Changeset 3847 for TI01-discovery

Timestamp:

02/05/08 15:17:46 (11 years ago)

Author:

cbyrom

Message:

Create new object, SpatioTemporalData?, to wrapper temporal and spatial data
+ change PostgresRecord? to include all spatiotemporal data from a moles file
using this new object.

Location:

TI01-discovery/branches/ingestAutomation-upgrade/OAIBatch

Files:

• PostgresRecord.py (modified) (4 diffs)
• SpatioTemporalData.py (added)

TI01-discovery/branches/ingestAutomation-upgrade/OAIBatch/PostgresRecord.py

@@ -19 +19 @@
 from ndgUtils.ndgObject import ndgObject
 from FileUtilities import FileUtilities
+from SpatioTemporalData import SpatioTemporalData
 
 class PostgresRecord:
@@ -269 +270 @@
         Extract spatio temporal data from the original document
         '''
-        #this is a fix to the  ElementTree namespace problem that namespaces are usually 
-        # represented as ns0, ns1, ns2 etc.
-        #cElementTree._namespace_map.update({'http://ndg.nerc.ac.uk/moles': 'moles', 'http://www.w3.org/1999/xlink':'xlink'})
-        self.east = 'null'
-        self.west = []
-        self.north = 'null'
-        self.south = 'null'
-        self.startdate='nostartdate'
-        self.enddate='noenddate'
+        # initialise the various spatiotemporal arrays used to extract data to
+        self.stData = SpatioTemporalData()
         
         molesFile = self._molesDir + self._shortFilename
-        
+        logging.info('Retrieving spatiotemporal info from moles file, %s' %molesFile)
+        
+        # load in the moles file and put this into an object for direct access to the xml elements
         dgMeta=MRW.dgMetadata()
         try:
@@ -286 +282 @@
         except Exception, detail:
             raise SystemError, 'Cannot parse the XML moles document %s. Detail:\n%s' %(molesFile, detail)
-        
-        bbox_list = []
+
+        # do quick checks to see if the relevant data exists
+        if not dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary:
+            logging.info("No data summary elements found - assuming no spatiotemporal data available")
+            return
+        
+        if not dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage:
+            logging.info("No data coverage elements found - assuming no spatiotemporal data available")
+            return
+        
+        if not dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage.dgSpatialCoverage:
+            logging.info("No spatial coverage elements found - assuming no spatial data available")
+        else:
+            self.getCoordData(dgMeta)
+
+        if not dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage.dgTemporalCoverage:
+            logging.info("No temporal coverage elements found - assuming no temporal data available")
+        else:
+            self.getTimeRangeData(dgMeta)
+
+
+    def getTimeRangeData(self, dgMeta):
+        '''
+        Parse an xml tree and add any time range data found
+        @param dgMeta: xml fragment for the time range
+        '''
+        logging.info("Extracting time range info")
         try:
-            logging.info("Extracting bounding box info")
-            bbox_list=self.listify(dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage.dgSpatialCoverage.BoundingBox)
+            dates = dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage.dgTemporalCoverage.DateRange
+            
+            if not dates:
+                logging.info("No temporal info found for document")
+                
+            dates_list = self.listify(dates)
+            for date in dates_list:
+                startdate=date.DateRangeStart
+                enddate= date.DateRangeEnd
+                if startdate==None or startdate=='None':
+                    startdate="nostartdate"
+                if enddate==None or enddate=='None':
+                    enddate="noenddate"
+                    
+                self.stData.addTimeRange(startdate, enddate)
+                logging.info("Temporal info: startdate " + \
+                             startdate + ", enddate " + enddate) 
+        except Exception, detail:
+            logging.info("Document does not contain temporal info.")
+            logging.info(detail)
+
+        
+    def getCoordData(self, dgMeta):
+        '''
+        Parse an xml tree and add any coord data found
+        @param dgMeta: xml fragment for the bounding boxes
+        '''
+        logging.info("Extracting bounding box info")
+        try:
+
+            bboxes = dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage.dgSpatialCoverage.BoundingBox
+            
+            if not bboxes:
+                logging.info("No bounding box info found for document")
+                return
+                
+            bbox_list=self.listify(bboxes)
             #parse the list of coordinates
             for bbox in bbox_list:
-                self.west.append(self.parseCoord(bbox.LimitWest, 'W', 'E'))
-                self.east = self.parseCoord(bbox.LimitEast, 'W', 'E')
-                self.north = self.parseCoord(bbox.LimitNorth, 'S', 'N')
-                self.south = self.parseCoord(bbox.LimitSouth, 'S', 'N')
+                north = self.parseCoord(bbox.LimitNorth, 'S', 'N')
+                south = self.parseCoord(bbox.LimitSouth, 'S', 'N')
+                east = self.parseCoord(bbox.LimitEast, 'W', 'E')
+                west = self.parseCoord(bbox.LimitWest, 'W', 'E')
+                self.stData.addCoords(north, south, east, west)
+                logging.info("Spatial info: west= " + west + ",south " + south + ", east " + \
+                    east + ", north " + north + "")
+                
         except Exception, detail:
-            logging.info("XML moles document " + molesFile + " does not contain a bounding box.")
-            logging.debug(detail)
-
-        try:
-            dates=dgMeta.dgMetadataRecord.dgDataEntity.dgDataSummary.dgDataCoverage.dgTemporalCoverage.DateRange
-            print "startdate = %s" %dates.DateRangeStart
-            print "enddate = %s" %dates.DateRangeEnd
-        except:
-            logging.info("XML moles document " + molesFile + " does not contain temporal info.")
-            no_dates=True
-
-        if no_bbox and no_dates:
-            logging.info("XML moles document " + molesFile + " does not contain any spatiotemporal info.")
-            return
-
-        if not no_dates:
-            startdate=dates.DateRangeStart
-            enddate= dates.DateRangeEnd
-            if startdate==None or startdate=='None':
-                startdate="nostartdate"
-            if enddate==None or enddate=='None':
-                enddate="noenddate"
-            self.startdate = startdate
-            self.enddate = enddate
-
-
-        logging.info("Spatial info: west= " + self.west + ",south " + self.south + ", east " + \
-                    self.east + ", north " + self.north + "")
-        logging.info("Temporal info: startdate " + self.startdate + ", enddate " + self.enddate) 
-
+            logging.warning("Problem encountered whilst parsing bounding box info - this may lead \n" + \
+                            "to an incomplete set of metadata being ingested. \nDetail: %s" %detail)
 
 
@@ -339 +371 @@
         @return: coord - the value of the coordinate as a string   
         '''
-        logging.info("Parsing document coordinates")
+        logging.debug("Parsing document coordinates")
         try:
             coord = coordValue.strip()