void cmark_parser_free(cmark_parser *parser) {
cmark_strbuf_free(parser->curline);
free(parser->curline);
cmark_strbuf_free(parser->linebuf);
free(parser->linebuf);
cmark_reference_map_free(parser->refmap);
free(parser);
}
char*
cmark_render(cmark_node *root,
int options,
int width,
void (*outc)(cmark_renderer*,
cmark_escaping,
int32_t,
unsigned char),
int (*render_node)(cmark_renderer *renderer,
cmark_node *node,
cmark_event_type ev_type,
int options))
{
cmark_strbuf pref = GH_BUF_INIT;
cmark_strbuf buf = GH_BUF_INIT;
cmark_node *cur;
cmark_event_type ev_type;
char *result;
cmark_iter *iter = cmark_iter_new(root);
cmark_renderer renderer = { &buf, &pref, 0, width,
0, 0, true, false, false,
outc, S_cr, S_blankline, S_out
};
while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
cur = cmark_iter_get_node(iter);
if (!render_node(&renderer, cur, ev_type, options)) {
// a false value causes us to skip processing
// the node's contents. this is used for
// autolinks.
cmark_iter_reset(iter, cur, CMARK_EVENT_EXIT);
}
}
// ensure final newline
if (renderer.buffer->ptr[renderer.buffer->size - 1] != '\n') {
cmark_strbuf_putc(renderer.buffer, '\n');
}
result = (char *)cmark_strbuf_detach(renderer.buffer);
cmark_iter_free(iter);
cmark_strbuf_free(renderer.prefix);
cmark_strbuf_free(renderer.buffer);
return result;
}
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);
}
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 void md_convert_hashes(cmark_iter *const iter) {
for(;;) {
cmark_event_type const event = cmark_iter_next(iter);
if(CMARK_EVENT_DONE == event) break;
if(CMARK_EVENT_EXIT != event) continue;
cmark_node *const node = cmark_iter_get_node(iter);
cmark_node_type const type = cmark_node_get_type(node);
if(CMARK_NODE_LINK != type && CMARK_NODE_IMAGE != type) continue;
char const *const URI = cmark_node_get_url(node);
if(!URI) continue;
if(0 != strncasecmp(URI, STR_LEN("hash:"))) continue;
cmark_node *sup = superscript("#", HASH_INFO_MSG, URI);
cmark_node_insert_after(node, sup);
cmark_iter_reset(iter, sup, CMARK_EVENT_EXIT);
char *escaped = QSEscape(URI, strlen(URI), true);
size_t const elen = strlen(escaped);
cmark_strbuf rel[1];
char const qpfx[] = "/sources/";
cmark_strbuf_init(&DEFAULT_MEM_ALLOCATOR, rel, sizeof(qpfx)-1+elen);
cmark_strbuf_put(rel, (unsigned char const *)qpfx, sizeof(qpfx)-1);
cmark_strbuf_put(rel, (unsigned char const *)escaped, elen);
free(escaped); escaped = NULL;
cmark_node_set_url(node, cmark_strbuf_cstr(rel));
cmark_strbuf_free(rel);
}
}
void cmark_free_doc_parser(cmark_doc_parser *parser)
{
cmark_strbuf_free(parser->curline);
free(parser->curline);
cmark_reference_map_free(parser->refmap);
free(parser);
}
static cmark_node *try_opening_code_block(cmark_syntax_extension *self,
bool indented,
cmark_parser *parser,
cmark_node *parent,
const char *input)
{
cmark_node *ret = NULL;
cmark_bufsize_t matched = scan_open_gtkdoc_code_block(input,
cmark_parser_get_first_nonspace(parser));
if (!indented && matched) {
ret = cmark_parser_add_child(parser, parent,
CMARK_NODE_CODE_BLOCK, cmark_parser_get_offset(parser));
cmark_node_set_syntax_extension(ret, self);
cmark_node_set_fenced(ret, true, 2,
cmark_parser_get_first_nonspace(parser) - cmark_parser_get_offset(parser),
0);
cmark_parser_advance_offset(parser, input, matched, false);
matched = scan_language_comment(input, matched);
if (matched) {
cmark_strbuf *lang = cmark_strbuf_new(32);
cmark_strbuf_put(lang, (unsigned char *) input + 17, matched - 20);
/* Will be transformed to fence info */
cmark_node_set_string_content(ret, cmark_strbuf_get(lang));
cmark_strbuf_free(lang);
cmark_parser_advance_offset(parser, input, matched, false);
}
}
return ret;
}
// Free a cmark_node list and any children.
