static link_type get_link_type(cmark_node *node) {
size_t title_len, url_len;
cmark_node *link_text;
char *realurl;
int realurllen;
bool isemail = false;
if (node->type != CMARK_NODE_LINK) {
return NO_LINK;
}
const char *url = cmark_node_get_url(node);
cmark_chunk url_chunk = cmark_chunk_literal(url);
if (url && *url == '#') {
return INTERNAL_LINK;
}
url_len = strlen(url);
if (url_len == 0 || scan_scheme(&url_chunk, 0) == 0) {
return NO_LINK;
}
const char *title = cmark_node_get_title(node);
title_len = strlen(title);
// if it has a title, we can't treat it as an autolink:
if (title_len == 0) {
link_text = node->first_child;
cmark_consolidate_text_nodes(link_text);
if (!link_text)
return NO_LINK;
realurl = (char *)url;
realurllen = (int)url_len;
if (strncmp(realurl, "mailto:", 7) == 0) {
realurl += 7;
realurllen -= 7;
isemail = true;
}
if (realurllen == link_text->as.literal.len &&
strncmp(realurl, (char *)link_text->as.literal.data,
link_text->as.literal.len) == 0) {
if (isemail) {
return EMAIL_AUTOLINK;
} else {
return URL_AUTOLINK;
}
}
}
return NORMAL_LINK;
}
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_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
accessors(test_batch_runner *runner)
{
static const char markdown[] =
"## Header\n"
"\n"
"* Item 1\n"
"* Item 2\n"
"\n"
"2. Item 1\n"
"\n"
"3. Item 2\n"
"\n"
"\n"
" code\n"
"\n"
"``` lang\n"
"fenced\n"
"```\n"
"\n"
"<div>html</div>\n"
"\n"
"[link](url 'title')\n";
cmark_node *doc = cmark_parse_document(markdown, sizeof(markdown) - 1, CMARK_OPT_DEFAULT);
// Getters
cmark_node *header = cmark_node_first_child(doc);
INT_EQ(runner, cmark_node_get_header_level(header), 2,
"get_header_level");
cmark_node *bullet_list = cmark_node_next(header);
INT_EQ(runner, cmark_node_get_list_type(bullet_list),
CMARK_BULLET_LIST, "get_list_type bullet");
INT_EQ(runner, cmark_node_get_list_tight(bullet_list), 1,
"get_list_tight tight");
cmark_node *ordered_list = cmark_node_next(bullet_list);
INT_EQ(runner, cmark_node_get_list_type(ordered_list),
CMARK_ORDERED_LIST, "get_list_type ordered");
INT_EQ(runner, cmark_node_get_list_delim(ordered_list),
CMARK_PERIOD_DELIM, "get_list_delim ordered");
INT_EQ(runner, cmark_node_get_list_start(ordered_list), 2,
"get_list_start");
INT_EQ(runner, cmark_node_get_list_tight(ordered_list), 0,
"get_list_tight loose");
cmark_node *code = cmark_node_next(ordered_list);
STR_EQ(runner, cmark_node_get_literal(code), "code\n",
"get_literal indented code");
cmark_node *fenced = cmark_node_next(code);
STR_EQ(runner, cmark_node_get_literal(fenced), "fenced\n",
"get_literal fenced code");
STR_EQ(runner, cmark_node_get_fence_info(fenced), "lang",
"get_fence_info");
cmark_node *html = cmark_node_next(fenced);
STR_EQ(runner, cmark_node_get_literal(html),
"<div>html</div>\n", "get_literal html");
cmark_node *paragraph = cmark_node_next(html);
INT_EQ(runner, cmark_node_get_start_line(paragraph), 19,
"get_start_line");
INT_EQ(runner, cmark_node_get_start_column(paragraph), 1,
"get_start_column");
INT_EQ(runner, cmark_node_get_end_line(paragraph), 19,
"get_end_line");
cmark_node *link = cmark_node_first_child(paragraph);
STR_EQ(runner, cmark_node_get_url(link), "url",
"get_url");
STR_EQ(runner, cmark_node_get_title(link), "title",
"get_title");
cmark_node *string = cmark_node_first_child(link);
STR_EQ(runner, cmark_node_get_literal(string), "link",
