static void
test_Clone_and_Reopen(TestBatchRunner *runner) {
String *foo = SSTR_WRAP_C("foo");
String *bar = SSTR_WRAP_C("bar");
RAMFile *file = RAMFile_new(NULL, false);
OutStream *outstream = OutStream_open((Obj*)file);
RAMFileHandle *fh;
InStream *instream;
InStream *clone;
InStream *reopened;
for (uint8_t i = 0; i < 26; i++) {
OutStream_Write_U8(outstream, 'a' + i);
}
OutStream_Close(outstream);
fh = RAMFH_open(foo, FH_READ_ONLY, file);
instream = InStream_open((Obj*)fh);
InStream_Seek(instream, 1);
TEST_TRUE(runner, Str_Equals(InStream_Get_Filename(instream), (Obj*)foo),
"Get_Filename");
clone = InStream_Clone(instream);
TEST_TRUE(runner, Str_Equals(InStream_Get_Filename(clone), (Obj*)foo),
"Clones have same filename");
TEST_TRUE(runner, InStream_Length(instream) == InStream_Length(clone),
"Clones have same length");
TEST_TRUE(runner, InStream_Read_U8(instream) == InStream_Read_U8(clone),
"Clones start at same file position");
reopened = InStream_Reopen(instream, bar, 25, 1);
TEST_TRUE(runner, Str_Equals(InStream_Get_Filename(reopened), (Obj*)bar),
"Reopened InStreams take new filename");
TEST_TRUE(runner, InStream_Read_U8(reopened) == 'z',
"Reopened stream starts at supplied offset");
TEST_TRUE(runner, InStream_Length(reopened) == 1,
"Reopened stream uses supplied length");
TEST_TRUE(runner, InStream_Tell(reopened) == 1,
"Tell() uses supplied offset for reopened stream");
InStream_Seek(reopened, 0);
TEST_TRUE(runner, InStream_Read_U8(reopened) == 'z',
"Seek() uses supplied offset for reopened stream");
DECREF(reopened);
DECREF(clone);
DECREF(instream);
DECREF(outstream);
DECREF(fh);
DECREF(file);
}
static void
test_Local_Open_In(TestBatchRunner *runner) {
Folder *real_folder = S_folder_with_contents();
CompoundFileReader *cf_reader = CFReader_open(real_folder);
InStream *instream;
instream = CFReader_Local_Open_In(cf_reader, foo);
TEST_TRUE(runner, instream != NULL,
"Local_Open_In for virtual file");
TEST_TRUE(runner,
Str_Starts_With(InStream_Get_Filename(instream), CFReader_Get_Path(cf_reader)),
"InStream's path includes directory");
DECREF(instream);
OutStream *outstream = CFReader_Open_Out(cf_reader, baz);
OutStream_Write_Bytes(outstream, "baz", 3);
OutStream_Close(outstream);
DECREF(outstream);
instream = CFReader_Local_Open_In(cf_reader, baz);
TEST_TRUE(runner, instream != NULL,
"Local_Open_In pass-through for real file");
DECREF(instream);
Err_set_error(NULL);
instream = CFReader_Local_Open_In(cf_reader, stuff);
TEST_TRUE(runner, instream == NULL,
"Local_Open_In for non-existent file returns NULL");
TEST_TRUE(runner, Err_get_error() != NULL,
"Local_Open_In for non-existent file sets global error");
DECREF(cf_reader);
DECREF(real_folder);
}
void
TextTermStepper_read_key_frame(TextTermStepper *self, InStream *instream)
{
const uint32_t text_len = InStream_Read_C32(instream);
CharBuf *value;
char *ptr;
// Allocate space.
if (self->value == NULL) {
self->value = (Obj*)CB_new(text_len);
}
value = (CharBuf*)self->value;
ptr = CB_Grow(value, text_len);
// Set the value text.
