-#include <stdio.h>
-#include <string.h>
+#include "vp4.h"
+
#include <algorithm>
-#include <unordered_map>
-#include <string>
-#include <vector>
+#include <arpa/inet.h>
#include <chrono>
-#include <unistd.h>
+#include <endian.h>
#include <fcntl.h>
+#include <stdio.h>
+#include <string.h>
+#include <string>
#include <sys/mman.h>
-#include <arpa/inet.h>
-#include <endian.h>
+#include <unistd.h>
+#include <unordered_map>
+#include <vector>
#include <zstd.h>
-#include "vp4.h"
-
-#define P4NENC_BOUND(n) ((n+127)/128+(n+32)*sizeof(uint32_t))
+#define P4NENC_BOUND(n) ((n + 127) / 128 + (n + 32) * sizeof(uint32_t))
using namespace std;
using namespace std::chrono;
#define dprintf(...)
//#define dprintf(...) fprintf(stderr, __VA_ARGS__);
-
+
static inline uint32_t read_unigram(const string &s, size_t idx)
{
if (idx < s.size()) {
static inline uint32_t read_trigram(const string &s, size_t start)
{
- return read_unigram(s, start) |
- (read_unigram(s, start + 1) << 8) |
+ return read_unigram(s, start) | (read_unigram(s, start + 1) << 8) |
(read_unigram(s, start + 2) << 16);
}
-bool has_access(const char *filename, unordered_map<string, bool> *access_rx_cache)
+bool has_access(const char *filename,
+ unordered_map<string, bool> *access_rx_cache)
{
const char *end = strchr(filename + 1, '/');
while (end != nullptr) {
Corpus(int fd);
~Corpus();
const Trigram *find_trigram(uint32_t trgm) const;
- const unsigned char *get_compressed_posting_list(const Trigram *trigram) const;
+ const unsigned char *
+ get_compressed_posting_list(const Trigram *trigram) const;
string_view get_compressed_filename_block(uint32_t docid) const;
private:
const Trigram *Corpus::find_trigram(uint32_t trgm) const
{
- const Trigram *trgmptr = lower_bound(trgm_begin, trgm_end, trgm, [](const Trigram &trgm, uint32_t t) {
- return trgm.trgm < t;
- });
+ const Trigram *trgmptr = lower_bound(
+ trgm_begin, trgm_end, trgm,
+ [](const Trigram &trgm, uint32_t t) { return trgm.trgm < t; });
if (trgmptr == trgm_end || trgmptr->trgm != trgm) {
return nullptr;
}
return trgmptr;
}
-const unsigned char *Corpus::get_compressed_posting_list(const Trigram *trgmptr) const
+const unsigned char *
+Corpus::get_compressed_posting_list(const Trigram *trgmptr) const
{
return reinterpret_cast<const unsigned char *>(data + trgmptr->offset);
}
string_view Corpus::get_compressed_filename_block(uint32_t docid) const
{
const char *compressed = (const char *)(data + filename_offsets[docid]);
- size_t compressed_size = filename_offsets[docid + 1] - filename_offsets[docid]; // Allowed we have a sentinel block at the end.
- return {compressed, compressed_size};
+ size_t compressed_size =
+ filename_offsets[docid + 1] -
+ filename_offsets[docid]; // Allowed we have a sentinel block at the end.
