// Return a link, an image, or a literal close bracket.
static cmark_node *handle_close_bracket(subject *subj) {
bufsize_t initial_pos, after_link_text_pos;
bufsize_t endurl, starttitle, endtitle, endall;
bufsize_t sps, n;
cmark_reference *ref = NULL;
cmark_chunk url_chunk, title_chunk;
cmark_chunk url, title;
bracket *opener;
cmark_node *inl;
cmark_chunk raw_label;
int found_label;
cmark_node *tmp, *tmpnext;
bool is_image;
advance(subj); // advance past ]
initial_pos = subj->pos;
// get last [ or ![
opener = subj->last_bracket;
if (opener == NULL) {
return make_str(subj->mem, cmark_chunk_literal("]"));
}
if (!opener->active) {
// take delimiter off stack
pop_bracket(subj);
return make_str(subj->mem, cmark_chunk_literal("]"));
}
// If we got here, we matched a potential link/image text.
// Now we check to see if it's a link/image.
is_image = opener->image;
after_link_text_pos = subj->pos;
// First, look for an inline link.
if (peek_char(subj) == '(' &&
((sps = scan_spacechars(&subj->input, subj->pos + 1)) > -1) &&
((n = manual_scan_link_url(&subj->input, subj->pos + 1 + sps, &url_chunk)) > -1)) {
// try to parse an explicit link:
endurl = subj->pos + 1 + sps + n;
starttitle = endurl + scan_spacechars(&subj->input, endurl);
// ensure there are spaces btw url and title
endtitle = (starttitle == endurl)
? starttitle
: starttitle + scan_link_title(&subj->input, starttitle);
endall = endtitle + scan_spacechars(&subj->input, endtitle);
if (peek_at(subj, endall) == ')') {
subj->pos = endall + 1;
title_chunk =
cmark_chunk_dup(&subj->input, starttitle, endtitle - starttitle);
url = cmark_clean_url(subj->mem, &url_chunk);
title = cmark_clean_title(subj->mem, &title_chunk);
cmark_chunk_free(subj->mem, &url_chunk);
cmark_chunk_free(subj->mem, &title_chunk);
goto match;
} else {
// it could still be a shortcut reference link
subj->pos = after_link_text_pos;
}
}
// Next, look for a following [link label] that matches in refmap.
// skip spaces
raw_label = cmark_chunk_literal("");
found_label = link_label(subj, &raw_label);
if (!found_label) {
// If we have a shortcut reference link, back up
// to before the spacse we skipped.
subj->pos = initial_pos;
}
if ((!found_label || raw_label.len == 0) && !opener->bracket_after) {
cmark_chunk_free(subj->mem, &raw_label);
raw_label = cmark_chunk_dup(&subj->input, opener->position,
initial_pos - opener->position - 1);
found_label = true;
}
if (found_label) {
ref = cmark_reference_lookup(subj->refmap, &raw_label);
cmark_chunk_free(subj->mem, &raw_label);
}
if (ref != NULL) { // found
url = chunk_clone(subj->mem, &ref->url);
title = chunk_clone(subj->mem, &ref->title);
goto match;
} else {
goto noMatch;
}
noMatch:
// If we fall through to here, it means we didn't match a link:
pop_bracket(subj); // remove this opener from delimiter list
subj->pos = initial_pos;
return make_str(subj->mem, cmark_chunk_literal("]"));
match:
inl = make_simple(subj->mem, is_image ? CMARK_NODE_IMAGE : CMARK_NODE_LINK);
inl->as.link.url = url;
inl->as.link.title = title;
cmark_node_insert_before(opener->inl_text, inl);
// Add link text:
tmp = opener->inl_text->next;
while (tmp) {
tmpnext = tmp->next;
cmark_node_append_child(inl, tmp);
tmp = tmpnext;
}
// Free the bracket [:
cmark_node_free(opener->inl_text);
process_emphasis(subj, opener->previous_delimiter);
pop_bracket(subj);
// Now, if we have a link, we also want to deactivate earlier link
// delimiters. (This code can be removed if we decide to allow links
// inside links.)
if (!is_image) {
opener = subj->last_bracket;
while (opener != NULL) {
if (!opener->image) {
if (!opener->active) {
break;
} else {
opener->active = false;
}
}
opener = opener->previous;
}
}
return NULL;
}
// Parse reference. Assumes string begins with '[' character.
// Modify refmap if a reference is encountered.