"get_literal string");
// Setters
OK(runner, cmark_node_set_header_level(header, 3),
"set_header_level");
OK(runner, cmark_node_set_list_type(bullet_list, CMARK_ORDERED_LIST),
"set_list_type ordered");
OK(runner, cmark_node_set_list_delim(bullet_list, CMARK_PAREN_DELIM),
"set_list_delim paren");
OK(runner, cmark_node_set_list_start(bullet_list, 3),
"set_list_start");
OK(runner, cmark_node_set_list_tight(bullet_list, 0),
"set_list_tight loose");
OK(runner, cmark_node_set_list_type(ordered_list, CMARK_BULLET_LIST),
"set_list_type bullet");
OK(runner, cmark_node_set_list_tight(ordered_list, 1),
"set_list_tight tight");
OK(runner, cmark_node_set_literal(code, "CODE\n"),
"set_literal indented code");
OK(runner, cmark_node_set_literal(fenced, "FENCED\n"),
"set_literal fenced code");
OK(runner, cmark_node_set_fence_info(fenced, "LANG"),
"set_fence_info");
OK(runner, cmark_node_set_literal(html, "<div>HTML</div>\n"),
"set_literal html");
OK(runner, cmark_node_set_url(link, "URL"),
"set_url");
OK(runner, cmark_node_set_title(link, "TITLE"),
"set_title");
OK(runner, cmark_node_set_literal(string, "LINK"),
"set_literal string");
char *rendered_html = cmark_render_html(doc, CMARK_OPT_DEFAULT);
static const char expected_html[] =
"<h3>Header</h3>\n"
"<ol start=\"3\">\n"
"<li>\n"
"<p>Item 1</p>\n"
"</li>\n"
"<li>\n"
"<p>Item 2</p>\n"
"</li>\n"
"</ol>\n"
"<ul>\n"
"<li>Item 1</li>\n"
"<li>Item 2</li>\n"
"</ul>\n"
"<pre><code>CODE\n"
"</code></pre>\n"
"<pre><code class=\"language-LANG\">FENCED\n"
"</code></pre>\n"
"<div>HTML</div>\n"
"<p><a href=\"URL\" title=\"TITLE\">LINK</a></p>\n";
STR_EQ(runner, rendered_html, expected_html, "setters work");
free(rendered_html);
// Getter errors
INT_EQ(runner, cmark_node_get_header_level(bullet_list), 0,
"get_header_level error");
INT_EQ(runner, cmark_node_get_list_type(header), CMARK_NO_LIST,
"get_list_type error");
INT_EQ(runner, cmark_node_get_list_start(code), 0,
"get_list_start error");
INT_EQ(runner, cmark_node_get_list_tight(fenced), 0,
"get_list_tight error");
OK(runner, cmark_node_get_literal(ordered_list) == NULL,
"get_literal error");
OK(runner, cmark_node_get_fence_info(paragraph) == NULL,
"get_fence_info error");
OK(runner, cmark_node_get_url(html) == NULL,
"get_url error");
OK(runner, cmark_node_get_title(header) == NULL,
"get_title error");
// Setter errors
OK(runner, !cmark_node_set_header_level(bullet_list, 3),
"set_header_level error");
OK(runner, !cmark_node_set_list_type(header, CMARK_ORDERED_LIST),
"set_list_type error");
OK(runner, !cmark_node_set_list_start(code, 3),
"set_list_start error");
OK(runner, !cmark_node_set_list_tight(fenced, 0),
"set_list_tight error");
OK(runner, !cmark_node_set_literal(ordered_list, "content\n"),
"set_literal error");
OK(runner, !cmark_node_set_fence_info(paragraph, "lang"),
"set_fence_info error");
OK(runner, !cmark_node_set_url(html, "url"),
"set_url error");
OK(runner, !cmark_node_set_title(header, "title"),
"set_title error");
OK(runner, !cmark_node_set_header_level(header, 0),
"set_header_level too small");
OK(runner, !cmark_node_set_header_level(header, 7),
"set_header_level too large");
OK(runner, !cmark_node_set_list_type(bullet_list, CMARK_NO_LIST),
"set_list_type invalid");
OK(runner, !cmark_node_set_list_start(bullet_list, -1),
"set_list_start negative");
cmark_node_free(doc);
}