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);
}
void cmark_consolidate_text_nodes(cmark_node *root) {
cmark_iter *iter = cmark_iter_new(root);
cmark_strbuf buf = GH_BUF_INIT;
cmark_event_type ev_type;
cmark_node *cur, *tmp, *next;
while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
cur = cmark_iter_get_node(iter);
if (ev_type == CMARK_EVENT_ENTER && cur->type == CMARK_NODE_TEXT &&
cur->next && cur->next->type == CMARK_NODE_TEXT) {
cmark_strbuf_clear(&buf);
cmark_strbuf_put(&buf, cur->as.literal.data, cur->as.literal.len);
tmp = cur->next;
while (tmp && tmp->type == CMARK_NODE_TEXT) {
cmark_iter_next(iter); // advance pointer
cmark_strbuf_put(&buf, tmp->as.literal.data, tmp->as.literal.len);
next = tmp->next;
cmark_node_free(tmp);
tmp = next;
}
cmark_chunk_free(&cur->as.literal);
cur->as.literal = cmark_chunk_buf_detach(&buf);
}
}
cmark_strbuf_free(&buf);
cmark_iter_free(iter);
}
// Like make_str, but parses entities.
// Returns an inline sequence consisting of str and entity elements.
static cmark_node *make_str_with_entities(cmark_chunk *content)
{
cmark_strbuf unescaped = GH_BUF_INIT;
if (houdini_unescape_html(&unescaped, content->data, (size_t)content->len)) {
return make_str(cmark_chunk_buf_detach(&unescaped));
} else {
return make_str(*content);
}
}
// Like make_str, but parses entities.
static cmark_node *make_str_with_entities(cmark_mem *mem,
cmark_chunk *content) {
cmark_strbuf unescaped = CMARK_BUF_INIT(mem);
if (houdini_unescape_html(&unescaped, content->data, content->len)) {
return make_str(mem, cmark_chunk_buf_detach(&unescaped));
} else {
return make_str(mem, *content);
}
}
static cmark_node *www_match(cmark_parser *parser, cmark_node *parent,
cmark_inline_parser *inline_parser) {
cmark_chunk *chunk = cmark_inline_parser_get_chunk(inline_parser);
size_t max_rewind = cmark_inline_parser_get_offset(inline_parser);
uint8_t *data = chunk->data + max_rewind;
size_t size = chunk->len - max_rewind;
int start = cmark_inline_parser_get_column(inline_parser);
size_t link_end;
if (max_rewind > 0 && strchr("*_~(", data[-1]) == NULL &&
!cmark_isspace(data[-1]))
return 0;
if (size < 4 || memcmp(data, "www.", strlen("www.")) != 0)
return 0;
link_end = check_domain(data, size, 0);
if (link_end == 0)
return NULL;
while (link_end < size && !cmark_isspace(data[link_end]))
link_end++;
link_end = autolink_delim(data, link_end);
if (link_end == 0)
return NULL;
cmark_inline_parser_set_offset(inline_parser, (int)(max_rewind + link_end));
cmark_node *node = cmark_node_new_with_mem(CMARK_NODE_LINK, parser->mem);
cmark_strbuf buf;
cmark_strbuf_init(parser->mem, &buf, 10);
cmark_strbuf_puts(&buf, "http://");
cmark_strbuf_put(&buf, data, (bufsize_t)link_end);
node->as.link.url = cmark_chunk_buf_detach(&buf);
cmark_node *text = cmark_node_new_with_mem(CMARK_NODE_TEXT, parser->mem);
text->as.literal =
cmark_chunk_dup(chunk, (bufsize_t)max_rewind, (bufsize_t)link_end);
cmark_node_append_child(node, text);
node->start_line = text->start_line =
node->end_line = text->end_line =
cmark_inline_parser_get_line(inline_parser);
node->start_column = text->start_column = start - 1;
node->end_column = text->end_column = cmark_inline_parser_get_column(inline_parser) - 1;
return node;
}
// Clean a URL: remove surrounding whitespace, 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;
}
houdini_unescape_html_f(&buf, url->data, url->len);
cmark_strbuf_unescape(&buf);
return cmark_chunk_buf_detach(&buf);
}
// Parse an entity or a regular "&" string.
// Assumes the subject has an '&' character at the current position.
static cmark_node *handle_entity(subject *subj) {
cmark_strbuf ent = CMARK_BUF_INIT(subj->mem);
bufsize_t len;
advance(subj);
len = houdini_unescape_ent(&ent, subj->input.data + subj->pos,
subj->input.len - subj->pos);
if (len == 0)
return make_str(subj->mem, cmark_chunk_literal("&"));
subj->pos += len;
return make_str(subj->mem, cmark_chunk_buf_detach(&ent));
}
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);
}
// Parse backtick code section or raw backticks, return an inline.