InStream_Read_Bytes(instream, ptr, text_len);
CB_Set_Size(value, text_len);
if (!StrHelp_utf8_valid(ptr, text_len)) {
THROW(ERR, "Invalid UTF-8 sequence in '%o' at byte %i64",
InStream_Get_Filename(instream),
InStream_Tell(instream) - text_len);
}
// Null-terminate.
ptr[text_len] = '\0';
}
void
TextTermStepper_Read_Delta_IMP(TextTermStepper *self, InStream *instream) {
TextTermStepperIVARS *const ivars = TextTermStepper_IVARS(self);
const uint32_t text_overlap = InStream_Read_C32(instream);
const uint32_t finish_chars_len = InStream_Read_C32(instream);
const uint32_t total_text_len = text_overlap + finish_chars_len;
// Allocate space.
CharBuf *charbuf = (CharBuf*)ivars->value;
char *ptr = CB_Grow(charbuf, total_text_len);
// Set the value text.
InStream_Read_Bytes(instream, ptr + text_overlap, finish_chars_len);
CB_Set_Size(charbuf, total_text_len);
if (!StrHelp_utf8_valid(ptr, total_text_len)) {
THROW(ERR, "Invalid UTF-8 sequence in '%o' at byte %i64",
InStream_Get_Filename(instream),
InStream_Tell(instream) - finish_chars_len);
}
// Null-terminate.
ptr[total_text_len] = '\0';
// Invalidate string.
DECREF(ivars->string);
ivars->string = NULL;
}
static void
S_fresh_flip(PostingPool *self, InStream *lex_temp_in,
InStream *post_temp_in) {
PostingPoolIVARS *const ivars = PostPool_IVARS(self);
if (ivars->flipped) { THROW(ERR, "Can't Flip twice"); }
ivars->flipped = true;
// Sort RawPostings in buffer, if any.
PostPool_Sort_Buffer(self);
// Bail if never flushed.
if (ivars->lex_end == 0) { return; }
// Get a Lexicon.
String *lex_alias = Str_newf("%o-%i64-to-%i64",
InStream_Get_Filename(lex_temp_in),
ivars->lex_start, ivars->lex_end);
InStream *lex_temp_in_dupe = InStream_Reopen(
lex_temp_in, lex_alias, ivars->lex_start,
ivars->lex_end - ivars->lex_start);
ivars->lexicon = (Lexicon*)RawLex_new(
ivars->schema, ivars->field, lex_temp_in_dupe, 0,
ivars->lex_end - ivars->lex_start);
DECREF(lex_alias);
DECREF(lex_temp_in_dupe);
// Get a PostingList.
String *post_alias
= Str_newf("%o-%i64-to-%i64", InStream_Get_Filename(post_temp_in),
ivars->post_start, ivars->post_end);
InStream *post_temp_in_dupe
= InStream_Reopen(post_temp_in, post_alias, ivars->post_start,
ivars->post_end - ivars->post_start);
ivars->plist
= (PostingList*)RawPList_new(ivars->schema, ivars->field,
post_temp_in_dupe, 0,
ivars->post_end - ivars->post_start);
DECREF(post_alias);
DECREF(post_temp_in_dupe);
}
static void
S_read_entry(LexIndex *self)
{
TermInfo *tinfo = self->tinfo;
i64_t offset = (i64_t)Math_decode_bigend_u64(self->offsets + self->tick);
char *data = self->data + offset;
size_t size = Math_decode_c32(&data);
ViewCB_Assign_Str(self->term, data, size);
data += size;
tinfo->doc_freq = Math_decode_c32(&data);
tinfo->post_filepos = Math_decode_c64(&data);
tinfo->skip_filepos = tinfo->doc_freq >= self->skip_interval
? Math_decode_c64(&data) : 0;
tinfo->lex_filepos = Math_decode_c64(&data);
/* Don't allow buffer overruns. */
if (data > self->limit) {
THROW("Buffer overrun in lexicon index for %o",
InStream_Get_Filename(self->ix_in));
}
}