// Return 0 if no reference found, otherwise position of subject
// after reference is parsed.
bufsize_t cmark_parse_reference_inline(cmark_mem *mem, cmark_strbuf *input,
cmark_reference_map *refmap) {
subject subj;
cmark_chunk lab;
cmark_chunk url;
cmark_chunk title;
bufsize_t matchlen = 0;
bufsize_t beforetitle;
subject_from_buf(mem, &subj, input, NULL);
// parse label:
if (!link_label(&subj, &lab) || lab.len == 0)
return 0;
// colon:
if (peek_char(&subj) == ':') {
advance(&subj);
} else {
return 0;
}
// parse link url:
spnl(&subj);
if ((matchlen = manual_scan_link_url(&subj.input, subj.pos, &url)) > -1 &&
url.len > 0) {
subj.pos += matchlen;
} else {
return 0;
}
// parse optional link_title
beforetitle = subj.pos;
spnl(&subj);
matchlen = scan_link_title(&subj.input, subj.pos);
if (matchlen) {
title = cmark_chunk_dup(&subj.input, subj.pos, matchlen);
subj.pos += matchlen;
} else {
subj.pos = beforetitle;
title = cmark_chunk_literal("");
}
// parse final spaces and newline:
skip_spaces(&subj);
if (!skip_line_end(&subj)) {
if (matchlen) { // try rewinding before title
subj.pos = beforetitle;
skip_spaces(&subj);
if (!skip_line_end(&subj)) {
return 0;
}
} else {
return 0;
}
}
// insert reference into refmap
cmark_reference_create(refmap, &lab, &url, &title);
return subj.pos;
}
static void process_emphasis(subject *subj, delimiter *stack_bottom) {
delimiter *closer = subj->last_delim;
delimiter *opener;
delimiter *old_closer;
bool opener_found;
int openers_bottom_index;
delimiter *openers_bottom[6] = {stack_bottom, stack_bottom, stack_bottom,
stack_bottom, stack_bottom, stack_bottom};
// move back to first relevant delim.
while (closer != NULL && closer->previous != stack_bottom) {
closer = closer->previous;
}
// now move forward, looking for closers, and handling each
while (closer != NULL) {
if (closer->can_close) {
switch (closer->delim_char) {
case '"':
openers_bottom_index = 0;
break;
case '\'':
openers_bottom_index = 1;
break;
case '_':
openers_bottom_index = 2;
break;
case '*':
openers_bottom_index = 3 + (closer->length % 3);
break;
default:
assert(false);
}
// Now look backwards for first matching opener:
opener = closer->previous;
opener_found = false;
while (opener != NULL && opener != openers_bottom[openers_bottom_index]) {
if (opener->can_open && opener->delim_char == closer->delim_char) {
// interior closer of size 2 can't match opener of size 1
// or of size 1 can't match 2
if (!(closer->can_open || opener->can_close) ||
((opener->length + closer->length) % 3) != 0) {
opener_found = true;
break;
}
}
opener = opener->previous;
}
old_closer = closer;
if (closer->delim_char == '*' || closer->delim_char == '_') {
if (opener_found) {
closer = S_insert_emph(subj, opener, closer);
} else {
closer = closer->next;
}
} else if (closer->delim_char == '\'') {
cmark_chunk_free(subj->mem, &closer->inl_text->as.literal);
closer->inl_text->as.literal = cmark_chunk_literal(RIGHTSINGLEQUOTE);
if (opener_found) {
cmark_chunk_free(subj->mem, &opener->inl_text->as.literal);
opener->inl_text->as.literal = cmark_chunk_literal(LEFTSINGLEQUOTE);
}
closer = closer->next;
} else if (closer->delim_char == '"') {
cmark_chunk_free(subj->mem, &closer->inl_text->as.literal);
closer->inl_text->as.literal = cmark_chunk_literal(RIGHTDOUBLEQUOTE);
if (opener_found) {
cmark_chunk_free(subj->mem, &opener->inl_text->as.literal);
opener->inl_text->as.literal = cmark_chunk_literal(LEFTDOUBLEQUOTE);
}
closer = closer->next;
}
if (!opener_found) {
// set lower bound for future searches for openers
openers_bottom[openers_bottom_index] = old_closer->previous;
if (!old_closer->can_open) {
// we can remove a closer that can't be an
// opener, once we've seen there's no
// matching opener:
remove_delimiter(subj, old_closer);
}
}
} else {
closer = closer->next;
}
}
// free all delimiters in list until stack_bottom:
while (subj->last_delim != NULL && subj->last_delim != stack_bottom) {
remove_delimiter(subj, subj->last_delim);
}
}
static void process_emphasis(subject *subj, delimiter *stack_bottom) {
delimiter *closer = subj->last_delim;
delimiter *opener;
delimiter *old_closer;
bool opener_found;
bool odd_match;
delimiter *openers_bottom[128];
// initialize openers_bottom:
openers_bottom['*'] = stack_bottom;
openers_bottom['_'] = stack_bottom;
openers_bottom['\''] = stack_bottom;
openers_bottom['"'] = stack_bottom;
// move back to first relevant delim.
