source: TI01-discovery/branches/ingestAutomation-upgrade/OAIBatch/SchemaNameSpace.py @ 3817

Subversion URL: http://proj.badc.rl.ac.uk/svn/ndg/TI01-discovery/branches/ingestAutomation-upgrade/OAIBatch/SchemaNameSpace.py@3817
Revision 3817, 1.2 KB checked in by cbyrom, 11 years ago (diff)

Add default logging support + create new version of ingest script, removing
all traces of the eXist DB + improve documentation and output.

Line 
1import sys, logging
2
3class SchemaNameSpace:
4    '''
5    Class to change/correct namespaces to the latest ones used by NDG discovery
6    NB, only currently handles correction of DIF files
7    '''
8    def __init__(self, infile, outfile, format):
9        '''
10        Constructor - with the logic to do the namespace change
11        @param infile: file to correct namespaces in
12        @param outfile: file to create with the corrected namespaces
13        @param format: Format of file being processed.  DIF is the only format which currently is processed.   
14        '''
15        logging.info("Correcting namespace in file, " + infile)
16           
17        self.ff=open(infile,'r')
18        self.ww=open(outfile,'w')
19        self.format= format
20        self.lines=self.ff.readlines()
21       
22        for self.line in self.lines:
23            if self.format== "DIF" and self.line.startswith('<DIF'):
24                logging.info("changing line for %s. output to %s" %(infile,outfile))
25                self.line='<DIF xmlns="http://gcmd.gsfc.nasa.gov/Aboutus/xml/dif/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">\n'
26            self.ww.write(self.line)
27        self.ff.close()
28        self.ww.close()
29
30        logging.info("Corrected file written to, " + outfile)
Note: See TracBrowser for help on using the repository browser.