| 1 | #! /usr/bin/perl -w |
|---|
| 2 | |
|---|
| 3 | # find_archives.pl |
|---|
| 4 | # Author: Albert Teoh |
|---|
| 5 | # Date: December 2005 |
|---|
| 6 | # ATNF, CSIRO |
|---|
| 7 | |
|---|
| 8 | # Crawls through data archive directories (given as an input parameter), |
|---|
| 9 | # searches for all observation and/or calibration files, enquires about |
|---|
| 10 | # their pulsar parameters and inputs these details as well as their |
|---|
| 11 | # location into a database |
|---|
| 12 | |
|---|
| 13 | |
|---|
| 14 | use strict; |
|---|
| 15 | use File::Find (); |
|---|
| 16 | use DBI qw(:sql_types); |
|---|
| 17 | use POSIX; |
|---|
| 18 | |
|---|
| 19 | # Set the variable $File::Find::dont_use_nlink if you're using AFS, |
|---|
| 20 | # since AFS cheats. |
|---|
| 21 | |
|---|
| 22 | # for the convenience of &wanted calls, including -eval statements: |
|---|
| 23 | use vars qw/*name *dir *prune @archive_dirs $archive_dir @files |
|---|
| 24 | $vap_cmd $vap_out $params $num_vap_params $db_handle |
|---|
| 25 | $filename $raj $decj $dm $period $bw $cnfg |
|---|
| 26 | $freq $inst $mjdint $mjdfrac $npol $nchan $nbin $nsub |
|---|
| 27 | $rcvr $site $length $obsrvr $data_loc $data_type |
|---|
| 28 | $file_size $MJD $raH $raM $raS $decD $decM $decS $rajd $decjd $decstring |
|---|
| 29 | $pi $pion180 $NGP_RA $NGP_DEC $ASC_NODE $draR $decR |
|---|
| 30 | $sinb $sinl $cosl $gb $gl $gl_raw |
|---|
| 31 | $bmaj $bmin $bpa $hdrver $survey $nbeam |
|---|
| 32 | $output $vap $archive_extensions $ext |
|---|
| 33 | $default |
|---|
| 34 | @result_params $FILE_SIZE_INDEX $FB $DFB $WBC $CPSR2 |
|---|
| 35 | $params_NAME |
|---|
| 36 | $params_PROJID |
|---|
| 37 | $params_RAJ |
|---|
| 38 | $params_DECJ |
|---|
| 39 | $params_FREQ |
|---|
| 40 | $params_BW |
|---|
| 41 | $params_LENGTH |
|---|
| 42 | $params_DATE |
|---|
| 43 | $params_TIME |
|---|
| 44 | $params_MJDINT |
|---|
| 45 | $params_MJDFRAC |
|---|
| 46 | $params_BMAJ |
|---|
| 47 | $params_BMIN |
|---|
| 48 | $params_BPA |
|---|
| 49 | $params_DM |
|---|
| 50 | $params_PERIOD |
|---|
| 51 | $params_NCHAN |
|---|
| 52 | $params_NPOL |
|---|
| 53 | $params_NBIN |
|---|
| 54 | $params_NSUB |
|---|
| 55 | $params_NBITS |
|---|
| 56 | $params_TSAMP |
|---|
| 57 | $params_NBEAM |
|---|
| 58 | $params_CNFG |
|---|
| 59 | $params_INST |
|---|
| 60 | $params_RCVR |
|---|
| 61 | $params_TELESCOP |
|---|
| 62 | $params_SITE |
|---|
| 63 | $params_OBSRVR |
|---|
| 64 | /; # $machine |
|---|
| 65 | |
|---|
| 66 | *name = *File::Find::name; |
|---|
| 67 | *dir = *File::Find::dir; |
|---|
| 68 | *prune = *File::Find::prune; |
|---|
| 69 | |
|---|
| 70 | # The data types |
|---|
| 71 | $FB = "FB"; |
|---|
| 72 | $WBC = "WBC"; |
|---|
| 73 | $DFB = "DFB"; |
|---|
| 74 | $CPSR2 = "CPSR2"; |
|---|
| 75 | |
|---|
| 76 | # Pulsar parameters (listed in "vap -H") |
|---|
| 77 | #$params = "name raj decj dm period bw cnfg freq inst intmjd fracmjd npol nchan nbin nsub rcvr site length obsrvr"; |
|---|
| 78 | #$params = "name projid ra dec freq bw length stt_date stt_time intmjd fracmjd bmaj bmin bpa dm period nchan npol nbin nsub nbits tsamp nbeam cnfg backend rcvr telescop site obsrvr"; |
|---|
| 79 | $params = "name projid ra dec freq bw length stt_date stt_time intmjd fracmjd mjd bmaj bmin bpa dm period nchan npol nbin nsub nbits tsamp nbeam cnfg backend rcvr telescop site obsrvr"; |
|---|
| 80 | |
|---|
| 81 | # vap result parameter indices. Used to locate the resulting |
|---|
| 82 | # parameter's location |
|---|
| 83 | $params_NAME = 0; |
|---|
| 84 | $params_PROJID = 1; |
|---|
| 85 | $params_RAJ = 2; |
|---|
| 86 | $params_DECJ = 3; |
|---|
| 87 | $params_FREQ = 4; |
|---|
| 88 | $params_BW = 5; |
|---|
| 89 | $params_LENGTH = 6; |
|---|
| 90 | $params_DATE = 7; |
|---|
| 91 | $params_TIME = 8; |
|---|
| 92 | $params_MJDINT = 9; |
|---|
| 93 | $params_MJDFRAC = 10; |
|---|
| 94 | $params_BMAJ = 11; |
|---|
| 95 | $params_BMIN = 12; |
|---|
| 96 | $params_BPA = 13; |
|---|
| 97 | $params_DM = 14; |
|---|
| 98 | $params_PERIOD = 15; |
|---|
| 99 | $params_NCHAN = 16; |
|---|
| 100 | $params_NPOL = 17; |
|---|
| 101 | $params_NBIN = 18; |
|---|
| 102 | $params_NSUB = 19; |
|---|
| 103 | $params_NBITS = 20; |
|---|
| 104 | $params_TSAMP = 21; |
|---|
| 105 | $params_NBEAM = 22; |
|---|
| 106 | $params_CNFG = 23; |
|---|