while (closer != NULL && closer->previous != stack_bottom) {
closer = closer->previous;
}
// now move forward, looking for closers, and handling each
while (closer != NULL) {
if (closer->can_close) {
// Now look backwards for first matching opener:
opener = closer->previous;
opener_found = false;
odd_match = false;
while (opener != NULL && opener != stack_bottom &&
opener != openers_bottom[closer->delim_char]) {
// interior closer of size 2 can't match opener of size 1
// or of size 1 can't match 2
odd_match = (closer->can_open || opener->can_close) &&
((opener->inl_text->as.literal.len +
closer->inl_text->as.literal.len) %
3 ==
0);
if (opener->delim_char == closer->delim_char && opener->can_open &&
!odd_match) {
opener_found = true;
break;
}
opener = opener->previous;
}
old_closer = closer;
if (closer->delim_char == '*' || closer->delim_char == '_') {
if (opener_found) {
closer = S_insert_emph(subj, opener, closer);
} else {
closer = closer->next;
}
} else if (closer->delim_char == '\'') {
cmark_chunk_free(subj->mem, &closer->inl_text->as.literal);
closer->inl_text->as.literal = cmark_chunk_literal(RIGHTSINGLEQUOTE);
if (opener_found) {
cmark_chunk_free(subj->mem, &opener->inl_text->as.literal);
opener->inl_text->as.literal = cmark_chunk_literal(LEFTSINGLEQUOTE);
}
closer = closer->next;
} else if (closer->delim_char == '"') {
cmark_chunk_free(subj->mem, &closer->inl_text->as.literal);
closer->inl_text->as.literal = cmark_chunk_literal(RIGHTDOUBLEQUOTE);
if (opener_found) {
cmark_chunk_free(subj->mem, &opener->inl_text->as.literal);
opener->inl_text->as.literal = cmark_chunk_literal(LEFTDOUBLEQUOTE);
}
closer = closer->next;
}
if (!opener_found && !odd_match) {
// set lower bound for future searches for openers
// (we don't do this with 'odd_match' set because
// a ** that didn't match an earlier * might turn into
// an opener, and the * might be matched by something
// else.
openers_bottom[old_closer->delim_char] = old_closer->previous;
if (!old_closer->can_open) {
// we can remove a closer that can't be an
// opener, once we've seen there's no
// matching opener:
remove_delimiter(subj, old_closer);
}
}
} else {
closer = closer->next;
}
}
// free all delimiters in list until stack_bottom:
while (subj->last_delim != stack_bottom) {
remove_delimiter(subj, subj->last_delim);
}
}
// Parse an inline, advancing subject, and add it as a child of parent.
// Return 0 if no inline can be parsed, 1 otherwise.
static int parse_inline(subject *subj, cmark_node *parent, int options) {
cmark_node *new_inl = NULL;
cmark_chunk contents;
unsigned char c;
bufsize_t endpos;
c = peek_char(subj);
if (c == 0) {
return 0;
}
switch (c) {
case '\r':
case '\n':
new_inl = handle_newline(subj);
break;
case '`':
new_inl = handle_backticks(subj);
break;
case '\\':
new_inl = handle_backslash(subj);
break;
case '&':
new_inl = handle_entity(subj);
break;
case '<':
new_inl = handle_pointy_brace(subj);
break;
case '*':
case '_':
case '\'':
case '"':
new_inl = handle_delim(subj, c, (options & CMARK_OPT_SMART) != 0);
break;
case '-':
new_inl = handle_hyphen(subj, (options & CMARK_OPT_SMART) != 0);
break;
case '.':
new_inl = handle_period(subj, (options & CMARK_OPT_SMART) != 0);
break;
case '[':
advance(subj);
new_inl = make_str(subj->mem, cmark_chunk_literal("["));
push_bracket(subj, false, new_inl);
break;
case ']':
new_inl = handle_close_bracket(subj);
break;
case '!':
advance(subj);
if (peek_char(subj) == '[') {
advance(subj);
new_inl = make_str(subj->mem, cmark_chunk_literal("!["));
push_bracket(subj, true, new_inl);
} else {
new_inl = make_str(subj->mem, cmark_chunk_literal("!"));
}
break;
default:
endpos = subject_find_special_char(subj, options);
contents = cmark_chunk_dup(&subj->input, subj->pos, endpos - subj->pos);
subj->pos = endpos;
// if we're at a newline, strip trailing spaces.
