georeference now takes a list of geotransforms and tries until one pass

config/emg_preprocessor-config.yml

@@ -69,7 +69,9 @@ preprocessing:
       data_file_globs:
         - "*imagery_*.tif"
       georeference:
-        type: gcp
+        geotransforms:
+          - type: gcp
+          - type: no_op
       calc:
         formulas:
           # complex b1/b2 to b1 uint16 for all 4 polarizations
@@ -132,11 +134,21 @@ preprocessing:
         - "*IMG_*.JP2"
       additional_file_globs:
         - "*DIM_*"
+        - "*RPC_*"
+      georeference:
+        geotransforms:
+          - type: rpc
+          - type: no_op
     SP07:
       data_file_globs:
         - "*IMG_*.JP2"
       additional_file_globs:
         - "*DIM_*"
+        - "*RPC_*"
+      georeference:
+        geotransforms:
+          - type: rpc
+          - type: no_op
 # this configuration is still a stub - not all product types are done 
 # https://gitlab.eox.at/esa/prism/vs/-/issues/56
 # https://gitlab.eox.at/esa/prism/vs/-/issues/23

documentation/operator-guide/configuration.rst

@@ -356,51 +356,55 @@ preprocessing
 
       How the extracted files shall be georeferenced.
 
-      type
+      geotransforms
 
-        The type of georeferencing to apply. One of ``gcp``, ``rpc``,
-        ``corner``, ``world``.
+        A list of georeference methods with options to try.
+      
+        type
 
-      options
+          The type of georeferencing to apply. One of ``gcp``, ``rpc``,
+          ``corner``, ``world``.
+
+        options
 
-        Additional options for the georeferencing. Depends on the type of
-        georeferencing.
+          Additional options for the georeferencing. Depends on the type of
+          georeferencing.
 
-        order
+          order
 
-          The polynomial order to use for GCP related georeferencing.
+            The polynomial order to use for GCP related georeferencing.
 
-        projection
+          projection
 
-          The projection to use for ungeoreferenced images.
+            The projection to use for ungeoreferenced images.
 
-        rpc_file_template
+          rpc_file_template
 
-          The file glob template to use to find the RPC file. Template
-          parameters are {filename}, {fileroot}, and {extension}.
+            The file glob template to use to find the RPC file. Template
+            parameters are {filename}, {fileroot}, and {extension}.
 
-        warp_options
+          warp_options
 
-          Warp options. See
-          https://gdal.org/python/osgeo.gdal-module.html#WarpOptions for
-          details
+            Warp options. See
+            https://gdal.org/python/osgeo.gdal-module.html#WarpOptions for
+            details
 
-        corner_names
+          corner_names
 
-          The metadata field name including the corner names. Tuple of four:
-          bottom-left, bottom-right, top-left and top-right
+            The metadata field name including the corner names. Tuple of four:
+            bottom-left, bottom-right, top-left and top-right
 
-        orbit_direction_name
+          orbit_direction_name
 
-          The metadata field name containing the orbit direction
+            The metadata field name containing the orbit direction
 
-        force_north_up
+          force_north_up
 
-          Circumvents the naming of corner names and assumes a north-up orientation of the image.
+            Circumvents the naming of corner names and assumes a north-up orientation of the image.
 
-        tps
+          tps
 
-          Whether to use TPS transformation instead of GCP polynomials.
+            Whether to use TPS transformation instead of GCP polynomials.
 
