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);
}
static void md_block_external_images(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);
if(CMARK_NODE_IMAGE != cmark_node_get_type(node)) continue;
char const *const URI = cmark_node_get_url(node);
if(URI) {
if(0 == strncasecmp(URI, STR_LEN("hash:"))) continue;
if(0 == strncasecmp(URI, STR_LEN("data:"))) continue;
}
cmark_node *link = cmark_node_new(CMARK_NODE_LINK);
cmark_node *text = cmark_node_new(CMARK_NODE_TEXT);
cmark_node_set_url(link, URI);
for(;;) {
cmark_node *child = cmark_node_first_child(node);
if(!child) break;
cmark_node_append_child(link, child);
}
if(cmark_node_first_child(link)) {
cmark_node_set_literal(text, " (external image)");
} else {
cmark_node_set_literal(text, "(external image)");
}
cmark_node_append_child(link, text);
cmark_node_insert_before(node, link);
cmark_node_free(node);
}
}
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);
}
}
static cmark_node *postprocess(cmark_syntax_extension *ext, cmark_parser *parser, cmark_node *root) {
cmark_iter *iter;
cmark_event_type ev;
cmark_node *node;
bool in_link = false;
cmark_consolidate_text_nodes(root);
iter = cmark_iter_new(root);
while ((ev = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
node = cmark_iter_get_node(iter);
if (in_link) {
if (ev == CMARK_EVENT_EXIT && node->type == CMARK_NODE_LINK) {
in_link = false;
}
continue;
}
if (ev == CMARK_EVENT_ENTER && node->type == CMARK_NODE_LINK) {
in_link = true;
continue;
}
if (ev == CMARK_EVENT_ENTER && node->type == CMARK_NODE_TEXT) {
postprocess_text(parser, node, 0);
}
}
cmark_iter_free(iter);
return root;
}
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;
}
void
cmark_iter_reset(cmark_iter *iter, cmark_node *current,
cmark_event_type event_type)
{
iter->next.ev_type = event_type;
iter->next.node = current;
cmark_iter_next(iter);
}
static void md_autolink(cmark_iter *const iter) {
regex_t linkify[1];
// <http://daringfireball.net/2010/07/improved_regex_for_matching_urls>
// Painstakingly ported to POSIX
int rc = regcomp(linkify, "([a-z][a-z0-9_-]+:(/{1,3}|[a-z0-9%])|www[0-9]{0,3}[.]|[a-z0-9.-]+[.][a-z]{2,4}/)([^[:space:]()<>]+|\\(([^[:space:]()<>]+|(\\([^[:space:]()<>]+\\)))*\\))+(\\(([^[:space:]()<>]+|(\\([^[:space:]()<>]+\\)))*\\)|[^][[:space:]`!(){};:'\".,<>?«»“”‘’])", REG_ICASE | REG_EXTENDED);
assert(0 == rc);
for(;;) {
cmark_event_type const event = cmark_iter_next(iter);
if(CMARK_EVENT_DONE == event) break;
if(CMARK_EVENT_ENTER != event) continue;
cmark_node *const node = cmark_iter_get_node(iter);
if(CMARK_NODE_TEXT != cmark_node_get_type(node)) continue;
char const *const str = cmark_node_get_literal(node);
char const *pos = str;
regmatch_t match;
while(0 == regexec(linkify, pos, 1, &match, 0)) {
regoff_t const loc = match.rm_so;
regoff_t const len = match.rm_eo - match.rm_so;
char *pfx = strndup(pos, loc);
char *link_abs = strndup(pos+loc, len);
char *link_rel = aasprintf("/history/%s", link_abs);
assert(pfx);
assert(link_abs);
assert(link_rel);
cmark_node *text = cmark_node_new(CMARK_NODE_TEXT);
cmark_node_set_literal(text, pfx);
cmark_node *link = cmark_node_new(CMARK_NODE_LINK);
cmark_node_set_url(link, link_rel);
cmark_node *sup = superscript("^", "", link_abs);
cmark_node *face = cmark_node_new(CMARK_NODE_TEXT);
cmark_node_set_literal(face, link_abs);
cmark_node_append_child(link, face);
cmark_node_insert_before(node, text);
cmark_node_insert_before(node, link);
cmark_node_insert_before(node, sup);
free(pfx); pfx = NULL;
free(link_abs); link_abs = NULL;
free(link_rel); link_rel = NULL;
pos += loc+len;
}
if(str != pos) {
cmark_node *text = cmark_node_new(CMARK_NODE_TEXT);
cmark_node_set_literal(text, pos);
cmark_node_insert_before(node, text);
cmark_node_free(node);
}
}
regfree(linkify);
}
static void md_escape_inline(cmark_iter *const iter) {
for(;;) {
cmark_event_type const event = cmark_iter_next(iter);
if(CMARK_EVENT_DONE == event) break;
if(CMARK_EVENT_ENTER != event) continue;
cmark_node *const node = cmark_iter_get_node(iter);
if(CMARK_NODE_INLINE_HTML != cmark_node_get_type(node)) continue;
char const *const str = cmark_node_get_literal(node);
cmark_node *text = cmark_node_new(CMARK_NODE_TEXT);
cmark_node_set_literal(text, str);
cmark_node_insert_before(node, text);
cmark_node_free(node);
}
}
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;
}
// Walk through node and all children, recursively, parsing
// string content into inline content where appropriate.
