String*
Highlighter_Highlight_IMP(Highlighter *self, String *text) {
HighlighterIVARS *const ivars = Highlighter_IVARS(self);
size_t size = Str_Get_Size(text)
+ Str_Get_Size(ivars->pre_tag)
+ Str_Get_Size(ivars->post_tag);
CharBuf *buf = CB_new(size);
CB_Cat(buf, ivars->pre_tag);
CB_Cat(buf, text);
CB_Cat(buf, ivars->post_tag);
String *retval = CB_Yield_String(buf);
DECREF(buf);
return retval;
}
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);
}
static void
test_Cat(TestBatch *batch) {
CharBuf *wanted = CB_newf("a%s", smiley);
CharBuf *got = S_get_cb("");
CB_Cat(got, wanted);
TEST_TRUE(batch, CB_Equals(wanted, (Obj*)got), "Cat");
DECREF(got);
got = S_get_cb("a");
CB_Cat_Char(got, 0x263A);
TEST_TRUE(batch, CB_Equals(wanted, (Obj*)got), "Cat_Char");
DECREF(got);
got = S_get_cb("a");
CB_Cat_Str(got, smiley, smiley_len);
TEST_TRUE(batch, CB_Equals(wanted, (Obj*)got), "Cat_Str");
DECREF(got);
got = S_get_cb("a");
CB_Cat_Trusted_Str(got, smiley, smiley_len);
TEST_TRUE(batch, CB_Equals(wanted, (Obj*)got), "Cat_Trusted_Str");
DECREF(got);
DECREF(wanted);
}
static void
S_cat_seg_name(CharBuf *seg_name, u32_t seg_num)
{
CharBuf *base_36 = StrHelp_to_base36(seg_num);
CB_Cat_Str(seg_name, "seg_", 4);
CB_Cat(seg_name, base_36);
DECREF(base_36);
}
CharBuf*
Compiler_to_string(Compiler *self) {
CharBuf *stringified_query = Query_To_String(self->parent);
CharBuf *string = CB_new_from_trusted_utf8("compiler(", 9);
CB_Cat(string, stringified_query);
CB_Cat_Trusted_Str(string, ")", 1);
DECREF(stringified_query);
return string;
}
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(0);
CB_Cat(buf, 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;
}
CharBuf*
PhraseQuery_to_string(PhraseQuery *self)
{
uint32_t i;
uint32_t num_terms = VA_Get_Size(self->terms);
CharBuf *retval = CB_Clone(self->field);
CB_Cat_Trusted_Str(retval, ":\"", 2);
for (i = 0; i < num_terms; i++) {
Obj *term = VA_Fetch(self->terms, i);
CharBuf *term_string = Obj_To_String(term);
CB_Cat(retval, term_string);
DECREF(term_string);
if (i < num_terms - 1) {
CB_Cat_Trusted_Str(retval, " ", 1);
}
}
CB_Cat_Trusted_Str(retval, "\"", 1);
return retval;
}
void
Err_Add_Frame_IMP(Err *self, const char *file, int line, const char *func) {
CharBuf *buf = CB_new(0);
CB_Cat(buf, self->mess);
if (!Str_Ends_With_Utf8(self->mess, "\n", 1)) {
CB_Cat_Char(buf, '\n');
}
if (func != NULL) {
CB_catf(buf, "\t%s at %s line %i32\n", func, file, (int32_t)line);
}
else {
CB_catf(buf, "\tat %s line %i32\n", file, (int32_t)line);
}
DECREF(self->mess);
self->mess = CB_Yield_String(buf);
DECREF(buf);
}
String*
ProximityQuery_To_String_IMP(ProximityQuery *self) {
ProximityQueryIVARS *const ivars = ProximityQuery_IVARS(self);
uint32_t num_terms = VA_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 = VA_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);
CB_catf(buf, "~%u32", ivars->within);
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
CB_vcatf(CharBuf *self, const char *pattern, va_list args) {
size_t pattern_len = strlen(pattern);
const char *pattern_start = pattern;
const char *pattern_end = pattern + pattern_len;
char buf[64];
for (; pattern < pattern_end; pattern++) {
const char *slice_end = pattern;
// Consume all characters leading up to a '%'.
while (slice_end < pattern_end && *slice_end != '%') { slice_end++; }
if (pattern != slice_end) {
size_t size = slice_end - pattern;
CB_Cat_Trusted_Str(self, pattern, size);
pattern = slice_end;
}
if (pattern < pattern_end) {
pattern++; // Move past '%'.