| 107 | $params_INST = 24; |
|---|
| 108 | $params_RCVR = 25; |
|---|
| 109 | $params_TELESCOP= 26; |
|---|
| 110 | $params_SITE = 27; |
|---|
| 111 | $params_OBSRVR = 28; |
|---|
| 112 | |
|---|
| 113 | # filename extensions |
|---|
| 114 | $archive_extensions = "ar rf cf cfb fb"; |
|---|
| 115 | |
|---|
| 116 | # The machine to do all the processing on |
|---|
| 117 | #$machine = "tycho"; |
|---|
| 118 | |
|---|
| 119 | # The location of vap |
|---|
| 120 | $vap = "/pulsar/psr/linux/bin/vap"; |
|---|
| 121 | |
|---|
| 122 | ##################### |
|---|
| 123 | # Begin main method |
|---|
| 124 | # |
|---|
| 125 | my @vap_params = split(/ /, $params); |
|---|
| 126 | $num_vap_params = scalar(@vap_params); |
|---|
| 127 | |
|---|
| 128 | # The column number (starting from 0) of the file |
|---|
| 129 | # size when calling ls -l |
|---|
| 130 | $FILE_SIZE_INDEX = 4; |
|---|
| 131 | |
|---|
| 132 | # Use the default operation which is to firstly delete any |
|---|
| 133 | # stale records, then crawl through and insert new records |
|---|
| 134 | $default = 1; |
|---|
| 135 | |
|---|
| 136 | print "ARGV = @ARGV\n"; |
|---|
| 137 | |
|---|
| 138 | if (scalar(@ARGV) == 0) { |
|---|
| 139 | print "find_archives.pl: A crawler that populates a relational database\n |
|---|
| 140 | with indexes to observational data along with their respective\n |
|---|
| 141 | cal files.\n\n |
|---|
| 142 | Usage: find_archives.pl [options] directory1 [directory2 ...]\n\n |
|---|
| 143 | If no options are provided, it will default to do both -d and -p\n |
|---|
| 144 | Options:\n |
|---|
| 145 | \t-d Delete stale records\n |
|---|
| 146 | \t-p Populate database with new records\n"; |
|---|
| 147 | |
|---|
| 148 | } |
|---|
| 149 | |
|---|
| 150 | # Connect to the database |
|---|
| 151 | $db_handle = connectdb(); |
|---|
| 152 | |
|---|
| 153 | ######################################################### |
|---|
| 154 | # Step 1. First remove any stale entries in the database |
|---|
| 155 | # (remove file locations that no longer exist) |
|---|
| 156 | ######################################################### |
|---|
| 157 | |
|---|
| 158 | if ($ARGV[0] eq "-d") { |
|---|
| 159 | deleteStaleRecords(); |
|---|
| 160 | $default = 0; |
|---|
| 161 | shift @ARGV; |
|---|
| 162 | } |
|---|
| 163 | |
|---|
| 164 | |
|---|
| 165 | ######################################################### |
|---|
| 166 | # Step 2. Crawl through each input directory and insert |
|---|
| 167 | # any new archives |
|---|
| 168 | |
|---|
| 169 | if ($ARGV[0] eq "-p") { |
|---|
| 170 | shift @ARGV; |
|---|
| 171 | @archive_dirs = @ARGV; |
|---|
| 172 | |
|---|
| 173 | foreach $archive_dir (@archive_dirs) { |
|---|
| 174 | |
|---|
| 175 | # Traverse desired filesystems |
|---|
| 176 | File::Find::find({wanted => \&wanted}, $archive_dir); |
|---|
| 177 | } |
|---|
| 178 | $default = 0; |
|---|
| 179 | } |
|---|
| 180 | |
|---|
| 181 | # Else do both |
|---|
| 182 | if ($default) { |
|---|
| 183 | |
|---|
| 184 | deleteStaleRecords(); |
|---|
| 185 | |
|---|
| 186 | @archive_dirs = @ARGV; |
|---|
| 187 | |
|---|
| 188 | foreach $archive_dir (@archive_dirs) { |
|---|
| 189 | |
|---|
| 190 | # Check that the input directory is legit |
|---|
| 191 | |
|---|
| 192 | # Traverse desired filesystems |
|---|
| 193 | File::Find::find({wanted => \&wanted}, $archive_dir); |
|---|
| 194 | } |
|---|
| 195 | |
|---|
| 196 | } |
|---|
| 197 | if ($db_handle->disconnect()) { |
|---|
| 198 | print "Successfully Disconnected from database\n"; |
|---|
| 199 | } |
|---|
| 200 | else { |
|---|
| 201 | print "Error: Failed to disconnect from the database\n"; |
|---|
| 202 | exit; |
|---|
| 203 | } |
|---|
| 204 | |
|---|
| 205 | ####################################################### |
|---|
| 206 | # Subroutine definitions |
|---|
| 207 | # |
|---|
| 208 | sub wanted { |
|---|
| 209 | |
|---|
| 210 | my @files; |
|---|
| 211 | my @all_files; |
|---|
| 212 | my @cpsr2_obs_dirs; |
|---|
| 213 | my $file; |
|---|
| 214 | my $cpsr2_obs_dir; |
|---|
| 215 | my $search_str = ""; |
|---|
| 216 | my $num_dumps = 0; |
|---|
| 217 | |
|---|
| 218 | if (-d and /^[a-zA-Z]{0,1}[0-9]{4}[+-][0-9]{4}.*/s ) { |
|---|
| 219 | |
|---|
| 220 | #print "In directory $_\n"; |
|---|
| 221 | |
|---|
| 222 | foreach $ext (split (/\s+/, $archive_extensions) ) { |
|---|
