Changeset 35384 for trunk/pstamp/scripts/pstampparse.pl

Timestamp:

Apr 9, 2013, 2:08:27 PM (13 years ago)

Author:

bills

Message:

beginning rework of the postage stamp parser

File:

• trunk/pstamp/scripts/pstampparse.pl (modified) (14 diffs)

trunk/pstamp/scripts/pstampparse.pl

@@ -211 +211 @@
 print "\n$nRows rows read from request file\n";
 
-# If this is a big request and it came from the upload page and doesn't have a specific label assigned
-# change it to the generic one that has lower with lower priority
-
-#
-# Loop over rows in the request file collecting consecutive rows that have the "same images of interest"
-# in the sense that their selection parameters will yield the same target "Runs".
-# Process the groups of rows together to reduce lookup time and to potentially make multiple
-# stamps from the same ppstamp process.
-
-my @rowList;
+
 my $num_jobs = 0;
 my $imageList;
 my $stage;
-my $need_magic;
 foreach my $row (@$rows) {
+
     if ($label eq 'WEB.UP' and ($nRows > 500 or $num_jobs > 500) and $req_id and !$no_update) {
+        # this is a big request and it came from the upload page and doesn't have a specific label assigned
+        # change it to the generic one that has lower with lower priority
         $label = 'WEB.BIG';
         print "\nChanging label for big WEB.UP request to $label\n";
@@ -237 +230 @@
         }
     }
-    # sanity check the paramaters
+
+    # validate the paramaters
     if (!checkRow($row)) {
         # when it enconters an error checkRow adds a fake job with an appropriate error code to the database
@@ -247 +241 @@
     $row->{error_code} = 0;
 
-    if (scalar @rowList == 0) {
-        push @rowList, $row;
-        next;
-    }
-
-    my $firstRow = $rowList[0];
-    if (same_images_of_interest($firstRow, $row)) {
-        # add this row to the list and move on
-        push @rowList, $row;
-        next;
-    }
-
-    # the images of interest for this new row doesn't match the list. 
-    # process the list ...
-    $num_jobs += processRows($action, \@rowList);
-
-    # and reset the list to contain just the new row
-    @rowList = ($row);
-}
-
-# out of rows process the list
-if (scalar @rowList > 0) {
-    $num_jobs += processRows($action, \@rowList);
+    $num_jobs += processRow($action, $row);
 }
 
@@ -283 +255 @@
     my $row = shift;
 
+    # check validity of the paramters in a request specification
     # If we encounter an error for a particular row add a job with the proper fault code.
+    # also adjust some paramterers like filter and set defaults for some others
 
     my $stage = $row->{IMG_TYPE};
@@ -315 +289 @@
     }
 
-
     my $component = $row->{COMPONENT};
     if (!defined $component or (lc($component) eq "null") or (lc($component) eq "all")) {
@@ -345 +318 @@
         return 0;
     }
+
+    my $fwhm_min = $row->{FWHM_MIN};
+    if (!defined $fwhm_min) {
+        $row->{FWHM_MIN} = 0;
+    } elsif (!validNumber($fwhm_min)) {
+        print STDERR "$fwhm_min is not a valid FWHM_MIN\n";
+        insertFakeJobForRow($row, 1, $PSTAMP_INVALID_REQUEST);
+        return 0;
+    }
+    my $fwhm_max = $row->{FWHM_MAX};
+    if (!defined $fwhm_max) {
+        $row->{FWHM_MAX} = 0;
+    } elsif (!validNumber($fwhm_max)) {
+        print STDERR "$fwhm_max is not a valid FWHM_MAX\n";
+        insertFakeJobForRow($row, 1, $PSTAMP_INVALID_REQUEST);
+        return 0;
+    }
+
     my $data_group = $row->{DATA_GROUP};
     if (!defined $data_group) {
@@ -410 +401 @@
     }
 