if (S_is_line_end_char(peek_char(subj))) {
cmark_chunk_rtrim(&contents);
}
new_inl = make_str(subj->mem, contents);
}
if (new_inl != NULL) {
cmark_node_append_child(parent, new_inl);
}
return 1;
}
// Return a link, an image, or a literal close bracket.
static cmark_node* handle_close_bracket(subject* subj, cmark_node *parent)
{
bufsize_t initial_pos;
bufsize_t starturl, endurl, starttitle, endtitle, endall;
bufsize_t n;
bufsize_t sps;
cmark_reference *ref;
bool is_image = false;
cmark_chunk url_chunk, title_chunk;
cmark_chunk url, title;
delimiter *opener;
cmark_node *link_text;
cmark_node *inl;
cmark_chunk raw_label;
int found_label;
advance(subj); // advance past ]
initial_pos = subj->pos;
// look through list of delimiters for a [ or !
opener = subj->last_delim;
while (opener) {
if (opener->delim_char == '[' || opener->delim_char == '!') {
break;
}
opener = opener->previous;
}
if (opener == NULL) {
return make_str(cmark_chunk_literal("]"));
}
if (!opener->active) {
// take delimiter off stack
remove_delimiter(subj, opener);
return make_str(cmark_chunk_literal("]"));
}
// If we got here, we matched a potential link/image text.
is_image = opener->delim_char == '!';
link_text = opener->inl_text->next;
// Now we check to see if it's a link/image.
// First, look for an inline link.
if (peek_char(subj) == '(' &&
((sps = scan_spacechars(&subj->input, subj->pos + 1)) > -1) &&
((n = scan_link_url(&subj->input, subj->pos + 1 + sps)) > -1)) {
// try to parse an explicit link:
starturl = subj->pos + 1 + sps; // after (
endurl = starturl + n;
starttitle = endurl + scan_spacechars(&subj->input, endurl);
// ensure there are spaces btw url and title
endtitle = (starttitle == endurl) ? starttitle :
starttitle + scan_link_title(&subj->input, starttitle);
endall = endtitle + scan_spacechars(&subj->input, endtitle);
if (peek_at(subj, endall) == ')') {
subj->pos = endall + 1;
url_chunk = cmark_chunk_dup(&subj->input, starturl, endurl - starturl);
title_chunk = cmark_chunk_dup(&subj->input, starttitle, endtitle - starttitle);
url = cmark_clean_url(&url_chunk);
title = cmark_clean_title(&title_chunk);
cmark_chunk_free(&url_chunk);
cmark_chunk_free(&title_chunk);
goto match;
} else {
goto noMatch;
}
}
// Next, look for a following [link label] that matches in refmap.
// skip spaces
subj->pos = subj->pos + scan_spacechars(&subj->input, subj->pos);
raw_label = cmark_chunk_literal("");
found_label = link_label(subj, &raw_label);
if (!found_label || raw_label.len == 0) {
cmark_chunk_free(&raw_label);
raw_label = cmark_chunk_dup(&subj->input, opener->position,
initial_pos - opener->position - 1);
}
if (!found_label) {
// If we have a shortcut reference link, back up
// to before the spacse we skipped.
subj->pos = initial_pos;
}
ref = cmark_reference_lookup(subj->refmap, &raw_label);
cmark_chunk_free(&raw_label);
if (ref != NULL) { // found
url = chunk_clone(&ref->url);
title = chunk_clone(&ref->title);
goto match;
} else {
goto noMatch;
}
noMatch:
// If we fall through to here, it means we didn't match a link:
remove_delimiter(subj, opener); // remove this opener from delimiter list
subj->pos = initial_pos;
return make_str(cmark_chunk_literal("]"));
match:
inl = opener->inl_text;
inl->type = is_image ? CMARK_NODE_IMAGE : CMARK_NODE_LINK;
cmark_chunk_free(&inl->as.literal);
inl->first_child = link_text;
process_emphasis(subj, opener);
inl->as.link.url = url;
inl->as.link.title = title;
inl->next = NULL;
if (link_text) {
cmark_node *tmp;
link_text->prev = NULL;
for (tmp = link_text; tmp->next != NULL; tmp = tmp->next) {
tmp->parent = inl;
}
tmp->parent = inl;
inl->last_child = tmp;
}
parent->last_child = inl;
// Now, if we have a link, we also want to deactivate earlier link
// delimiters. (This code can be removed if we decide to allow links
// inside links.)