| 223 | $search_str .= $File::Find::name . "/*.$ext " |
|---|
| 224 | } |
|---|
| 225 | |
|---|
| 226 | @all_files = glob($search_str); |
|---|
| 227 | |
|---|
| 228 | if (scalar(@all_files) == 0) { |
|---|
| 229 | |
|---|
| 230 | #print "This is a CPSR2 directory: $_\n"; |
|---|
| 231 | |
|---|
| 232 | # Foreach observation |
|---|
| 233 | @cpsr2_obs_dirs = glob($File::Find::name . "/????-??-??-??:??:??"); |
|---|
| 234 | |
|---|
| 235 | foreach $cpsr2_obs_dir (@cpsr2_obs_dirs) { |
|---|
| 236 | #print "dir is $cpsr2_obs_dir\n"; |
|---|
| 237 | |
|---|
| 238 | # Need to check if this entry already exists |
|---|
| 239 | $cpsr2_obs_dir =~ s/\s//g; |
|---|
| 240 | |
|---|
| 241 | if (isDuplicate($cpsr2_obs_dir) == 1) { |
|---|
| 242 | print "$cpsr2_obs_dir is a duplicate entry. Skipping....\n\n"; |
|---|
| 243 | next; |
|---|
| 244 | } |
|---|
| 245 | #else { |
|---|
| 246 | #print "$cpsr2_obs_dir NOT a duplicate. Inserting into database\n"; |
|---|
| 247 | #} |
|---|
| 248 | |
|---|
| 249 | |
|---|
| 250 | @all_files = glob($cpsr2_obs_dir . "/m*fb " . $cpsr2_obs_dir . "/m*ar"); |
|---|
| 251 | |
|---|
| 252 | if (scalar(@all_files) == 0) { |
|---|
| 253 | next; |
|---|
| 254 | } |
|---|
| 255 | |
|---|
| 256 | #print "first file is " . $all_files[0] . "\n"; |
|---|
| 257 | |
|---|
| 258 | # Work out the length of each dump |
|---|
| 259 | # $vap_cmd = "ssh $machine $vap -n -c \\\"$params\\\" ".$all_files[0]. " |"; |
|---|
| 260 | $vap_cmd = "$vap -n -c \"$params\" ".$all_files[0]. " | grep -v filename | grep -v -x \"\" |"; |
|---|
| 261 | |
|---|
| 262 | print "Calling vap...$vap_cmd\n"; |
|---|
| 263 | |
|---|
| 264 | open(VAP,$vap_cmd); |
|---|
| 265 | |
|---|
| 266 | while ($vap_out = <VAP>) { |
|---|
| 267 | |
|---|
| 268 | print "vap_out = $vap_out\n"; |
|---|
| 269 | |
|---|
| 270 | #@result_params = split(/\s+/, $vap_out); |
|---|
| 271 | @result_params = split(" ", $vap_out); |
|---|
| 272 | |
|---|
| 273 | $length = $result_params[$params_LENGTH+1]; |
|---|
| 274 | $result_params[$params_BW+1] *= 2; |
|---|
| 275 | |
|---|
| 276 | #print "length = $length secs\n"; |
|---|
| 277 | |
|---|
| 278 | $num_dumps = `ls -1 $cpsr2_obs_dir/m*fb $cpsr2_obs_dir/m*ar | wc -l`; |
|---|
| 279 | |
|---|
| 280 | #print "num dumps = $num_dumps\n"; |
|---|
| 281 | |
|---|
| 282 | $result_params[$params_LENGTH+1] = $length*$num_dumps; |
|---|
| 283 | |
|---|
| 284 | #print "total length = " . $result_params[$params_LENGTH+1] . "\n"; |
|---|
| 285 | |
|---|
| 286 | #print "new bw = " . $result_params[$params_BW+1] . "\n"; |
|---|
| 287 | |
|---|
| 288 | # +1 because first column is filename |
|---|
| 289 | if (scalar(@result_params) == ($num_vap_params + 1)) { |
|---|
| 290 | print "@result_params\n"; |
|---|
| 291 | |
|---|
| 292 | populate_observations($cpsr2_obs_dir, $num_dumps); |
|---|
| 293 | } |
|---|
| 294 | else { |
|---|
| 295 | print "Incorrect number of cols = ".scalar(@result_params).". Expected " . ($num_vap_params+1) ."\n"; |
|---|
| 296 | } |
|---|
| 297 | |
|---|
| 298 | } |
|---|
| 299 | print "\n"; |
|---|
| 300 | |
|---|
| 301 | |
|---|
| 302 | # Count the number of files |
|---|
| 303 | } |
|---|
| 304 | } |
|---|
| 305 | else { |
|---|
| 306 | foreach $file (@all_files) { |
|---|
| 307 | |
|---|
| 308 | # Need to check if this entry already exists |
|---|
| 309 | $_ =~ s/\s//g; |
|---|
| 310 | $file =~ s/\s//g; |
|---|
| 311 | |
|---|
| 312 | if (isDuplicate($file) == 1) { |
|---|
| 313 | #print "$file is a duplicate entry. Skipping....\n\n"; |
|---|
| 314 | next; |
|---|
| 315 | } |
|---|
| 316 | else { |
|---|
| 317 | #print "$file NOT a duplicate. Inserting into database\n"; |
|---|
| 318 | } |
|---|
| 319 | |
|---|
| 320 | # $vap_cmd = "ssh $machine $vap -n -c \\\"$params\\\" $file |"; |
|---|
| 321 | $vap_cmd = "$vap -n -c \"$params\" $file |"; |
|---|
| 322 | |
|---|
| 323 | print "Calling vap...$vap_cmd\n"; |
|---|
| 324 | |
|---|
| 325 | open(VAP,$vap_cmd); |
|---|
| 326 | |
|---|
| 327 | while ($vap_out = <VAP>) { |
|---|
| 328 | @result_params = split(/\s+/, $vap_out); |
|---|
| 329 | |
|---|
| 330 | # +1 because first column is filename |
|---|
| 331 | if (scalar(@result_params) == ($num_vap_params + 1)) { |
|---|
| 332 | print "@result_params\n"; |
|---|
| 333 | |
|---|
| 334 | populate_observations($File::Find::name); |
|---|
| 335 | } |
