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_puts(&buf, cmark_node_get_literal(cur));
tmp = cur->next;
while (tmp && tmp->type == CMARK_NODE_TEXT) {
cmark_iter_next(iter); // advance pointer
cmark_strbuf_puts(&buf, cmark_node_get_literal(tmp));
next = tmp->next;
cmark_node_free(tmp);
tmp = next;
}
cmark_node_set_literal(cur, (char *)cmark_strbuf_detach(&buf));
}
}
cmark_iter_free(iter);
}
char *cmark_render_xml(cmark_node *root, long options)
{
char *result;
cmark_strbuf xml = GH_BUF_INIT;
cmark_event_type ev_type;
cmark_node *cur;
struct render_state state = { &xml, 0 };
if (options & CMARK_OPT_NORMALIZE) {
cmark_consolidate_text_nodes(root);
}
cmark_iter *iter = cmark_iter_new(root);
cmark_strbuf_puts(state.xml,
"<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
cmark_strbuf_puts(state.xml,
"<!DOCTYPE CommonMark SYSTEM \"CommonMark.dtd\">\n");
while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
cur = cmark_iter_get_node(iter);
S_render_node(cur, ev_type, &state, options);
}
result = (char *)cmark_strbuf_detach(&xml);
cmark_iter_free(iter);
cmark_strbuf_free(&xml);
return result;
}
static cmark_node *symbol_link_match(cmark_syntax_extension *self,
cmark_parser *parser,
cmark_node *parent,
cmark_inline_parser *inline_parser) {
cmark_node *link = NULL;
char *symbol_name = NULL;
NamedLink *named_link = NULL;
int start_offset = cmark_inline_parser_get_offset(inline_parser);
ParsingContext context;
context.parser = inline_parser;
context.allow_dashes = 0;
if (start_offset > 0) {
char prev_char = cmark_inline_parser_peek_at(
inline_parser,
start_offset - 1);
if (prev_char && prev_char != ' ' && prev_char != '\t' && prev_char != '\n')
return NULL;
}
cmark_inline_parser_advance_offset(inline_parser);
symbol_name = cmark_inline_parser_take_while(inline_parser,
(CMarkInlinePredicate) is_valid_symbol_name, &context);
if (!symbol_name)
goto done;
named_link = PRIV(self)->link_resolve_func(symbol_name);
if (!named_link || !named_link->ref) {
int actual_line, actual_col;
translate_sourcepos(get_first_parent_block(parent),
start_offset, &actual_line, &actual_col);
cmark_strbuf *message = cmark_strbuf_new(0);
cmark_strbuf_puts(message, "Trying to link to non-existing symbol ‘");
cmark_strbuf_puts(message, symbol_name);
cmark_strbuf_puts(message, "’");
diagnose("gtk-doc-bad-link", cmark_strbuf_get(message), actual_line - 1, actual_col - 1);
cmark_strbuf_free(message);
link = cmark_node_new (CMARK_NODE_TEXT);
cmark_node_set_literal (link, symbol_name);
} else {
link = cmark_node_new(CMARK_NODE_LINK);
}
cmark_node_set_url(link, symbol_name);
done:
free(symbol_name);
free_named_link(named_link);
return link;
}
static void html_render(cmark_syntax_extension *extension,
cmark_html_renderer *renderer, cmark_node *node,
cmark_event_type ev_type, int options) {
bool entering = (ev_type == CMARK_EVENT_ENTER);
if (entering) {
cmark_strbuf_puts(renderer->html, "<del>");
} else {
cmark_strbuf_puts(renderer->html, "</del>");
}
}
int houdini_escape_href(cmark_strbuf *ob, const uint8_t *src, bufsize_t size) {
static const uint8_t hex_chars[] = "0123456789ABCDEF";
bufsize_t i = 0, org;
uint8_t hex_str[3];
hex_str[0] = '%';
while (i < size) {
org = i;
while (i < size && HREF_SAFE[src[i]] != 0)
i++;
if (likely(i > org))
cmark_strbuf_put(ob, src + org, i - org);
/* escaping */
if (i >= size)
break;
switch (src[i]) {
/* amp appears all the time in URLs, but needs
* HTML-entity escaping to be inside an href */
case '&':
cmark_strbuf_puts(ob, "&");
break;
/* the single quote is a valid URL character
* according to the standard; it needs HTML
* entity escaping too */
case '\'':
cmark_strbuf_puts(ob, "'");
break;
/* the space can be escaped to %20 or a plus
* sign. we're going with the generic escape
* for now. the plus thing is more commonly seen
* when building GET strings */
#if 0
case ' ':
cmark_strbuf_putc(ob, '+');
break;
#endif
/* every other character goes with a %XX escaping */
default:
hex_str[1] = hex_chars[(src[i] >> 4) & 0xF];
hex_str[2] = hex_chars[src[i] & 0xF];
cmark_strbuf_put(ob, hex_str, 3);
}
i++;
}
return 1;
}
static bool S_put_footnote_backref(cmark_html_renderer *renderer, cmark_strbuf *html) {
if (renderer->written_footnote_ix >= renderer->footnote_ix)
return false;
renderer->written_footnote_ix = renderer->footnote_ix;
cmark_strbuf_puts(html, "<a href=\"#fnref");
char n[32];
snprintf(n, sizeof(n), "%d", renderer->footnote_ix);
cmark_strbuf_puts(html, n);
cmark_strbuf_puts(html, "\" class=\"footnote-backref\">↩</a>");
return true;
}
char *cmark_render_html_with_mem(cmark_node *root, int options, cmark_llist *extensions, cmark_mem *mem) {
char *result;
cmark_strbuf html = CMARK_BUF_INIT(mem);
cmark_event_type ev_type;
cmark_node *cur;
cmark_html_renderer renderer = {&html, NULL, NULL, 0, 0, NULL};
cmark_iter *iter = cmark_iter_new(root);
for (; extensions; extensions = extensions->next)
if (((cmark_syntax_extension *) extensions->data)->html_filter_func)
renderer.filter_extensions = cmark_llist_append(
mem,
renderer.filter_extensions,
(cmark_syntax_extension *) extensions->data);
while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
cur = cmark_iter_get_node(iter);
S_render_node(&renderer, cur, ev_type, options);
}
if (renderer.footnote_ix) {
cmark_strbuf_puts(&html, "</ol>\n</section>\n");
}
result = (char *)cmark_strbuf_detach(&html);
cmark_llist_free(mem, renderer.filter_extensions);
cmark_iter_free(iter);
return result;
}
static inline void outc(cmark_renderer *renderer, cmark_escaping escape,
int32_t c, unsigned char nextc) {
bool needs_escaping = false;
char encoded[20];
needs_escaping =
escape != LITERAL &&
((escape == NORMAL &&
(c == '*' || c == '_' || c == '[' || c == ']' || c == '#' || c == '<' ||
c == '>' || c == '\\' || c == '`' || c == '!' ||
(c == '&' && isalpha(nextc)) || (c == '!' && nextc == '[') ||
(renderer->begin_line && (c == '-' || c == '+' || c == '=')) ||
((c == '.' || c == ')') &&
isdigit(renderer->buffer->ptr[renderer->buffer->size - 1])))) ||
