mas01cr@239: #include "audioDB.h"
mas01cr@385: extern "C" {
mas01cr@385: #include "audioDB_API.h"
mas01cr@385: }
mas01cr@410: #include "audioDB-internals.h"
mas01cr@410: 
mas01cr@239: /* Make a new database.
mas01cr@239: 
mas01cr@407: IF size(featuredata) < O2_LARGE_ADB_SIZE
mas01cr@239:    The database consists of:
mas01cr@239: 
mas01cr@239:    * a header (see dbTableHeader struct definition);
mas01cr@239:    * keyTable: list of keys of tracks;
mas01cr@239:    * trackTable: Maps implicit feature index to a feature vector
mas01cr@239:      matrix (sizes of tracks)
mas01cr@239:    * featureTable: Lots of doubles;
mas01cr@239:    * timesTable: (start,end) time points for each feature vector;
mas01cr@239:    * powerTable: associated power for each feature vector;
mas01cr@239:    * l2normTable: squared l2norms for each feature vector.
mas01cr@407: 
mas01mc@324: ELSE the database consists of:
mas01cr@407: 
mas01mc@324:    * a header (see dbTableHeader struct definition);
mas01mc@324:    * keyTable: list of keys of tracks
mas01mc@324:    * trackTable: sizes of tracks
mas01mc@324:    * featureTable: list of feature file names
mas01mc@324:    * timesTable: list of times file names
mas01mc@324:    * powerTable: list of power file names
mas01mc@324: 
mas01cr@239: */
mas01cr@239: 
mas01cr@407: adb_t *audiodb_create(const char *path, unsigned datasize, unsigned ntracks, unsigned datadim) {
mas01cr@407:   int fd;
mas01cr@407:   adb_header_t *header = 0;
mas01cr@407:   off_t databytes, auxbytes;
mas01cr@407:   if(datasize == 0) {
mas01cr@407:     datasize = O2_DEFAULT_DATASIZE;
mas01cr@407:   }
mas01cr@407:   if(ntracks == 0) {
mas01cr@407:     ntracks = O2_DEFAULT_NTRACKS;
mas01cr@407:   }
mas01cr@407:   if(datadim == 0) {
mas01cr@407:     datadim = O2_DEFAULT_DATADIM;
mas01cr@407:   }
mas01cr@239: 
mas01cr@407:   if ((fd = open(path, O_RDWR|O_CREAT|O_EXCL, S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP|S_IROTH|S_IWOTH)) < 0) {
mas01cr@407:     goto error;
mas01cr@407:   }
mas01cr@407:   if (acquire_lock(fd, true)) {
mas01cr@407:     goto error;
mas01cr@407:   }
mas01cr@389: 
mas01cr@407:   header = (adb_header_t *) malloc(sizeof(adb_header_t));
mas01cr@407:   if(!header) {
mas01cr@407:     goto error;
mas01cr@407:   }
mas01cr@389: 
mas01cr@407:   // Initialize header
mas01cr@407:   header->magic = O2_MAGIC;
mas01cr@407:   header->version = O2_FORMAT_VERSION;
mas01cr@407:   header->numFiles = 0;
mas01cr@407:   header->dim = 0;
mas01cr@407:   header->flags = 0;
mas01cr@407:   header->headerSize = O2_HEADERSIZE;
mas01cr@407:   header->length = 0;
mas01cr@407:   header->fileTableOffset = ALIGN_PAGE_UP(O2_HEADERSIZE);
mas01cr@407:   header->trackTableOffset = ALIGN_PAGE_UP(header->fileTableOffset + O2_FILETABLE_ENTRY_SIZE*ntracks);
mas01cr@407:   header->dataOffset = ALIGN_PAGE_UP(header->trackTableOffset + O2_TRACKTABLE_ENTRY_SIZE*ntracks);
mas01cr@407: 
mas01cr@407:   databytes = ((off_t) datasize) * 1024 * 1024;
mas01cr@407:   auxbytes = databytes / datadim;
mas01cr@407: 
