Rename Corpus to EncodingCorpus, and make it private.

We don't really want two different classes named Corpus; that's pretty
confusing, even though they don't live in the same binary.

diff --git a/database-builder.cpp b/database-builder.cpp
index 89a06d6c59a52399a11c05a6513df780f4edb281..42fc186bf3689792f6bd85519303c5134c99a4ed 100644 (file)
--- a/database-builder.cpp
+++ b/database-builder.cpp
@@ -175,7 +175,44 @@ string DictionaryBuilder::train(size_t buf_size)
        return buf;
 }
 
-Corpus::Corpus(FILE *outfp, size_t block_size, ZSTD_CDict *cdict, bool store_dir_times)
+class EncodingCorpus : public DatabaseReceiver {
+public:
+       EncodingCorpus(FILE *outfp, size_t block_size, ZSTD_CDict *cdict, bool store_dir_times);
+       ~EncodingCorpus();
+
+       void add_file(std::string filename, dir_time dt) override;
+       void flush_block() override;
+       void finish() override;
+
+       std::vector<uint64_t> filename_blocks;
+       size_t num_files = 0, num_files_in_block = 0, num_blocks = 0;
+       bool seen_trigram(uint32_t trgm)
+       {
+               return invindex[trgm] != nullptr;
+       }
+       size_t num_files_seen() const override { return num_files; }
+       PostingListBuilder &get_pl_builder(uint32_t trgm);
+       size_t num_trigrams() const;
+       std::string get_compressed_dir_times();
+
+private:
+       void compress_dir_times(size_t allowed_slop);
+
+       std::unique_ptr<PostingListBuilder *[]> invindex;
+       FILE *outfp;
+       std::string current_block;
+       std::string tempbuf;
+       const size_t block_size;
+       const bool store_dir_times;
+       ZSTD_CDict *cdict;
+
+       ZSTD_CStream *dir_time_ctx = nullptr;
+       std::string dir_times;  // Buffer of still-uncompressed data.
+       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)
 {
        fill(invindex.get(), invindex.get() + NUM_TRIGRAMS, nullptr);
@@ -185,14 +222,14 @@ Corpus::Corpus(FILE *outfp, size_t block_size, ZSTD_CDict *cdict, bool store_dir
        }
 }
 
-Corpus::~Corpus()
+EncodingCorpus::~EncodingCorpus()
 {
        for (unsigned i = 0; i < NUM_TRIGRAMS; ++i) {
                delete invindex[i];
        }
 }
 
-PostingListBuilder &Corpus::get_pl_builder(uint32_t trgm)
+PostingListBuilder &EncodingCorpus::get_pl_builder(uint32_t trgm)
 {
        if (invindex[trgm] == nullptr) {
                invindex[trgm] = new PostingListBuilder;
@@ -200,7 +237,7 @@ PostingListBuilder &Corpus::get_pl_builder(uint32_t trgm)
        return *invindex[trgm];
 }
 
-void Corpus::add_file(string filename, dir_time dt)
+void EncodingCorpus::add_file(string filename, dir_time dt)
 {
        ++num_files;
        if (!current_block.empty()) {
@@ -224,7 +261,7 @@ void Corpus::add_file(string filename, dir_time dt)
        }
 }
 
-void Corpus::compress_dir_times(size_t allowed_slop)
+void EncodingCorpus::compress_dir_times(size_t allowed_slop)
 {
        while (dir_times.size() >= allowed_slop) {
                size_t old_size = dir_times_compressed.size();
@@ -256,7 +293,7 @@ void Corpus::compress_dir_times(size_t allowed_slop)
        }
 }
 
-void Corpus::flush_block()
+void EncodingCorpus::flush_block()
 {
        if (current_block.empty()) {
                return;
@@ -290,12 +327,12 @@ void Corpus::flush_block()
        ++num_blocks;
 }
 
-void Corpus::finish()
+void EncodingCorpus::finish()
 {
        flush_block();
 }
 
-size_t Corpus::num_trigrams() const
+size_t EncodingCorpus::num_trigrams() const
 {
        size_t num = 0;
        for (unsigned trgm = 0; trgm < NUM_TRIGRAMS; ++trgm) {
@@ -306,7 +343,7 @@ size_t Corpus::num_trigrams() const
        return num;
 }
 
-string Corpus::get_compressed_dir_times()
+string EncodingCorpus::get_compressed_dir_times()
 {
        if (!store_dir_times) {
                return "";
@@ -385,7 +422,7 @@ uint32_t next_prime(uint32_t x)
        return x;
 }
 
-unique_ptr<Trigram[]> create_hashtable(Corpus &corpus, const vector<uint32_t> &all_trigrams, uint32_t ht_size, uint32_t num_overflow_slots)
+unique_ptr<Trigram[]> create_hashtable(EncodingCorpus &corpus, const vector<uint32_t> &all_trigrams, uint32_t ht_size, uint32_t num_overflow_slots)
 {
        unique_ptr<Trigram[]> ht(new Trigram[ht_size + num_overflow_slots + 1]);  // 1 for the sentinel element at the end.
        for (unsigned i = 0; i < ht_size + num_overflow_slots + 1; ++i) {
@@ -490,10 +527,10 @@ DatabaseBuilder::DatabaseBuilder(const char *outfile, gid_t owner, int block_siz
        hdr.conf_block_offset_bytes = 0;
 }
 