(escape == URL && (c == '`' || c == '<' || c == '>' || isspace(c) ||
c == '\\' || c == ')' || c == '(')) ||
(escape == TITLE &&
(c == '`' || c == '<' || c == '>' || c == '"' || c == '\\')));
if (needs_escaping) {
if (isspace(c)) {
// use percent encoding for spaces
sprintf(encoded, "%%%2x", c);
cmark_strbuf_puts(renderer->buffer, encoded);
renderer->column += 3;
} else {
cmark_render_ascii(renderer, "\\");
cmark_render_code_point(renderer, c);
}
} else {
cmark_render_code_point(renderer, c);
}
}
int
houdini_escape_html0(cmark_strbuf *ob, const uint8_t *src, size_t size, int secure)
{
size_t i = 0, org, esc = 0;
while (i < size) {
org = i;
while (i < size && (esc = HTML_ESCAPE_TABLE[src[i]]) == 0)
i++;
if (i > org)
cmark_strbuf_put(ob, src + org, i - org);
/* escaping */
if (unlikely(i >= size))
break;
/* The forward slash is only escaped in secure mode */
if ((src[i] == '/' || src[i] == '\'') && !secure) {
cmark_strbuf_putc(ob, src[i]);
} else {
cmark_strbuf_puts(ob, HTML_ESCAPES[esc]);
}
i++;
}
return 1;
}
// Assumes no newlines, assumes ascii content:
void
cmark_render_ascii(cmark_renderer* renderer, const char* s)
{
int origsize = renderer->buffer->size;
cmark_strbuf_puts(renderer->buffer, s);
renderer->column += renderer->buffer->size - origsize;
}
static CMARK_INLINE void outc(cmark_renderer *renderer, cmark_escaping escape,
int32_t c, unsigned char nextc) {
bool needs_escaping = false;
bool follows_digit =
renderer->buffer->size > 0 &&
cmark_isdigit(renderer->buffer->ptr[renderer->buffer->size - 1]);
char encoded[ENCODED_SIZE];
needs_escaping =
c < 0x80 && escape != LITERAL &&
((escape == NORMAL &&
(c < 0x20 ||
c == '*' || c == '_' || c == '[' || c == ']' || c == '#' || c == '<' ||
c == '>' || c == '\\' || c == '`' || c == '!' ||
(c == '&' && cmark_isalpha(nextc)) || (c == '!' && nextc == '[') ||
(renderer->begin_content && (c == '-' || c == '+' || c == '=') &&
// begin_content doesn't get set to false til we've passed digits
// at the beginning of line, so...
!follows_digit) ||
(renderer->begin_content && (c == '.' || c == ')') && follows_digit &&
(nextc == 0 || cmark_isspace(nextc))))) ||
(escape == URL &&
(c == '`' || c == '<' || c == '>' || cmark_isspace(c) || c == '\\' ||
c == ')' || c == '(')) ||
(escape == TITLE &&
(c == '`' || c == '<' || c == '>' || c == '"' || c == '\\')));
if (needs_escaping) {
if (escape == URL && cmark_isspace(c)) {
// use percent encoding for spaces
snprintf(encoded, ENCODED_SIZE, "%%%2X", c);
cmark_strbuf_puts(renderer->buffer, encoded);
renderer->column += 3;
} else if (cmark_ispunct(c)) {
cmark_render_ascii(renderer, "\\");
cmark_render_code_point(renderer, c);
} else { // render as entity
snprintf(encoded, ENCODED_SIZE, "&#%d;", c);
cmark_strbuf_puts(renderer->buffer, encoded);
renderer->column += strlen(encoded);
}
} else {
cmark_render_code_point(renderer, c);
}
}
static cmark_node *fixup_nodes(cmark_inline_parser *inline_parser,
cmark_node *parent,
int size)
{
int node_text_len;
cmark_node *prev = NULL;
cmark_node *tmp;
int name_size = size;
cmark_strbuf *name;
for (prev = cmark_node_last_child(parent); prev; prev = cmark_node_previous(prev)) {
if (cmark_node_get_type(prev) == CMARK_NODE_TEXT) {
const char *text = cmark_node_get_literal(prev);
node_text_len = strlen(text);
size -= node_text_len;
if (size <= 0) {
if (size < 0) {
char *split_text = my_strndup(text, size * -1);
cmark_node *split = cmark_node_new(CMARK_NODE_TEXT);
cmark_node_set_literal(split, split_text);
free(split_text);
split_text = my_strndup(text + (size * - 1), node_text_len - size);
cmark_node_set_literal(prev, split_text);
free(split_text);
cmark_node_insert_before(prev, split);
}
break;
}
} else {
return NULL;
}
}
name = cmark_strbuf_new(name_size + 1);
tmp = prev;
while (tmp) {
cmark_node *next = cmark_node_next(tmp);
cmark_strbuf_puts(name, cmark_node_get_literal(tmp));
if (tmp != prev)
cmark_node_free(tmp);
tmp = next;
}
cmark_node_set_type(prev, CMARK_NODE_LINK);
cmark_node_set_url(prev, cmark_strbuf_get(name));
cmark_strbuf_free(name);
return prev;
}
// 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));
}
static void escape_man(cmark_strbuf *dest, const unsigned char *source, int length)
{
int i;
unsigned char c;
for (i = 0; i < length; i++) {
c = source[i];
if (c == '.' && i == 0) {
cmark_strbuf_puts(dest, "\\&.");
} else if (c == '\'' && i == 0) {
cmark_strbuf_puts(dest, "\\&'");
} else if (c == '-') {
cmark_strbuf_puts(dest, "\\-");
} else if (c == '\\') {
cmark_strbuf_puts(dest, "\\e");
} else {
cmark_strbuf_putc(dest, source[i]);
}
}
}
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;
}
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 void filter_html_block(cmark_html_renderer *renderer, uint8_t *data, size_t len) {
cmark_strbuf *html = renderer->html;
cmark_llist *it;
cmark_syntax_extension *ext;
bool filtered;
uint8_t *match;
while (len) {
match = (uint8_t *) memchr(data, '<', len);
if (!match)
break;
if (match != data) {
cmark_strbuf_put(html, data, (bufsize_t)(match - data));
len -= (match - data);
data = match;
}
filtered = false;
for (it = renderer->filter_extensions; it; it = it->next) {
ext = ((cmark_syntax_extension *) it->data);
if (!ext->html_filter_func(ext, data, len)) {
filtered = true;
break;
}
}
if (!filtered) {
cmark_strbuf_putc(html, '<');
} else {
cmark_strbuf_puts(html, "<");
}
++data;
--len;
}
if (len)
cmark_strbuf_put(html, data, (bufsize_t)len);
}
static int S_render_node(cmark_renderer *renderer, cmark_node *node,
cmark_event_type ev_type, int options) {
cmark_node *tmp;
int list_number;
cmark_delim_type list_delim;
int numticks;
bool extra_spaces;
int i;
bool entering = (ev_type == CMARK_EVENT_ENTER);
const char *info, *code, *title;
char fencechar[2] = {'\0', '\0'};
size_t info_len, code_len;
char listmarker[LISTMARKER_SIZE];
char *emph_delim;
bool first_in_list_item;
bufsize_t marker_width;
bool allow_wrap = renderer->width > 0 && !(CMARK_OPT_NOBREAKS & options) &&
!(CMARK_OPT_HARDBREAKS & options);
// Don't adjust tight list status til we've started the list.
