source: TI07-MOLES/trunk/PythonCode/wsgi/DIF.py @ 2278

Subversion URL: http://proj.badc.rl.ac.uk/svn/ndg/TI07-MOLES/trunk/PythonCode/wsgi/DIF.py@2278
Revision 2278, 7.1 KB checked in by lawrence, 12 years ago (diff)

Better handling of DIF parameters as currently reported by
mini-MOLES

Line 
1# python class to support methods on a DIF ... to conform with
2# renderEntity etc ...
3#
4from Utilities import *
5from geoUtilities import *
6from People import *
7from AccessControl import AccessControl
8from ETxmlView import loadET, nsdumb
9from renderEntity import renderEntity
10try: #python 2.5
11    from xml.etree import ElementTree as ET
12except ImportError:
13    try:
14        # if you've installed it yourself it comes this way
15        import ElementTree as ET
16    except ImportError:
17        # if you've egged it this is the way it comes
18        from elementtree import ElementTree as ET
19debug=1
20class DIFService:
21    ''' A DIF only knows about a related URL '''
22    def __init__(self,c,u,d):
23        ''' Take a related url tuple (content_type,url,description) and store it, using
24        an ndgModifier if necessary '''
25        self.contentType,self.url,self.description=c,u,d
26    def __str__(self):
27        return '<a href="%s" title="%s">%s</a>'%self.contentType,self.url,self.description
28   
29class DIF:
30    ''' Supports the NASA GCMD DIF format for python operations,
31    note ... not a complete implementation, currently minimum to
32    show a reasonable piece of content '''
33    def __init__(self,xml,et=0,debug=0,ndgObject=None):
34       
35        '''Initialise a python dif instance based on an xml document (expected
36        to be an input string if et=0, otherwise an ElementTree instance) '''
37        if et:
38            try:
39                self.tree=xml
40                self.xml=ET.tostring(xml)
41            except:
42                raise TypeError,'DIF input is not a valid ElementTree instance'
43        else:
44            try:
45                self.tree=loadET(xml)
46                self.xml=xml
47            except: # for some reason we can't parse the document, and it's worth knowing why
48                if isinstance(xml,str):
49                    l=min(len(xml),300)-1
50                    if l!=len(xml):xml=xml[0:l]+'\n...'
51                    raise ValueError,'DIF input cannot be parsed into an ElementTree instance:\n%s'%xml
52                else:
53                    raise TypeError,'DIF input of type [%s] needs to be a string!'%type(xml)
54       
55        self.debug=debug
56       
57        # if this is an ndgObject that'll have been sorted externally ...
58        self.ndgObject=ndgObject
59       
60        # now try and interpret it
61       
62        helper=nsdumb(self.tree)
63        self.metadataType='DIF'
64        if helper.strip(self.tree.tag)!=self.metadataType: 
65            self.tree=helper.find(self.tree,self.metadataType)
66            if self.tree is None: 
67                raise ValueError, 'DIF input does not include a DIF element:\n%s'%self.xml
68       
69        self.entryID=helper.getText(self.tree,'Entry_ID')
70        self.abstract=helper.getText(self.tree,'Summary')
71        self.name=helper.getText(self.tree,'Entry_Title')
72        self.abbreviation=self.name[0:min(5,len(self.name))]
73       
74        #Note that entity.constraints.html is about access control on the metadata,
75        #and so we don't populate this here ...
76        self.constraints=AccessControl(None)
77       
78        #need entity.parameters, entity.bbox, entity.timeCoverage, entity.curator, entity.creators
79        # nb, two ways of parameters being stored
80        self.parameters=[]
81        for parameter in helper.findall(self.tree,'Parameters'):
82            name=parameter.text
83            if name is not None: 
84                self.parameters.append(name.replace('>','/'))
85            else:
86                name=''
87                for level in ['Category','Topic','Term','Variable','Detailed_Variable']:
