// linear search for domain (and * if available)
static const tldnode *
findTldNode(const tldnode *parent, const char *seg_start, const char *seg_end)
{
const tldnode *allNode = 0;
for (unsigned int i = 0; i < parent->num_children; i++)
{
if (!allNode && subnodes(parent)[i]->attr == ALL)
allNode = subnodes(parent)[i];
else
{
size_t m = seg_end - seg_start;
size_t n = strlen(subnodes(parent)[i]->label);
if (m == n && !memcmp(subnodes(parent)[i]->label, seg_start, n))
return subnodes(parent)[i];
}
}
return allNode;
}
static void
printTldTreeI(const tldnode *node, const char *spacer)
{
if (node->attr)
printf("%s%s: %c\n", spacer, node->label, node->attr);
else
printf("%s%s:\n", spacer, node->label);
if (node->num_children > 0)
{
size_t n = strlen(spacer);
char nspacer[n+2+1];
memcpy(nspacer, spacer, n);
nspacer[n] = ' ';
nspacer[n+1] = ' ';
nspacer[n+2] = '\0';
for (unsigned int i = 0; i < node->num_children; i++)
printTldTreeI(subnodes(node)[i], nspacer);
}
}
/*
* Get next rendered chunk
*/
PyObject *
tdi_render_next(tdi_render_t *self)
{
PyObject *tmp, *endtag, *nodes, *esc;
tdi_node_t *node;
int done, asked;
switch (self->stage) {
case TDI_RE_STAGE_ONLY_CONTENT:
self->stage = TDI_RE_STAGE_DONE;
Py_INCREF(self->model);
return self->model;
case TDI_RE_STAGE_BEGIN:
self->stage = TDI_RE_STAGE_NEXTNODE;
case TDI_RE_STAGE_NEXTNODE:
tdi_re_stage_nextnode:
if (!self->stack) {
self->stage = TDI_RE_STAGE_DONE;
goto exit;
}
if (!(tmp = PyIter_Next(self->stack->iter))) {
if (PyErr_Occurred())
goto exit;
self->done = self->stack->done;
self->endtag = self->stack->endtag;
Py_XINCREF(self->endtag);
tdi_render_stack_pop(&self->stack);
goto tdi_re_stage_endtag;
}
node = (tdi_node_t *)tmp;
if (node->kind == TEXT_NODE) {
tmp = self->emit_escaped ?
node->content->with_escapes : node->content->clean;
Py_INCREF(tmp);
Py_DECREF(node);
return tmp;
}
if (node->flags & NODE_REMOVED) {
Py_DECREF(node);
goto tdi_re_stage_nextnode;
}
if ((asked = ask_model(self, node, &done)) == -1) {
Py_DECREF(node);
goto exit;
}
if (asked == 1) {
Py_DECREF(node);
goto tdi_re_stage_nextnode;
}
if (!(node->flags & (NODE_NOELEMENT | NODE_MASKED))) {
if (!(tmp = ENCODE_STARTTAG(node))) {
Py_DECREF(node);
goto exit;
}
asked = 1;
Py_INCREF(node->endtag);
endtag = node->endtag;
}
else
endtag = NULL;
if (node->content) {
if (self->emit_escaped) {
if (node->content->with_escapes) {
Py_INCREF(node->content->with_escapes);
self->content = node->content->with_escapes;
}
else {
if (!(esc = ENCODE_ESCAPE(node, node->content->clean))) {
Py_DECREF(node);
goto exit;
}
self->content = esc;
}
}
else {
Py_INCREF(node->content->clean);
self->content = node->content->clean;
}
Py_DECREF(node);
self->endtag = endtag;
self->stage = TDI_RE_STAGE_CONTENT;
if (asked != 1)
goto tdi_re_stage_content;
}
else {
nodes = subnodes(node, (tdi_adapter_t *)self->model);
Py_DECREF(node);
if (!nodes)
goto exit;
if (tdi_render_stack_push(&self->stack, nodes, endtag,
self->done) == -1)
goto exit;
if (done)
self->done = 1;
if (asked != 1)
goto tdi_re_stage_nextnode;
}
return tmp;
case TDI_RE_STAGE_CONTENT:
tdi_re_stage_content:
self->stage = self->endtag
? TDI_RE_STAGE_ENDTAG
: TDI_RE_STAGE_NEXTNODE;
tmp = self->content;
self->content = NULL;
return tmp;
case TDI_RE_STAGE_ENDTAG:
tdi_re_stage_endtag:
self->stage = TDI_RE_STAGE_NEXTNODE;
if (!self->endtag)
goto tdi_re_stage_nextnode;
tmp = self->endtag;
self->endtag = NULL;
return tmp;
case TDI_RE_STAGE_DONE:
goto exit;
}
exit:
return NULL;
}
static char *
getRegisteredDomainDropI(const char *hostname, const tldnode *tree,
int drop_unknown)
{
// Eliminate some special (always-fail) cases first.
if (hostname[0] == '.' || hostname[0] == '\0')
return 0;
// The registered domain will always be a suffix of the input hostname.
// Start at the end of the name and work backward.
const char *head = hostname;
const char *seg_end = hostname + strlen(hostname);
const char *seg_start;
if (seg_end[-1] == '.')
seg_end--;
seg_start = seg_end;
for (;;) {
while (seg_start > head && *seg_start != '.')
seg_start--;
if (*seg_start == '.')
seg_start++;
// [seg_start, seg_end) is one label.
const tldnode *subtree = findTldNode(tree, seg_start, seg_end);
if (!subtree
|| (subtree->num_children == 1
&& subnodes(subtree)[0]->attr == THIS))
// Match found.
break;
if (seg_start == head)
// No match, i.e. the input name is too short to be a
// registered domain.
return 0;
// Advance to the next label.
tree = subtree;
if (seg_start[-1] != '.')
abort();
seg_end = seg_start - 1;
seg_start = seg_end - 1;
}
// Ensure the stripped domain contains at least two labels.
if (!strchr(seg_start, '.'))
{
if (seg_start == head || drop_unknown)
return 0;
seg_start -= 2;
while (seg_start > head && *seg_start != '.')
seg_start--;
if (*seg_start == '.')
seg_start++;
}
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wcast-qual"
return (char *)seg_start;
#pragma GCC diagnostic pop
}