|---|
| 336 | else { |
|---|
| 337 | print "Incorrect number of cols = ".scalar(@result_params).". Expected " . $num_vap_params+1 ."\n"; |
|---|
| 338 | } |
|---|
| 339 | } |
|---|
| 340 | print "\n"; |
|---|
| 341 | } |
|---|
| 342 | } |
|---|
| 343 | # don't traverse into this directory |
|---|
| 344 | $File::Find::prune = 1; |
|---|
| 345 | } |
|---|
| 346 | } |
|---|
| 347 | |
|---|
| 348 | sub isDuplicate { |
|---|
| 349 | |
|---|
| 350 | #DEBUGGING |
|---|
| 351 | return 0; |
|---|
| 352 | |
|---|
| 353 | my $filename = $_[0]; |
|---|
| 354 | my $search_data_loc; |
|---|
| 355 | my $search_filename; |
|---|
| 356 | my $pulsar_name; |
|---|
| 357 | |
|---|
| 358 | if ( $filename =~ m/(.*(J[0-9]{4}[+-][0-9]{1,4}.*))[\/]+([a-z][0-9]{6}_[0-9]{6}\.([a-z]{2,}))$/i ) { |
|---|
| 359 | $search_data_loc = $1; |
|---|
| 360 | $search_filename = $3; |
|---|
| 361 | $pulsar_name = $2; |
|---|
| 362 | } |
|---|
| 363 | |
|---|
| 364 | elsif ($filename =~ m/(.*(J[0-9]{4}[+-][0-9]{1,4}.*))[\/]+([0-9]{4}-[0-9]{2}-[0-9]{2}-[0-9]{2}:[0-9]{2}:[0-9]{2})$/i ) { |
|---|
| 365 | $search_data_loc = $1; |
|---|
| 366 | $search_filename = $3; |
|---|
| 367 | $pulsar_name = $2; |
|---|
| 368 | } |
|---|
| 369 | else { |
|---|
| 370 | return 1; |
|---|
| 371 | } |
|---|
| 372 | |
|---|
| 373 | |
|---|
| 374 | my $sql; |
|---|
| 375 | |
|---|
| 376 | if ($pulsar_name =~ m/.*_R$/) { |
|---|
| 377 | |
|---|
| 378 | $sql = qq{ SELECT * from cals |
|---|
| 379 | WHERE data_loc LIKE ? AND |
|---|
| 380 | filename = ? |
|---|
| 381 | }; |
|---|
| 382 | } |
|---|
| 383 | else { |
|---|
| 384 | print "This is an observation\n"; |
|---|
| 385 | $sql = qq{ SELECT * from observations |
|---|
| 386 | WHERE data_loc LIKE ? AND |
|---|
| 387 | filename = ? |
|---|
| 388 | }; |
|---|
| 389 | } |
|---|
| 390 | |
|---|
| 391 | my $sth = $db_handle->prepare( $sql ); |
|---|
| 392 | |
|---|
| 393 | eval { |
|---|
| 394 | print "like $search_data_loc and filename = $search_filename \n"; |
|---|
| 395 | |
|---|
| 396 | $sth->bind_param( 1, $search_data_loc, SQL_VARCHAR ); |
|---|
| 397 | $sth->bind_param( 2, $search_filename, SQL_VARCHAR ); |
|---|
| 398 | $sth->execute(); |
|---|
| 399 | |
|---|
| 400 | }; |
|---|
| 401 | |
|---|
| 402 | if( $@ ) { |
|---|
| 403 | warn "Database error: $DBI::errstr\n"; |
|---|
| 404 | $db_handle->rollback(); #just die if rollback is failing |
|---|
| 405 | } |
|---|
| 406 | |
|---|
| 407 | # read the records |
|---|
| 408 | my @data = $sth->fetchrow_array(); |
|---|
| 409 | |
|---|
| 410 | $sth->finish(); |
|---|
| 411 | |
|---|
| 412 | if (scalar(@data) > 0) { |
|---|
| 413 | return 1; |
|---|
| 414 | } |
|---|
| 415 | |
|---|
| 416 | return 0; |
|---|
| 417 | } |
|---|
| 418 | |
|---|
| 419 | sub populate_observations { |
|---|
| 420 | |
|---|
| 421 | |
|---|
| 422 | print "populate_obs(".$_[0].")\n"; |
|---|
| 423 | $data_loc = $_[0]; |
|---|
| 424 | my $display_data_loc = $data_loc; |
|---|
| 425 | |
|---|
| 426 | if ($data_loc =~ m/(.*)\/[0-9]{4}-[0-9]{2}-[0-9]{2}-[0-9]{2}:[0-9]{2}:[0-9]{2}$/i) { |
|---|
| 427 | print "Actual data loc is $1\n"; |
|---|
| 428 | $display_data_loc = $1; |
|---|
| 429 | } |
|---|
| 430 | |
|---|
| 431 | $filename = shift(@result_params); |
|---|
| 432 | |
|---|
| 433 | # remove leading and trailing whitespace |
|---|
| 434 | $filename =~ s/ //g; |
|---|
| 435 | |
|---|
| 436 | my $display_filename = $filename; |
|---|
| 437 | |
|---|
| 438 | my $ls = `ls -l $data_loc/$filename`; |
|---|
| 439 | |
|---|
| 440 | my @file_details = split(/\s+/, $ls); |
|---|
| 441 | |
|---|
| 442 | $file_size = $file_details[$FILE_SIZE_INDEX]; |
|---|
| 443 | |
|---|
| 444 | # Get the total filesize |
|---|
| 445 | if (scalar (@_) > 1) { |
|---|
| 446 | my $num_dumps = $_[1]; |
|---|
| 447 | |
|---|
| 448 | $file_size *= $num_dumps*2; # *2 because of other band |
|---|
| 449 | print "Multiplying filesize by number files = $num_dumps with total $file_size\n"; |
|---|
| 450 | } |
|---|
| 451 | |
|---|
| 452 | # Try to find the data type |
|---|
| 453 | if ($filename =~ /\.ar$/) { |
|---|
| 454 | $data_type = $FB; |
|---|
| 455 | } |
|---|
| 456 | elsif ($filename =~ /\.rf/ || $filename =~ /fb/) { |
|---|
| 457 | if ($filename =~ /^a/) { # DFB file |
|---|
| 458 | $data_type = $DFB; |
|---|
| 459 | } |
|---|
| 460 | |
|---|
| 461 | elsif ($filename =~ /^w/) { # Wide Band Correlator file |