// Otherwise we loose the blank line between a paragraph and
// a following list.
if (!(node->type == CMARK_NODE_ITEM && node->prev == NULL && entering)) {
tmp = get_containing_block(node);
renderer->in_tight_list_item =
tmp && // tmp might be NULL if there is no containing block
((tmp->type == CMARK_NODE_ITEM &&
cmark_node_get_list_tight(tmp->parent)) ||
(tmp && tmp->parent && tmp->parent->type == CMARK_NODE_ITEM &&
cmark_node_get_list_tight(tmp->parent->parent)));
}
switch (node->type) {
case CMARK_NODE_DOCUMENT:
break;
case CMARK_NODE_BLOCK_QUOTE:
if (entering) {
LIT("> ");
renderer->begin_content = true;
cmark_strbuf_puts(renderer->prefix, "> ");
} else {
cmark_strbuf_truncate(renderer->prefix, renderer->prefix->size - 2);
BLANKLINE();
}
break;
case CMARK_NODE_LIST:
if (!entering && node->next && (node->next->type == CMARK_NODE_CODE_BLOCK ||
node->next->type == CMARK_NODE_LIST)) {
// this ensures that a following indented code block or list will be
// inteprereted correctly.
CR();
LIT("<!-- end list -->");
BLANKLINE();
}
break;
case CMARK_NODE_ITEM:
if (cmark_node_get_list_type(node->parent) == CMARK_BULLET_LIST) {
marker_width = 4;
} else {
list_number = cmark_node_get_list_start(node->parent);
list_delim = cmark_node_get_list_delim(node->parent);
tmp = node;
while (tmp->prev) {
tmp = tmp->prev;
list_number += 1;
}
// we ensure a width of at least 4 so
// we get nice transition from single digits
// to double
snprintf(listmarker, LISTMARKER_SIZE, "%d%s%s", list_number,
list_delim == CMARK_PAREN_DELIM ? ")" : ".",
list_number < 10 ? " " : " ");
marker_width = strlen(listmarker);
}
if (entering) {
if (cmark_node_get_list_type(node->parent) == CMARK_BULLET_LIST) {
LIT(" - ");
renderer->begin_content = true;
} else {
LIT(listmarker);
renderer->begin_content = true;
}
for (i = marker_width; i--;) {
cmark_strbuf_putc(renderer->prefix, ' ');
}
} else {
cmark_strbuf_truncate(renderer->prefix,
renderer->prefix->size - marker_width);
CR();
}
break;
case CMARK_NODE_HEADING:
if (entering) {
for (i = cmark_node_get_heading_level(node); i > 0; i--) {
LIT("#");
}
LIT(" ");
renderer->begin_content = true;
renderer->no_linebreaks = true;
} else {
renderer->no_linebreaks = false;
BLANKLINE();
}
break;
case CMARK_NODE_CODE_BLOCK:
first_in_list_item = node->prev == NULL && node->parent &&
node->parent->type == CMARK_NODE_ITEM;
if (!first_in_list_item) {
BLANKLINE();
}
info = cmark_node_get_fence_info(node);
info_len = strlen(info);
fencechar[0] = strchr(info, '`') == NULL ? '`' : '~';
code = cmark_node_get_literal(node);
code_len = strlen(code);
// use indented form if no info, and code doesn't
// begin or end with a blank line, and code isn't
// first thing in a list item
if (info_len == 0 && (code_len > 2 && !cmark_isspace(code[0]) &&
!(cmark_isspace(code[code_len - 1]) &&
cmark_isspace(code[code_len - 2]))) &&
!first_in_list_item) {
LIT(" ");
cmark_strbuf_puts(renderer->prefix, " ");
OUT(cmark_node_get_literal(node), false, LITERAL);
cmark_strbuf_truncate(renderer->prefix, renderer->prefix->size - 4);
} else {
numticks = longest_backtick_sequence(code) + 1;
if (numticks < 3) {
numticks = 3;
}
for (i = 0; i < numticks; i++) {
LIT(fencechar);
}
LIT(" ");
OUT(info, false, LITERAL);
CR();
OUT(cmark_node_get_literal(node), false, LITERAL);
CR();
for (i = 0; i < numticks; i++) {
LIT(fencechar);
}
}
BLANKLINE();
break;
case CMARK_NODE_HTML_BLOCK:
BLANKLINE();
OUT(cmark_node_get_literal(node), false, LITERAL);
BLANKLINE();
break;
case CMARK_NODE_CUSTOM_BLOCK:
BLANKLINE();
OUT(entering ? cmark_node_get_on_enter(node) : cmark_node_get_on_exit(node),
false, LITERAL);
BLANKLINE();
break;
case CMARK_NODE_THEMATIC_BREAK:
BLANKLINE();
LIT("-----");
BLANKLINE();
break;
case CMARK_NODE_PARAGRAPH:
if (!entering) {
BLANKLINE();
}
break;
case CMARK_NODE_TEXT:
OUT(cmark_node_get_literal(node), allow_wrap, NORMAL);
break;
case CMARK_NODE_LINEBREAK:
if (!(CMARK_OPT_HARDBREAKS & options)) {
LIT(" ");
}
CR();
break;
case CMARK_NODE_SOFTBREAK:
if (CMARK_OPT_HARDBREAKS & options) {
LIT(" ");
CR();
} else if (!renderer->no_linebreaks && renderer->width == 0 &&
!(CMARK_OPT_HARDBREAKS & options) &&
!(CMARK_OPT_NOBREAKS & options)) {
CR();
} else {
OUT(" ", allow_wrap, LITERAL);
}
break;
case CMARK_NODE_CODE:
code = cmark_node_get_literal(node);
code_len = strlen(code);
numticks = shortest_unused_backtick_sequence(code);
extra_spaces = code_len == 0 ||
code[0] == '`' || code[code_len - 1] == '`' ||
code[0] == ' ' || code[code_len - 1] == ' ';
for (i = 0; i < numticks; i++) {
LIT("`");
}
if (extra_spaces) {
LIT(" ");
}
OUT(cmark_node_get_literal(node), allow_wrap, LITERAL);
if (extra_spaces) {
LIT(" ");
}
for (i = 0; i < numticks; i++) {
LIT("`");
}
break;
case CMARK_NODE_HTML_INLINE:
OUT(cmark_node_get_literal(node), false, LITERAL);
break;
case CMARK_NODE_CUSTOM_INLINE:
OUT(entering ? cmark_node_get_on_enter(node) : cmark_node_get_on_exit(node),
false, LITERAL);
break;
case CMARK_NODE_STRONG:
if (entering) {
LIT("**");
} else {
LIT("**");
}
break;
case CMARK_NODE_EMPH:
// If we have EMPH(EMPH(x)), we need to use *_x_*
// because **x** is STRONG(x):
if (node->parent && node->parent->type == CMARK_NODE_EMPH &&
node->next == NULL && node->prev == NULL) {
emph_delim = "_";
} else {
emph_delim = "*";
}
if (entering) {
LIT(emph_delim);
} else {
LIT(emph_delim);
}
break;
case CMARK_NODE_LINK:
if (is_autolink(node)) {
if (entering) {
LIT("<");
if (strncmp(cmark_node_get_url(node), "mailto:", 7) == 0) {
LIT((const char *)cmark_node_get_url(node) + 7);
} else {
LIT((const char *)cmark_node_get_url(node));
}
LIT(">");
// return signal to skip contents of node...
