int ws_utf8_set_data(WsUtf8String *string, const unsigned char *data,
size_t len)
{
size_t num_chars;
if (!ws_utf8_verify(data, len, &num_chars))
/* Malformed data. */
return 0;
/* Init `string' to empty. */
ws_free(string->data);
string->data = NULL;
string->len = 0;
string->num_chars = 0;
/* Set the new data. */
string->data = ws_memdup(data, len);
if (string->data == NULL)
return 0;
string->len = len;
string->num_chars = num_chars;
return 1;
}
WsBool ws_bc_add_function(WsBc *bc, WsUInt8 *index_return, char *name,
WsUInt8 num_arguments, WsUInt8 num_locals,
WsUInt32 code_size, unsigned char *code)
{
WsBcFunction *nf;
/* First, add the function to the function pool. */
nf = ws_realloc(bc->functions,
(bc->num_functions + 1) * sizeof(WsBcFunction));
if (nf == NULL)
return WS_FALSE;
bc->functions = nf;
bc->functions[bc->num_functions].num_arguments = num_arguments;
bc->functions[bc->num_functions].num_locals = num_locals;
bc->functions[bc->num_functions].code_size = code_size;
bc->functions[bc->num_functions].code = ws_memdup(code, code_size);
if (bc->functions[bc->num_functions].code == NULL)
return WS_FALSE;
/* Save the index of the function. */
*index_return = bc->num_functions++;
/* For external functions (which have name), add a name entry to the
function name pool. */
if (name) {
WsBcFunctionName *nfn;
nfn = ws_realloc(bc->function_names,
((bc->num_function_names + 1)
* sizeof(WsBcFunctionName)));
if (nfn == NULL)
return WS_FALSE;
bc->function_names = nfn;
bc->function_names[bc->num_function_names].index = *index_return;
bc->function_names[bc->num_function_names].name = ws_strdup(name);
if (bc->function_names[bc->num_function_names].name == NULL)
return WS_FALSE;
bc->num_function_names++;
}
/* All done. */
return WS_TRUE;
}
WsBool ws_bc_add_const_utf8_string(WsBc *bc, WsUInt16 *index_return,
const unsigned char *data, size_t len)
{
WsUInt16 i;
WsBcConstant *nc;
/* Do we already have a suitable UFT-8 constant? */
for (i = 0; i < bc->num_constants; i++) {
if (bc->constants[i].type == WS_BC_CONST_TYPE_UTF8_STRING
&& bc->constants[i].u.v_string.len == len
&& memcmp(bc->constants[i].u.v_string.data,
data, len) == 0) {
*index_return = i;
return WS_TRUE;
}
}
/* Must add a new constant. */
nc = ws_realloc(bc->constants,
(bc->num_constants + 1) * sizeof(WsBcConstant));
if (nc == NULL)
return WS_FALSE;
bc->constants = nc;
bc->constants[bc->num_constants].type = WS_BC_CONST_TYPE_UTF8_STRING;
bc->constants[bc->num_constants].u.v_string.len = len;
bc->constants[bc->num_constants].u.v_string.data
= ws_memdup(data, len);
if (bc->constants[bc->num_constants].u.v_string.data == NULL)
return WS_FALSE;
*index_return = bc->num_constants++;
return WS_TRUE;
}
WsBc *ws_bc_decode(const unsigned char *data, size_t data_len)
{
WsBc *bc = ws_bc_alloc(WS_BC_STRING_ENC_ISO_8859_1);
WsByte b;
WsUInt32 ui32;
WsUInt16 ui16, j;
WsUInt16 ui16b;
WsUInt8 ui8, num_functions, k, l;
WsInt8 i8;
WsInt16 i16;
WsInt32 i32;
WsIeee754Result ieee754;
unsigned char *ucp;
size_t decoded;
/* Decode the byte-code header. */
decoded = ws_decode_buffer(data, data_len,
WS_ENC_BYTE, &b,
WS_ENC_MB_UINT32, &ui32,
WS_ENC_END);
if (!decoded
|| b != WS_BC_VERSION
|| ui32 != data_len - decoded)