static void process_inlines(cmark_node* root, cmark_reference_map *refmap, int options)
{
cmark_iter *iter = cmark_iter_new(root);
cmark_node *cur;
cmark_event_type ev_type;
while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
cur = cmark_iter_get_node(iter);
if (ev_type == CMARK_EVENT_ENTER) {
if (cur->type == NODE_PARAGRAPH ||
cur->type == NODE_HEADER) {
cmark_parse_inlines(cur, refmap, options);
}
}
}
cmark_iter_free(iter);
}
char *cmark_render_html(cmark_node *root, int options)
{
char *result;
cmark_strbuf html = GH_BUF_INIT;
cmark_event_type ev_type;
cmark_node *cur;
struct render_state state = { &html, NULL };
cmark_iter *iter = cmark_iter_new(root);
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(&html);
cmark_iter_free(iter);
return result;
}
static void
iterator(test_batch_runner *runner) {
cmark_node *doc = cmark_parse_document("> a *b*\n\nc", 10, CMARK_OPT_DEFAULT);
int parnodes = 0;
cmark_event_type ev_type;
cmark_iter *iter = cmark_iter_new(doc);
cmark_node *cur;
while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
cur = cmark_iter_get_node(iter);
if (cur->type == CMARK_NODE_PARAGRAPH &&
ev_type == CMARK_EVENT_ENTER) {
parnodes += 1;
}
}
INT_EQ(runner, parnodes, 2, "iterate correctly counts paragraphs");
cmark_iter_free(iter);
cmark_node_free(doc);
}
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;
}
static void
iterator_delete(test_batch_runner *runner) {
static const char md[] =
"a *b* c\n"
"\n"
"* item1\n"
"* item2\n"
"\n"
"a `b` c\n"
"\n"
"* item1\n"
"* item2\n";
cmark_node *doc = cmark_parse_document(md, sizeof(md) - 1,
CMARK_OPT_DEFAULT);
cmark_iter *iter = cmark_iter_new(doc);
cmark_event_type ev_type;
while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
cmark_node *node = cmark_iter_get_node(iter);
// Delete list, emph, and code nodes.
if ((ev_type == CMARK_EVENT_EXIT &&
node->type == CMARK_NODE_LIST) ||
(ev_type == CMARK_EVENT_EXIT &&
node->type == CMARK_NODE_EMPH) ||
(ev_type == CMARK_EVENT_ENTER &&
node->type == CMARK_NODE_CODE)) {
cmark_node_free(node);
}
}
char *html = cmark_render_html(doc, CMARK_OPT_DEFAULT);
static const char expected[] =
"<p>a c</p>\n"
"<p>a c</p>\n";
STR_EQ(runner, html, expected, "iterate and delete nodes");
free(html);
cmark_iter_free(iter);
cmark_node_free(doc);
}
// Convert a single node.
