view create.cpp @ 509:cc2b97d020b1

Code rearrangements to tease apart library code from C++ audioDB code.

There should be precisely no functional changes in this commit.
Instead, the only thing that has happened is that all the abstraction
violation and other horribleness is concentrated in one place: the
include of "audioDB-internals.h" in audioDB.h -- the separation will be
complete once that include can be removed.

This include is necessary because the command-line binary / SOAP server
still does some things directly rather than through an API: not least of
which the operations that have not yet been integrated into the API yet,
but also some messing around with constants, flags and nominally
internal functions.  The intent is to remove as many of these as
possible and think quite hard about the rest.

In the meantime, the library is now much more self-contained: the only
things it uses are in the audioDB_API.h and audioDB-internals.h headers;
thus there are fewer nasty surprises lurking for readers of the code.

The Makefile has been adjusted to take advantage of this rearrangement
in the dependencies.

extern "C" {
#include "audioDB_API.h"
}
#include "audioDB-internals.h"

/* Make a new database.

(FIXME: this text, in particular the conditional, will not be true 
once we implement create flags rather than defaulting on format based
on the requested size arguments)

IF size(featuredata) < ADB_FIXME_LARGE_ADB_SIZE
   The database consists of:

   * a header (see adb_header_t definition);
   * keyTable: list of keys of tracks;
   * trackTable: Maps implicit feature index to a feature vector
     matrix (sizes of tracks)
   * featureTable: Lots of doubles;
   * timesTable: (start,end) time points for each feature vector;
   * powerTable: associated power for each feature vector;
   * l2normTable: squared l2norms for each feature vector.

ELSE the database consists of:

   * a header (see adb_header_t definition);
   * keyTable: list of keys of tracks
   * trackTable: sizes of tracks
   * featureTable: list of feature file names
   * timesTable: list of times file names
   * powerTable: list of power file names

*/

adb_t *audiodb_create(const char *path, unsigned datasize, unsigned ntracks, unsigned datadim) {
  int fd;
  adb_header_t *header = 0;
  off_t databytes, auxbytes;
  if(datasize == 0) {
    datasize = ADB_DEFAULT_DATASIZE;
  }
  if(ntracks == 0) {
    ntracks = ADB_DEFAULT_NTRACKS;
  }
  if(datadim == 0) {
    datadim = ADB_DEFAULT_DATADIM;
  }

  if ((fd = open(path, O_RDWR|O_CREAT|O_EXCL, S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP|S_IROTH|S_IWOTH)) < 0) {
    goto error;
  }
  if (acquire_lock(fd, true)) {
    goto error;
  }

  header = (adb_header_t *) malloc(sizeof(adb_header_t));
  if(!header) {
    goto error;
  }

  // Initialize header
  header->magic = ADB_MAGIC;
  header->version = ADB_FORMAT_VERSION;
  header->numFiles = 0;
  header->dim = 0;
  header->flags = 0;
  header->headerSize = ADB_HEADER_SIZE;
  header->length = 0;
  header->fileTableOffset = align_page_up(ADB_HEADER_SIZE);
  header->trackTableOffset = align_page_up(header->fileTableOffset + ADB_FILETABLE_ENTRY_SIZE*ntracks); //
  header->dataOffset = align_page_up(header->trackTableOffset + ADB_TRACKTABLE_ENTRY_SIZE*ntracks);

  databytes = ((off_t) datasize) * 1024 * 1024;
  auxbytes = databytes / datadim;

  // For backward-compatibility, Record the point-encoding parameter for LSH indexing in the adb header
  // If this value is 0 then it will be set to 14

#if ADB_FIXME_LSH_N_POINT_BITS > 15
#error "consistency check of ADB_FIXME_LSH_N_POINT_BITS failed (>31)"
#endif

  header->flags |= ADB_FIXME_LSH_N_POINT_BITS << 28;

  // If database will fit in a single file the vectors are copied into the AudioDB instance
  // Else all the vectors are left on the FileSystem and we use the dataOffset as storage
  // for the location of the features, powers and times files (assuming that arbitrary keys are used for the fileTable)
  if(ntracks < ADB_FIXME_LARGE_ADB_NTRACKS && datasize < ADB_FIXME_LARGE_ADB_SIZE) {
    header->timesTableOffset = align_page_up(header->dataOffset + databytes);
    header->powerTableOffset = align_page_up(header->timesTableOffset + 2*auxbytes);
    header->l2normTableOffset = align_page_up(header->powerTableOffset + auxbytes);
    header->dbSize = align_page_up(header->l2normTableOffset + auxbytes);
  } else { // Create REFERENCES ADB, features and powers kept on filesystem
    header->flags |= ADB_HEADER_FLAG_REFERENCES;
    header->timesTableOffset = align_page_up(header->dataOffset + ADB_FILETABLE_ENTRY_SIZE*ntracks);
    header->powerTableOffset = align_page_up(header->timesTableOffset + ADB_FILETABLE_ENTRY_SIZE*ntracks);
    header->l2normTableOffset = align_page_up(header->powerTableOffset + ADB_FILETABLE_ENTRY_SIZE*ntracks);
    header->dbSize = header->l2normTableOffset;
  }

  write_or_goto_error(fd, header, ADB_HEADER_SIZE);

  // go to the location corresponding to the last byte
  if (lseek (fd, header->dbSize - 1, SEEK_SET) == -1) {
    goto error;
  }

  // write a dummy byte at the last location
  write_or_goto_error(fd, "", 1);

  free(header);
  return audiodb_open(path, O_RDWR);

 error:
  if(header) {
    free(header);
  }
  return NULL;
}