remove_delimiter(subj, opener);
if (!is_image) {
opener = subj->last_delim;
while (opener != NULL) {
if (opener->delim_char == '[') {
if (!opener->active) {
break;
} else {
opener->active = false;
}
}
opener = opener->previous;
}
}
return NULL;
}
static void process_emphasis(subject *subj, delimiter *stack_bottom)
{
delimiter *closer = subj->last_delim;
delimiter *opener;
delimiter *old_closer;
bool opener_found;
delimiter *openers_bottom[128];
// initialize openers_bottom:
openers_bottom['*'] = stack_bottom;
openers_bottom['_'] = stack_bottom;
openers_bottom['\''] = stack_bottom;
openers_bottom['"'] = stack_bottom;
// move back to first relevant delim.
while (closer != NULL && closer->previous != stack_bottom) {
closer = closer->previous;
}
// now move forward, looking for closers, and handling each
while (closer != NULL) {
if (closer->can_close &&
(closer->delim_char == '*' || closer->delim_char == '_' ||
closer->delim_char == '"' || closer->delim_char == '\'')) {
// Now look backwards for first matching opener:
opener = closer->previous;
opener_found = false;
while (opener != NULL && opener != stack_bottom &&
opener != openers_bottom[closer->delim_char]) {
if (opener->delim_char == closer->delim_char &&
opener->can_open) {
opener_found = true;
break;
}
opener = opener->previous;
}
old_closer = closer;
if (closer->delim_char == '*' || closer->delim_char == '_') {
if (opener_found) {
closer = S_insert_emph(subj, opener, closer);
} else {
closer = closer->next;
}
} else if (closer->delim_char == '\'') {
cmark_chunk_free(&closer->inl_text->as.literal);
closer->inl_text->as.literal =
cmark_chunk_literal(RIGHTSINGLEQUOTE);
if (opener_found) {
cmark_chunk_free(&opener->inl_text->as.literal);
opener->inl_text->as.literal =
cmark_chunk_literal(LEFTSINGLEQUOTE);
}
closer = closer->next;
} else if (closer->delim_char == '"') {
cmark_chunk_free(&closer->inl_text->as.literal);
closer->inl_text->as.literal =
cmark_chunk_literal(RIGHTDOUBLEQUOTE);
if (opener_found) {
cmark_chunk_free(&opener->inl_text->as.literal);
opener->inl_text->as.literal =
cmark_chunk_literal(LEFTDOUBLEQUOTE);
}
closer = closer->next;
}
if (!opener_found) {
// set lower bound for future searches for openers:
openers_bottom[old_closer->delim_char] = old_closer->previous;
if (!old_closer->can_open) {
// we can remove a closer that can't be an
// opener, once we've seen there's no
// matching opener:
remove_delimiter(subj, old_closer);
}
}
} else {
closer = closer->next;
}
}
// free all delimiters in list until stack_bottom:
while (subj->last_delim != stack_bottom) {
remove_delimiter(subj, subj->last_delim);
}
}
static void
S_process_line(cmark_parser *parser, const unsigned char *buffer, size_t bytes)
{
cmark_node* last_matched_container;
int offset = 0;
int matched = 0;
int lev = 0;
int i;
cmark_list *data = NULL;
bool all_matched = true;
cmark_node* container;
cmark_node* cur = parser->current;
bool blank = false;
int first_nonspace;
int indent;
cmark_chunk input;
utf8proc_detab(parser->curline, buffer, bytes);
// Add a newline to the end if not present:
// TODO this breaks abstraction:
if (parser->curline->ptr[parser->curline->size - 1] != '\n') {
cmark_strbuf_putc(parser->curline, '\n');
}
input.data = parser->curline->ptr;
input.len = parser->curline->size;
// container starts at the document root.
container = parser->root;
parser->line_number++;
// for each containing cmark_node, try to parse the associated line start.
// bail out on failure: container will point to the last matching cmark_node.