static char*
S_node_to_pod(cmark_node *node, CFCClass *klass, int header_level) {
char *result = CFCUtil_strdup("");
if (node == NULL) {
return result;
}
int found_matching_code_block = false;
cmark_iter *iter = cmark_iter_new(node);
cmark_event_type ev_type;
while (CMARK_EVENT_DONE != (ev_type = cmark_iter_next(iter))) {
cmark_node *node = cmark_iter_get_node(iter);
cmark_node_type type = cmark_node_get_type(node);
switch (type) {
case CMARK_NODE_DOCUMENT:
break;
case CMARK_NODE_PARAGRAPH:
if (ev_type == CMARK_EVENT_EXIT) {
result = CFCUtil_cat(result, "\n\n", NULL);
}
break;
case CMARK_NODE_BLOCK_QUOTE:
case CMARK_NODE_LIST:
if (ev_type == CMARK_EVENT_ENTER) {
result = CFCUtil_cat(result, "=over\n\n", NULL);
}
else {
result = CFCUtil_cat(result, "=back\n\n", NULL);
}
break;
case CMARK_NODE_ITEM:
// TODO: Ordered lists.
if (ev_type == CMARK_EVENT_ENTER) {
result = CFCUtil_cat(result, "=item *\n\n", NULL);
}
break;
case CMARK_NODE_HEADER:
if (ev_type == CMARK_EVENT_ENTER) {
int extra_level = cmark_node_get_header_level(node) - 1;
char *header = CFCUtil_sprintf("=head%d ",
header_level + extra_level);
result = CFCUtil_cat(result, header, NULL);
FREEMEM(header);
}
else {
result = CFCUtil_cat(result, "\n\n", NULL);
}
break;
case CMARK_NODE_CODE_BLOCK: {
int is_host = CFCMarkdown_code_block_is_host(node, "perl");
if (is_host) {
found_matching_code_block = true;
const char *content = cmark_node_get_literal(node);
char *copy = CFCUtil_strdup(content);
// Chomp trailing newline.
size_t len = strlen(copy);
if (len > 0 && copy[len-1] == '\n') {
copy[len-1] = '\0';
}
char *indented
= CFCUtil_global_replace(copy, "\n", "\n ");
result
= CFCUtil_cat(result, " ", indented, "\n\n", NULL);
FREEMEM(indented);
FREEMEM(copy);
}
if (CFCMarkdown_code_block_is_last(node)) {
if (!found_matching_code_block) {
result = CFCUtil_cat(result,
" Code example for Perl is missing\n\n");
}
else {
// Reset.
found_matching_code_block = false;
}
}
break;
}
case CMARK_NODE_HTML: {
const char *html = cmark_node_get_literal(node);
result = CFCUtil_cat(result, "=begin html\n\n", html,
"\n=end\n\n", NULL);
break;
}
case CMARK_NODE_HRULE:
break;
case CMARK_NODE_TEXT: {
const char *content = cmark_node_get_literal(node);
char *escaped = S_pod_escape(content);
result = CFCUtil_cat(result, escaped, NULL);
FREEMEM(escaped);
break;
}
case CMARK_NODE_LINEBREAK:
// POD doesn't support line breaks. Start a new paragraph.
result = CFCUtil_cat(result, "\n\n", NULL);
break;
case CMARK_NODE_SOFTBREAK:
result = CFCUtil_cat(result, "\n", NULL);
break;
case CMARK_NODE_CODE: {
const char *content = cmark_node_get_literal(node);
char *escaped = S_pod_escape(content);
result = CFCUtil_cat(result, "C<", escaped, ">", NULL);
FREEMEM(escaped);
break;
}
case CMARK_NODE_INLINE_HTML: {
const char *html = cmark_node_get_literal(node);
CFCUtil_warn("Inline HTML not supported in POD: %s", html);
break;
}
case CMARK_NODE_LINK:
if (ev_type == CMARK_EVENT_ENTER) {
char *pod = S_convert_link(node, klass, header_level);
result = CFCUtil_cat(result, pod, NULL);
FREEMEM(pod);
cmark_iter_reset(iter, node, CMARK_EVENT_EXIT);
}
break;
case CMARK_NODE_IMAGE:
CFCUtil_warn("Images not supported in POD");
break;
case CMARK_NODE_STRONG:
if (ev_type == CMARK_EVENT_ENTER) {
result = CFCUtil_cat(result, "B<", NULL);
}
else {
result = CFCUtil_cat(result, ">", NULL);
}
break;
case CMARK_NODE_EMPH:
if (ev_type == CMARK_EVENT_ENTER) {
result = CFCUtil_cat(result, "I<", NULL);
}
else {
result = CFCUtil_cat(result, ">", NULL);
}
break;
default:
CFCUtil_die("Invalid cmark node type: %d", (int)type);
break;
}
}
cmark_iter_free(iter);
return result;
}