|---|
| 462 | $data_type = $WBC; |
|---|
| 463 | } |
|---|
| 464 | elsif ($filename =~ /^m/ || $filename =~ /^n/ || $filename =~ /^r/) { |
|---|
| 465 | $data_type = $CPSR2; |
|---|
| 466 | |
|---|
| 467 | # Use the directory of observation instead of filename instance |
|---|
| 468 | my @dirs = split(/\//, $data_loc); |
|---|
| 469 | |
|---|
| 470 | print "The new filename will be " . $dirs[$#dirs] . " from $data_loc\n"; |
|---|
| 471 | |
|---|
| 472 | $display_filename = $dirs[$#dirs]; |
|---|
| 473 | |
|---|
| 474 | # For some reason, the name of the pulsar cals for CPSR2 files are all "CAL" |
|---|
| 475 | if ($data_loc =~ /\_R/) { |
|---|
| 476 | my @jname_parts = split(/J/i, $dirs[$#dirs-1]); |
|---|
| 477 | $result_params[$params_NAME] = $jname_parts[1]; |
|---|
| 478 | } |
|---|
| 479 | |
|---|
| 480 | } |
|---|
| 481 | else { |
|---|
| 482 | $data_type = undef; |
|---|
| 483 | } |
|---|
| 484 | } |
|---|
| 485 | |
|---|
| 486 | # Check that N/A values are set to NULL |
|---|
| 487 | |
|---|
| 488 | # Get the total MJD |
|---|
| 489 | if ( $result_params[$params_MJDINT] eq "N/A" && |
|---|
| 490 | $result_params[$params_MJDFRAC] eq "N/A" ) { |
|---|
| 491 | $MJD = undef; |
|---|
| 492 | } |
|---|
| 493 | else { |
|---|
| 494 | $MJD = $result_params[$params_MJDINT] + $result_params[$params_MJDFRAC]; |
|---|
| 495 | } |
|---|
| 496 | |
|---|
| 497 | print "MJD = $MJD\n"; |
|---|
| 498 | |
|---|
| 499 | #### Calculate the RA and Dec in decimal degrees. |
|---|
| 500 | |
|---|
| 501 | ($raH,$raM,$raS) = split(':', $result_params[$params_RAJ]); |
|---|
| 502 | $rajd = ($raH + $raM/60. + $raS/3600.) * 15.; |
|---|
| 503 | |
|---|
| 504 | ($decD,$decM,$decS) = split(':', $result_params[$params_DECJ]); |
|---|
| 505 | $decjd = (abs($decD) + $decM/60. + $decS/3600.); |
|---|
| 506 | my @decstring = split(/ */,$result_params[$params_DECJ]); |
|---|
| 507 | if($decstring[0] eq '-'){ |
|---|
| 508 | $decjd = -1. * $decjd; |
|---|
| 509 | } |
|---|
| 510 | |
|---|
| 511 | ### Calculate the Galactic Longitude and latitude |
|---|
| 512 | $pi=asin(1) * 2.; |
|---|
| 513 | $pion180 = $pi/180.; |
|---|
| 514 | $NGP_RA = 192.859508 * $pion180; # location of NGP |
|---|
| 515 | $NGP_DEC= 27.128336 * $pion180; |
|---|
| 516 | $ASC_NODE=32.932; |
|---|
| 517 | |
|---|
| 518 | $draR = $rajd*$pion180 - $NGP_RA; |
|---|
| 519 | $decR = $decjd*$pion180; |
|---|
| 520 | $sinb = cos($decR) * cos($NGP_DEC) * cos($draR) + sin($decR) * sin($NGP_DEC); |
|---|
| 521 | $gb = asin($sinb); # this is the latitude, but in radians. |
|---|
| 522 | |
|---|
| 523 | $sinl = (sin($decR) * cos($NGP_DEC) - cos($decR) * cos($draR) * sin($NGP_DEC)) / cos($gb); |
|---|
| 524 | $cosl = cos($decR) * sin($draR) / cos($gb); |
|---|
| 525 | |
|---|
| 526 | # Need to get the correct quadrant, as this isn't preserved by |
|---|
| 527 | # atan, which returns angle between -90 and 90. |
|---|
| 528 | $gl_raw = atan($sinl/$cosl); |
|---|
| 529 | if($sinl > 0){ |
|---|
| 530 | if($cosl > 0 ){ $gl = $gl_raw; } |
|---|
| 531 | else { $gl = $gl_raw + $pi; } |
|---|
| 532 | } |
|---|
| 533 | else { |
|---|
| 534 | if($cosl > 0){ $gl = $gl_raw + 2.*$pi; } |
|---|
| 535 | else{ $gl = $gl_raw + $pi; } |
|---|
| 536 | } |
|---|
| 537 | # Now put them into degrees. |
|---|
| 538 | $gb = $gb / $pion180; |
|---|
| 539 | $gl = ($gl / $pion180) + $ASC_NODE; |
|---|
| 540 | |
|---|
| 541 | print "rajd = $rajd, decjd=$decjd, gl=$gl, gb=$gb\n"; |
|---|
| 542 | |
|---|
| 543 | if($result_params[$params_BMAJ] eq "UNDEF"){ |
|---|
| 544 | $bmaj = (1.2*(299792458./($result_params[$params_FREQ] *1.e6))/64.) / $pion180; } |
|---|
| 545 | if($result_params[$params_BMAJ] eq "UNDEF"){ |
|---|
| 546 | $bmin = (1.2*(299792458./($result_params[$params_FREQ] *1.e6))/64.) / $pion180; } |
|---|
| 547 | if($result_params[$params_BMAJ] eq "UNDEF"){ |
|---|
| 548 | $bpa = 0.; } |
|---|
| 549 | |
|---|
| 550 | # Dud things that aren't required for this set of data. |
|---|
| 551 | $hdrver = NULL; |
|---|
| 552 | $survey = NULL; |
|---|
| 553 | $nbeam = 1; |
|---|
| 554 | |
|---|
| 555 | my $i; |
|---|
| 556 | for ($i = 0; $i <= $#result_params; $i++) { |
|---|
| 557 | if ($result_params[$i] eq "N/A") { |
|---|
| 558 | $result_params[$i] = undef; |
|---|
| 559 | } |
|---|
| 560 | } |
|---|
| 561 | |
|---|