while (container->last_child && container->last_child->open) {
container = container->last_child;
first_nonspace = offset;
while (peek_at(&input, first_nonspace) == ' ') {
first_nonspace++;
}
indent = first_nonspace - offset;
blank = peek_at(&input, first_nonspace) == '\n';
if (container->type == NODE_BLOCK_QUOTE) {
matched = indent <= 3 && peek_at(&input, first_nonspace) == '>';
if (matched) {
offset = first_nonspace + 1;
if (peek_at(&input, offset) == ' ')
offset++;
} else {
all_matched = false;
}
} else if (container->type == NODE_LIST_ITEM) {
if (indent >= container->as.list.marker_offset +
container->as.list.padding) {
offset += container->as.list.marker_offset +
container->as.list.padding;
} else if (blank) {
offset = first_nonspace;
} else {
all_matched = false;
}
} else if (container->type == NODE_CODE_BLOCK) {
if (!container->as.code.fenced) { // indented
if (indent >= CODE_INDENT) {
offset += CODE_INDENT;
} else if (blank) {
offset = first_nonspace;
} else {
all_matched = false;
}
} else {
// skip optional spaces of fence offset
i = container->as.code.fence_offset;
while (i > 0 && peek_at(&input, offset) == ' ') {
offset++;
i--;
}
}
} else if (container->type == NODE_HEADER) {
// a header can never contain more than one line
all_matched = false;
if (blank) {
container->last_line_blank = true;
}
} else if (container->type == NODE_HTML) {
if (blank) {
container->last_line_blank = true;
all_matched = false;
}
} else if (container->type == NODE_PARAGRAPH) {
if (blank) {
container->last_line_blank = true;
all_matched = false;
}
}
if (!all_matched) {
container = container->parent; // back up to last matching cmark_node
break;
}
}
last_matched_container = container;
// check to see if we've hit 2nd blank line, break out of list:
if (blank && container->last_line_blank) {
break_out_of_lists(parser, &container, parser->line_number);
}
// unless last matched container is code cmark_node, try new container starts:
while (container->type != NODE_CODE_BLOCK &&
container->type != NODE_HTML) {
first_nonspace = offset;
while (peek_at(&input, first_nonspace) == ' ')
first_nonspace++;
indent = first_nonspace - offset;
blank = peek_at(&input, first_nonspace) == '\n';
if (indent >= CODE_INDENT) {
if (cur->type != NODE_PARAGRAPH && !blank) {
offset += CODE_INDENT;
container = add_child(parser, container, NODE_CODE_BLOCK, parser->line_number, offset + 1);
container->as.code.fenced = false;
container->as.code.fence_char = 0;
container->as.code.fence_length = 0;
container->as.code.fence_offset = 0;
container->as.code.info = cmark_chunk_literal("");
} else { // indent > 4 in lazy line
break;
}
} else if (peek_at(&input, first_nonspace) == '>') {
offset = first_nonspace + 1;
// optional following character
if (peek_at(&input, offset) == ' ')
offset++;
container = add_child(parser, container, NODE_BLOCK_QUOTE, parser->line_number, offset + 1);
} else if ((matched = scan_atx_header_start(&input, first_nonspace))) {
offset = first_nonspace + matched;
container = add_child(parser, container, NODE_HEADER, parser->line_number, offset + 1);
int hashpos = cmark_chunk_strchr(&input, '#', first_nonspace);
int level = 0;
while (peek_at(&input, hashpos) == '#') {
level++;
hashpos++;
}
container->as.header.level = level;
container->as.header.setext = false;
} else if ((matched = scan_open_code_fence(&input, first_nonspace))) {
container = add_child(parser, container, NODE_CODE_BLOCK, parser->line_number, first_nonspace + 1);
container->as.code.fenced = true;
container->as.code.fence_char = peek_at(&input, first_nonspace);
container->as.code.fence_length = matched;
container->as.code.fence_offset = first_nonspace - offset;
container->as.code.info = cmark_chunk_literal("");
offset = first_nonspace + matched;
} else if ((matched = scan_html_block_tag(&input, first_nonspace))) {
container = add_child(parser, container, NODE_HTML, parser->line_number, first_nonspace + 1);
// note, we don't adjust offset because the tag is part of the text
} else if (container->type == NODE_PARAGRAPH &&
(lev = scan_setext_header_line(&input, first_nonspace)) &&
// check that there is only one line in the paragraph:
cmark_strbuf_strrchr(&container->string_content, '\n',
cmark_strbuf_len(&container->string_content) - 2) < 0) {
container->type = NODE_HEADER;
container->as.header.level = lev;
container->as.header.setext = true;
offset = input.len - 1;
} else if (!(container->type == NODE_PARAGRAPH && !all_matched) &&
(matched = scan_hrule(&input, first_nonspace))) {
// it's only now that we know the line is not part of a setext header:
container = add_child(parser, container, NODE_HRULE, parser->line_number, first_nonspace + 1);
container = finalize(parser, container,
parser->line_number);
offset = input.len - 1;
} else if ((matched = parse_list_marker(&input, first_nonspace, &data))) {
// compute padding:
offset = first_nonspace + matched;
i = 0;
while (i <= 5 && peek_at(&input, offset + i) == ' ') {
i++;
}
// i = number of spaces after marker, up to 5
if (i >= 5 || i < 1 || peek_at(&input, offset) == '\n') {
data->padding = matched + 1;
if (i > 0) {
offset += 1;
}
} else {
data->padding = matched + i;
offset += i;
}
// check container; if it's a list, see if this list item
// can continue the list; otherwise, create a list container.
