static void
test_overlap(TestBatchRunner *runner) {
size_t result;
result = StrHelp_overlap("", "", 0, 0);
TEST_UINT_EQ(runner, result, 0, "two empty strings");
result = StrHelp_overlap("", "foo", 0, 3);
TEST_UINT_EQ(runner, result, 0, "first string is empty");
result = StrHelp_overlap("foo", "", 3, 0);
TEST_UINT_EQ(runner, result, 0, "second string is empty");
result = StrHelp_overlap("foo", "foo", 3, 3);
TEST_UINT_EQ(runner, result, 3, "equal strings");
result = StrHelp_overlap("foo bar", "foo", 7, 3);
TEST_UINT_EQ(runner, result, 3, "first string is longer");
result = StrHelp_overlap("foo", "foo bar", 3, 7);
TEST_UINT_EQ(runner, result, 3, "second string is longer");
result = StrHelp_overlap("bar", "baz", 3, 3);
TEST_UINT_EQ(runner, result, 2, "different byte");
}
void
TextTermStepper_Write_Delta_IMP(TextTermStepper *self, OutStream *outstream,
Obj *value) {
TextTermStepperIVARS *const ivars = TextTermStepper_IVARS(self);
CharBuf *charbuf = (CharBuf*)ivars->value;
const char *last_text = CB_Get_Ptr8(charbuf);
size_t last_size = CB_Get_Size(charbuf);
const char *new_text = NULL;
size_t new_size = 0;
if (Obj_is_a(value, STRING)) {
String *new_string = (String*)value;
new_text = Str_Get_Ptr8(new_string);
new_size = Str_Get_Size(new_string);
}
else if (Obj_is_a(value, CHARBUF)) {
CharBuf *new_charbuf = (CharBuf*)value;
new_text = CB_Get_Ptr8(new_charbuf);
new_size = CB_Get_Size(new_charbuf);
}
else {
THROW(ERR, "'value' must be a String or CharBuf");
}
// Count how many bytes the strings share at the top.
const int32_t overlap = StrHelp_overlap(last_text, new_text,
last_size, new_size);
const char *const diff_start_str = new_text + overlap;
const size_t diff_len = new_size - overlap;
// Write number of common bytes and common bytes.
OutStream_Write_C32(outstream, overlap);
OutStream_Write_String(outstream, diff_start_str, diff_len);
// Update value.
CB_Mimic_Utf8(charbuf, new_text, new_size);
// Invalidate string.
DECREF(ivars->string);
ivars->string = NULL;
}
void
TextTermStepper_write_delta(TextTermStepper *self, OutStream *outstream,
Obj *value)
{
CharBuf *new_value = (CharBuf*)CERTIFY(value, CHARBUF);
CharBuf *last_value = (CharBuf*)self->value;
char *new_text = (char*)CB_Get_Ptr8(new_value);
size_t new_size = CB_Get_Size(new_value);
char *last_text = (char*)CB_Get_Ptr8(last_value);
size_t last_size = CB_Get_Size(last_value);
// Count how many bytes the strings share at the top.
const int32_t overlap = StrHelp_overlap(last_text, new_text,
last_size, new_size);
const char *const diff_start_str = new_text + overlap;
const size_t diff_len = new_size - overlap;
// Write number of common bytes and common bytes.
OutStream_Write_C32(outstream, overlap);
OutStream_Write_String(outstream, diff_start_str, diff_len);
// Update value.
CB_Mimic((CharBuf*)self->value, value);
}
ByteBuf*
HLWriter_TV_Buf_IMP(HighlightWriter *self, Inversion *inversion) {
const char *last_text = "";
size_t last_len = 0;
ByteBuf *tv_buf = BB_new(20 + Inversion_Get_Size(inversion) * 8);
uint32_t num_postings = 0;
Token **tokens;
uint32_t freq;
UNUSED_VAR(self);
// Leave space for a c32 indicating the number of postings.
BB_Set_Size(tv_buf, C32_MAX_BYTES);
Inversion_Reset(inversion);
while ((tokens = Inversion_Next_Cluster(inversion, &freq)) != NULL) {
Token *token = *tokens;
char *const token_text = Token_Get_Text(token);
const int32_t token_len = Token_Get_Len(token);
int32_t overlap = StrHelp_overlap(last_text, token_text,
last_len, token_len);
char *ptr;
char *orig;
size_t old_size = BB_Get_Size(tv_buf);
size_t new_size = old_size
+ C32_MAX_BYTES // overlap
+ C32_MAX_BYTES // length of string diff
+ (token_len - overlap) // diff char data
+ C32_MAX_BYTES // num prox
+ (C32_MAX_BYTES * freq * 3); // pos data
// Allocate for worst-case scenario.
ptr = BB_Grow(tv_buf, new_size);
orig = ptr;
ptr += old_size;
// Track number of postings.
num_postings += 1;
// Append the string diff to the tv_buf.
NumUtil_encode_c32(overlap, &ptr);
NumUtil_encode_c32((token_len - overlap), &ptr);
memcpy(ptr, (token_text + overlap), (token_len - overlap));
ptr += token_len - overlap;
// Save text and text_len for comparison next loop.
last_text = token_text;
last_len = token_len;
// Append the number of positions for this term.
NumUtil_encode_c32(freq, &ptr);
do {
// Add position, start_offset, and end_offset to tv_buf.
NumUtil_encode_c32(Token_Get_Pos(token), &ptr);
NumUtil_encode_c32(Token_Get_Start_Offset(token), &ptr);
NumUtil_encode_c32(Token_Get_End_Offset(token), &ptr);
} while (--freq && (token = *++tokens));
// Set new byte length.
BB_Set_Size(tv_buf, ptr - orig);
}
// Go back and start the term vector string with the posting count.
char *dest = BB_Get_Buf(tv_buf);
NumUtil_encode_padded_c32(num_postings, &dest);
return tv_buf;
}