| 562 | print "filename = $display_filename\n"; |
|---|
| 563 | print "src_name = " . $result_params[$params_NAME] . "\n"; |
|---|
| 564 | print "projid = " . $result_params[$params_PROJID] . "\n"; |
|---|
| 565 | print "raj = " . $result_params[$params_RAJ] . "\n"; |
|---|
| 566 | print "dec = " . $result_params[$params_DECJ] . "\n"; |
|---|
| 567 | print "data_type = $data_type\n"; |
|---|
| 568 | print "freq = " . $result_params[$params_FREQ] . "\n"; |
|---|
| 569 | print "bw = " . $result_params[$params_BW] . "\n"; |
|---|
| 570 | print "scanlen = " . $result_params[$params_LENGTH] . "\n"; |
|---|
| 571 | print "date = " . $result_params[$params_DATE] ."\n"; |
|---|
| 572 | print "ut = " . $result_params[$params_TIME] ."\n"; |
|---|
| 573 | print "MJD = " . $MJD ."\n"; |
|---|
| 574 | print "rajd = " . $rajd ."\n"; |
|---|
| 575 | print "decjd = " . $decjd ."\n"; |
|---|
| 576 | print "gl = " . $gl ."\n"; |
|---|
| 577 | print "gb = " . $gb ."\n"; |
|---|
| 578 | print "bmaj = " . $bmaj ."\n"; |
|---|
| 579 | print "bmin = " . $bmin ."\n"; |
|---|
| 580 | print "bpa = " . $bpa ."\n"; |
|---|
| 581 | print "dm = " . $result_params[$params_DM] . "\n"; |
|---|
| 582 | print "period = " . $result_params[$params_PERIOD] . "\n"; |
|---|
| 583 | print "nchan = " . $result_params[$params_NCHAN] . "\n"; |
|---|
| 584 | print "npol = " . $result_params[$params_NPOL] . "\n"; |
|---|
| 585 | print "nbin = " . $result_params[$params_NBIN] . "\n"; |
|---|
| 586 | print "nsub = " . $result_params[$params_NSUB] . "\n"; |
|---|
| 587 | print "tsamp = " . $result_params[$params_TSAMP] . "\n"; |
|---|
| 588 | print "nbits = " . $result_params[$params_NBITS] . "\n"; |
|---|
| 589 | print "nbeam = " . $result_params[$params_NBEAM] . "\n"; |
|---|
| 590 | print "cnfg = " . $result_params[$params_CNFG] . "\n"; |
|---|
| 591 | print "inst = " . $result_params[$params_INST] . "\n"; |
|---|
| 592 | print "rcvr = " . $result_params[$params_RCVR] . "\n"; |
|---|
| 593 | print "hdrver = " . $hdrver . "\n"; |
|---|
| 594 | print "survey = " . $survey . "\n"; |
|---|
| 595 | print "telescope = " . $result_params[$params_TELESCOP] . "\n"; |
|---|
| 596 | print "site = " . $result_params[$params_SITE] . "\n"; |
|---|
| 597 | print "obsrvr = " . $result_params[$params_OBSRVR] . "\n"; |
|---|
| 598 | print "data_loc = $display_data_loc\n"; |
|---|
| 599 | print "file_size = $file_size bytes\n"; |
|---|
| 600 | |
|---|
| 601 | my $sql; |
|---|
| 602 | |
|---|
| 603 | if ($data_loc =~ /\_R/) { |
|---|
| 604 | print "Inserting cal file $display_filename\n"; |
|---|
| 605 | |
|---|
| 606 | $sql = qq{ INSERT INTO cals |
|---|
| 607 | (filename, src_name, project_id, |
|---|
| 608 | raj, decj, data_type, obsfreq, bw, |
|---|
| 609 | scanlen, date, ut, MJD, |
|---|
| 610 | rajd, decjd, gl, gb, bmaj, bmin, bpa, |
|---|
| 611 | dm, period, nchan, npol, nbin, nsub, |
|---|
| 612 | tsamp, nbits, nbeam, |
|---|
| 613 | cnfg, inst, rcvr, hdrver, survey, |
|---|
| 614 | telescope, site, obsrvr, |
|---|
| 615 | data_loc, file_size_bytes |
|---|
| 616 | ) |
|---|
| 617 | |
|---|
| 618 | VALUES |
|---|
| 619 | ( ?, ?, ? |
|---|
| 620 | ?, ?, ?, ?, ?, |
|---|
| 621 | ?, ?, ?, ?, |
|---|
| 622 | ?, ?, ?, ?, ?, ?, ?, |
|---|
| 623 | ?, ?, ?, ?, ?, ?, |
|---|
| 624 | ?, ?, ?, |
|---|
| 625 | ?, ?, ?, ?, ?, |
|---|
| 626 | ?, ?, ?, |
|---|
| 627 | ?, ?, |
|---|
| 628 | ) |
|---|
| 629 | }; |
|---|
| 630 | } |
|---|
| 631 | else { |
|---|
| 632 | |
|---|
| 633 | print "Inserting obs file $display_filename\n"; |
|---|
| 634 | |
|---|
| 635 | $sql = qq{ INSERT INTO observations |
|---|
| 636 | (filename, src_name, project_id, |
|---|
| 637 | raj, decj, data_type, obsfreq, bw, |
|---|
| 638 | scanlen, date, ut, MJD, |
|---|
| 639 | rajd, decjd, gl, gb, bmaj, bmin, bpa, |
|---|
| 640 | dm, period, nchan, npol, nbin, nsub, |
|---|
| 641 | tsamp, nbits, nbeam, |
|---|
| 642 | cnfg, inst, rcvr, hdrver, survey, |
|---|
| 643 | telescope, site, obsrvr, |
|---|
| 644 | data_loc, file_size_bytes |
|---|
| 645 | ) |
|---|
| 646 | |
|---|
| 647 | VALUES |
|---|
| 648 | ( ?, ?, ? |
|---|
| 649 | ?, ?, ?, ?, ?, |
|---|
| 650 | ?, ?, ?, ?, |
|---|
| 651 | ?, ?, ?, ?, ?, ?, ?, |
|---|
| 652 | ?, ?, ?, ?, ?, ?, |
|---|
| 653 | ?, ?, ?, |
|---|
| 654 | ?, ?, ?, ?, ?, |
|---|
| 655 | ?, ?, ?, |
|---|
| 656 | ?, ?, |
|---|
| 657 | ) |
|---|