     calc
 

preprocessor/preprocessor/config-schema.yaml

@@ -117,44 +117,47 @@ definitions:
             ".*":
               type: string
     georeference:
-      description: The definition of a georeferencing step.
-      type: object
+      type: object  
       properties:
-        type:
-          description: The type of georeferencing to apply.
-          type: string
-          enum: [gcp, rpc, corner, world] # TODO: more
-        options:
-          description: Additional options for the georeferencing. Depends on the type of georeferencing.
-          type: object
-          properties:
-            order:
-              description: The polynomial order to use for GCP reprojection.
-              type: number
-            projection:
-              description: The projection to use for ungeoreferenced images.
-              type: string
-            rpc_file_template:
-              description: The file glob template to use to find the RPC file. Template parameters are {filename}, {fileroot}, and {extension}.
-              type: string
-            warp_options:
-              description: "Warp options. See https://gdal.org/python/osgeo.gdal-module.html#WarpOptions for details"
-            corner_names:
-              description: "The metadata field name including the corner names. Tuple of four: bottom-left, bottom-right, top-left and top-right"
-              type: array
-              items:
+        geotransforms:
+          description: A list of geotransform methods to use
+          type: array
+          items:
+            type: object
+            properties:
+              type:
+                description: The type of georeferencing to apply.
                 type: string
-            orbit_direction_name:
-              description: The metadata field name containing the orbit direction
-              type: string
-            force_north_up:
-              description:
-              type: boolean
-            tps:
-              description: Whether to use TPS transformation instead of GCP polynomials.
-              type: boolean
-
-      required: [type]
+                enum: [gcp, rpc, corner, world, no_op] # TODO: more
+              options:
+                description: Additional options for the georeferencing. Depends on the type of georeferencing.
+                type: object
+                properties:
+                  order:
+                    description: The polynomial order to use for GCP reprojection.
+                    type: number
+                  projection:
+                    description: The projection to use for ungeoreferenced images.
+                    type: string
+                  rpc_file_template:
+                    description: The file glob template to use to find the RPC file. Template parameters are {filename}, {fileroot}, and {extension}.
+                    type: string
+                  warp_options:
+                    description: "Warp options. See https://gdal.org/python/osgeo.gdal-module.html#WarpOptions for details"
+                  corner_names:
+                    description: "The metadata field name including the corner names. Tuple of four: bottom-left, bottom-right, top-left and top-right"
+                    type: array
+                    items:
+                      type: string
+                  orbit_direction_name:
+                    description: The metadata field name containing the orbit direction
+                    type: string
+                  force_north_up:
+                    description:
+                    type: boolean
+                  tps:
+                    description: Whether to use TPS transformation instead of GCP polynomials.
+                    type: boolean
     calc:
       description: Definition of a calculation step.
       type: object

preprocessor/preprocessor/steps/georeference.py

@@ -11,25 +11,39 @@ from ..util import gdal, osr, replace_ext
 logger = logging.getLogger(__name__)
 
 
-def georeference_step(source_dir: os.PathLike, target_dir: os.PathLike, type: str, **options: dict):
-    type_name = type.lower()
-    if type_name == 'gcp':
-        georef_func = gcp_georef
-    elif type_name == 'rpc':
-        georef_func = rpc_georef
-    elif type_name == 'world':
-        georef_func = world_georef
-    elif type_name == 'corners':
-        georef_func = corner_georef
-    else:
-        raise Exception('Invalid georeference type %s' % type_name)
-    for filename in [path for path in glob(join(source_dir, '**'), recursive=True) if not os.path.isdir(path)]:
-        target_filename = join(target_dir, basename(filename))
-        georef_func(filename, target_filename, **options)
+def georeference_step(source_dir: os.PathLike, target_dir: os.PathLike, geotransforms: List[dict]):
+    success = False
+    for options in geotransforms:
+        type_name = options['type'].lower()
+        if type_name == 'gcp':
+            georef_func = gcp_georef
+        elif type_name == 'rpc':
+            georef_func = rpc_georef
+        elif type_name == 'world':
+            georef_func = world_georef
+        elif type_name == 'corners':
+            georef_func = corner_georef
+        elif type_name == 'no_op':
+            georef_func = no_op
+        else:
+            raise Exception('Invalid georeference type %s' % type_name)
+        try:
+            for filename in [path for path in glob(join(source_dir, '**'), recursive=True) if not os.path.isdir(path)]:
+                target_filename = join(target_dir, basename(filename))
+                georef_func(filename, target_filename, **options)
+        except:
+            # if any file fails, try another method
+            continue
+        else:
+            # all files georeferenced without exception, no need to try another methods
+            success = True
+            break
+    if not success and len(types) > 0:
+        raise Exception('No configured georeference method passed: %s' % ','.join([item['type'] for item in geotransforms]))
 
 
 def gcp_georef(input_filename: os.PathLike, target_filename: os.PathLike, order: int=1, projection: str='EPSG:4326',
-               tps: bool=False):
+               tps: bool=False, warp_options: dict=None):
     succeded = False
 
     # simple case: get the geotransform from some GCPs
@@ -68,9 +82,11 @@ def gcp_georef(input_filename: os.PathLike, target_filename: os.PathLike, order:
             target_filename,
             input_filename,
             dstSRS=projection,
+            **(warp_options or {}),
             **options,
         )
 
+
 def rpc_georef(input_filename: os.PathLike, target_filename: os.PathLike, rpc_file_template: str='{fileroot}.RPC', warp_options: dict=None):
     fileroot, extension = splitext(input_filename)
     rpc_file_glob = rpc_file_template.format(
@@ -141,6 +157,10 @@ def world_georef():
     pass
 
 
+def no_op(input_filename: os.PathLike, target_filename: os.PathLike):
+    # does not perform any operation
+    # configure as last step if you want other geotransform method to either pass or fail and then perform no other
+    shutil.move(input_filename, target_filename)
 
 
 def gcps_from_borders(size: Tuple[float, float], coords: List[Tuple[float, float]], orbit_direction: str, force_north_up: bool=False):