Hopiu/linkchecker
1
0
Fork 0
mirror of https://github.com/Hopiu/linkchecker.git synced 2026-04-12 18:40:57 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

only call set_encoding for meta tags, and syntax cleanup

Browse source 
git-svn-id: https://linkchecker.svn.sourceforge.net/svnroot/linkchecker/trunk/linkchecker@2153 e7d03fd6-7b0d-0410-9947-9c21f3af8025
This commit is contained in:
calvin 2005-01-18 15:54:11 +00:00
parent 83020982ba
commit cec0a01f4b
1 changed files with 72 additions and 61 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

133
linkcheck/HtmlParser/htmlparse.y
View file

@ -50,39 +50,40 @@ static PyObject* list_dict;
static PyObject* set_encoding;
/* set_doctype helper function */
static PyObject* set_doctype;
/* the unicode string u'meta' */
static PyObject* u_meta;
/* macros for easier scanner state manipulation */
/* clear buffer b, returning NULL on error */
#define CLEAR_BUF(b) \
b = PyMem_Resize(b, char, 1); \
if (b==NULL) return NULL; \
if (b == NULL) return NULL; \
(b)[0] = '\0'
/* clear buffer b, returning NULL and decref self on error */
#define CLEAR_BUF_DECREF(self, b) \
b = PyMem_Resize(b, char, 1); \
if (b==NULL) { Py_DECREF(self); return NULL; } \
if (b == NULL) { Py_DECREF(self); return NULL; } \
(b)[0] = '\0'
#define CHECK_ERROR(ud, label) \
if (ud->error && PyObject_HasAttrString(ud->handler, "error")==1) { \
if (ud->error && PyObject_HasAttrString(ud->handler, "error") == 1) { \
callback = PyObject_GetAttrString(ud->handler, "error"); \
if (!callback) { error=1; goto label; } \
if (!callback) { error = 1; goto label; } \
result = PyObject_CallFunction(callback, "O", ud->error); \
if (!result) { error=1; goto label; } \
if (!result) { error = 1; goto label; } \
}
/* generic callback macro */
#define CALLBACK(ud, attr, format, arg, label) \
if (PyObject_HasAttrString(ud->handler, attr)==1) { \
if (PyObject_HasAttrString(ud->handler, attr) == 1) { \
callback = PyObject_GetAttrString(ud->handler, attr); \
if (callback==NULL) { error=1; goto label; } \
if (callback == NULL) { error = 1; goto label; } \
result = PyObject_CallFunction(callback, format, arg); \
if (result==NULL) { error=1; goto label; } \
Py_DECREF(callback); \
Py_DECREF(result); \
callback=result=NULL; \
if (result == NULL) { error = 1; goto label; } \
Py_CLEAR(callback); \
Py_CLEAR(result); \
}
/* set old line and column */
@ -115,10 +116,10 @@ static int html_end_tag (PyObject* ptag, PyObject* parser) {
char* doctype;
int ret = 1;
pdoctype = PyObject_GetAttrString(parser, "doctype");
if (pdoctype==NULL) return -1;
if (pdoctype == NULL) return -1;
doctype = PyString_AsString(pdoctype);
if (doctype == NULL) { Py_DECREF(pdoctype); return -1; }
if (strcmp(doctype, "HTML")==0) {
if (strcmp(doctype, "HTML") == 0) {
char* tag = PyString_AsString(ptag);
if (tag == NULL) { Py_DECREF(pdoctype); return -1; }
ret = strcmp(tag, "area")!=0 &&
@ -188,19 +189,23 @@ element: T_WAIT { YYACCEPT; /* wait for more lexer input */ }
PyObject* tag = PyTuple_GET_ITEM($1, 0);
PyObject* attrs = PyTuple_GET_ITEM($1, 1);
int error = 0;
if (tag==NULL || attrs==NULL) { error = 1; goto finish_start; }
/* set encoding */
result = PyObject_CallFunction(set_encoding, "OOO", ud->parser, tag, attrs);
if (result==NULL) { error=1; goto finish_start; }
Py_DECREF(result); result = NULL;
if (PyObject_HasAttrString(ud->handler, "start_element")==1) {
int cmp;
if (tag == NULL || attrs == NULL) { error = 1; goto finish_start; }
cmp = PyObject_RichCompareBool(tag, u_meta, Py_EQ);
if (cmp == -1) { error = 1; goto finish_start; }
if (cmp == 1) {