data->marker_offset = indent;
if (container->type != NODE_LIST ||
!lists_match(&container->as.list, data)) {
container = add_child(parser, container, NODE_LIST, parser->line_number,
first_nonspace + 1);
memcpy(&container->as.list, data, sizeof(*data));
}
// add the list item
container = add_child(parser, container, NODE_LIST_ITEM, parser->line_number,
first_nonspace + 1);
/* TODO: static */
memcpy(&container->as.list, data, sizeof(*data));
free(data);
} else {
break;
}
if (accepts_lines(container->type)) {
// if it's a line container, it can't contain other containers
break;
}
}
// what remains at offset is a text line. add the text to the
// appropriate container.
first_nonspace = offset;
while (peek_at(&input, first_nonspace) == ' ')
first_nonspace++;
indent = first_nonspace - offset;
blank = peek_at(&input, first_nonspace) == '\n';
// cmark_node quote lines are never blank as they start with >
// and we don't count blanks in fenced code for purposes of tight/loose
// lists or breaking out of lists. we also don't set last_line_blank
// on an empty list item.
container->last_line_blank = (blank &&
container->type != NODE_BLOCK_QUOTE &&
container->type != NODE_HEADER &&
!(container->type == NODE_CODE_BLOCK &&
container->as.code.fenced) &&
!(container->type == NODE_LIST_ITEM &&
container->first_child == NULL &&
container->start_line == parser->line_number));
cmark_node *cont = container;
while (cont->parent) {
cont->parent->last_line_blank = false;
cont = cont->parent;
}
if (cur != last_matched_container &&
container == last_matched_container &&
!blank &&
cur->type == NODE_PARAGRAPH &&
cmark_strbuf_len(&cur->string_content) > 0) {
add_line(cur, &input, offset);
} else { // not a lazy continuation
// finalize any blocks that were not matched and set cur to container:
while (cur != last_matched_container) {
cur = finalize(parser, cur, parser->line_number);
assert(cur != NULL);
}
if (container->type == NODE_CODE_BLOCK &&
!container->as.code.fenced) {
add_line(container, &input, offset);
} else if (container->type == NODE_CODE_BLOCK &&
container->as.code.fenced) {
matched = 0;
if (indent <= 3 &&
peek_at(&input, first_nonspace) == container->as.code.fence_char) {
int fence_len = scan_close_code_fence(&input, first_nonspace);
if (fence_len > container->as.code.fence_length)
matched = 1;
}
if (matched) {
// if closing fence, don't add line to container; instead, close it:
container = finalize(parser, container,
parser->line_number);
} else {
add_line(container, &input, offset);
}
} else if (container->type == NODE_HTML) {
add_line(container, &input, offset);
} else if (blank) {
// ??? do nothing
} else if (container->type == NODE_HEADER) {
chop_trailing_hashtags(&input);
add_line(container, &input, first_nonspace);
container = finalize(parser, container,
parser->line_number);
} else if (accepts_lines(container->type)) {
add_line(container, &input, first_nonspace);
} else if (container->type != NODE_HRULE &&
container->type != NODE_HEADER) {
// create paragraph container for line
container = add_child(parser, container, NODE_PARAGRAPH, parser->line_number, first_nonspace + 1);
add_line(container, &input, first_nonspace);
} else {
assert(false);
}
parser->current = container;
}
cmark_strbuf_clear(parser->curline);
}
// Parse an inline, advancing subject, and add it as a child of parent.
// Return 0 if no inline can be parsed, 1 otherwise.
static int parse_inline(subject* subj, cmark_node * parent)
{
cmark_node* new_inl = NULL;
cmark_chunk contents;
unsigned char c;
int endpos;
c = peek_char(subj);
if (c == 0) {
return 0;
}
switch(c) {
case '\n':
new_inl = handle_newline(subj);
break;
case '`':
new_inl = handle_backticks(subj);
break;
case '\\':
new_inl = handle_backslash(subj);
break;
case '&':
new_inl = handle_entity(subj);
break;
case '<':
new_inl = handle_pointy_brace(subj);
break;
case '*':
case '_':
new_inl = handle_strong_emph(subj, c);
break;
case '[':
advance(subj);
new_inl = make_str(cmark_chunk_literal("["));
push_delimiter(subj, '[', true, false, new_inl);
break;
case ']':
new_inl = handle_close_bracket(subj, parent);
break;
case '!':
advance(subj);
if (peek_char(subj) == '[') {
advance(subj);
new_inl = make_str(cmark_chunk_literal("!["));
push_delimiter(subj, '!', false, true, new_inl);
} else {
new_inl = make_str(cmark_chunk_literal("!"));
}
break;
default:
endpos = subject_find_special_char(subj);
contents = cmark_chunk_dup(&subj->input, subj->pos, endpos - subj->pos);
subj->pos = endpos;
// if we're at a newline, strip trailing spaces.