/* This is not a valid (or supported) byte-code header. */
goto error;
WS_UPDATE_DATA;
/* Constant pool. */
decoded = ws_decode_buffer(data, data_len,
WS_ENC_MB_UINT16, &ui16,
WS_ENC_MB_UINT16, &ui16b,
WS_ENC_END);
if (!decoded)
goto error;
bc->string_encoding = ui16b;
bc->constants = ws_calloc(ui16, sizeof(WsBcConstant));
if (bc->constants == NULL)
goto error;
bc->num_constants = ui16;
WS_UPDATE_DATA;
for (j = 0; j < bc->num_constants; j++) {
WsBcConstant *c = &bc->constants[j];
decoded = ws_decode_buffer(data, data_len,
WS_ENC_UINT8, &ui8,
WS_ENC_END);
if (decoded != 1)
goto error;
WS_UPDATE_DATA;
switch (ui8) {
case WS_BC_CONST_INT8:
decoded = ws_decode_buffer(data, data_len,
WS_ENC_INT8, &i8,
WS_ENC_END);
if (decoded != 1)
goto error;
WS_UPDATE_DATA;
c->type = WS_BC_CONST_TYPE_INT;
c->u.v_int = i8;
break;
case WS_BC_CONST_INT16:
decoded = ws_decode_buffer(data, data_len,
WS_ENC_INT16, &i16,
WS_ENC_END);
if (decoded != 2)
goto error;
WS_UPDATE_DATA;
c->type = WS_BC_CONST_TYPE_INT;
c->u.v_int = i16;
break;
case WS_BC_CONST_INT32:
decoded = ws_decode_buffer(data, data_len,
WS_ENC_INT32, &i32,
WS_ENC_END);
if (decoded != 4)
goto error;
WS_UPDATE_DATA;
c->type = WS_BC_CONST_TYPE_INT;
c->u.v_int = i32;
break;
case WS_BC_CONST_FLOAT32:
decoded = ws_decode_buffer(data, data_len,
WS_ENC_DATA, &ucp, (size_t) 4,
WS_ENC_END);
if (decoded != 4)
goto error;
WS_UPDATE_DATA;
ieee754 = ws_ieee754_decode_single(ucp, &c->u.v_float);
switch (ieee754) {
case WS_IEEE754_OK:
c->type = WS_BC_CONST_TYPE_FLOAT32;
break;
case WS_IEEE754_NAN:
c->type = WS_BC_CONST_TYPE_FLOAT32_NAN;
break;
case WS_IEEE754_POSITIVE_INF:
c->type = WS_BC_CONST_TYPE_FLOAT32_POSITIVE_INF;
break;
case WS_IEEE754_NEGATIVE_INF:
c->type = WS_BC_CONST_TYPE_FLOAT32_NEGATIVE_INF;
break;
}
break;
case WS_BC_CONST_UTF8_STRING:
decoded = ws_decode_buffer(data, data_len,
WS_ENC_MB_UINT32, &ui32,
WS_ENC_END);
if (decoded == 0)
goto error;
WS_UPDATE_DATA;
c->type = WS_BC_CONST_TYPE_UTF8_STRING;
c->u.v_string.len = ui32;
decoded = ws_decode_buffer(data, data_len,
WS_ENC_DATA, &ucp, c->u.v_string.len,
WS_ENC_END);
if (decoded != ui32)
goto error;
WS_UPDATE_DATA;
c->u.v_string.data = ws_memdup(ucp, ui32);
if (c->u.v_string.data == NULL)
goto error;
/* Check the validity of the data. */
if (!ws_utf8_verify(c->u.v_string.data, c->u.v_string.len,
&c->u.v_string.num_chars))
goto error;
break;
case WS_BC_CONST_EMPTY_STRING:
c->type = WS_BC_CONST_TYPE_EMPTY_STRING;
break;
case WS_BC_CONST_EXT_ENC_STRING:
ws_fatal("external character encoding not implemented yet");
break;
default:
/* Reserved. */
goto error;
break;
}
}
/* Pragma pool. */
decoded = ws_decode_buffer(data, data_len,
WS_ENC_MB_UINT16, &ui16,
WS_ENC_END);
if (!decoded)
goto error;
bc->pragmas = ws_calloc(ui16, sizeof(WsBcPragma));
if (bc->pragmas == NULL)
goto error;
bc->num_pragmas = ui16;
WS_UPDATE_DATA;
for (j = 0; j < bc->num_pragmas; j++) {
WsBcPragma *p = &bc->pragmas[j];
decoded = ws_decode_buffer(data, data_len,
WS_ENC_UINT8, &ui8,
WS_ENC_END);
if (decoded != 1)
goto error;
WS_UPDATE_DATA;
p->type = ui8;
switch (ui8) {
case WS_BC_PRAGMA_ACCESS_DOMAIN:
decoded = ws_decode_buffer(data, data_len,
WS_ENC_MB_UINT16, &p->index_1,
