Project

General

Profile

« Previous | Next » 

Revision 32

Added by Aaron Marcuse-Kubitza about 13 years ago

data2xml: Added support for * abbrs for backward (child-to-parent) pointers

View differences:

scripts/data2xml/NYBG-VegBank-VegX_mapping.csv
4 4 
"InstitutionCode","/taxonInterpretation/museum_ID->/party/OrganizationName/_name/firstName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept/partyWithRole/*ID->/parties/party/organizationName/_name/firstName"
5 5 
"CollectionCode","/taxonInterpretation/museum_ID->/party/OrganizationName/_name/lastName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept/partyWithRole/*ID->/parties/party/organizationName/_name/lastName"
6 6 
"CatalogNumber","/taxonInterpretation/museumAccessionNumber","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept/voucher"
7 
"ScientificName","/plantName/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonName/Simple"
7 
"ScientificName","/plantName/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonName/Simple"
8 8 
"BasisOfRecord",,
9 
"Kingdom","/plantName[plantStatus/plantLevel=Kingdom]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=reg]/Name"
10 
"Phylum","/plantName[plantStatus/plantLevel=Subkingdom]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=phyl_div]/Name"
11 
"Class","/plantName[plantStatus/plantLevel=Class]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=cl]/Name"
12 
"Order","/plantName[plantStatus/plantLevel=Order]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=ord]/Name"
13 
"Family","/plantName[plantStatus/plantLevel=Family]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=fam]/Name"
14 
"Genus","/plantName[plantStatus/plantLevel=Genus]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=gen]/Name"
15 
"Species","/plantName[plantStatus/plantLevel=Species]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=sp]/Name"
16 
"Subspecies","/plantName[plantStatus/plantLevel=Subspecies]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=ssp]/Name"
17 
"ScientificNameAuthor","/plantConcept/reference_ID->/referenceParty/{givenName,surname}","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept/AccordingTo/Simple"
18 
"IdentifiedBy","/taxonInterpretation/PARTY_ID->/party/{givenName,middleName,surName}","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/partyWithRole[role=identifier]/*ID->/parties/party/individualName/{givenName,surName}"
19 
"YearIdentified","/taxonInterpretation/interpretationDate/_date/year","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/date/_date/year"
20 
"MonthIdentified","/taxonInterpretation/interpretationDate/_date/month","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/date/_date/month"
21 
"DayIdentified","/taxonInterpretation/interpretationDate/_date/day","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/date/_date/day"
9 
"Kingdom","/plantName[plantStatus/plantLevel=Kingdom]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=reg]/Name"
10 
"Phylum","/plantName[plantStatus/plantLevel=Subkingdom]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=phyl_div]/Name"
11 
"Class","/plantName[plantStatus/plantLevel=Class]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=cl]/Name"
12 
"Order","/plantName[plantStatus/plantLevel=Order]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=ord]/Name"
13 
"Family","/plantName[plantStatus/plantLevel=Family]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=fam]/Name"
14 
"Genus","/plantName[plantStatus/plantLevel=Genus]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=gen]/Name"
15 
"Species","/plantName[plantStatus/plantLevel=Species]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=sp]/Name"
16 
"Subspecies","/plantName[plantStatus/plantLevel=Subspecies]/plantName","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept[Rank/@code=ssp]/Name"
17 
"ScientificNameAuthor","/plantConcept/reference_ID->/referenceParty/{givenName,surname}","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/*s/taxonRelationshipAssertion!/assertion/*ID->/*s/taxonConcept/AccordingTo/Simple"
18 
"IdentifiedBy","/taxonInterpretation/PARTY_ID->/party/{givenName,middleName,surName}","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/partyWithRole[role=identifier]/*ID->/parties/party/individualName/{givenName,surName}"
19 
"YearIdentified","/taxonInterpretation/interpretationDate/_date/year","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/date/_date/year"
20 
"MonthIdentified","/taxonInterpretation/interpretationDate/_date/month","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/date/_date/month"
21 
"DayIdentified","/taxonInterpretation/interpretationDate/_date/day","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/date/_date/day"
22 22 
"TypeStatus",,
23 23 
"CollectorNumber",,
24 24 
"FieldNumber","/taxonInterpretation/collectionNumber","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept/authorCode"
25 
"Collector","/taxonInterpretation/collector_ID->/party/{givenName,middleName,surName}","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/partyWithRole[role=collector]/*ID->/parties/party/individualName/{givenName,surName}"
25 
"Collector","/taxonInterpretation/collector_ID->/party/{givenName,middleName,surName}","/*s/individualOrganismObservation/*ID->/*s/individualOrganism/*sID->/*s/taxonNameUsageConcept->/*s/taxonDetermination[*ID]/partyWithRole[role=collector]/*ID->/parties/party/individualName/{givenName,surName}"
26 26 
"YearCollected","/taxonInterpretation/collectionDate/_date/year","/*s/individualOrganismObservation/*ID->/*s/plotObservation/obsStartDate/_date/year"
27 27 
"MonthCollected","/taxonInterpretation/collectionDate/_date/month","/*s/individualOrganismObservation/*ID->/*s/plotObservation/obsStartDate/_date/month"
28 28 
"DayCollected","/taxonInterpretation/collectionDate/_date/day","/*s/individualOrganismObservation/*ID->/*s/plotObservation/obsStartDate/_date/day"
scripts/data2xml/xpath.py
31 31 

  		




  32
  32
  
    
def set_value(path, value): path[-1].value = value


  		




  33
  33
  
    

  		




  
  34
  
    
def backward_id(elem):


  		




  
  35
  
    
    if len(elem.attrs) >= 1 and value(elem.attrs[0]) == None:


  		




  
  36
  
    
        return elem.attrs[0]


  		




  
  37
  
    
    else: return None


  		




  
  38
  
    

  		




  34
  39
  
    
class XpathParser(Parser):


  		




  35
  40
  
    
    def _main(self):


  		




  36
  41
  
    
        self._match_str('/', required=True)


  		




  ......