if (peek_char(subj) == '\n') {
cmark_chunk_rtrim(&contents);
}
new_inl = make_str(contents);
}
if (new_inl != NULL) {
cmark_node_append_child(parent, new_inl);
}
return 1;
}
static void S_out(cmark_renderer *renderer, const char *source, bool wrap,
cmark_escaping escape) {
int length = cmark_strbuf_safe_strlen(source);
unsigned char nextc;
int32_t c;
int i = 0;
int last_nonspace;
int len;
cmark_chunk remainder = cmark_chunk_literal("");
int k = renderer->buffer->size - 1;
wrap = wrap && !renderer->no_wrap;
if (renderer->in_tight_list_item && renderer->need_cr > 1) {
renderer->need_cr = 1;
}
while (renderer->need_cr) {
if (k < 0 || renderer->buffer->ptr[k] == '\n') {
k -= 1;
} else {
cmark_strbuf_putc(renderer->buffer, '\n');
if (renderer->need_cr > 1) {
cmark_strbuf_put(renderer->buffer, renderer->prefix->ptr,
renderer->prefix->size);
}
}
renderer->column = 0;
renderer->begin_line = true;
renderer->begin_content = true;
renderer->need_cr -= 1;
}
while (i < length) {
if (renderer->begin_line) {
cmark_strbuf_put(renderer->buffer, renderer->prefix->ptr,
renderer->prefix->size);
// note: this assumes prefix is ascii:
renderer->column = renderer->prefix->size;
}
len = cmark_utf8proc_iterate((const uint8_t *)source + i, length - i, &c);
if (len == -1) { // error condition
return; // return without rendering rest of string
}
nextc = source[i + len];
if (c == 32 && wrap) {
if (!renderer->begin_line) {
last_nonspace = renderer->buffer->size;
cmark_strbuf_putc(renderer->buffer, ' ');
renderer->column += 1;
renderer->begin_line = false;
renderer->begin_content = false;
// skip following spaces
while (source[i + 1] == ' ') {
i++;
}
// We don't allow breaks that make a digit the first character
// because this causes problems with commonmark output.
if (!cmark_isdigit(source[i + 1])) {
renderer->last_breakable = last_nonspace;
}
}
} else if (c == 10) {
cmark_strbuf_putc(renderer->buffer, '\n');
renderer->column = 0;
renderer->begin_line = true;
renderer->begin_content = true;
renderer->last_breakable = 0;
} else if (escape == LITERAL) {
cmark_render_code_point(renderer, c);
renderer->begin_line = false;
// we don't set 'begin_content' to false til we've
// finished parsing a digit. Reason: in commonmark
// we need to escape a potential list marker after
// a digit:
renderer->begin_content = renderer->begin_content &&
cmark_isdigit(c) == 1;
} else {
(renderer->outc)(renderer, escape, c, nextc);
renderer->begin_line = false;
renderer->begin_content = renderer->begin_content &&
cmark_isdigit(c) == 1;
}
// If adding the character went beyond width, look for an
// earlier place where the line could be broken:
if (renderer->width > 0 && renderer->column > renderer->width &&
!renderer->begin_line && renderer->last_breakable > 0) {
// copy from last_breakable to remainder
cmark_chunk_set_cstr(&remainder, (char *)renderer->buffer->ptr +
renderer->last_breakable + 1);
// truncate at last_breakable
cmark_strbuf_truncate(renderer->buffer, renderer->last_breakable);
// add newline, prefix, and remainder
cmark_strbuf_putc(renderer->buffer, '\n');
cmark_strbuf_put(renderer->buffer, renderer->prefix->ptr,
renderer->prefix->size);
cmark_strbuf_put(renderer->buffer, remainder.data, remainder.len);
renderer->column = renderer->prefix->size + remainder.len;
cmark_chunk_free(&remainder);
renderer->last_breakable = 0;
renderer->begin_line = false;
renderer->begin_content = false;
}
i += len;
}
}