return 0;
}
} else {
if (entering) {
LIT("[");
} else {
LIT("](");
OUT(cmark_node_get_url(node), false, URL);
title = cmark_node_get_title(node);
if (strlen(title) > 0) {
LIT(" \"");
OUT(title, false, TITLE);
LIT("\"");
}
LIT(")");
}
}
break;
case CMARK_NODE_IMAGE:
if (entering) {
LIT("![");
} else {
LIT("](");
OUT(cmark_node_get_url(node), false, URL);
title = cmark_node_get_title(node);
if (strlen(title) > 0) {
OUT(" \"", allow_wrap, LITERAL);
OUT(title, false, TITLE);
LIT("\"");
}
LIT(")");
}
break;
default:
assert(false);
break;
}
return 1;
}
static int
S_render_node(cmark_node *node, cmark_event_type ev_type,
struct render_state *state)
{
cmark_node *tmp;
cmark_strbuf *man = state->man;
int list_number;
bool entering = (ev_type == CMARK_EVENT_ENTER);
if (state->plain == node) { // back at original node
state->plain = NULL;
}
if (state->plain != NULL) {
switch(node->type) {
case CMARK_NODE_TEXT:
case CMARK_NODE_CODE:
escape_man(man, node->as.literal.data,
node->as.literal.len);
break;
case CMARK_NODE_LINEBREAK:
case CMARK_NODE_SOFTBREAK:
cmark_strbuf_putc(man, ' ');
break;
default:
break;
}
return 1;
}
switch (node->type) {
case CMARK_NODE_DOCUMENT:
break;
case CMARK_NODE_BLOCK_QUOTE:
if (entering) {
cr(man);
cmark_strbuf_puts(man, ".RS");
cr(man);
} else {
cr(man);
cmark_strbuf_puts(man, ".RE");
cr(man);
}
break;
case CMARK_NODE_LIST:
break;
case CMARK_NODE_ITEM:
if (entering) {
cr(man);
cmark_strbuf_puts(man, ".IP ");
if (cmark_node_get_list_type(node->parent) ==
CMARK_BULLET_LIST) {
cmark_strbuf_puts(man, "\\[bu] 2");
} else {
list_number = cmark_node_get_list_start(node->parent);
tmp = node;
while (tmp->prev) {
tmp = tmp->prev;
list_number += 1;
}
cmark_strbuf_printf(man, "\"%d.\" 4", list_number);
}
cr(man);
} else {
cr(man);
}
break;
case CMARK_NODE_HEADER:
if (entering) {
cr(man);
cmark_strbuf_puts(man,
cmark_node_get_header_level(node) == 1 ?
".SH" : ".SS");
cr(man);
} else {
cr(man);
}
break;
case CMARK_NODE_CODE_BLOCK:
cr(man);
cmark_strbuf_puts(man, ".IP\n.nf\n\\f[C]\n");
escape_man(man, node->as.code.literal.data,
node->as.code.literal.len);
cr(man);
cmark_strbuf_puts(man, "\\f[]\n.fi");
cr(man);
break;
case CMARK_NODE_HTML:
break;
case CMARK_NODE_HRULE:
cr(man);
cmark_strbuf_puts(man, ".PP\n * * * * *");
cr(man);
break;
case CMARK_NODE_PARAGRAPH:
if (entering) {
// no blank line if first paragraph in list:
if (node->parent &&
node->parent->type == CMARK_NODE_ITEM &&
node->prev == NULL) {
// no blank line or .PP
} else {
cr(man);
cmark_strbuf_puts(man, ".PP\n");
}
} else {
cr(man);
}
break;
case CMARK_NODE_TEXT:
escape_man(man, node->as.literal.data,
node->as.literal.len);
break;
case CMARK_NODE_LINEBREAK:
cmark_strbuf_puts(man, ".PD 0\n.P\n.PD");
cr(man);
break;
case CMARK_NODE_SOFTBREAK:
cmark_strbuf_putc(man, '\n');
break;
case CMARK_NODE_CODE:
cmark_strbuf_puts(man, "\\f[C]");
escape_man(man, node->as.literal.data, node->as.literal.len);
cmark_strbuf_puts(man, "\\f[]");
break;
case CMARK_NODE_INLINE_HTML:
break;
case CMARK_NODE_STRONG:
if (entering) {
cmark_strbuf_puts(man, "\\f[B]");
} else {
cmark_strbuf_puts(man, "\\f[]");
}
break;
case CMARK_NODE_EMPH:
if (entering) {
cmark_strbuf_puts(man, "\\f[I]");
} else {
cmark_strbuf_puts(man, "\\f[]");
}
break;
case CMARK_NODE_LINK:
if (!entering) {
cmark_strbuf_printf(man, " (%s)",
cmark_node_get_url(node));
}
break;
case CMARK_NODE_IMAGE:
if (entering) {
cmark_strbuf_puts(man, "[IMAGE: ");
state->plain = node;
} else {
cmark_strbuf_puts(man, "]");
}
break;
default:
assert(false);
break;
}
// cmark_strbuf_putc(man, 'x');
return 1;
}
static int
S_render_node(cmark_node *node, cmark_event_type ev_type,
struct render_state *state, long options)
{
cmark_strbuf *xml = state->xml;
bool literal = false;
cmark_delim_type delim;
bool entering = (ev_type == CMARK_EVENT_ENTER);
if (entering) {
indent(state);
cmark_strbuf_printf(xml, "<%s",
cmark_node_get_type_string(node));
if (options & CMARK_OPT_SOURCEPOS && node->start_line != 0) {
cmark_strbuf_printf(xml, " sourcepos=\"%d:%d-%d:%d\"",
node->start_line,
node->start_column,
node->end_line,
node->end_column);
}
literal = false;
switch (node->type) {
case CMARK_NODE_TEXT:
case CMARK_NODE_CODE:
case CMARK_NODE_HTML:
case CMARK_NODE_INLINE_HTML:
cmark_strbuf_puts(xml, ">");
escape_xml(xml, node->as.literal.data,
node->as.literal.len);
cmark_strbuf_puts(xml, "</");
cmark_strbuf_puts(xml,
cmark_node_get_type_string(node));
literal = true;
break;
case CMARK_NODE_LIST:
switch (cmark_node_get_list_type(node)) {
case CMARK_ORDERED_LIST:
cmark_strbuf_puts(xml, " type=\"ordered\"");
cmark_strbuf_printf(xml, " start=\"%d\"",
cmark_node_get_list_start(node));
delim = cmark_node_get_list_delim(node);
if (delim == CMARK_PAREN_DELIM) {
cmark_strbuf_puts(xml,
" delim=\"paren\"");
} else if (delim == CMARK_PERIOD_DELIM) {
cmark_strbuf_puts(xml,
" delim=\"period\"");
}
break;
case CMARK_BULLET_LIST:
cmark_strbuf_puts(xml, " type=\"bullet\"");
break;
default:
break;
}
cmark_strbuf_printf(xml, " tight=\"%s\"",
(cmark_node_get_list_tight(node) ?
