Advice on loading OpenTopography data into a PostGIS database apreciated

thiemo at gelassene-pferde.biz thiemo at gelassene-pferde.biz
Thu Oct 31 16:34:47 PDT 2024


Hi

I have problems to load OpenTopography data into a PostGIS  
installation. This is within my project 33  
https://sourceforge.net/projects/treintaytres/ . You can get the  
latest source code from in case you are interested in. My problem is  
at revision 32. I want to load OpenTopography raster files into a  
PostGIS database using Python3. I defined the following function.

def process_files(data_directory, file_name_regexp, conn, source_id, logger):
     loaded_files =  
get_loaded_files(conn=conn,source_id=source_id,logger=logger,)
     existing_hashes = {file_hash for file_hash, _ in loaded_files}

     # "next" inhibits recursion, so only the top level is retrieved
     logger.info(f"Looking into '{data_directory}'")
     try:
         cur = conn.cursor()
         cur.execute("set postgis.gdal_enabled_drivers = 'ENABLE_ALL';")
         root, dirs, files = next(os.walk(data_directory))
         for file_name in files:
             if re.match(file_name_regexp, file_name):
                 file_path = os.path.join(root, file_name)
                 logger.info(f"Processing '{file_path}'")
                 file_hash = calculate_file_sha3_512_hash(file_path)
                 file_creation_time =  
datetime.fromtimestamp(os.path.getctime(file_path))

                 # If the hash is alread present, skip this file FIXME  
check on file names
                 if file_hash in existing_hashes:
                     continue

                 # Get the raster data
                 with open(file_path, 'rb') as f:
                     raster_data = f.read()

                 statement = """insert into TOPO_FILES (TILE
                        ,FILE_NAME
                        ,FILE_CREATION_PIT
                        ,FILE_HASH
                        ,SOURCE_ID)
    values (ST_FromGDALRaster(%s)
           ,%s
           ,%s
           ,%s
           ,%s::uuid);"""
                 logger.debug("statement")
                 logger.debug(statement)
                 logger.debug("First 100 bytes of raster_data")
                 logger.debug(f"{raster_data[:100]}")
                 logger.debug(f"file_name:{file_name}")
                 logger.debug(f"file_creation_time:{file_creation_time}")
                 logger.debug(f"file_hash:{file_hash}")
                 logger.debug(f"source_id:{source_id}")
                 params = (psycopg2.Binary(raster_data), file_name,  
file_creation_time, file_hash, source_id)
                 # params = (raster_data
                         #  ,file_name
                         #  ,file_creation_time
                         #  ,file_hash
                         #  ,source_id)
                 # logger.debug(f"params:{params}")
                 cur.execute(statement, params)
                 conn.commit()
                 cur.close()
     except StopIteration:
         logger.error(f"Error: '{data_directory}' could not be walked.  
Directory might be empty or inaccessible.")

However, it results in the following error, I am unable to iron out.

2024-11-01 00:28:33 - root - DEBUG -  
source_id:4f68d890-a08c-4c06-8aa5-741ad36b6abe
Traceback (most recent call last):
   File  
"/home/thiemo/external_projects/svn/33/trunk/code_files/data_storage/load_OpenTopography_data.py", line 737, in  
<module>
     main()
   File  
"/home/thiemo/external_projects/svn/33/trunk/code_files/data_storage/load_OpenTopography_data.py", line 714, in  
main
     process_files(
   File  
"/home/thiemo/external_projects/svn/33/trunk/code_files/data_storage/load_OpenTopography_data.py", line 442, in  
process_files
     cur.execute(statement, params)
psycopg2.errors.InternalError_: RASTER_fromGDALRaster: Could not open  
bytea with GDAL. Check that the bytea is of a GDAL supported format

I tried a variety of things, among other not conversion to binary, or  
explicit conversion to bytea, i.e. "ST_FromGDALRaster(%s::bytea)". I  
have run out of ideas and would very much appreciate, if someone could  
nudge me into the proper direction.

Kind regards

Thiemo





More information about the postgis-users mailing list