/* set encoding */
result = PyObject_CallFunction(set_encoding, "OO", ud->parser, attrs);
if (result == NULL) { error = 1; goto finish_start; }
Py_CLEAR(result);
}
if (PyObject_HasAttrString(ud->handler, "start_element") == 1) {
callback = PyObject_GetAttrString(ud->handler, "start_element");
if (!callback) { error=1; goto finish_start; }
if (!callback) { error = 1; goto finish_start; }
result = PyObject_CallFunction(callback, "OO", tag, attrs);
if (!result) { error=1; goto finish_start; }
Py_DECREF(callback);
Py_DECREF(result);
callback = result = NULL;
if (!result) { error = 1; goto finish_start; }
Py_CLEAR(callback);
Py_CLEAR(result);
}
CHECK_ERROR(ud, finish_start);
finish_start:
@ -227,27 +232,31 @@ finish_start:
PyObject* tag = PyTuple_GET_ITEM($1, 0);
PyObject* attrs = PyTuple_GET_ITEM($1, 1);
int error = 0;
int cmp;
char* fname;
PyObject* tagname;
if (tag==NULL || attrs==NULL) { error = 1; goto finish_start_end; }
if (tag == NULL || attrs == NULL) { error = 1; goto finish_start_end; }
tagname = PyUnicode_AsEncodedString(tag, "ascii", "ignore");
if (tagname==NULL) { error=1; goto finish_start_end; }
/* set encoding */
result = PyObject_CallFunction(set_encoding, "OOO", ud->parser, tag, attrs);
if (result==NULL) { error=1; goto finish_start_end; }
Py_DECREF(result); result = NULL;
if (tagname == NULL) { error = 1; goto finish_start_end; }
cmp = PyObject_RichCompareBool(tag, u_meta, Py_EQ);
if (cmp == -1) { error = 1; goto finish_start; }
if (cmp == 1) {
/* set encoding */
result = PyObject_CallFunction(set_encoding, "OO", ud->parser, attrs);
if (result == NULL) { error = 1; goto finish_start_end; }
Py_CLEAR(result);
}
if (html_end_tag(tagname, ud->parser))
fname = "start_end_element";
else
fname = "start_element";
if (PyObject_HasAttrString(ud->handler, fname)==1) {
if (PyObject_HasAttrString(ud->handler, fname) == 1) {
callback = PyObject_GetAttrString(ud->handler, fname);
if (!callback) { error=1; goto finish_start_end; }
if (!callback) { error = 1; goto finish_start_end; }
result = PyObject_CallFunction(callback, "OO", tag, attrs);
if (!result) { error=1; goto finish_start_end; }
Py_DECREF(callback);
Py_DECREF(result);
callback = result = NULL;
if (!result) { error = 1; goto finish_start_end; }
Py_CLEAR(callback);
Py_CLEAR(result);
}
CHECK_ERROR(ud, finish_start_end);
finish_start_end:
@ -273,16 +282,15 @@ finish_start_end:
int error = 0;
/* encode tagname in ASCII, ignoring any unknown chars */
PyObject* tagname = PyUnicode_AsEncodedString($1, "ascii", "ignore");
if (tagname==NULL) { error=1; goto finish_end; }
if (PyObject_HasAttrString(ud->handler, "end_element")==1 &&
if (tagname == NULL) { error = 1; goto finish_end; }
if (PyObject_HasAttrString(ud->handler, "end_element") == 1 &&
html_end_tag(tagname, ud->parser)) {
callback = PyObject_GetAttrString(ud->handler, "end_element");
if (callback==NULL) { error=1; goto finish_end; }
if (callback == NULL) { error = 1; goto finish_end; }
result = PyObject_CallFunction(callback, "O", $1);
if (result==NULL) { error=1; goto finish_end; }
Py_DECREF(callback);
Py_DECREF(result);
callback = result = NULL;
if (result == NULL) { error = 1; goto finish_end; }
Py_CLEAR(callback);
Py_CLEAR(result);
}
CHECK_ERROR(ud, finish_end);
finish_end:
@ -370,8 +378,8 @@ finish_cdata:
int error = 0;
/* set encoding */
result = PyObject_CallFunction(set_doctype, "OO", ud->parser, $1);
if (result==NULL) { error=1; goto finish_doctype; }
Py_DECREF(result); result = NULL;
if (result == NULL) { error = 1; goto finish_doctype; }
Py_CLEAR(result);
CALLBACK(ud, "doctype", "O", $1, finish_doctype);