WS_ENC_END);
if (!decoded)
goto error;
WS_CHECK_STRING(p->index_1);
break;
case WS_BC_PRAGMA_ACCESS_PATH:
decoded = ws_decode_buffer(data, data_len,
WS_ENC_MB_UINT16, &p->index_1,
WS_ENC_END);
if (!decoded)
goto error;
WS_CHECK_STRING(p->index_1);
break;
case WS_BC_PRAGMA_USER_AGENT_PROPERTY:
decoded = ws_decode_buffer(data, data_len,
WS_ENC_MB_UINT16, &p->index_1,
WS_ENC_MB_UINT16, &p->index_2,
WS_ENC_END);
if (!decoded)
goto error;
WS_CHECK_STRING(p->index_1);
WS_CHECK_STRING(p->index_2);
break;
case WS_BC_PRAGMA_USER_AGENT_PROPERTY_AND_SCHEME:
decoded = ws_decode_buffer(data, data_len,
WS_ENC_MB_UINT16, &p->index_1,
WS_ENC_MB_UINT16, &p->index_2,
WS_ENC_MB_UINT16, &p->index_3,
WS_ENC_END);
if (!decoded)
goto error;
WS_CHECK_STRING(p->index_1);
WS_CHECK_STRING(p->index_2);
WS_CHECK_STRING(p->index_3);
break;
default:
goto error;
break;
}
WS_UPDATE_DATA;
}
/* Function pool. */
decoded = ws_decode_buffer(data, data_len,
WS_ENC_UINT8, &num_functions,
WS_ENC_END);
if (decoded != 1)
goto error;
WS_UPDATE_DATA;
/* Function names. */
decoded = ws_decode_buffer(data, data_len,
WS_ENC_UINT8, &ui8,
WS_ENC_END);
if (decoded != 1)
goto error;
WS_UPDATE_DATA;
if (ui8) {
/* We have function names. */
bc->function_names = ws_calloc(ui8, sizeof(WsBcFunctionName));
if (bc->function_names == NULL)
goto error;
bc->num_function_names = ui8;
for (k = 0; k < bc->num_function_names; k++) {
WsBcFunctionName *n = &bc->function_names[k];
decoded = ws_decode_buffer(data, data_len,
WS_ENC_UINT8, &n->index,
WS_ENC_UINT8, &ui8,
WS_ENC_END);
if (decoded != 2)
goto error;
WS_UPDATE_DATA;
decoded = ws_decode_buffer(data, data_len,
WS_ENC_DATA, &ucp, (size_t) ui8,
WS_ENC_END);
if (decoded != ui8)
goto error;
WS_UPDATE_DATA;
n->name = ws_memdup(ucp, ui8);
if (n->name == NULL)
goto error;
/* Check the validity of the name. */
if (!ws_utf8_verify((unsigned char *) n->name, ui8, NULL))
goto error;
/* Just check that the data contains only valid characters. */
for (l = 0; l < ui8; l++) {
unsigned int ch = (unsigned char) n->name[l];
if (('a' <= ch && ch <= 'z')
|| ('A' <= ch && ch <= 'Z')
|| ch == '_'
|| (l > 0 && ('0' <= ch && ch <= '9')))
/* Ok. */
continue;
/* Invalid character in the function name. */
goto error;
}
/* Is the index valid? */
if (n->index >= num_functions)
goto error;
}
}
/* Functions. */
if (num_functions) {
/* We have functions. */
bc->functions = ws_calloc(num_functions, sizeof(WsBcFunction));
if (bc->functions == NULL)
goto error;
bc->num_functions = num_functions;
for (k = 0; k < bc->num_functions; k++) {
WsBcFunction *f = &bc->functions[k];
decoded = ws_decode_buffer(data, data_len,
WS_ENC_UINT8, &f->num_arguments,
WS_ENC_UINT8, &f->num_locals,
WS_ENC_MB_UINT32, &f->code_size,
WS_ENC_END);
if (!decoded)
goto error;
WS_UPDATE_DATA;
decoded = ws_decode_buffer(data, data_len,
WS_ENC_DATA, &ucp, f->code_size,
WS_ENC_END);
if (decoded != f->code_size)
goto error;
WS_UPDATE_DATA;
if (f->code_size) {
/* It is not an empty function. */
f->code = ws_memdup(ucp, f->code_size);
if (f->code == NULL)
goto error;
}
}
}
/* Did we process it all? */
if (data_len != 0)
goto error;
/* All done. */
return bc;
/*
* Error handling.
*/
error:
ws_bc_free(bc);
return NULL;
}