"true" : "false"));
break;
case CMARK_NODE_HEADER:
cmark_strbuf_printf(xml, " level=\"%d\"",
node->as.header.level);
break;
case CMARK_NODE_CODE_BLOCK:
if (node->as.code.info.len > 0) {
cmark_strbuf_puts(xml, " info=\"");
escape_xml(xml, node->as.code.info.data,
node->as.code.info.len);
cmark_strbuf_putc(xml, '"');
}
cmark_strbuf_puts(xml, ">");
escape_xml(xml, node->as.code.literal.data,
node->as.code.literal.len);
cmark_strbuf_puts(xml, "</");
cmark_strbuf_puts(xml,
cmark_node_get_type_string(node));
literal = true;
break;
case CMARK_NODE_LINK:
case CMARK_NODE_IMAGE:
cmark_strbuf_puts(xml, " destination=\"");
escape_xml(xml, node->as.link.url, -1);
cmark_strbuf_putc(xml, '"');
cmark_strbuf_puts(xml, " title=\"");
escape_xml(xml, node->as.link.title, -1);
cmark_strbuf_putc(xml, '"');
break;
default:
break;
}
if (node->first_child) {
state->indent += 2;
} else if (!literal) {
cmark_strbuf_puts(xml, " /");
}
cmark_strbuf_puts(xml, ">\n");
} else if (node->first_child) {
state->indent -= 2;
indent(state);
cmark_strbuf_printf(xml, "</%s>\n",
cmark_node_get_type_string(node));
}
return 1;
}
static int S_render_node(cmark_html_renderer *renderer, cmark_node *node,
cmark_event_type ev_type, int options) {
cmark_node *parent;
cmark_node *grandparent;
cmark_strbuf *html = renderer->html;
cmark_llist *it;
cmark_syntax_extension *ext;
char start_heading[] = "<h0";
char end_heading[] = "</h0";
bool tight;
bool filtered;
char buffer[BUFFER_SIZE];
bool entering = (ev_type == CMARK_EVENT_ENTER);
if (renderer->plain == node) { // back at original node
renderer->plain = NULL;
}
if (renderer->plain != NULL) {
switch (node->type) {
case CMARK_NODE_TEXT:
case CMARK_NODE_CODE:
case CMARK_NODE_HTML_INLINE:
escape_html(html, node->as.literal.data, node->as.literal.len);
break;
case CMARK_NODE_LINEBREAK:
case CMARK_NODE_SOFTBREAK:
cmark_strbuf_putc(html, ' ');
break;
default:
break;
}
return 1;
}
if (node->extension && node->extension->html_render_func) {
node->extension->html_render_func(node->extension, renderer, node, ev_type, options);
return 1;
}
switch (node->type) {
case CMARK_NODE_DOCUMENT:
break;
case CMARK_NODE_BLOCK_QUOTE:
if (entering) {
cmark_html_render_cr(html);
cmark_strbuf_puts(html, "<blockquote");
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, ">\n");
} else {
cmark_html_render_cr(html);
cmark_strbuf_puts(html, "</blockquote>\n");
}
break;
case CMARK_NODE_LIST: {
cmark_list_type list_type = node->as.list.list_type;
int start = node->as.list.start;
if (entering) {
cmark_html_render_cr(html);
if (list_type == CMARK_BULLET_LIST) {
cmark_strbuf_puts(html, "<ul");
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, ">\n");
} else if (start == 1) {
cmark_strbuf_puts(html, "<ol");
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, ">\n");
} else {
snprintf(buffer, BUFFER_SIZE, "<ol start=\"%d\"", start);
cmark_strbuf_puts(html, buffer);
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, ">\n");
}
} else {
cmark_strbuf_puts(html,
list_type == CMARK_BULLET_LIST ? "</ul>\n" : "</ol>\n");
}
break;
}
case CMARK_NODE_ITEM:
if (entering) {
cmark_html_render_cr(html);
cmark_strbuf_puts(html, "<li");
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_putc(html, '>');
} else {
cmark_strbuf_puts(html, "</li>\n");
}
break;
case CMARK_NODE_HEADING:
if (entering) {
cmark_html_render_cr(html);
start_heading[2] = (char)('0' + node->as.heading.level);
cmark_strbuf_puts(html, start_heading);
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_putc(html, '>');
} else {
end_heading[3] = (char)('0' + node->as.heading.level);
cmark_strbuf_puts(html, end_heading);
cmark_strbuf_puts(html, ">\n");
}
break;
case CMARK_NODE_CODE_BLOCK:
cmark_html_render_cr(html);
if (node->as.code.info.len == 0) {
cmark_strbuf_puts(html, "<pre");
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, "><code>");
} else {
bufsize_t first_tag = 0;
while (first_tag < node->as.code.info.len &&
!cmark_isspace(node->as.code.info.data[first_tag])) {
first_tag += 1;
}
if (options & CMARK_OPT_GITHUB_PRE_LANG) {
cmark_strbuf_puts(html, "<pre");
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, " lang=\"");
escape_html(html, node->as.code.info.data, first_tag);
if (first_tag < node->as.code.info.len && (options & CMARK_OPT_FULL_INFO_STRING)) {
cmark_strbuf_puts(html, "\" data-meta=\"");
escape_html(html, node->as.code.info.data + first_tag + 1, node->as.code.info.len - first_tag - 1);
}
cmark_strbuf_puts(html, "\"><code>");
} else {
cmark_strbuf_puts(html, "<pre");
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, "><code class=\"language-");
escape_html(html, node->as.code.info.data, first_tag);
if (first_tag < node->as.code.info.len && (options & CMARK_OPT_FULL_INFO_STRING)) {
cmark_strbuf_puts(html, "\" data-meta=\"");
escape_html(html, node->as.code.info.data + first_tag + 1, node->as.code.info.len - first_tag - 1);
}
cmark_strbuf_puts(html, "\">");
}
}
escape_html(html, node->as.code.literal.data, node->as.code.literal.len);
cmark_strbuf_puts(html, "</code></pre>\n");
break;
case CMARK_NODE_HTML_BLOCK:
cmark_html_render_cr(html);
if (!(options & CMARK_OPT_UNSAFE)) {
cmark_strbuf_puts(html, "<!-- raw HTML omitted -->");
} else if (renderer->filter_extensions) {
filter_html_block(renderer, node->as.literal.data, node->as.literal.len);
} else {
cmark_strbuf_put(html, node->as.literal.data, node->as.literal.len);
}
cmark_html_render_cr(html);
break;
case CMARK_NODE_CUSTOM_BLOCK:
cmark_html_render_cr(html);
if (entering) {
cmark_strbuf_put(html, node->as.custom.on_enter.data,
node->as.custom.on_enter.len);
} else {
cmark_strbuf_put(html, node->as.custom.on_exit.data,
node->as.custom.on_exit.len);
}
cmark_html_render_cr(html);
break;
case CMARK_NODE_THEMATIC_BREAK:
cmark_html_render_cr(html);
cmark_strbuf_puts(html, "<hr");
