void finish();
vector<unsigned char> encoded;
- size_t get_num_docids() const {
+ size_t get_num_docids() const
+ {
// Updated only when we flush, so check that we're finished.
assert(pending_deltas.empty());
return num_docids;
std::unique_ptr<PostingListBuilder *[]> invindex;
FILE *outfp;
+ off_t outfp_pos; // Cheaper than calling ftell(outfp) all the time.
std::string current_block;
std::string tempbuf;
const size_t block_size;
std::string dir_times_compressed;
};
-
EncodingCorpus::EncodingCorpus(FILE *outfp, size_t block_size, ZSTD_CDict *cdict, bool store_dir_times)
- : invindex(new PostingListBuilder *[NUM_TRIGRAMS]), outfp(outfp), block_size(block_size), store_dir_times(store_dir_times), cdict(cdict)
+ : invindex(new PostingListBuilder *[NUM_TRIGRAMS]), outfp(outfp), outfp_pos(ftell(outfp)), block_size(block_size), store_dir_times(store_dir_times), cdict(cdict)
{
fill(invindex.get(), invindex.get() + NUM_TRIGRAMS, nullptr);
if (store_dir_times) {
ptr += 3;
continue;
}
- for ( ;; ) {
+ for (;;) {
// NOTE: Will read one byte past the end of the trigram, but it's OK,
// since we always call it from contexts where there's a terminating zero byte.
uint32_t trgm;
}
// Compress and add the filename block.
- filename_blocks.push_back(ftell(outfp));
+ filename_blocks.push_back(outfp_pos);
string compressed = zstd_compress(current_block, cdict, &tempbuf);
if (fwrite(compressed.data(), compressed.size(), 1, outfp) != 1) {
perror("fwrite()");
exit(1);
}
+ outfp_pos += compressed.size();
current_block.clear();
num_files_in_block = 0;
int fd = -1;
#ifdef O_TMPFILE
fd = open(path.c_str(), O_WRONLY | O_TMPFILE, 0640);
- if (fd == -1 && errno != EOPNOTSUPP) {
+ if (fd == -1 && errno != EOPNOTSUPP && errno != EISDIR) {
perror(path.c_str());
exit(1);
}