(subitem) &&
(item->next || subitem->next)) {
b->as.list.tight = false;
break;
}
subitem = subitem->next;
}
if (!(b->as.list.tight)) {
break;
}
item = item->next;
}
break;
default:
break;
}
return parent;
}
static void finalize(cmark_doc_parser *parser, cmark_node* b, int line_number)
{
int firstlinelen;
int pos;
cmark_node* item;
cmark_node* subitem;
if (!b->open)
return; // don't do anything if the cmark_node is already closed
b->open = false;
if (line_number > b->start_line) {
b->end_line = line_number - 1;
} else {
b->end_line = line_number;
}
switch (b->type) {
case NODE_PARAGRAPH:
pos = 0;
while (strbuf_at(&b->string_content, 0) == '[' &&
(pos = parse_reference_inline(&b->string_content, parser->refmap))) {
strbuf_drop(&b->string_content, pos);
}
if (is_blank(&b->string_content, 0)) {
b->type = NODE_REFERENCE_DEF;
}
break;
case NODE_INDENTED_CODE:
remove_trailing_blank_lines(&b->string_content);
strbuf_putc(&b->string_content, '\n');
break;
case NODE_FENCED_CODE:
// first line of contents becomes info
firstlinelen = strbuf_strchr(&b->string_content, '\n', 0);
strbuf_init(&b->as.code.info, 0);
houdini_unescape_html_f(
&b->as.code.info,
b->string_content.ptr,
firstlinelen
);
strbuf_drop(&b->string_content, firstlinelen + 1);
strbuf_trim(&b->as.code.info);
strbuf_unescape(&b->as.code.info);
break;
case NODE_LIST: // determine tight/loose status
b->as.list.tight = true; // tight by default
item = b->first_child;
while (item) {
// check for non-final non-empty list item ending with blank line:
if (item->last_line_blank && item->next) {
b->as.list.tight = false;
break;
}
// recurse into children of list item, to see if there are
// spaces between them:
subitem = item->first_child;
while (subitem) {
if (ends_with_blank_line(subitem) &&
(item->next || subitem->next)) {
b->as.list.tight = false;
break;
}
subitem = subitem->next;
}
if (!(b->as.list.tight)) {
break;
}
item = item->next;
}
break;
default:
break;
}
}
static cmark_node*
finalize(cmark_parser *parser, cmark_node* b, int line_number)
{
int firstlinelen;
int pos;
cmark_node* item;
cmark_node* subitem;
cmark_node* parent;
parent = b->parent;
// don't do anything if the cmark_node is already closed
if (!b->open)
return parent;
b->open = false;
if (line_number > b->start_line) {
b->end_line = line_number - 1;
} else {
b->end_line = line_number;
}
switch (b->type) {
case NODE_PARAGRAPH:
while (cmark_strbuf_at(&b->string_content, 0) == '[' &&
(pos = cmark_parse_reference_inline(&b->string_content, parser->refmap))) {
cmark_strbuf_drop(&b->string_content, pos);
}
if (is_blank(&b->string_content, 0)) {
// remove blank node (former reference def)
cmark_node_free(b);
}
break;
case NODE_CODE_BLOCK:
if (!b->as.code.fenced) { // indented code
remove_trailing_blank_lines(&b->string_content);
cmark_strbuf_putc(&b->string_content, '\n');
} else {
// first line of contents becomes info
firstlinelen = cmark_strbuf_strchr(&b->string_content, '\n', 0);
cmark_strbuf tmp = GH_BUF_INIT;
houdini_unescape_html_f(
&tmp,
b->string_content.ptr,
firstlinelen
);
cmark_strbuf_trim(&tmp);
cmark_strbuf_unescape(&tmp);
b->as.code.info = cmark_chunk_buf_detach(&tmp);
cmark_strbuf_drop(&b->string_content, firstlinelen + 1);
}
b->as.code.literal = cmark_chunk_buf_detach(&b->string_content);
break;
case NODE_HTML:
b->as.literal = cmark_chunk_buf_detach(&b->string_content);
break;
case NODE_LIST: // determine tight/loose status
b->as.list.tight = true; // tight by default
item = b->first_child;
while (item) {
// check for non-final non-empty list item ending with blank line:
if (item->last_line_blank && item->next) {
b->as.list.tight = false;
break;
}
// recurse into children of list item, to see if there are
// spaces between them:
subitem = item->first_child;
while (subitem) {
if (ends_with_blank_line(subitem) &&
(item->next || subitem->next)) {
b->as.list.tight = false;
break;
}
subitem = subitem->next;
}
if (!(b->as.list.tight)) {
break;
}
item = item->next;
}
break;
default:
break;
}
return parent;
}