switch (*pattern) {
case '%': {
CB_Cat_Trusted_Str(self, "%", 1);
}
break;
case 'o': {
Obj *obj = va_arg(args, Obj*);
if (!obj) {
CB_Cat_Trusted_Str(self, "[NULL]", 6);
}
else if (Obj_Is_A(obj, CHARBUF)) {
CB_Cat(self, (CharBuf*)obj);
}
else {
CharBuf *string = Obj_To_String(obj);
CB_Cat(self, string);
DECREF(string);
}
}
break;
case 'i': {
int64_t val = 0;
size_t size;
if (pattern[1] == '8') {
val = va_arg(args, int32_t);
pattern++;
}
else if (pattern[1] == '3' && pattern[2] == '2') {
val = va_arg(args, int32_t);
pattern += 2;
}
else if (pattern[1] == '6' && pattern[2] == '4') {
val = va_arg(args, int64_t);
pattern += 2;
}
else {
S_die_invalid_pattern(pattern_start);
}
size = sprintf(buf, "%" I64P, val);
CB_Cat_Trusted_Str(self, buf, size);
}
break;
case 'u': {
uint64_t val = 0;
size_t size;
if (pattern[1] == '8') {
val = va_arg(args, uint32_t);
pattern += 1;
}
else if (pattern[1] == '3' && pattern[2] == '2') {
val = va_arg(args, uint32_t);
pattern += 2;
}
else if (pattern[1] == '6' && pattern[2] == '4') {
val = va_arg(args, uint64_t);
pattern += 2;
}
else {
S_die_invalid_pattern(pattern_start);
}
size = sprintf(buf, "%" U64P, val);
CB_Cat_Trusted_Str(self, buf, size);
}
break;
case 'f': {
if (pattern[1] == '6' && pattern[2] == '4') {
double num = va_arg(args, double);
char bigbuf[512];
size_t size = sprintf(bigbuf, "%g", num);
CB_Cat_Trusted_Str(self, bigbuf, size);
pattern += 2;
}
else {
S_die_invalid_pattern(pattern_start);
}
}
break;
case 'x': {
if (pattern[1] == '3' && pattern[2] == '2') {
unsigned long val = va_arg(args, uint32_t);
size_t size = sprintf(buf, "%.8lx", val);
CB_Cat_Trusted_Str(self, buf, size);
pattern += 2;
}
else {
S_die_invalid_pattern(pattern_start);
}
}
break;
case 's': {
char *string = va_arg(args, char*);
if (string == NULL) {
CB_Cat_Trusted_Str(self, "[NULL]", 6);
}
else {
size_t size = strlen(string);
if (StrHelp_utf8_valid(string, size)) {
CB_Cat_Trusted_Str(self, string, size);
}
else {
CB_Cat_Trusted_Str(self, "[INVALID UTF8]", 14);
}
}
}
break;
default: {
// Assume NULL-terminated pattern string, which
// eliminates the need for bounds checking if '%' is
// the last visible character.
S_die_invalid_pattern(pattern_start);
}
}
void
Err_cat_mess(Err *self, const CharBuf *mess) {
CB_Cat(self->mess, mess);
}
String*
Highlighter_Highlight_Excerpt_IMP(Highlighter *self, VArray *spans,
String *raw_excerpt, int32_t top) {
int32_t hl_start = 0;
int32_t hl_end = 0;
StringIterator *iter = Str_Top(raw_excerpt);
StringIterator *temp = Str_Top(raw_excerpt);
CharBuf *buf = CB_new(Str_Get_Size(raw_excerpt) + 32);
CharBuf *encode_buf = NULL;
int32_t raw_excerpt_end = top + Str_Length(raw_excerpt);
for (uint32_t i = 0, max = VA_Get_Size(spans); i < max; i++) {
Span *span = (Span*)VA_Fetch(spans, i);
int32_t offset = Span_Get_Offset(span);
if (offset < top) {
continue;
}
else if (offset >= raw_excerpt_end) {
break;
}
else {
int32_t relative_start = offset - top;
int32_t relative_end = relative_start + Span_Get_Length(span);
if (relative_start <= hl_end) {
if (relative_end > hl_end) {
hl_end = relative_end;
}
}
else {
if (hl_start < hl_end) {
// Highlight previous section
int32_t highlighted_len = hl_end - hl_start;
StrIter_Assign(temp, iter);
StrIter_Advance(iter, highlighted_len);
String *to_cat = StrIter_substring(temp, iter);
String *encoded = S_do_encode(self, to_cat, &encode_buf);