  50
  55
  
    
            tree.append(elem)


  		




  51
  56
  
    
            if not self._match_str('/'): break


  		




  52
  57
  
    
            elem_idx += 1


  		




  
  58
  
    
        


  		




  
  59
  
    
        # Expand * abbrs


  		




  
  60
  
    
        elem_idx = 0


  		




  
  61
  
    
        for elem in tree:


  		




  
  62
  
    
            id_ = backward_id(elem)


  		




  
  63
  
    
            if id_ != None: elem = id_[0]; offset = -2


  		




  
  64
  
    
            elif elem.is_ptr: offset = 2


  		




  
  65
  
    
            else: offset = 1


  		




  
  66
  
    
            before, abbr, after = elem.name.partition('*')


  		




  
  67
  
    
            if abbr != '':


  		




  
  68
  
    
                try: elem.name = before+tree[elem_idx+offset].name+after


  		




  
  69
  
    
                except IndexError: pass # no replacement elem


  		




  
  70
  
    
            elem_idx += 1


  		




  
  71
  
    
        


  		




  53
  72
  
    
        # Add lookahead assertion for rest of path


  		




  54
  73
  
    
        if fork_idx != None: tree[fork_idx].attrs.append(tree[fork_idx+1:])


  		




  
  74
  
    
        


  		




  55
  75
  
    
        return tree


  		




  56
  76
  
    
    


  		




  57
  77
  
    
    def _fields(self):


  		




  ......




  78
  98
  
    
    def _field(self):


  		




  79
  99
  
    
        return self._name()


  		




  80
  100
  
    
    


  		




  81
  
  
    
    def _name(self): return self._match_re(r'[\w.]+', required=True)


  		




  
  101
  
    
    def _name(self): return self._match_re(r'[\w.*]+', required=True)


  		




  82
  102
  
    
    


  		




  83
  103
  
    
    def _value(self): return self._match_re(r'[\w.|]+', required=True)


  		




  84
  104
  
    

  		




  ......




  95
  115
  
    
    path[id_level].attrs.append([XpathElem('id', id_, is_attr=True)])


  		




  96
  116
  
    

  		




  97
  117
  
    
def get(doc, path, create=False, last_only=None, parent=None):


  		




  98
  
  
    
    # Warning: The last_only optimization may split data between multiple nodes


  		




  
  118
  
    
    # Warning: The last_only optimization may put data that should be together


  		




  
  119
  
    
    # into separate nodes


  		




  99
  120
  
    
    if parent == None: parent = doc.documentElement


  		




  100
  121
  
    
    if last_only == None: last_only = create


  		




  101
  122
  
    
    elem_idx = 0


  		




  ......




  130
  151
  
    
        # Follow pointer


  		




  131
  152
  
    
        if elem.is_ptr:


  		




  132
  153
  
    
            path = deepcopy(path[elem_idx+1:]) # rest of path


  		




  133
  
  
    
            attrs = path[instance_level].attrs


  		




  134
  
  
    
            if len(attrs) >= 1 and value(attrs[0]) == None:


  		




  
  154
  
    
            id_elem = backward_id(path[instance_level])


  		




  
  155
  
    
            if id_elem != None:


  		




  135
  156
  
    
                # backward (child-to-parent) pointer with target ID attr


  		




  136
  
  
    
                set_value(attrs[0], xml_util.get_id(node))


  		




  
  157
  
    
                set_value(id_elem, xml_util.get_id(node))


  		




  137
  158
  
    
            else: # forward (parent-to-child) pointer


  		




  138
  159
  
    
                id_ = xml_util.value(node)


  		




  139
  160
  
    
                obj_path = obj(path) # target object


  		












  

    
      scripts/data2xml/data2xml
    
  





  29
  29
  
    
        name = row[0]


  		




  30
  30
  
    
        path = row[dest_idx]


  		




  31
  31
  
    
        if name != '' and path != '':


  		




  32
  
  
    
            if path.startswith('?'): path = path[1:]


  		




  33
  32
  
    
            if path.startswith('/*s/'): has_types = True # *s used for type elem


  		




  34
  33
  
    
            path = path.replace('<name>', name)


  		




  35
  
  
    
            path = re.sub(r'\*(?=\w*(?:->/[^/]+)?/(\w+))', r'\1', path)


  		




  36
  34
  
    
            mappings[name] = xpath.XpathParser(path).parse()


  		




  37
  35
  
    
    stream.close()


  		




  38
  36
  
    
    


  		












Also available in:  Unified diff