-Corpus *DatabaseBuilder::start_corpus(bool store_dir_times)
+DatabaseReceiver *DatabaseBuilder::start_corpus(bool store_dir_times)
 {
        corpus_start = steady_clock::now();
-       corpus = new Corpus(outfp, block_size, cdict, store_dir_times);
+       corpus = new EncodingCorpus(outfp, block_size, cdict, store_dir_times);
        return corpus;
 }
 

diff --git a/database-builder.h b/database-builder.h
index 2a56e11093cbe26c4823c82642ebb59ddb4386cc..0c25476145086f1793f7e6324a41e6a097f25de0 100644 (file)
--- a/database-builder.h
+++ b/database-builder.h
@@ -41,6 +41,9 @@ public:
        virtual void add_file(std::string filename, dir_time dt) = 0;
        virtual void flush_block() = 0;
        virtual void finish() { flush_block(); }
+
+       // EncodingCorpus only.
+       virtual size_t num_files_seen() const { return -1; }
 };
 
 class DictionaryBuilder : public DatabaseReceiver {
@@ -65,45 +68,12 @@ private:
        std::vector<size_t> lengths;
 };
 
-class Corpus : public DatabaseReceiver {
-public:
-       Corpus(FILE *outfp, size_t block_size, ZSTD_CDict *cdict, bool store_dir_times);
-       ~Corpus();
-
-       void add_file(std::string filename, dir_time dt) override;
-       void flush_block() override;
-       void finish() override;
-
-       std::vector<uint64_t> filename_blocks;
-       size_t num_files = 0, num_files_in_block = 0, num_blocks = 0;
-       bool seen_trigram(uint32_t trgm)
-       {
-               return invindex[trgm] != nullptr;
-       }
-       PostingListBuilder &get_pl_builder(uint32_t trgm);
-       size_t num_trigrams() const;
-       std::string get_compressed_dir_times();
-
-private:
-       void compress_dir_times(size_t allowed_slop);
-
-       std::unique_ptr<PostingListBuilder *[]> invindex;
-       FILE *outfp;
-       std::string current_block;
-       std::string tempbuf;
-       const size_t block_size;
-       const bool store_dir_times;
-       ZSTD_CDict *cdict;
-
-       ZSTD_CStream *dir_time_ctx = nullptr;
-       std::string dir_times;  // Buffer of still-uncompressed data.
-       std::string dir_times_compressed;
-};
+class EncodingCorpus;
 
 class DatabaseBuilder {
 public:
        DatabaseBuilder(const char *outfile, gid_t owner, int block_size, std::string dictionary, bool check_visibility);
-       Corpus *start_corpus(bool store_dir_times);
+       DatabaseReceiver *start_corpus(bool store_dir_times);
        void set_next_dictionary(std::string next_dictionary);
        void set_conf_block(std::string conf_block);
        void finish_corpus();
@@ -117,7 +87,7 @@ private:
        Header hdr;
        const int block_size;
        std::chrono::steady_clock::time_point corpus_start;
-       Corpus *corpus = nullptr;
+       EncodingCorpus *corpus = nullptr;
        ZSTD_CDict *cdict = nullptr;
        std::string next_dictionary, conf_block;
 };

diff --git a/plocate-build.cpp b/plocate-build.cpp
index cb765ed742a738bb91507b5fefa910a85b26ac59..028a1ee06c620d8274ae707ba1905041ac288a17 100644 (file)
--- a/plocate-build.cpp
+++ b/plocate-build.cpp
@@ -167,7 +167,7 @@ void do_build(const char *infile, const char *outfile, int block_size, bool plai
        string dictionary = builder.train(1024);
 
        DatabaseBuilder db(outfile, /*owner=*/-1, block_size, dictionary, /*check_visibility=*/true);
-       Corpus *corpus = db.start_corpus(/*store_dir_times=*/false);
+       DatabaseReceiver *corpus = db.start_corpus(/*store_dir_times=*/false);
        if (plaintext) {
                read_plaintext(infp, corpus);
        } else {
@@ -175,7 +175,7 @@ void do_build(const char *infile, const char *outfile, int block_size, bool plai
        }
        fclose(infp);
 
-       dprintf("Read %zu files from %s\n", corpus->num_files, infile);
+       dprintf("Read %zu files from %s\n", corpus->num_files_seen(), infile);
        db.finish_corpus();
 }
 

diff --git a/updatedb.cpp b/updatedb.cpp
index 1465b5b54fc11fd80a11060ffad511ecc1e0cb1a..895b0715b05afce634a8916f576ccdf833230654 100644 (file)
--- a/updatedb.cpp
+++ b/updatedb.cpp
@@ -530,7 +530,7 @@ string ExistingDB::read_next_dictionary() const
 // “parent_dev” must be the device of the parent directory of “path”.
 //
 // Takes ownership of fd.
-int scan(const string &path, int fd, dev_t parent_dev, dir_time modified, dir_time db_modified, ExistingDB *existing_db, Corpus *corpus, DictionaryBuilder *dict_builder)
+int scan(const string &path, int fd, dev_t parent_dev, dir_time modified, dir_time db_modified, ExistingDB *existing_db, DatabaseReceiver *corpus, DictionaryBuilder *dict_builder)
 {
        if (string_list_contains_dir_path(&conf_prunepaths, &conf_prunepaths_index, path)) {
                if (conf_debug_pruning) {
@@ -782,7 +782,7 @@ int main(int argc, char **argv)
 
        DatabaseBuilder db(conf_output.c_str(), owner, conf_block_size, existing_db.read_next_dictionary(), conf_check_visibility);
        db.set_conf_block(conf_block);
-       Corpus *corpus = db.start_corpus(/*store_dir_times=*/true);
+       DatabaseReceiver *corpus = db.start_corpus(/*store_dir_times=*/true);
 
        int root_fd = opendir_noatime(AT_FDCWD, conf_scan_root);
        if (root_fd == -1) {