void
ErrMsg_set_with_errno(const char *fmt, ...) {
int cur_errno = errno;
CharBuf *buf = CB_new(0);
va_list args;
va_start(args, fmt);
CB_VCatF(buf, fmt, args);
va_end(args);
CB_Cat_Trusted_Utf8(buf, ": ", 2);
const char *msg = ErrMsg_strerror(cur_errno);
if (msg != NULL) {
CB_Cat_Trusted_Utf8(buf, msg, strlen(msg));
}
else {
CB_catf(buf, "Unknown error: %i32", (int32_t)cur_errno);
}
Err_set_error(Err_new(CB_Yield_String(buf)));
DECREF(buf);
}
static String*
S_encode_entities(String *text, CharBuf *buf) {
StringIterator *iter = Str_Top(text);
size_t space = 0;
const int MAX_ENTITY_BYTES = 9; // &#dddddd;
// Scan first so that we only allocate once.
int32_t code_point;
while (STRITER_DONE != (code_point = StrIter_Next(iter))) {
if (code_point > 127
|| (!isgraph(code_point) && !isspace(code_point))
|| code_point == '<'
|| code_point == '>'
|| code_point == '&'
|| code_point == '"'
) {
space += MAX_ENTITY_BYTES;
}
else {
space += 1;
}
}
CB_Grow(buf, space);
CB_Set_Size(buf, 0);
DECREF(iter);
iter = Str_Top(text);
while (STRITER_DONE != (code_point = StrIter_Next(iter))) {
if (code_point > 127
|| (!isgraph(code_point) && !isspace(code_point))
) {
CB_catf(buf, "&#%u32;", code_point);
}
else if (code_point == '<') {
CB_Cat_Trusted_Utf8(buf, "<", 4);
}
else if (code_point == '>') {
CB_Cat_Trusted_Utf8(buf, ">", 4);
}
else if (code_point == '&') {
CB_Cat_Trusted_Utf8(buf, "&", 5);
}
else if (code_point == '"') {
CB_Cat_Trusted_Utf8(buf, """, 6);
}
else {
CB_Cat_Char(buf, code_point);
}
}
DECREF(iter);
return CB_To_String(buf);
}
static void
test_Cat(TestBatchRunner *runner) {
String *wanted = Str_newf("a%s", smiley);
CharBuf *got = S_get_cb("");
CB_Cat(got, wanted);
TEST_TRUE(runner, S_cb_equals(got, wanted), "Cat");
DECREF(got);
got = S_get_cb("a");
CB_Cat_Char(got, 0x263A);
TEST_TRUE(runner, S_cb_equals(got, wanted), "Cat_Char");
DECREF(got);
got = S_get_cb("a");
CB_Cat_Utf8(got, smiley, smiley_len);
TEST_TRUE(runner, S_cb_equals(got, wanted), "Cat_Utf8");
DECREF(got);
got = S_get_cb("a");
CB_Cat_Trusted_Utf8(got, smiley, smiley_len);
TEST_TRUE(runner, S_cb_equals(got, wanted), "Cat_Trusted_Utf8");
DECREF(got);
DECREF(wanted);
}
String*
PhraseQuery_To_String_IMP(PhraseQuery *self) {
PhraseQueryIVARS *const ivars = PhraseQuery_IVARS(self);
uint32_t num_terms = Vec_Get_Size(ivars->terms);
CharBuf *buf = CB_new_from_str(ivars->field);
CB_Cat_Trusted_Utf8(buf, ":\"", 2);
for (uint32_t i = 0; i < num_terms; i++) {
Obj *term = Vec_Fetch(ivars->terms, i);
String *term_string = Obj_To_String(term);
CB_Cat(buf, term_string);
DECREF(term_string);
if (i < num_terms - 1) {
CB_Cat_Trusted_Utf8(buf, " ", 1);
}
}
CB_Cat_Trusted_Utf8(buf, "\"", 1);
String *retval = CB_Yield_String(buf);
DECREF(buf);
return retval;
}
String*
ORQuery_To_String_IMP(ORQuery *self) {
ORQueryIVARS *const ivars = ORQuery_IVARS(self);
uint32_t num_kids = Vec_Get_Size(ivars->children);
if (!num_kids) { return Str_new_from_trusted_utf8("()", 2); }
else {
CharBuf *buf = CB_new_from_trusted_utf8("(", 1);
uint32_t last_kid = num_kids - 1;
for (uint32_t i = 0; i < num_kids; i++) {
String *kid_string = Obj_To_String(Vec_Fetch(ivars->children, i));
CB_Cat(buf, kid_string);
DECREF(kid_string);
if (i == last_kid) {
CB_Cat_Trusted_Utf8(buf, ")", 1);
}
else {
CB_Cat_Trusted_Utf8(buf, " OR ", 4);
}
}
String *retval = CB_Yield_String(buf);
DECREF(buf);
return retval;
}
}
void
ErrMsg_set_with_win_error(const char *fmt, ...) {
char *win_error = Err_win_error();
CharBuf *buf = CB_new(0);
va_list args;
va_start(args, fmt);
CB_VCatF(buf, fmt, args);
va_end(args);
CB_Cat_Trusted_Utf8(buf, ": ", 2);
CB_Cat_Utf8(buf, win_error, strlen(win_error));
Err_set_error(Err_new(CB_Yield_String(buf)));
DECREF(buf);
FREEMEM(win_error);
}
static Hash*
S_extract_tv_cache(ByteBuf *field_buf) {
Hash *tv_cache = Hash_new(0);
const char *tv_string = BB_Get_Buf(field_buf);
int32_t num_terms = NumUtil_decode_c32(&tv_string);
CharBuf *text_buf = CB_new(0);
// Read the number of highlightable terms in the field.
for (int32_t i = 0; i < num_terms; i++) {
size_t overlap = NumUtil_decode_c32(&tv_string);
size_t len = NumUtil_decode_c32(&tv_string);
// Decompress the term text.
CB_Set_Size(text_buf, overlap);
CB_Cat_Trusted_Utf8(text_buf, tv_string, len);
tv_string += len;
// Get positions & offsets string.
const char *bookmark_ptr = tv_string;
int32_t num_positions = NumUtil_decode_c32(&tv_string);
while (num_positions--) {
// Leave nums compressed to save a little mem.
NumUtil_skip_cint(&tv_string);
NumUtil_skip_cint(&tv_string);
NumUtil_skip_cint(&tv_string);
}
len = tv_string - bookmark_ptr;
// Store the $text => $posdata pair in the output hash.
String *text = CB_To_String(text_buf);
Hash_Store(tv_cache, (Obj*)text,
(Obj*)BB_new_bytes(bookmark_ptr, len));
DECREF(text);
}
DECREF(text_buf);
return tv_cache;
}