+ return { compressed, compressed_size };
}
-size_t scan_docid(const string &needle, uint32_t docid, const Corpus &corpus, unordered_map<string, bool> *access_rx_cache)
+size_t scan_docid(const string &needle, uint32_t docid, const Corpus &corpus,
+ unordered_map<string, bool> *access_rx_cache)
{
string_view compressed = corpus.get_compressed_filename_block(docid);
size_t matched = 0;
string block;
- block.resize(ZSTD_getFrameContentSize(compressed.data(), compressed.size()) + 1);
+ block.resize(ZSTD_getFrameContentSize(compressed.data(), compressed.size()) +
+ 1);
- ZSTD_decompress(&block[0], block.size(), compressed.data(), compressed.size());
+ ZSTD_decompress(&block[0], block.size(), compressed.data(),
+ compressed.size());
block[block.size() - 1] = '\0';
for (const char *filename = block.data();
- filename != block.data() + block.size();
- filename += strlen(filename) + 1) {
+ filename != block.data() + block.size();
+ filename += strlen(filename) + 1) {
if (strstr(filename, needle.c_str()) == nullptr) {
continue;
}
exit(EXIT_FAILURE);
}
- //steady_clock::time_point start = steady_clock::now();
+ // steady_clock::time_point start = steady_clock::now();
if (access("/", R_OK | X_OK)) {
// We can't find anything, no need to bother...
return;
auto last = unique(trigrams.begin(), trigrams.end());
trigrams.erase(last, trigrams.end());
}
- sort(trigrams.begin(), trigrams.end(), [&](const Trigram *a, const Trigram *b) {
- return a->num_docids < b->num_docids;
- });
+ sort(trigrams.begin(), trigrams.end(),
+ [&](const Trigram *a, const Trigram *b) {
+ return a->num_docids < b->num_docids;
+ });
vector<uint32_t> in1, in2, out;
for (const Trigram *trgmptr : trigrams) {
- //uint32_t trgm = trgmptr->trgm;
+ // uint32_t trgm = trgmptr->trgm;
size_t num = trgmptr->num_docids;
const unsigned char *pldata = corpus.get_compressed_posting_list(trgmptr);
if (in1.empty()) {
in1.resize(num + 128);
p4nd1dec128v32(const_cast<unsigned char *>(pldata), num, &in1[0]);
in1.resize(num);
- dprintf("trigram '%c%c%c' decoded to %zu entries\n", trgm & 0xff, (trgm >> 8) & 0xff, (trgm >> 16) & 0xff, num);
+ dprintf("trigram '%c%c%c' decoded to %zu entries\n", trgm & 0xff,
+ (trgm >> 8) & 0xff, (trgm >> 16) & 0xff, num);
} else {
if (num > in1.size() * 100) {
- dprintf("trigram '%c%c%c' has %zu entries, ignoring the rest (will weed out false positives later)\n",
- trgm & 0xff, (trgm >> 8) & 0xff, (trgm >> 16) & 0xff, num);
+ dprintf("trigram '%c%c%c' has %zu entries, ignoring the rest (will "
+ "weed out false positives later)\n",
+ trgm & 0xff, (trgm >> 8) & 0xff, (trgm >> 16) & 0xff, num);
break;
}
p4nd1dec128v32(const_cast<unsigned char *>(pldata), num, &in2[0]);
out.clear();
- set_intersection(in1.begin(), in1.end(), in2.begin(), in2.begin() + num, back_inserter(out));
+ set_intersection(in1.begin(), in1.end(), in2.begin(), in2.begin() + num,
+ back_inserter(out));
swap(in1, out);
- dprintf("trigram '%c%c%c' decoded to %zu entries, %zu left\n", trgm & 0xff, (trgm >> 8) & 0xff, (trgm >> 16) & 0xff, num, in1.size());
+ dprintf("trigram '%c%c%c' decoded to %zu entries, %zu left\n",
+ trgm & 0xff, (trgm >> 8) & 0xff, (trgm >> 16) & 0xff, num,
+ in1.size());
if (in1.empty()) {
dprintf("no matches (intersection list is empty)\n");
break;
steady_clock::time_point end = steady_clock::now();
dprintf("Intersection took %.1f ms. Doing final verification and printing:\n",
- 1e3 * duration<float>(end - start).count());
+ 1e3 * duration<float>(end - start).count());
unordered_map<string, bool> access_rx_cache;
}
end = steady_clock::now();
dprintf("Done in %.1f ms, found %d matches.\n",
- 1e3 * duration<float>(end - start).count(), matched);
+ 1e3 * duration<float>(end - start).count(), matched);
}
int main(int argc, char **argv)
projects / plocate / blobdiff