88                    name+=helper.getText(parameter,level)+'/'
89                self.parameters.append(name.rstrip('/'))
90
91        #load up information about spatial bounding box
92        self.bbox=Bounding(self.tree,entity='DIF',getter=helper.getText)
93       
94        #load up information about temporal extent
95        tc=(
96            helper.getText(self.tree,'Temporal_Coverage/Start_Date'),
97            helper.getText(self.tree,'Temporal_Coverage/Stop_Date'),
98            helper.getText(self.tree,'Data_Set_Progress') )
99        self.timeCoverage=TimeCoverage(tc)
100       
101        #load up those silly paleo keywords
102        self.paleoKeywords=[]
103        paleoElements=helper.findall(self.tree,'Paleo_Temporal_Coverage')
104        for e in paleoElements:
105            self.paleoKeywords.append(helper.getText(e,'Chronostratigraphic_Unit'))
106           
107        #Data curator information
108        self.centre=DIFcontact(helper.find(self.tree,'Data_Center'),ctype='centre',helper=helper)
109        self.curator=DIFcontact(self.tree)
110
111        #Data Creators
112        self.creators=[]
113        # use author here because a full dif entry for creator wont necessarily exist in citation ...
114        self.authors=DIFAuthors(self.tree,helper)
115        self.date=dateParse(helper.getText(self.tree,'Data_Set_Citation/Dataset_Release_Date'),'YYYY')
116        if self.date=='': self.date='XXXX'
117        self.title=helper.getText(self.tree,'Data_Set_Citation/Dataset_Title')
118        self.briefCitation=None
119        if (self.authors!='' and self.date!='' and self.title!=''):
120            self.briefCitation='%s (%s): %s'%(self.authors,self.date,self.title)
121
122        #services
123        self.services=[]
124
125        for item in helper.findall(self.tree,'Related_URL'):
126            self.services.append(
127                DIFService(
128                 helper.getText(item,'URL_Content_type'),
129                 helper.getText(item,'URL'),
130                 helper.getText(item,'Description') ))
131       
132        if self.ndgObject is None:
133            self.binding=None
134        else:
135            if self.ndgObject.discoveryURL is not None:
136                self.binding=DIFService('DISCOVERY',self.ndgObject.discoveryURL,'Discovery record')
137            else: self.binding=None
138
139        if self.debug:
140           f=open('difs.log','a')
141           f.write('%s##\n%s\n##################################\n'%(self.entryID,self.xml))
142           f.close()
143           
144    def toHTML(self,config):
145
146        if self.tree is not None:
147            renderer=renderEntity(config)
148            return renderer.render(self)
149        else:
150            return '<p>No Valid DIF</p>'
151
152
153import unittest
154
155class TestCase(unittest.TestCase):
156    """
157    """
158
159    inputFile = 'examples/neodc.eg1.dif'
160    inputFile2= 'examples/bodc.eg2.edmed.dif'
161    configFile='examples/example.config'
162   
163    def setUp(self):
164        ''' Load example config and DIF files for testing '''
165        f=file(self.inputFile2,'r')
166        xml=f.read()
167        self.dif=DIF(xml)
168        self.config=myConfig(self.configFile)
169
170    def testEntries(self):
171        ''' Testing the DIF object can be loaded and some key entries extracted '''
172        print 'Entry ID [%s]'%self.dif.entryID
173        print 'Author [%s]'%self.dif.authors
174       
175    def testrenderDIF(self):
176        ''' Testing the conversion to html '''
177        print self.dif.timeCoverage
178        html=self.dif.toHTML(self.config)
179        g=file('difOutput.html','w')
180        g.write(html)
181   
182    def testParameters(self):
183        ''' Testing obtaining parameters from an edmed dif in original format '''
184        print 'Parameters %s'%self.dif.parameters
185
186if __name__=="__main__":
187    unittest.main()
188
189       
190       
Note: See TracBrowser for help on using the repository browser.