mirror of
https://forge.fsky.io/lda/Parsee.git
synced 2026-03-13 15:15:10 +00:00
302 lines
6.7 KiB
C
302 lines
6.7 KiB
C
#include <XEP393.h>
|
|
|
|
#include <Cytoplasm/Memory.h>
|
|
#include <Cytoplasm/Array.h>
|
|
#include <Cytoplasm/Log.h>
|
|
|
|
#include <string.h>
|
|
#include <ctype.h>
|
|
|
|
static XEP393Element *
|
|
CreateElementVessel(XEP393Element *parent, XEP393Type type)
|
|
{
|
|
XEP393Element *ret = Malloc(sizeof(*ret));
|
|
ret->parent = parent;
|
|
ret->type = type;
|
|
ret->children = ArrayCreate();
|
|
ret->text_data = NULL;
|
|
|
|
if (parent)
|
|
{
|
|
ArrayAdd(parent->children, ret);
|
|
}
|
|
return ret;
|
|
}
|
|
static void
|
|
XEP393FreeElementBase(XEP393Element *element, bool unlink)
|
|
{
|
|
size_t i, len;
|
|
if (!element)
|
|
{
|
|
return;
|
|
}
|
|
|
|
len = ArraySize(element->children);
|
|
for (i = 0; i < len; i++)
|
|
{
|
|
XEP393FreeElement(ArrayGet(element->children, i));
|
|
}
|
|
|
|
if (element->parent && unlink)
|
|
{
|
|
XEP393Element *parent = element->parent;
|
|
len = ArraySize(parent->children);
|
|
for (i = 0; i < len; i++)
|
|
{
|
|
XEP393Element *c = ArrayGet(parent->children, i);
|
|
if (c == element)
|
|
{
|
|
ArrayDelete(parent->children, i);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
ArrayFree(element->children);
|
|
Free(element->text_data);
|
|
Free(element);
|
|
}
|
|
void
|
|
XEP393FreeElement(XEP393Element *element)
|
|
{
|
|
XEP393FreeElementBase(element, false);
|
|
}
|
|
|
|
typedef struct StrView {
|
|
char *start;
|
|
char *end;
|
|
|
|
bool heap_free;
|
|
} StrView;
|
|
#define ViewLength(v) ((size_t) ((v.end) - (v.start)))
|
|
static char *
|
|
StringifyView(StrView v)
|
|
{
|
|
char *r;
|
|
size_t len;
|
|
if (!v.start || v.start > v.end)
|
|
{
|
|
return NULL;
|
|
}
|
|
|
|
len = ViewLength(v);
|
|
r = Malloc(len + 1);
|
|
memcpy(r, v.start, len);
|
|
r[len] = '\0';
|
|
|
|
return r;
|
|
}
|
|
static StrView
|
|
CreateStaticView(char *str)
|
|
{
|
|
StrView view = {
|
|
.start = str,
|
|
.end = str + strlen(str),
|
|
.heap_free = false
|
|
};
|
|
|
|
return view;
|
|
}
|
|
static bool
|
|
IdentifySpan(char span_tag, StrView in, StrView *view)
|
|
{
|
|
size_t length;
|
|
bool found = false;
|
|
char prev = '\0';
|
|
if (in.start >= in.end)
|
|
{
|
|
return false;
|
|
}
|
|
if (ViewLength(in) < 2)
|
|
{
|
|
return false;
|
|
}
|
|
|
|
if (*in.start != span_tag || isspace(*(in.start + 1)))
|
|
{
|
|
/* The opening styling directive MUST NOT be followed
|
|
* by a whitespace character */
|
|
return false;
|
|
}
|
|
view->start = in.start + 1;
|
|
in.start += 1;
|
|
|
|
for (length = 0; ViewLength(in) > 0; length++, in.start++)
|
|
{
|
|
if (*in.start == span_tag)
|
|
{
|
|
found = true;
|
|
break;
|
|
}
|
|
|
|
prev = *in.start;
|
|
}
|
|
if (!found || !length || (prev && isspace(prev)))
|
|
{
|
|
/* the closing styling directive MUST NOT be preceeded
|
|
* by a whitespace character. */
|
|
return false;
|
|
}
|
|
|
|
view->end = in.start;
|
|
return true;
|
|
}
|
|
|
|
|
|
static void
|
|
XEP393Decode(StrView view, XEP393Element *root)
|
|
{
|
|
StrView subview = view;
|
|
StrView textview = view;
|
|
XEP393Element *text, *span;
|
|
bool managed = false;
|
|
char prev = '\0', curr = '\0';
|
|
|
|
textview.end = subview.start;
|
|
for (; subview.start < subview.end; subview.start++)
|
|
{
|
|