| 658 | }; |
|---|
| 659 | } |
|---|
| 660 | # my $sth = $db_handle->prepare( $sql ); |
|---|
| 661 | #DEBUGGING |
|---|
| 662 | my $sth; |
|---|
| 663 | |
|---|
| 664 | #for( @records ) { |
|---|
| 665 | eval { |
|---|
| 666 | |
|---|
| 667 | $sth->bind_param( 1, $display_filename, SQL_VARCHAR ); |
|---|
| 668 | $sth->bind_param( 2, $result_params[$params_NAME], SQL_VARCHAR ); |
|---|
| 669 | $sth->bind_param( 3, $result_params[$params_RAJ], SQL_VARCHAR ); |
|---|
| 670 | $sth->bind_param( 4, $result_params[$params_DECJ], SQL_VARCHAR ); |
|---|
| 671 | $sth->bind_param( 5, $data_type, SQL_VARCHAR ); |
|---|
| 672 | $sth->bind_param( 6, $result_params[$params_FREQ], SQL_NUMERIC ); |
|---|
| 673 | $sth->bind_param( 7, $result_params[$params_BW], SQL_NUMERIC ); |
|---|
| 674 | $sth->bind_param( 8, $result_params[$params_LENGTH], SQL_NUMERIC ); |
|---|
| 675 | $sth->bind_param( 9, $result_params[$params_DATE], SQL_NUMERIC ); |
|---|
| 676 | $sth->bind_param( 10, $result_params[$params_TIME], SQL_NUMERIC ); |
|---|
| 677 | $sth->bind_param( 11, $MJD, SQL_NUMERIC ); |
|---|
| 678 | $sth->bind_param( 12, $rajd, SQL_NUMERIC ); |
|---|
| 679 | $sth->bind_param( 13, $decjd, SQL_NUMERIC ); |
|---|
| 680 | $sth->bind_param( 14, $gl, SQL_NUMERIC ); |
|---|
| 681 | $sth->bind_param( 15, $gb, SQL_NUMERIC ); |
|---|
| 682 | $sth->bind_param( 16, $bmaj, SQL_NUMERIC ); |
|---|
| 683 | $sth->bind_param( 17, $bmin, SQL_NUMERIC ); |
|---|
| 684 | $sth->bind_param( 18, $bpa, SQL_NUMERIC ); |
|---|
| 685 | $sth->bind_param( 19, $result_params[$params_DM], SQL_NUMERIC ); |
|---|
| 686 | $sth->bind_param( 20, $result_params[$params_PERIOD], SQL_NUMERIC ); |
|---|
| 687 | $sth->bind_param( 21, $result_params[$params_NCHAN], SQL_INTEGER ); |
|---|
| 688 | $sth->bind_param( 22, $result_params[$params_NPOL], SQL_INTEGER ); |
|---|
| 689 | $sth->bind_param( 23, $result_params[$params_NBIN], SQL_INTEGER ); |
|---|
| 690 | $sth->bind_param( 24, $result_params[$params_NSUB], SQL_INTEGER ); |
|---|
| 691 | $sth->bind_param( 25, $result_params[$params_NBITS], SQL_INTEGER ); |
|---|
| 692 | $sth->bind_param( 26, $result_params[$params_TSAMP], SQL_INTEGER ); |
|---|
| 693 | $sth->bind_param( 27, $nbeam, SQL_INTEGER ); |
|---|
| 694 | $sth->bind_param( 28, $result_params[$params_CNFG], SQL_VARCHAR ); |
|---|
| 695 | $sth->bind_param( 29, $result_params[$params_INST], SQL_VARCHAR ); |
|---|
| 696 | $sth->bind_param( 30, $result_params[$params_RCVR], SQL_VARCHAR ); |
|---|
| 697 | $sth->bind_param( 31, $hdrver, SQL_VARCHAR ); |
|---|
| 698 | $sth->bind_param( 32, $survey, SQL_VARCHAR ); |
|---|
| 699 | $sth->bind_param( 33, $result_params[$params_TELESCOP], SQL_VARCHAR ); |
|---|
| 700 | $sth->bind_param( 34, $result_params[$params_SITE], SQL_INTEGER ); |
|---|
| 701 | $sth->bind_param( 35, $result_params[$params_OBSRVR], SQL_VARCHAR ); |
|---|
| 702 | $sth->bind_param( 13, $display_data_loc, SQL_VARCHAR ); |
|---|
| 703 | $sth->bind_param( 34, $file_size, SQL_INTEGER ); |
|---|
| 704 | |
|---|
| 705 | # $sth->execute(); |
|---|
| 706 | #DEBUGGING |
|---|
| 707 | DBI::dump_results($sth); |
|---|
| 708 | |
|---|
| 709 | # $db_handle->commit(); # Autocommit is already ON so no need for a manual commit |
|---|
| 710 | }; |
|---|
| 711 | |
|---|
| 712 | if( $@ ) { |
|---|
| 713 | print "warning statement\n"; |
|---|
| 714 | #DEBUGGING |
|---|
| 715 | # warn "Database error: $DBI::errstr\n"; |
|---|
| 716 | # $db_handle->rollback(); #just die if rollback is failing |
|---|
| 717 | } |
|---|
| 718 | |
|---|
| 719 | #} |
|---|
| 720 | |
|---|
| 721 | #DEBUGGING |
|---|
| 722 | # $sth->finish(); |
|---|
| 723 | } |
|---|
| 724 | |
|---|
| 725 | # Removes |
|---|
| 726 | sub deleteStaleRecords { |
|---|
| 727 | |
|---|
| 728 | my $sql, my $sth, my @data, my $file; |
|---|
| 729 | |
|---|
| 730 | $sql = qq{ SELECT data_loc, filename from cals }; |
|---|
| 731 | $sth = $db_handle->prepare( $sql ); |
|---|
| 732 | |
|---|
| 733 | eval { |
|---|
| 734 | |
|---|
| 735 | $sth->execute(); |
|---|
| 736 | |
|---|
| 737 | }; |
|---|
| 738 | |
|---|
| 739 | if( $@ ) { |
|---|
| 740 | warn "Database error: $DBI::errstr\n"; |
|---|
| 741 | $db_handle->rollback(); #just die if rollback is failing |
|---|
| 742 | } |
|---|
| 743 | |
|---|
| 744 | my @cals2delete; |
|---|
| 745 | |
|---|
| 746 | while (@data = $sth->fetchrow_array()) { |