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, " />\n");
break;
case CMARK_NODE_PARAGRAPH:
parent = cmark_node_parent(node);
grandparent = cmark_node_parent(parent);
if (grandparent != NULL && grandparent->type == CMARK_NODE_LIST) {
tight = grandparent->as.list.tight;
} else {
tight = false;
}
if (!tight) {
if (entering) {
cmark_html_render_cr(html);
cmark_strbuf_puts(html, "<p");
cmark_html_render_sourcepos(node, html, options);
cmark_strbuf_putc(html, '>');
} else {
if (parent->type == CMARK_NODE_FOOTNOTE_DEFINITION && node->next == NULL) {
cmark_strbuf_putc(html, ' ');
S_put_footnote_backref(renderer, html);
}
cmark_strbuf_puts(html, "</p>\n");
}
}
break;
case CMARK_NODE_TEXT:
escape_html(html, node->as.literal.data, node->as.literal.len);
break;
case CMARK_NODE_LINEBREAK:
cmark_strbuf_puts(html, "<br />\n");
break;
case CMARK_NODE_SOFTBREAK:
if (options & CMARK_OPT_HARDBREAKS) {
cmark_strbuf_puts(html, "<br />\n");
} else if (options & CMARK_OPT_NOBREAKS) {
cmark_strbuf_putc(html, ' ');
} else {
cmark_strbuf_putc(html, '\n');
}
break;
case CMARK_NODE_CODE:
cmark_strbuf_puts(html, "<code>");
escape_html(html, node->as.literal.data, node->as.literal.len);
cmark_strbuf_puts(html, "</code>");
break;
case CMARK_NODE_HTML_INLINE:
if (!(options & CMARK_OPT_UNSAFE)) {
cmark_strbuf_puts(html, "<!-- raw HTML omitted -->");
} else {
filtered = false;
for (it = renderer->filter_extensions; it; it = it->next) {
ext = (cmark_syntax_extension *) it->data;
if (!ext->html_filter_func(ext, node->as.literal.data, node->as.literal.len)) {
filtered = true;
break;
}
}
if (!filtered) {
cmark_strbuf_put(html, node->as.literal.data, node->as.literal.len);
} else {
cmark_strbuf_puts(html, "<");
cmark_strbuf_put(html, node->as.literal.data + 1, node->as.literal.len - 1);
}
}
break;
case CMARK_NODE_CUSTOM_INLINE:
if (entering) {
cmark_strbuf_put(html, node->as.custom.on_enter.data,
node->as.custom.on_enter.len);
} else {
cmark_strbuf_put(html, node->as.custom.on_exit.data,
node->as.custom.on_exit.len);
}
break;
case CMARK_NODE_STRONG:
if (entering) {
cmark_strbuf_puts(html, "<strong>");
} else {
cmark_strbuf_puts(html, "</strong>");
}
break;
case CMARK_NODE_EMPH:
if (entering) {
cmark_strbuf_puts(html, "<em>");
} else {
cmark_strbuf_puts(html, "</em>");
}
break;
case CMARK_NODE_LINK:
if (entering) {
cmark_strbuf_puts(html, "<a href=\"");
if (!(!(options & CMARK_OPT_UNSAFE) &&
scan_dangerous_url(&node->as.link.url, 0))) {
houdini_escape_href(html, node->as.link.url.data,
node->as.link.url.len);
}
if (node->as.link.title.len) {
cmark_strbuf_puts(html, "\" title=\"");
escape_html(html, node->as.link.title.data, node->as.link.title.len);
}
cmark_strbuf_puts(html, "\">");
} else {
cmark_strbuf_puts(html, "</a>");
}
break;
case CMARK_NODE_IMAGE:
if (entering) {
cmark_strbuf_puts(html, "<img src=\"");
if (!(!(options & CMARK_OPT_UNSAFE) &&
scan_dangerous_url(&node->as.link.url, 0))) {
houdini_escape_href(html, node->as.link.url.data,
node->as.link.url.len);
}
cmark_strbuf_puts(html, "\" alt=\"");
renderer->plain = node;
} else {
if (node->as.link.title.len) {
cmark_strbuf_puts(html, "\" title=\"");
escape_html(html, node->as.link.title.data, node->as.link.title.len);
}
cmark_strbuf_puts(html, "\" />");
}
break;
case CMARK_NODE_FOOTNOTE_DEFINITION:
if (entering) {
if (renderer->footnote_ix == 0) {
cmark_strbuf_puts(html, "<section class=\"footnotes\">\n<ol>\n");
}
++renderer->footnote_ix;
cmark_strbuf_puts(html, "<li id=\"fn");
char n[32];
snprintf(n, sizeof(n), "%d", renderer->footnote_ix);
cmark_strbuf_puts(html, n);
cmark_strbuf_puts(html, "\">\n");
} else {
if (S_put_footnote_backref(renderer, html)) {
cmark_strbuf_putc(html, '\n');
}
cmark_strbuf_puts(html, "</li>\n");
}
break;
case CMARK_NODE_FOOTNOTE_REFERENCE:
if (entering) {
cmark_strbuf_puts(html, "<sup class=\"footnote-ref\"><a href=\"#fn");
cmark_strbuf_put(html, node->as.literal.data, node->as.literal.len);
cmark_strbuf_puts(html, "\" id=\"fnref");
cmark_strbuf_put(html, node->as.literal.data, node->as.literal.len);
cmark_strbuf_puts(html, "\">");
cmark_strbuf_put(html, node->as.literal.data, node->as.literal.len);
cmark_strbuf_puts(html, "</a></sup>");
}
break;
default:
assert(false);
break;
}
return 1;
}
static int
S_render_node(cmark_node *node, cmark_event_type ev_type,
struct render_state *state, int options)
{
cmark_node *parent;
cmark_node *grandparent;
cmark_strbuf *html = state->html;
char start_header[] = "<h0";
char end_header[] = "</h0";
bool tight;
bool entering = (ev_type == CMARK_EVENT_ENTER);
if (state->plain == node) { // back at original node
state->plain = NULL;
}
if (state->plain != NULL) {
switch(node->type) {
case CMARK_NODE_TEXT:
case CMARK_NODE_CODE:
case CMARK_NODE_INLINE_HTML:
escape_html(html, node->as.literal.data,
node->as.literal.len);
break;
case CMARK_NODE_LINEBREAK:
case CMARK_NODE_SOFTBREAK:
cmark_strbuf_putc(html, ' ');
break;
default:
break;
}
return 1;
}
switch (node->type) {
case CMARK_NODE_DOCUMENT:
break;
case CMARK_NODE_BLOCK_QUOTE:
if (entering) {
cr(html);
cmark_strbuf_puts(html, "<blockquote");
S_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, ">\n");
} else {
cr(html);
cmark_strbuf_puts(html, "</blockquote>\n");
}
break;
case CMARK_NODE_LIST: {
cmark_list_type list_type = node->as.list.list_type;
int start = node->as.list.start;
if (entering) {
cr(html);
if (list_type == CMARK_BULLET_LIST) {
cmark_strbuf_puts(html, "<ul");
S_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, ">\n");
} else if (start == 1) {
cmark_strbuf_puts(html, "<ol");
S_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, ">\n");
} else {
cmark_strbuf_printf(html,
"<ol start=\"%d\"",
start);
S_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, ">\n");
}
} else {
cmark_strbuf_puts(html,
list_type == CMARK_BULLET_LIST ?