StrView span_view;
|
|
managed = false;
|
|
curr = *subview.start;
|
|
if (prev == '\0' || prev == '\n')
|
|
{
|
|
/* TODO: Start of line, start parsing blocks. */
|
|
}
|
|
#define Spanify(xep_symbol) \
|
|
managed = true; \
|
|
textview.end = subview.start; \
|
|
text = CreateElementVessel( \
|
|
root, XEP393_TEXT \
|
|
); \
|
|
text->text_data = StringifyView(textview); \
|
|
\
|
|
/* Found a span. */ \
|
|
span = CreateElementVessel( \
|
|
root, xep_symbol \
|
|
); \
|
|
\
|
|
XEP393Decode(span_view, span); \
|
|
\
|
|
/* Update subview */ \
|
|
subview.start = span_view.end + 1; \
|
|
\
|
|
/* Update textview */ \
|
|
textview.start = subview.start; \
|
|
textview.end = subview.start
|
|
if (IdentifySpan('_', subview, &span_view))
|
|
{
|
|
Spanify(XEP393_ITALIC);
|
|
}
|
|
else if (IdentifySpan('*', subview, &span_view))
|
|
{
|
|
Spanify(XEP393_EMPH);
|
|
}
|
|
else if (IdentifySpan('`', subview, &span_view))
|
|
{
|
|
Spanify(XEP393_MONO);
|
|
}
|
|
else
|
|
{
|
|
/* Text character: update end */
|
|
textview.end = subview.start;
|
|
}
|
|
|
|
prev = curr;
|
|
}
|
|
|
|
if (!managed)
|
|
{
|
|
textview.end = subview.start;
|
|
text = CreateElementVessel(
|
|
root, XEP393_TEXT
|
|
);
|
|
text->text_data = StringifyView(textview);
|
|
}
|
|
}
|
|
|
|
XEP393Element *
|
|
XEP393(char *message)
|
|
{
|
|
StrView view = CreateStaticView(message);
|
|
XEP393Element *root = CreateElementVessel(NULL, XEP393_ROOT);
|
|
|
|
/* TODO: Parse blocks first, *then* spans. Considering the
|
|
* current architecture, this shouldn't be too hard to integrate,
|
|
* given how string views already manage boundaries, and elements
|
|
* can already be used to contain blocks I think.
|
|
*
|
|
* Actually, nevermind, these would be pure pain. Nested blocks,
|
|
* unterminated ones, QUOTES. Just hell. I hate parsing this shit. */
|
|
XEP393Decode(view, root);
|
|
return root;
|
|
}
|
|
|
|
#include <XML.h>
|
|
|
|
static void
|
|
ShoveXML(XEP393Element *element, XMLElement *xmlparent)
|
|
{
|
|
XMLElement *head = xmlparent;
|
|
size_t i;
|
|
if (!element || !xmlparent)
|
|
{
|
|
return;
|
|
}
|
|
|
|
switch (element->type)
|
|
{
|
|
case XEP393_ITALIC:
|
|
head = XMLCreateTag("i");
|
|
XMLAddChild(xmlparent, head);
|
|
break;
|
|
case XEP393_EMPH:
|
|
head = XMLCreateTag("strong");
|
|
XMLAddChild(xmlparent, head);
|
|
break;
|
|
case XEP393_MONO:
|
|
head = XMLCreateTag("code");
|
|
XMLAddChild(xmlparent, head);
|
|
break;
|
|
default: break;
|
|
}
|
|
if (ArraySize(element->children) == 0 && element->text_data)
|
|
{
|
|
XMLElement *text = XMLCreateText(element->text_data);
|
|
XMLAddChild(head, text);
|
|
}
|
|
|
|
for (i = 0; i < ArraySize(element->children); i++)
|
|
{
|
|
XEP393Element *sub = ArrayGet(element->children, i);
|
|
ShoveXML(sub, head);
|
|
}
|
|
}
|
|
|
|
#include <StringStream.h>
|
|
char *
|
|
XEP393ToXMLString(XEP393Element *xepd)
|
|
{
|
|
XMLElement *root;
|
|
|
|
Stream *writer;
|
|
char *ret = NULL;
|
|
size_t i;
|
|
if (!xepd)
|
|
{
|
|
return NULL;
|
|
}
|
|
|
|
root = XMLCreateTag("ROOT");
|
|
ShoveXML(xepd, root);
|
|
|
|
writer = StrStreamWriter(&ret);
|
|
for (i = 0; i < ArraySize(root->children); i++)
|
|
{
|
|
XMLEncode(writer, ArrayGet(root->children, i));
|
|
}
|
|
XMLFreeElement(root);
|
|
StreamFlush(writer);
|
|
StreamClose(writer);
|
|
|
|
return ret;
|
|
}
|