// Assumes that the subject has a backtick at the current position.
static cmark_node* handle_backticks(subject *subj)
{
cmark_chunk openticks = take_while(subj, isbacktick);
int startpos = subj->pos;
int endpos = scan_to_closing_backticks(subj, openticks.len);
if (endpos == 0) { // not found
subj->pos = startpos; // rewind
return make_str(openticks);
} else {
cmark_strbuf buf = GH_BUF_INIT;
cmark_strbuf_set(&buf, subj->input.data + startpos, endpos - startpos - openticks.len);
cmark_strbuf_trim(&buf);
cmark_strbuf_normalize_whitespace(&buf);
return make_code(cmark_chunk_buf_detach(&buf));
}
}
// Assumes we have a hyphen at the current position.
static cmark_node *handle_hyphen(subject *subj, bool smart) {
cmark_strbuf buf = CMARK_BUF_INIT(NULL);
int startpos = subj->pos;
int en_count = 0;
int em_count = 0;
int numhyphens;
int i;
advance(subj);
if (!smart || peek_char(subj) != '-') {
return make_str(subj->mem, cmark_chunk_literal("-"));
}
while (smart && peek_char(subj) == '-') {
advance(subj);
}
numhyphens = subj->pos - startpos;
buf.mem = subj->mem;
if (numhyphens % 3 == 0) { // if divisible by 3, use all em dashes
em_count = numhyphens / 3;
} else if (numhyphens % 2 == 0) { // if divisible by 2, use all en dashes
en_count = numhyphens / 2;
} else if (numhyphens % 3 == 2) { // use one en dash at end
en_count = 1;
em_count = (numhyphens - 2) / 3;
} else { // use two en dashes at the end
en_count = 2;
em_count = (numhyphens - 4) / 3;
}
for (i = em_count; i > 0; i--) {
cmark_strbuf_puts(&buf, EMDASH);
}
for (i = en_count; i > 0; i--) {
cmark_strbuf_puts(&buf, ENDASH);
}
return make_str(subj->mem, cmark_chunk_buf_detach(&buf));
}
// Clean a URL: remove surrounding whitespace and surrounding <>,
// and remove \ that escape punctuation.
cmark_chunk cmark_clean_url(cmark_chunk *url)
{
cmark_strbuf buf = GH_BUF_INIT;
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);
}
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 postprocess_text(cmark_parser *parser, cmark_node *text, int offset) {
size_t link_end;
uint8_t *data = text->as.literal.data,
*at;
size_t size = text->as.literal.len;
int rewind, max_rewind,
nb = 0, np = 0, ns = 0;
if (offset < 0 || (size_t)offset >= size)
return;
data += offset;
size -= offset;
at = (uint8_t *)memchr(data, '@', size);
if (!at)
return;
max_rewind = (int)(at - data);
data += max_rewind;
size -= max_rewind;
for (rewind = 0; rewind < max_rewind; ++rewind) {
uint8_t c = data[-rewind - 1];
if (cmark_isalnum(c))
continue;
if (strchr(".+-_", c) != NULL)
continue;
if (c == '/')
ns++;
break;
}
if (rewind == 0 || ns > 0) {
postprocess_text(parser, text, max_rewind + 1 + offset);
return;
}
for (link_end = 0; link_end < size; ++link_end) {
uint8_t c = data[link_end];
if (cmark_isalnum(c))
continue;
if (c == '@')
nb++;
else if (c == '.' && link_end < size - 1 && cmark_isalnum(data[link_end + 1]))
np++;
else if (c != '-' && c != '_')
break;
}
if (link_end < 2 || nb != 1 || np == 0 ||
(!cmark_isalpha(data[link_end - 1]) && data[link_end - 1] != '.')) {
postprocess_text(parser, text, max_rewind + 1 + offset);
return;
}
link_end = autolink_delim(data, link_end);
if (link_end == 0) {
postprocess_text(parser, text, max_rewind + 1 + offset);
return;
}
cmark_chunk_to_cstr(parser->mem, &text->as.literal);
cmark_node *link_node = cmark_node_new_with_mem(CMARK_NODE_LINK, parser->mem);
cmark_strbuf buf;
cmark_strbuf_init(parser->mem, &buf, 10);
cmark_strbuf_puts(&buf, "mailto:");
cmark_strbuf_put(&buf, data - rewind, (bufsize_t)(link_end + rewind));
link_node->as.link.url = cmark_chunk_buf_detach(&buf);
cmark_node *link_text = cmark_node_new_with_mem(CMARK_NODE_TEXT, parser->mem);
cmark_chunk email = cmark_chunk_dup(
&text->as.literal,
offset + max_rewind - rewind,
(bufsize_t)(link_end + rewind));
cmark_chunk_to_cstr(parser->mem, &email);
link_text->as.literal = email;
cmark_node_append_child(link_node, link_text);
cmark_node_insert_after(text, link_node);
cmark_node *post = cmark_node_new_with_mem(CMARK_NODE_TEXT, parser->mem);
post->as.literal = cmark_chunk_dup(&text->as.literal,
(bufsize_t)(offset + max_rewind + link_end),
(bufsize_t)(size - link_end));
cmark_chunk_to_cstr(parser->mem, &post->as.literal);
cmark_node_insert_after(link_node, post);
text->as.literal.len = offset + max_rewind - rewind;
text->as.literal.data[text->as.literal.len] = 0;
postprocess_text(parser, post, 0);
}
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;
}
