Project

General

Profile

« Previous | Next » 

Revision 25

Added by Aaron Marcuse-Kubitza about 13 years ago

In data2xml, added shortcut for lookahead assertion using ! symbol

View differences:

scripts/data2xml/NYBG-VegBank-VegX_mapping.csv
4 4 
"InstitutionCode","/taxonInterpretation/museum_ID->/party/OrganizationName/_name/firstName","/*s/taxonNameUsageConcept/partyWithRole/partyID->/parties/party/organizationName/_name/firstName"
5 5 
"CollectionCode","/taxonInterpretation/museum_ID->/party/OrganizationName/_name/lastName","/*s/taxonNameUsageConcept/partyWithRole/partyID->/parties/party/organizationName/_name/lastName"
6 6 
"CatalogNumber","/taxonInterpretation/museumAccessionNumber","/*s/taxonNameUsageConcept/voucher"
7 
"ScientificName","/plantName/plantName","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion/assertion/taxonNameID->/*s/taxonName/Simple"
7 
"ScientificName","/plantName/plantName","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/taxonNameID->/*s/taxonName/Simple"
8 8 
"BasisOfRecord",,
9 
"Kingdom","/plantName[plantStatus/plantLevel=Kingdom]/plantName","/*s/taxonConcept[Rank/@code=reg]/Name"
10 
"Phylum","/plantName[plantStatus/plantLevel=Subkingdom]/plantName","/*s/taxonConcept[Rank/@code=phyl_div]/Name"
11 
"Class","/plantName[plantStatus/plantLevel=Class]/plantName","/*s/taxonConcept[Rank/@code=cl]/Name"
12 
"Order","/plantName[plantStatus/plantLevel=Order]/plantName","/*s/taxonConcept[Rank/@code=ord]/Name"
13 
"Family","/plantName[plantStatus/plantLevel=Family]/plantName","/*s/taxonConcept[Rank/@code=fam]/Name"
14 
"Genus","/plantName[plantStatus/plantLevel=Genus]/plantName","/*s/taxonConcept[Rank/@code=gen]/Name"
15 
"Species","/plantName[plantStatus/plantLevel=Species]/plantName","/*s/taxonConcept[Rank/@code=sp]/Name"
16 
"Subspecies","/plantName[plantStatus/plantLevel=Subspecies]/plantName","/*s/taxonConcept[Rank/@code=ssp]/Name"
17 
"ScientificNameAuthor","/plantConcept/reference_ID->/referenceParty/{givenName,surname}","/*s/taxonConcept/AccordingTo/Simple"
9 
"Kingdom","/plantName[plantStatus/plantLevel=Kingdom]/plantName","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/taxonConceptID->/*s/taxonConcept[Rank/@code=reg]/Name"
10 
"Phylum","/plantName[plantStatus/plantLevel=Subkingdom]/plantName","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/taxonConceptID->/*s/taxonConcept[Rank/@code=phyl_div]/Name"
11 
"Class","/plantName[plantStatus/plantLevel=Class]/plantName","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/taxonConceptID->/*s/taxonConcept[Rank/@code=cl]/Name"
12 
"Order","/plantName[plantStatus/plantLevel=Order]/plantName","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/taxonConceptID->/*s/taxonConcept[Rank/@code=ord]/Name"
13 
"Family","/plantName[plantStatus/plantLevel=Family]/plantName","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/taxonConceptID->/*s/taxonConcept[Rank/@code=fam]/Name"
14 
"Genus","/plantName[plantStatus/plantLevel=Genus]/plantName","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/taxonConceptID->/*s/taxonConcept[Rank/@code=gen]/Name"
15 
"Species","/plantName[plantStatus/plantLevel=Species]/plantName","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/taxonConceptID->/*s/taxonConcept[Rank/@code=sp]/Name"
16 
"Subspecies","/plantName[plantStatus/plantLevel=Subspecies]/plantName","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/taxonConceptID->/*s/taxonConcept[Rank/@code=ssp]/Name"
17 
"ScientificNameAuthor","/plantConcept/reference_ID->/referenceParty/{givenName,surname}","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/*s/taxonRelationshipAssertion!/assertion/taxonConceptID->/*s/taxonConcept/AccordingTo/Simple"
18 18 
"IdentifiedBy","/taxonInterpretation/PARTY_ID->/party/{givenName,middleName,surName}","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/partyWithRole[role=identifier]/partyID->/parties/party/individualName/{givenName,surName}"
19 19 
"YearIdentified","/taxonInterpretation/interpretationDate/_date/year","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/date/_date/year"
20 20 
"MonthIdentified","/taxonInterpretation/interpretationDate/_date/month","/*s/taxonNameUsageConcept->/*s/taxonDetermination[taxonNameUsageConceptID]/date/_date/month"
......
41 41 
"MaximumElevation","/plot/elevation/max,plot/elevationRange/to","/*s/plot/geospatial/maximumElevationInMeters"
42 42 
"MinimumDepth","/observation/waterDepth/min","/*s/plot/geospatial/minimumDepthInMeters"
43 43 
"MaximumDepth","/observation/waterDepth/max","/*s/plot/geospatial/maximumDepthInMeters"
44 
"Sex","/userDefined[tableName=taxonInterpretation,userDefinedName=<name>]/definedValue ","/*s/individualOrganismObservation/simpleUserdefined[name=sex]/value"
44 
"Sex","/userDefined[tableName=taxonInterpretation,userDefinedName=sex]/definedValue ","/*s/individualOrganismObservation/simpleUserdefined[name=sex]/value"
45 45 
"PreparationType",,
46 46 
"IndividualCount",,
47 47 
