source: TI01-discovery/branches/ingestAutomation-upgrade/OAIBatch/DIF.py @ 4027

Subversion URL: http://proj.badc.rl.ac.uk/svn/ndg/TI01-discovery/branches/ingestAutomation-upgrade/OAIBatch/DIF.py@4027
Revision 4027, 6.7 KB checked in by cbyrom, 11 years ago (diff)

Add setup script for oai ingester + remove dependencies in this project
on local ndgUtils and csml project files - and remove these from this
codebase, subsequently.

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 ndgUtils.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
19
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
80        self.parameters=[]
81        for parameter in helper.findall(self.tree,'Parameters'):
82            name=''
83            for level in ['Category','Topic','Term','Variable','Detailed_Variable']:
84                name+=helper.getText(parameter,level)+'/'
85            self.parameters.append(name.rstrip('/'))
86           
87
88        #load up information about spatial bounding box
89        self.bbox=Bounding(self.tree,entity='DIF',getter=helper.getText)
90       
91        #load up information about temporal extent
92        tc=(
93            helper.getText(self.tree,'Temporal_Coverage/Start_Date'),
94            helper.getText(self.tree,'Temporal_Coverage/Stop_Date'),
95            helper.getText(self.tree,'Data_Set_Progress') )
96        self.timeCoverage=TimeCoverage(tc)
97       
98        #load up those silly paleo keywords
99        self.paleoKeywords=[]
100        paleoElements=helper.findall(self.tree,'Paleo_Temporal_Coverage')
101        for e in paleoElements:
102            self.paleoKeywords.append(helper.getText(e,'Chronostratigraphic_Unit'))
103           
104        #Data curator information
105        self.centre=DIFcontact(helper.find(self.tree,'Data_Center'),ctype='centre',helper=helper)
106        self.curator=DIFcontact(self.tree)
107
108        #Data Creators
109        self.creators=[]
110        # use author here because a full dif entry for creator wont necessarily exist in citation ...
111        self.authors=DIFAuthors(self.tree,helper)
112        self.date=dateParse(helper.getText(self.tree,'Data_Set_Citation/Dataset_Release_Date'),'YYYY')
113        if self.date=='': self.date='XXXX'
114        self.title=helper.getText(self.tree,'Data_Set_Citation/Dataset_Title')
115        self.briefCitation=None
116        if (self.authors!='' and self.date!='' and self.title!=''):
117            self.briefCitation='%s (%s): %s'%(self.authors,self.date,self.title)
118
119        #services
120        self.services=[]
121
122        for item in helper.findall(self.tree,'Related_URL'):
123            self.services.append(
124                DIFService(
125                 helper.getText(item,'URL_Content_type'),
126                 helper.getText(item,'URL'),
127                 helper.getText(item,'Description') ))
128       
129        if self.ndgObject is None:
130            self.binding=None
131        else:
132            if self.ndgObject.discoveryURL is not None:
133                self.binding=DIFService('DISCOVERY',self.ndgObject.discoveryURL,'Discovery record')
134            else: self.binding=None
135
136        if self.debug:
137           f=open('difs.log','a')
138           f.write('%s##\n%s\n##################################\n'%(self.entryID,self.xml))
139           f.close()
140           
141    def toHTML(self,config):
142
143        if self.tree is not None:
144            renderer=renderEntity(config)
145            return renderer.render(self)
146        else:
147            return '<p>No Valid DIF</p>'
148
149
150import unittest
151
152class TestCase(unittest.TestCase):
153    """
154    """
155
156    inputFile = 'examples/neodc.eg1.dif'
157    configFile='examples/example.config'
158   
159    def setUp(self):
160        ''' Load example config and DIF files for testing '''
161        f=file(self.inputFile,'r')
162        xml=f.read()
163        self.dif=DIF(xml)
164        self.config=myConfig(self.configFile)
165
166    def testEntries(self):
167        ''' Testing the DIF object can be loaded and some key entries extracted '''
168        print 'Entry ID [%s]'%self.dif.entryID
169        print 'Author [%s]'%self.dif.authors
170       
171    def testrenderDIF(self):
172        ''' Testing the conversion to html '''
173        print self.dif.timeCoverage
174        html=self.dif.toHTML(self.config)
175        g=file('difOutput.html','w')
176        g.write(html)
177
178if __name__=="__main__":
179    unittest.main()
180
181       
182       
Note: See TracBrowser for help on using the repository browser.