static
void S_free_nodes(cmark_node *e)
{
cmark_node *next;
while (e != NULL) {
cmark_strbuf_free(&e->string_content);
switch (e->type) {
case NODE_CODE_BLOCK:
cmark_chunk_free(&e->as.code.info);
cmark_chunk_free(&e->as.code.literal);
break;
case NODE_TEXT:
case NODE_INLINE_HTML:
case NODE_CODE:
case NODE_HTML:
cmark_chunk_free(&e->as.literal);
break;
case NODE_LINK:
case NODE_IMAGE:
free(e->as.link.url);
free(e->as.link.title);
break;
default:
break;
}
if (e->last_child) {
// Splice children into list
e->last_child->next = e->next;
e->next = e->first_child;
}
next = e->next;
free(e);
e = next;
}
}
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;
}
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;
}
cmark_node *cmark_parser_finish(cmark_parser *parser)
{
if (parser->linebuf->size) {
S_process_line(parser, parser->linebuf->ptr,
parser->linebuf->size);
cmark_strbuf_clear(parser->linebuf);
}
finalize_document(parser);
cmark_strbuf_free(parser->curline);
#if CMARK_DEBUG_NODES
if (cmark_node_check(parser->root, stderr)) {
abort();
}
#endif
return parser->root;
}
char *cmark_render_man(cmark_node *root, long options)
{
char *result;
cmark_strbuf man = GH_BUF_INIT;
struct render_state state = { &man, NULL };
cmark_node *cur;
cmark_event_type ev_type;
cmark_iter *iter = cmark_iter_new(root);
if (options == 0) options = 0; // avoid warning about unused parameters
while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
cur = cmark_iter_get_node(iter);
S_render_node(cur, ev_type, &state);
}
result = (char *)cmark_strbuf_detach(&man);
cmark_iter_free(iter);
cmark_strbuf_free(&man);
return result;
}
cmark_node *cmark_parser_finish(cmark_parser *parser) {
if (parser->linebuf->size) {
S_process_line(parser, parser->linebuf->ptr, parser->linebuf->size);
cmark_strbuf_clear(parser->linebuf);
}
finalize_document(parser);
if (parser->options & CMARK_OPT_NORMALIZE) {
cmark_consolidate_text_nodes(parser->root);
}
cmark_strbuf_free(parser->curline);
#if CMARK_DEBUG_NODES
if (cmark_node_check(parser->root, stderr)) {
abort();
}
#endif
return parser->root;
}
// Free a cmark_node list and any children.
static void S_free_nodes(cmark_node *e) {
cmark_node *next;
while (e != NULL) {
cmark_strbuf_free(&e->content);
switch (e->type) {
case CMARK_NODE_CODE_BLOCK:
cmark_chunk_free(NODE_MEM(e), &e->as.code.info);
cmark_chunk_free(NODE_MEM(e), &e->as.code.literal);
break;
case CMARK_NODE_TEXT:
case CMARK_NODE_HTML_INLINE:
case CMARK_NODE_CODE:
case CMARK_NODE_HTML_BLOCK:
cmark_chunk_free(NODE_MEM(e), &e->as.literal);
break;
case CMARK_NODE_LINK:
case CMARK_NODE_IMAGE:
cmark_chunk_free(NODE_MEM(e), &e->as.link.url);
cmark_chunk_free(NODE_MEM(e), &e->as.link.title);
break;
case CMARK_NODE_CUSTOM_BLOCK:
case CMARK_NODE_CUSTOM_INLINE:
cmark_chunk_free(NODE_MEM(e), &e->as.custom.on_enter);
cmark_chunk_free(NODE_MEM(e), &e->as.custom.on_exit);
break;
default:
break;
}
if (e->last_child) {
// Splice children into list
e->last_child->next = e->next;
e->next = e->first_child;
}
next = e->next;
NODE_MEM(e)->free(e);
e = next;
}
}
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;
}