"PreviousCatalogNumber",,
48 48 
"RelationshipType",,
49 49 
"RelatedCatalogItem",,
50 50 
"Notes","/taxonInterpretation/notes","/*s/taxonNameUsageConcept/note/text"
51 
"Habitat","/userDefined[tableName=stemLocation,userDefinedName=<name>]/definedValue ","/*s/individualOrganismObservation/simpleUserdefined[name=habitat]/value"
52 
"PlantFungusDescription","/userDefined[tableName=stemLocation,userDefinedName=<name>]/definedValue ","/*s/individualOrganismObservation/simpleUserdefined[name=plantFungusDescription]/value"
53 
"Substrate","/userDefined[tableName=stemLocation,userDefinedName=<name>]/definedValue ","/*s/individualOrganismObservation/simpleUserdefined[name=substrate]/value"
54 
"Vegetation","/userDefined[tableName=stemLocation,userDefinedName=<name>]/definedValue ","/*s/individualOrganismObservation/simpleUserdefined[name=vegetation]/value"
51 
"Habitat","/userDefined[tableName=stemLocation,userDefinedName=habitat]/definedValue ","/*s/individualOrganismObservation/simpleUserdefined[name=habitat]/value"
52 
"PlantFungusDescription","/userDefined[tableName=stemLocation,userDefinedName=plantFungusDescription]/definedValue ","/*s/individualOrganismObservation/simpleUserdefined[name=plantFungusDescription]/value"
53 
"Substrate","/userDefined[tableName=stemLocation,userDefinedName=substrate]/definedValue ","/*s/individualOrganismObservation/simpleUserdefined[name=substrate]/value"
54 
"Vegetation","/userDefined[tableName=stemLocation,userDefinedName=vegetation]/definedValue ","/*s/individualOrganismObservation/simpleUserdefined[name=vegetation]/value"
55 55 
"UniqueNYInternalRecordNumber",,
scripts/data2xml/xpath.py
19 19 
    def __repr__(self):
20 20 
        str_ = ''
21 21 
        if self.is_attr: str_ += '@'
22 
        str_ += self.name+repr(self.attrs)+'='+repr(self.value)
22 
        str_ += self.name
23 
        if self.attrs != []: str_ += repr(self.attrs)
24 
        if self.value != None: str_ += '='+repr(self.value)
23 25 
        if self.is_ptr: str_ += '->'
24 26 
        return str_
25 27
......
36 38
37 39 
    def _path(self):
38 40 
        tree = []
41 
        fork_idx = None
42 
        elem_idx = 0
39 43 
        while True:
40 44 
            elem = XpathElem(is_attr=self._match_str('@'), name=self._fields())
41 45 
            if self._match_str('['):
42 46 
                elem.attrs = self._attrs()
43 47 
                self._match_str(']', required=True)
44 48 
            elem.is_ptr = self._match_str('->')
49 
            if not elem.is_ptr and self._match_str('!'): fork_idx = elem_idx
45 50 
            tree.append(elem)
46 51 
            if not self._match_str('/'): break
52 
            elem_idx += 1
53 
        # Add lookahead assertion for rest of path
54 
        if fork_idx != None: tree[fork_idx].attrs.append(tree[fork_idx+1:])
47 55 
        return tree
48 56
49 57 
    def _fields(self):
......
116 124
117 125 
        # Follow pointer
118 126 
        if elem.is_ptr:
119 
            target = deepcopy(path[elem_idx+1:]) # rest of path
120 
            attrs = target[types_id_level].attrs
127 
            path = deepcopy(path[elem_idx+1:]) # rest of path
128 
            attrs = path[types_id_level].attrs
121 129 
            if len(attrs) >= 1 and value(attrs[0]) == None:
122 130 
                # backward (child-to-parent) pointer with target ID attr
123 131 
                set_value(attrs[0], xml_util.get_id(node))
124 132 
            else: # forward (parent-to-child) pointer
125 133 
                id_ = xml_util.value(node)
126 
                if id_ == None:
134 
                obj_path = path[:types_id_level+1] # target object
135 
                if id_ == None or get(doc, obj_path, False, True) == None:
136 
                    # no target or target attrs don't match
127 137 
                    if not create: return None
138
128 139 
                    # Use last target object's ID + 1
129 
                    obj = get(doc, target[:types_id_level+1], False, True)
130 
                    if obj != None: id_ = str(int(xml_util.get_id(obj)) + 1)
140 
                    last_path = deepcopy(obj_path)
141 
                    last_path[-1].attrs = [] # just get by tag name
142 
                    last = get(doc, last_path, False, True)
143 
                    if last != None: id_ = str(int(xml_util.get_id(last)) + 1)
131 144 
                    else: id_ = '0'
145
146 
                    # Will append if target attrs didn't match. Place ! in XPath
147 
                    # after element to fork at to avoid this.
132 148 
                    xml_util.set_value(doc, node, id_)
133 149 
                else: last_only = False
134 
                set_id(target, id_)
135 
            return get(doc, target, create, last_only)
150 
                set_id(path, id_)
151 
            return get(doc, path, create, last_only)
136 152
137 153 
        parent = node
138 154 
        elem_idx += 1

Also available in: Unified diff