|---|
| 747 | |
|---|
| 748 | if (scalar(@data) == 2) { |
|---|
| 749 | $file = $data[0]."/".$data[1]; |
|---|
| 750 | |
|---|
| 751 | if (!-e $file) { |
|---|
| 752 | print "$file does NOT exist\n\n"; |
|---|
| 753 | |
|---|
| 754 | # Push the row into the cals to delete |
|---|
| 755 | # So array will look like: |
|---|
| 756 | # {/full/path/to/dir, filename.cf, /full/path/to/dir, filename2.cf, etc..} |
|---|
| 757 | push (@cals2delete, @data); |
|---|
| 758 | } |
|---|
| 759 | } |
|---|
| 760 | |
|---|
| 761 | } |
|---|
| 762 | |
|---|
| 763 | $sql = qq{ SELECT data_loc, filename from observations }; |
|---|
| 764 | |
|---|
| 765 | $sth = $db_handle->prepare( $sql ); |
|---|
| 766 | eval { |
|---|
| 767 | |
|---|
| 768 | $sth->execute(); |
|---|
| 769 | |
|---|
| 770 | }; |
|---|
| 771 | |
|---|
| 772 | if( $@ ) { |
|---|
| 773 | warn "Database error: $DBI::errstr\n"; |
|---|
| 774 | $db_handle->rollback(); #just die if rollback is failing |
|---|
| 775 | } |
|---|
| 776 | |
|---|
| 777 | my @obs2delete; |
|---|
| 778 | |
|---|
| 779 | while (@data = $sth->fetchrow_array()) { |
|---|
| 780 | |
|---|
| 781 | if (scalar(@data) == 2) { |
|---|
| 782 | $file = $data[0]."/".$data[1]; |
|---|
| 783 | |
|---|
| 784 | if (!-e $file) { |
|---|
| 785 | print "$file does NOT exist\n\n"; |
|---|
| 786 | |
|---|
| 787 | # Push the row into the cals to delete |
|---|
| 788 | # So array will look like: |
|---|
| 789 | # {/full/path/to/dir, filename.cf, /full/path/to/dir, filename2.cf, etc..} |
|---|
| 790 | push (@obs2delete, @data); |
|---|
| 791 | } |
|---|
| 792 | } |
|---|
| 793 | } |
|---|
| 794 | |
|---|
| 795 | print "There are " . scalar(@cals2delete)/2 . " stale cals (" . |
|---|
| 796 | scalar(@cals2delete) . "/2) and " . |
|---|
| 797 | scalar(@obs2delete)/2 . " stale obs to delete\n"; |
|---|
| 798 | |
|---|
| 799 | # Delete the rows |
|---|
| 800 | my $data_loc, my $filename; |
|---|
| 801 | |
|---|
| 802 | while (@cals2delete) { |
|---|
| 803 | $data_loc = shift @cals2delete; |
|---|
| 804 | $filename = shift @cals2delete; |
|---|
| 805 | |
|---|
| 806 | $sql = qq{ DELETE FROM cals WHERE data_loc = ? AND filename = ?}; |
|---|
| 807 | |
|---|
| 808 | $sth = $db_handle->prepare( $sql ); |
|---|
| 809 | eval { |
|---|
| 810 | |
|---|
| 811 | $sth->bind_param( 1, $data_loc, SQL_VARCHAR ); |
|---|
| 812 | $sth->bind_param( 2, $filename, SQL_VARCHAR ); |
|---|
| 813 | $sth->execute(); |
|---|
| 814 | |
|---|
| 815 | }; |
|---|
| 816 | |
|---|
| 817 | if( $@ ) { |
|---|
| 818 | warn "Database error: $DBI::errstr\n"; |
|---|
| 819 | $db_handle->rollback(); #just die if rollback is failing |
|---|
| 820 | } |
|---|
| 821 | else { |
|---|
| 822 | print "Successfully deleted $data_loc/$filename\n"; |
|---|
| 823 | } |
|---|
| 824 | |
|---|
| 825 | } |
|---|
| 826 | |
|---|
| 827 | while (@obs2delete) { |
|---|
| 828 | $data_loc = shift @obs2delete; |
|---|
| 829 | $filename = shift @obs2delete; |
|---|
| 830 | |
|---|
| 831 | $sql = qq{ DELETE FROM observations WHERE data_loc = ? AND filename = ?}; |
|---|
| 832 | |
|---|
| 833 | $sth = $db_handle->prepare( $sql ); |
|---|
| 834 | |
|---|
| 835 | eval { |
|---|
| 836 | |
|---|
| 837 | $sth->bind_param( 1, $data_loc, SQL_VARCHAR ); |
|---|
| 838 | $sth->bind_param( 2, $filename, SQL_VARCHAR ); |
|---|
| 839 | $sth->execute(); |
|---|
| 840 | |
|---|
| 841 | }; |
|---|
| 842 | |
|---|
| 843 | if( $@ ) { |
|---|
| 844 | warn "Database error: $DBI::errstr\n"; |
|---|
| 845 | $db_handle->rollback(); #just die if rollback is failing |
|---|
| 846 | } |
|---|
| 847 | else { |
|---|
| 848 | print "Successfully deleted $data_loc/$filename\n"; |
|---|
| 849 | } |
|---|
| 850 | } |
|---|
| 851 | |
|---|
| 852 | $sth->finish(); |
|---|
| 853 | |
|---|
| 854 | } |
|---|
| 855 | |
|---|
| 856 | sub connectdb { |
|---|
| 857 | |
|---|
| 858 | # Connect to the MySQL server |
|---|
| 859 | # my $dbh = DBI->connect("dbi:mysql:database=psrchive;host=localhost", "psrdba", "lighthouse") |
|---|
| 860 | # or die "Couldn't connect to database: $DBI::errstr\n"; |
|---|
| 861 | |
|---|
| 862 | # print "Successfully connected to db\n" if $dbh; |
|---|
| 863 | |
|---|
| 864 | # DEBUGGING |
|---|
| 865 | print "We would normally connect to the DB here\n"; |
|---|
| 866 | return 1; |
|---|
| 867 | # return $dbh; |
|---|
| 868 | } |
|---|