mas01cr@458:   /* FIXME: what's going on here?  There are two distinct preprocessor
mas01cr@458:      constants (O2_LSH_N_POINT_BITS, LSH_N_POINT_BITS); a third is
mas01cr@458:      presumably some default (O2_DEFAULT_LSH_N_POINT_BITS), and then
mas01cr@458:      there's this magic 28 bits [which needs to be the same as the 28
mas01cr@458:      in audiodb_lsh_n_point_bits()].  Should this really be part of
mas01cr@458:      the flags structure at all?  Putting it elsewhere will of course
mas01cr@458:      break backwards compatibility, unless 14 is the only value that's
mas01cr@458:      been used anywhere... */
mas01cr@407: 
mas01cr@407:   // For backward-compatibility, Record the point-encoding parameter for LSH indexing in the adb header
mas01cr@407:   // If this value is 0 then it will be set to 14
mas01cr@407: 
mas01mc@324: #if O2_LSH_N_POINT_BITS > 15
mas01mc@324: #error "AudioDB Compile ERROR: consistency check of O2_LSH_POINT_BITS failed (>15)"
mas01mc@324: #endif
mas01cr@407: 
mas01cr@407:   header->flags |= LSH_N_POINT_BITS << 28;
mas01cr@407: 
mas01cr@407:   // If database will fit in a single file the vectors are copied into the AudioDB instance
mas01cr@407:   // Else all the vectors are left on the FileSystem and we use the dataOffset as storage
mas01cr@407:   // for the location of the features, powers and times files (assuming that arbitrary keys are used for the fileTable)
mas01cr@407:   if(ntracks<O2_LARGE_ADB_NTRACKS && datasize<O2_LARGE_ADB_SIZE){
mas01cr@407:     header->timesTableOffset = ALIGN_PAGE_UP(header->dataOffset + databytes);
mas01cr@407:     header->powerTableOffset = ALIGN_PAGE_UP(header->timesTableOffset + 2*auxbytes);
mas01cr@407:     header->l2normTableOffset = ALIGN_PAGE_UP(header->powerTableOffset + auxbytes);
mas01cr@407:     header->dbSize = ALIGN_PAGE_UP(header->l2normTableOffset + auxbytes);
mas01cr@407:   } else { // Create LARGE_ADB, features and powers kept on filesystem
mas01cr@407:     header->flags |= O2_FLAG_LARGE_ADB;
mas01cr@407:     header->timesTableOffset = ALIGN_PAGE_UP(header->dataOffset + O2_FILETABLE_ENTRY_SIZE*ntracks);
mas01cr@407:     header->powerTableOffset = ALIGN_PAGE_UP(header->timesTableOffset + O2_FILETABLE_ENTRY_SIZE*ntracks);
mas01cr@407:     header->l2normTableOffset = ALIGN_PAGE_UP(header->powerTableOffset + O2_FILETABLE_ENTRY_SIZE*ntracks);
mas01cr@407:     header->dbSize = header->l2normTableOffset;
mas01cr@407:   }
mas01cr@407: 
mas01cr@410:   write_or_goto_error(fd, header, O2_HEADERSIZE);
mas01cr@407: 
mas01cr@407:   // go to the location corresponding to the last byte
mas01cr@407:   if (lseek (fd, header->dbSize - 1, SEEK_SET) == -1) {
mas01cr@407:     goto error;
mas01cr@407:   }
mas01cr@407: 
mas01cr@407:   // write a dummy byte at the last location
mas01cr@410:   write_or_goto_error(fd, "", 1);
mas01cr@389: 
mas01cr@407:   free(header);
mas01cr@407:   return audiodb_open(path, O_RDWR);
mas01cr@389: 
mas01cr@407:  error:
mas01cr@407:   if(header) {
mas01cr@390:     free(header);
mas01mc@324:   }
mas01cr@407:   return NULL;
mas01cr@239: }