CHECK_ERROR(ud, finish_doctype);
finish_doctype:
@ -394,7 +402,7 @@ finish_doctype:
PyObject* result = NULL;
int error = 0;
PyObject* script = PyUnicode_DecodeASCII("script", 6, "ignore");
if (script==NULL) { error=1; goto finish_script; }
if (script == NULL) { error = 1; goto finish_script; }
CALLBACK(ud, "characters", "O", $1, finish_script);
CALLBACK(ud, "end_element", "O", script, finish_script);
CHECK_ERROR(ud, finish_script);
@ -419,7 +427,7 @@ finish_script:
PyObject* result = NULL;
int error = 0;
PyObject* style = PyUnicode_DecodeASCII("style", 5, "ignore");
if (style==NULL) { error=1; goto finish_style; }
if (style == NULL) { error = 1; goto finish_style; }
CALLBACK(ud, "characters", "O", $1, finish_style);
CALLBACK(ud, "end_element", "O", style, finish_style);
CHECK_ERROR(ud, finish_style);
@ -529,7 +537,7 @@ static int parser_init (parser_object* self, PyObject* args, PyObject* kwds) {
if (!PyArg_ParseTupleAndKeywords(args, kwds, "|O", kwlist, &handler)) {
return -1;
}
if (handler==NULL) {
if (handler == NULL) {
return 0;
}
Py_DECREF(self->handler);
@ -633,18 +641,18 @@ static PyObject* parser_flush (parser_object* self, PyObject* args) {
PyObject* result = NULL;
/* reset buffer */
CLEAR_BUF(self->userData->buf);
if (s==NULL) { error=1; goto finish_flush; }
if (PyObject_HasAttrString(self->handler, "characters")==1) {
if (s == NULL) { error = 1; goto finish_flush; }
if (PyObject_HasAttrString(self->handler, "characters") == 1) {
callback = PyObject_GetAttrString(self->handler, "characters");
if (callback==NULL) { error=1; goto finish_flush; }
if (callback == NULL) { error = 1; goto finish_flush; }
result = PyObject_CallFunction(callback, "O", s);
if (result==NULL) { error=1; goto finish_flush; }
if (result == NULL) { error = 1; goto finish_flush; }
}
finish_flush:
Py_XDECREF(callback);
Py_XDECREF(result);
Py_XDECREF(s);
if (error==1) {
if (error == 1) {
return NULL;
}
}
@ -933,37 +941,40 @@ PyMODINIT_FUNC inithtmlsax (void) {
if (PyType_Ready(&parser_type) < 0) {
return;
}
if ((m = Py_InitModule3("htmlsax", htmlsax_methods, "SAX HTML parser routines"))==NULL) {
if ((m = Py_InitModule3("htmlsax", htmlsax_methods, "SAX HTML parser routines")) == NULL) {
return;
}
Py_INCREF(&parser_type);
if (PyModule_AddObject(m, "parser", (PyObject *)&parser_type)==-1) {
if (PyModule_AddObject(m, "parser", (PyObject *)&parser_type) == -1) {
/* init error */
PyErr_Print();
}
if ((m = PyImport_ImportModule("linkcheck.HtmlParser"))==NULL) {
if ((m = PyImport_ImportModule("linkcheck.HtmlParser")) == NULL) {
return;
}
if ((resolve_entities = PyObject_GetAttrString(m, "resolve_entities"))==NULL) {
if ((resolve_entities = PyObject_GetAttrString(m, "resolve_entities")) == NULL) {
Py_DECREF(m);
return;
}
if ((set_encoding = PyObject_GetAttrString(m, "set_encoding"))==NULL) {
if ((set_encoding = PyObject_GetAttrString(m, "set_encoding")) == NULL) {
Py_DECREF(resolve_entities);
Py_DECREF(m);
return;
}
if ((set_doctype = PyObject_GetAttrString(m, "set_doctype"))==NULL) {
if ((set_doctype = PyObject_GetAttrString(m, "set_doctype")) == NULL) {
Py_DECREF(resolve_entities);
Py_DECREF(set_encoding);
Py_DECREF(m);
return;
}
Py_DECREF(m);
if ((m = PyImport_ImportModule("linkcheck.containers"))==NULL) {
if ((u_meta = PyString_Decode("meta", 4, "ascii", "ignore")) == NULL) {
return;
}
if ((list_dict = PyObject_GetAttrString(m, "ListDict"))==NULL) {
if ((m = PyImport_ImportModule("linkcheck.containers")) == NULL) {
return;
}
if ((list_dict = PyObject_GetAttrString(m, "ListDict")) == NULL) {
Py_DECREF(m);
return;
}