"</ul>\n" : "</ol>\n");
}
break;
}
case CMARK_NODE_ITEM:
if (entering) {
cr(html);
cmark_strbuf_puts(html, "<li");
S_render_sourcepos(node, html, options);
cmark_strbuf_putc(html, '>');
} else {
cmark_strbuf_puts(html, "</li>\n");
}
break;
case CMARK_NODE_HEADER:
if (entering) {
cr(html);
start_header[2] = '0' + node->as.header.level;
cmark_strbuf_puts(html, start_header);
S_render_sourcepos(node, html, options);
cmark_strbuf_putc(html, '>');
} else {
end_header[3] = '0' + node->as.header.level;
cmark_strbuf_puts(html, end_header);
cmark_strbuf_puts(html, ">\n");
}
break;
case CMARK_NODE_CODE_BLOCK:
cr(html);
if (!node->as.code.fenced || node->as.code.info.len == 0) {
cmark_strbuf_puts(html, "<pre");
S_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, "><code>");
} else {
int first_tag = 0;
while (first_tag < node->as.code.info.len &&
node->as.code.info.data[first_tag] != ' ') {
first_tag += 1;
}
cmark_strbuf_puts(html, "<pre");
S_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, "><code class=\"language-");
escape_html(html, node->as.code.info.data, first_tag);
cmark_strbuf_puts(html, "\">");
}
escape_html(html, node->as.code.literal.data,
node->as.code.literal.len);
cmark_strbuf_puts(html, "</code></pre>\n");
break;
case CMARK_NODE_HTML:
cr(html);
cmark_strbuf_put(html, node->as.literal.data, node->as.literal.len);
break;
case CMARK_NODE_HRULE:
cr(html);
cmark_strbuf_puts(html, "<hr");
S_render_sourcepos(node, html, options);
cmark_strbuf_puts(html, " />\n");
break;
case CMARK_NODE_PARAGRAPH:
parent = cmark_node_parent(node);
grandparent = cmark_node_parent(parent);
if (grandparent != NULL &&
grandparent->type == CMARK_NODE_LIST) {
tight = grandparent->as.list.tight;
} else {
tight = false;
}
if (!tight) {
if (entering) {
cr(html);
cmark_strbuf_puts(html, "<p");
S_render_sourcepos(node, html, options);
cmark_strbuf_putc(html, '>');
} else {
cmark_strbuf_puts(html, "</p>\n");
}
}
break;
case CMARK_NODE_TEXT:
escape_html(html, node->as.literal.data, node->as.literal.len);
break;
case CMARK_NODE_LINEBREAK:
cmark_strbuf_puts(html, "<br />\n");
break;
case CMARK_NODE_SOFTBREAK:
if (options & CMARK_OPT_HARDBREAKS) {
cmark_strbuf_puts(html, "<br />\n");
} else {
cmark_strbuf_putc(html, '\n');
}
break;
case CMARK_NODE_CODE:
cmark_strbuf_puts(html, "<code>");
escape_html(html, node->as.literal.data, node->as.literal.len);
cmark_strbuf_puts(html, "</code>");
break;
case CMARK_NODE_INLINE_HTML:
cmark_strbuf_put(html, node->as.literal.data, node->as.literal.len);
break;
case CMARK_NODE_STRONG:
if (entering) {
cmark_strbuf_puts(html, "<strong>");
} else {
cmark_strbuf_puts(html, "</strong>");
}
break;
case CMARK_NODE_EMPH:
if (entering) {
cmark_strbuf_puts(html, "<em>");
} else {
cmark_strbuf_puts(html, "</em>");
}
break;
case CMARK_NODE_LINK:
if (entering) {
cmark_strbuf_puts(html, "<a href=\"");
if (node->as.link.url)
escape_href(html, node->as.link.url, -1);
if (node->as.link.title) {
cmark_strbuf_puts(html, "\" title=\"");
escape_html(html, node->as.link.title, -1);
}
cmark_strbuf_puts(html, "\">");
} else {
cmark_strbuf_puts(html, "</a>");
}
break;
case CMARK_NODE_IMAGE:
if (entering) {
cmark_strbuf_puts(html, "<img src=\"");
if (node->as.link.url)
escape_href(html, node->as.link.url, -1);
cmark_strbuf_puts(html, "\" alt=\"");
state->plain = node;
} else {
if (node->as.link.title) {
cmark_strbuf_puts(html, "\" title=\"");
escape_html(html, node->as.link.title, -1);
}
cmark_strbuf_puts(html, "\" />");
}
break;
default:
assert(false);
break;
}
// cmark_strbuf_putc(html, 'x');
return 1;
}
static cmark_node *fixup_nodes(cmark_syntax_extension *self,
cmark_parser *parser,
cmark_inline_parser *inline_parser,
cmark_node *parent,
int start_offset,
int size)
{
int node_text_len;
cmark_node *prev = NULL;
cmark_node *tmp;
int name_size = size;
cmark_strbuf *name;
NamedLink *named_link;
for (prev = cmark_node_last_child(parent); prev; prev = cmark_node_previous(prev)) {
if (cmark_node_get_type(prev) == CMARK_NODE_TEXT) {
const char *text = cmark_node_get_literal(prev);
node_text_len = strlen(text);
size -= node_text_len;
if (size <= 0) {
if (size < 0) {
char *split_text = my_strndup(text, size * -1);
cmark_node *split = cmark_node_new(CMARK_NODE_TEXT);
cmark_node_set_literal(split, split_text);
free(split_text);
split_text = my_strndup(text + (size * - 1), node_text_len - size);
cmark_node_set_literal(prev, split_text);
free(split_text);
cmark_node_insert_before(prev, split);
}
break;
}
} else {
return NULL;
}
}
name = cmark_strbuf_new(name_size + 1);
tmp = prev;
while (tmp) {
cmark_node *next = cmark_node_next(tmp);
cmark_strbuf_puts(name, cmark_node_get_literal(tmp));
if (tmp != prev)
cmark_node_free(tmp);
tmp = next;
}
named_link = PRIV(self)->link_resolve_func(cmark_strbuf_get(name));
if (!named_link || !named_link->ref) {
int actual_line, actual_col;
translate_sourcepos(get_first_parent_block(parent),
start_offset, &actual_line, &actual_col);
cmark_strbuf *message = cmark_strbuf_new(0);
cmark_strbuf_puts(message, "Trying to link to non-existing symbol ‘");
cmark_strbuf_puts(message, cmark_strbuf_get(name));
cmark_strbuf_puts(message, "’");
diagnose("gtk-doc-bad-link", cmark_strbuf_get(message), actual_line - 1,
actual_col - 1);
cmark_strbuf_free(message);
cmark_node_set_literal(prev, cmark_strbuf_get(name));
cmark_strbuf_free(name);
return prev;
}
free_named_link(named_link);
cmark_node_set_type(prev, CMARK_NODE_LINK);
cmark_node_set_url(prev, cmark_strbuf_get(name));
cmark_strbuf_free(name);
return prev;
}
static int
S_render_node(cmark_renderer *renderer,
cmark_node *node,
cmark_event_type ev_type,
int options)
{
cmark_node *tmp;
int list_number;
cmark_delim_type list_delim;
int numticks;
int i;
bool entering = (ev_type == CMARK_EVENT_ENTER);
const char *info, *code, *title;
size_t info_len, code_len;
cmark_strbuf listmarker = GH_BUF_INIT;
char *emph_delim;
bufsize_t marker_width;
// Don't adjust tight list status til we've started the list.
