added geoms manipulate attributes script

geomsma.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+
+import sys,argparse,os
+import pyhdf.SD as SD
+import h5py
+import shutil
+import datetime as dt
+import numpy as np
+
+user=os.environ['USER']
+
+def main():
+  # Manage arguments
+    parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+        description = """Manipulate attributes in GEOMS hdf files""",
+        epilog      = """Written by Bavo Langerock, BIRA-IASB 2024
+
+Example to add a DOI to an existing GEOMS file and increment its DATA_FILE_VERSION
+  ./geomsma.py groundbased_ftir.o3_bira.iasb003_irwg2023_la.reunion.maido_20240608t045021z_20240608t060345z_001.hdf groundbased_ftir.o3_bira.iasb003_irwg2023_la.reunion.maido_20240608t045021z_20240608t060345z_002.hdf DATA_MODIFICATIONS "Modifications file version 002: added DOI" FILE_DOI 10.60897/ndacc.la.reunion.maido_ftir.o3_bira.iasb003_irwg2023""")
+    parser.add_argument('infile',help='source GEOMS file') 
+    parser.add_argument('outfile',help='target GEOMS file, from which the FILE_NAME,DATA_SOURCE,DATA_FILE_VERSION attributes are derived')
+    parser.add_argument('attributes',help='pairs <GEOMSfileattribute> "<new value>"', nargs=argparse.REMAINDER)
+    args = parser.parse_args()
+    infile=args.infile
+    outfile=args.outfile
+    assert os.path.realpath(outfile)!=os.path.realpath(infile),f'output {outfile} will overwrite input source {infile}'
+    nkeys=args.attributes[0::2];nvalues=args.attributes[1::2]
+    assert len(nkeys)==len(nvalues),f'not all keys have a value'
+    newsource='_'.join(os.path.basename(outfile).split('_')[1:-4])
+    newversion=os.path.basename(outfile).split('_')[-1].rsplit('.',1)[0]
+    shutil.copyfile(infile,outfile)
+    if h5py.is_hdf5(infile):
+      with h5py.File(outfile,'a') as fid:
+        fid.attrs['DATA_SOURCE']=np.array(newsource.upper(),dtype='S')
+        fid.attrs['FILE_NAME']=np.array(os.path.basename(outfile),dtype='S')
+        fid.attrs['DATA_FILE_VERSION']=np.array(newversion,dtype='S')
+        if 'DATA_MODIFICATIONS' in fid.attrs: X=np.string_(fid.attrs['DATA_MODIFICATIONS']) if isinstance(fid.attrs['DATA_MODIFICATIONS'],str) else fid.attrs['DATA_MODIFICATIONS']
+        else: X=np.string_('')
+        X=';'.join([x for x in (X.astype(dtype='U'),'changed DATA_SOURCE by %s on %s, origin is %s'%(user,dt.datetime.now().strftime('%Y%m%dT%H%M%Sz'),infile)) if x.lower() not in ('','none')])
+        fid.attrs['DATA_MODIFICATIONS']=np.array(X,dtype='S')
+        for nk,nv in zip(nkeys,nvalues): fid.attrs[nk]=np.array(nv,dtype='S')
+    else:
+      hdf4id=SD.SD(outfile,SD.SDC.WRITE)
+      att4=hdf4id.attr('DATA_SOURCE');att4.set(SD.SDC.CHAR8,newsource.upper())
+      att4=hdf4id.attr('FILE_NAME');att4.set(SD.SDC.CHAR8,os.path.basename(outfile))
+      att4=hdf4id.attr('DATA_FILE_VERSION');att4.set(SD.SDC.CHAR8,newversion)
+      try: X=getattr(hdf4id,'DATA_MODIFICATIONS');
+      except: X=''
+      finally: att4=hdf4id.attr('DATA_MODIFICATIONS')
+      if X: X=';'.join([x for x in (X,'changed DATA_SOURCE by %s on %s, origin is %s'%(user,dt.datetime.now().strftime('%Y%m%dT%H%M%Sz'),infile)) if x.lower() not in ('','none')]);att4.set(SD.SDC.CHAR8,X)
+      for nk,nv in zip(nkeys,nvalues):
+        att4=hdf4id.attr(nk);att4.set(SD.SDC.CHAR8,nv)
+
+if __name__ == '__main__': main()