+
     return 1;
 }
@@ -418 +410 @@
 
     $num_jobs = 1;
+    return $num_jobs;
+}
+
+sub processRow {
+    my $action  = shift;
+    my $row     = shift;
+
+    my $num_jobs = 0;
+
+    my $project  = $row->{PROJECT};
+
+    # note: resolve_project avoids running pstamptool every time by remembering the
+    # last project resolved
+    my $proj_hash = resolve_project($ipprc, $project, $dbname, $dbserver);
+    if (!$proj_hash) {
+        insertFakeJobForRow($row, 1, $PSTAMP_UNKNOWN_PROJECT);
+        $num_jobs++;
+        return $num_jobs;
+    }
+    my $image_db   = $proj_hash->{dbname};
+    my $camera     = $proj_hash->{camera};
+
+    my $req_type    = $row->{REQ_TYPE};
+    my $rownum      = $row->{ROWNUM};
+
+    # Since user can get unmagicked data "by coordinate" requests can go back in time
+    # to dredge unusable data from the "dark days"...
+    if ($req_type eq 'bycoord' and $row->{MJD_MIN} == 0) {
+        # ... so unless the user sets mjd_min clamp it to 2009-04-01
+        # XXX: This value should live in the pstampProject table not be hardcoded here
+        $row->{MJD_MIN} = 54922;
+    }
+
+    if (($req_type eq 'byskycell') or ($req_type eq 'bycoord')) {
+        # avoid error from print below if $id isn't needed
+        $row->{ID} = "" if !$row->{ID};
+    }
+    
+    # Call PS::IPP::PStamp::Job locate_images subroutine to get the images for this
+    # request specification. An array reference is returned.
+    my $start_locate = gettimeofday();
+
+    print "\nCalling new_locate_images for row: $rownum\n";
+
+    $imageList = new_locate_images($ipprc, $image_db, $camera, $row, $verbose);
+
+    my $dtime_locate = gettimeofday() - $start_locate;
+    print "Time to locate_images for row $rownum $dtime_locate\n";
+
+    my @rowList = ($row);
+    $num_jobs += queueJobs($action, \@rowList, $imageList);
+
+    # if this row slipped through without a job being added add one. 
+    if ($row->{job_num} == 0) {
+        print "row $row->{ROWNUM} produced no jobs\n";
+        print STDERR "row $row->{ROWNUM} produced no jobs\n";
+        my $error_code = $row->{error_code};
+        $error_code =  $PSTAMP_NO_IMAGE_MATCH if !$error_code;
+        insertFakeJobForRow($row, ++$row->{job_num}, $error_code);
+    }
+
     return $num_jobs;
 }
@@ -447 +500 @@
     }
     my $req_type  = $row->{REQ_TYPE};
-    $stage        = $row->{IMG_TYPE};
+    my $stage     = $row->{IMG_TYPE};
     my $id        = $row->{ID};
     my $component = $row->{COMPONENT};
@@ -463 +516 @@
     my $image_db   = $proj_hash->{dbname};
     my $camera     = $proj_hash->{camera};
-    $need_magic    = $proj_hash->{need_magic};
+    my $need_magic    = $proj_hash->{need_magic};
     # Since user can get unmagicked data "by coordinate" requests can go back in time
     # to dredge unusable data from the "dark days"...
@@ -523 +576 @@
     # information required is contained there
 
-    $imageList = locate_images($ipprc, $image_db, \@rowList, $req_type, $stage, $id, $tess_id, $component,
+    $imageList = locate_images($ipprc, $image_db, $rowList, $req_type, $stage, $id, $tess_id, $component,
                 $option_mask, $need_magic, $mjd_min, $mjd_max, $filter, $data_group, $verbose);
 
@@ -533 +586 @@
     $row->{need_magic} = $need_magic;
 
-    $num_jobs += queueJobs($action, \@rowList, $imageList);
+    $num_jobs += queueJobs($action, $rowList, $imageList);
 
     # if a row slipped through with no jobs add a faulted one
-    foreach my $row (@rowList) {
+    foreach my $row (@$rowList) {
         if ($row->{job_num} == 0) {
             print "row $row->{ROWNUM} produced no jobs\n";
@@ -675 +728 @@
     my $imageList = shift;
 
-    my $firstRow = $rowList[0];
+    my $firstRow = $rowList->[0];
     my $stage    = $firstRow->{IMG_TYPE};
     my $job_type = $firstRow->{JOB_TYPE};
@@ -1004 +1057 @@
     my $r_fault = shift;
     my $r_dep_id = shift;
+
+    my $need_magic = 0;
 
     if ($stage ne 'raw') {