String *hl_frag = Highlighter_Highlight(self, encoded);
CB_Cat(buf, hl_frag);
DECREF(hl_frag);
DECREF(encoded);
DECREF(to_cat);
}
int32_t non_highlighted_len = relative_start - hl_end;
StrIter_Assign(temp, iter);
StrIter_Advance(iter, non_highlighted_len);
String *to_cat = StrIter_substring(temp, iter);
String *encoded = S_do_encode(self, to_cat, &encode_buf);
CB_Cat(buf, (String*)encoded);
DECREF(encoded);
DECREF(to_cat);
hl_start = relative_start;
hl_end = relative_end;
}
}
}
if (hl_start < hl_end) {
// Highlight final section
int32_t highlighted_len = hl_end - hl_start;
StrIter_Assign(temp, iter);
StrIter_Advance(iter, highlighted_len);
String *to_cat = StrIter_substring(temp, iter);
String *encoded = S_do_encode(self, to_cat, &encode_buf);
String *hl_frag = Highlighter_Highlight(self, encoded);
CB_Cat(buf, hl_frag);
DECREF(hl_frag);
DECREF(encoded);
DECREF(to_cat);
}
// Last text, beyond last highlight span.
if (StrIter_Has_Next(iter)) {
String *to_cat = StrIter_substring(iter, NULL);
String *encoded = S_do_encode(self, to_cat, &encode_buf);
CB_Cat(buf, encoded);
DECREF(encoded);
DECREF(to_cat);
}
String *highlighted = CB_Yield_String(buf);
DECREF(encode_buf);
DECREF(buf);
DECREF(temp);
DECREF(iter);
return highlighted;
}
String*
Highlighter_Raw_Excerpt_IMP(Highlighter *self, String *field_val,
int32_t *start_ptr, HeatMap *heat_map) {
HighlighterIVARS *const ivars = Highlighter_IVARS(self);
// Find start of excerpt.
StringIterator *top = Str_Top(field_val);
int32_t best_location = S_hottest(heat_map);
int32_t start;
uint32_t max_skip;
if ((uint32_t)best_location <= ivars->slop) {
// If the beginning of the string falls within the window centered
// around the hottest point in the field, start the fragment at the
// beginning.
start = 0;
max_skip = best_location;
}
else {
start = best_location - ivars->slop;
max_skip = ivars->slop;
StrIter_Advance(top, start);
}
uint32_t num_skipped;
bool found_starting_edge
= S_find_starting_boundary(top, max_skip, &num_skipped);
start += num_skipped;
// Find end of excerpt.
StringIterator *tail = StrIter_Clone(top);
uint32_t max_len = ivars->excerpt_length;
if (!found_starting_edge) {
// Leave space for starting ellipsis and space character.
max_len -= 2;
}
bool found_ending_edge = true;
uint32_t excerpt_len = StrIter_Advance(tail, max_len);
// Skip up to slop code points but keep at least max_len - slop.
if (excerpt_len > max_len - ivars->slop) {
max_skip = excerpt_len - (max_len - ivars->slop);
found_ending_edge
= S_find_ending_boundary(tail, max_skip, &num_skipped);
if (num_skipped >= excerpt_len) {
excerpt_len = 0;
}
else {
excerpt_len -= num_skipped;
}
}
// Extract excerpt.
String *raw_excerpt;
if (!excerpt_len) {
raw_excerpt = Str_new_from_trusted_utf8("", 0);
}
else {
String *substring = StrIter_substring(top, tail);
CharBuf *buf = CB_new(Str_Get_Size(substring) + 8);
// If not starting on a sentence boundary, prepend an ellipsis.
if (!found_starting_edge) {
CB_Cat_Char(buf, ELLIPSIS_CODE_POINT);
CB_Cat_Char(buf, ' ');
start -= 2;
}
CB_Cat(buf, substring);
// If not ending on a sentence boundary, append an ellipsis.
if (!found_ending_edge) {
CB_Cat_Char(buf, ELLIPSIS_CODE_POINT);
}
raw_excerpt = CB_Yield_String(buf);
DECREF(buf);
DECREF(substring);
}
*start_ptr = start;
DECREF(top);
DECREF(tail);
return raw_excerpt;
}