// Otherwise we loose the blank line between a paragraph and
// a following list.
if (!(node->type == CMARK_NODE_ITEM && node->prev == NULL &&
entering)) {
tmp = get_containing_block(node);
renderer->in_tight_list_item =
(tmp->type == CMARK_NODE_ITEM &&
cmark_node_get_list_tight(tmp->parent)) ||
(tmp &&
tmp->parent &&
tmp->parent->type == CMARK_NODE_ITEM &&
cmark_node_get_list_tight(tmp->parent->parent));
}
switch (node->type) {
case CMARK_NODE_DOCUMENT:
break;
case CMARK_NODE_BLOCK_QUOTE:
if (entering) {
LIT("> ");
cmark_strbuf_puts(renderer->prefix, "> ");
} else {
cmark_strbuf_truncate(renderer->prefix,
renderer->prefix->size - 2);
BLANKLINE();
}
break;
case CMARK_NODE_LIST:
if (!entering && node->next &&
(node->next->type == CMARK_NODE_CODE_BLOCK ||
node->next->type == CMARK_NODE_LIST)) {
// this ensures 2 blank lines after list,
// if before code block or list:
LIT("\n");
}
break;
case CMARK_NODE_ITEM:
if (cmark_node_get_list_type(node->parent) ==
CMARK_BULLET_LIST) {
marker_width = 2;
} else {
list_number = cmark_node_get_list_start(node->parent);
list_delim = cmark_node_get_list_delim(node->parent);
tmp = node;
while (tmp->prev) {
tmp = tmp->prev;
list_number += 1;
}
// we ensure a width of at least 4 so
// we get nice transition from single digits
// to double
cmark_strbuf_printf(&listmarker,
"%d%s%s", list_number,
list_delim == CMARK_PAREN_DELIM ?
")" : ".",
list_number < 10 ? " " : " ");
marker_width = listmarker.size;
}
if (entering) {
if (cmark_node_get_list_type(node->parent) ==
CMARK_BULLET_LIST) {
LIT("* ");
cmark_strbuf_puts(renderer->prefix, " ");
} else {
LIT((char *)listmarker.ptr);
for (i = marker_width; i--;) {
cmark_strbuf_putc(renderer->prefix, ' ');
}
}
} else {
cmark_strbuf_truncate(renderer->prefix,
renderer->prefix->size -
marker_width);
CR();
}
cmark_strbuf_free(&listmarker);
break;
case CMARK_NODE_HEADER:
if (entering) {
for (int i = cmark_node_get_header_level(node); i > 0; i--) {
LIT("#");
}
LIT(" ");
renderer->no_wrap = true;
} else {
renderer->no_wrap = false;
BLANKLINE();
}
break;
case CMARK_NODE_CODE_BLOCK:
BLANKLINE();
info = cmark_node_get_fence_info(node);
info_len = safe_strlen(info);
code = cmark_node_get_literal(node);
code_len = safe_strlen(code);
// use indented form if no info, and code doesn't
// begin or end with a blank line, and code isn't
// first thing in a list item
if (info_len == 0 &&
(code_len > 2 &&
!isspace(code[0]) &&
!(isspace(code[code_len - 1]) &&
isspace(code[code_len - 2]))) &&
!(node->prev == NULL && node->parent &&
node->parent->type == CMARK_NODE_ITEM)) {
LIT(" ");
cmark_strbuf_puts(renderer->prefix, " ");
OUT(cmark_node_get_literal(node), false, LITERAL);
cmark_strbuf_truncate(renderer->prefix,
renderer->prefix->size - 4);
} else {
numticks = longest_backtick_sequence(code) + 1;
if (numticks < 3) {
numticks = 3;
}
for (i = 0; i < numticks; i++) {
LIT("`");
}
LIT(" ");
OUT(info, false, LITERAL);
CR();
OUT(cmark_node_get_literal(node), false, LITERAL);
CR();
for (i = 0; i < numticks; i++) {
LIT("`");
}
}
BLANKLINE();
break;
case CMARK_NODE_HTML:
BLANKLINE();
OUT(cmark_node_get_literal(node), false, LITERAL);
BLANKLINE();
break;
case CMARK_NODE_HRULE:
BLANKLINE();
LIT("-----");
BLANKLINE();
break;
case CMARK_NODE_PARAGRAPH:
if (!entering) {
BLANKLINE();
}
break;
case CMARK_NODE_TEXT:
OUT(cmark_node_get_literal(node), true, NORMAL);
break;
case CMARK_NODE_LINEBREAK:
if (!(CMARK_OPT_HARDBREAKS & options)) {
LIT("\\");
}
CR();
break;
case CMARK_NODE_SOFTBREAK:
if (renderer->width == 0 &&
!(CMARK_OPT_HARDBREAKS & options)) {
CR();
} else {
OUT(" ", true, LITERAL);
}
break;
case CMARK_NODE_CODE:
code = cmark_node_get_literal(node);
code_len = safe_strlen(code);
numticks = shortest_unused_backtick_sequence(code);
for (i = 0; i < numticks; i++) {
LIT("`");
}
if (code_len == 0 || code[0] == '`') {
LIT(" ");
}
OUT(cmark_node_get_literal(node), true, LITERAL);
if (code_len == 0 || code[code_len - 1] == '`') {
LIT(" ");
}
for (i = 0; i < numticks; i++) {
LIT("`");
}
break;
case CMARK_NODE_INLINE_HTML:
OUT(cmark_node_get_literal(node), false, LITERAL);
break;
case CMARK_NODE_STRONG:
if (entering) {
LIT("**");
} else {
LIT("**");
}
break;
case CMARK_NODE_EMPH:
// If we have EMPH(EMPH(x)), we need to use *_x_*
// because **x** is STRONG(x):
if (node->parent && node->parent->type == CMARK_NODE_EMPH &&
node->next == NULL && node->prev == NULL) {
emph_delim = "_";
} else {
emph_delim = "*";
}
if (entering) {
LIT(emph_delim);
} else {
LIT(emph_delim);
}
break;
case CMARK_NODE_LINK:
if (is_autolink(node)) {
if (entering) {
LIT("<");
if (strncmp(cmark_node_get_url(node),
"mailto:", 7) == 0) {
LIT((char *)cmark_node_get_url(node) + 7);
} else {
LIT((char *)cmark_node_get_url(node));
}
LIT(">");
// return signal to skip contents of node...
return 0;
}
} else {
if (entering) {
LIT("[");
} else {
LIT("](");
OUT(cmark_node_get_url(node), false, URL);
title = cmark_node_get_title(node);
if (safe_strlen(title) > 0) {
LIT(" \"");
OUT(title, false, TITLE);
LIT("\"");
}
LIT(")");
}
}
break;
case CMARK_NODE_IMAGE:
if (entering) {
LIT("![");
} else {
LIT("](");
OUT(cmark_node_get_url(node), false, URL);
title = cmark_node_get_title(node);
if (safe_strlen(title) > 0) {
OUT(" \"", true, LITERAL);
OUT(title, false, TITLE);
LIT("\"");
}
LIT(")");
}
break;
default:
assert(false);
break;
}
return 1;
}
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);
}
