feat: standardising 16bits TIFF TDE-628 (#333)

* feat: standardising 16bits TIFF TDE-628

* fix: formatting fails

* refactor: improve argument passing to get_gdal_command()

* chore: remove useless empty line

scripts/gdal/gdal_bands.py

@@ -64,3 +64,20 @@ def get_gdal_band_offset(file: str, info: Optional[GdalInfo] = None) -> List[str
         return ["-b", "1", "-b", "2", "-b", "3"] + alpha_band_info
 
     return ["-b", str(band_red["band"]), "-b", str(band_green["band"]), "-b", str(band_blue["band"])] + alpha_band_info
+
+
+def get_gdal_band_type(file: str, info: Optional[GdalInfo] = None) -> str:
+    """Get the band type of the first band
+
+    Args:
+        file: file to check
+        info: optional precomputed gdalinfo
+
+    Returns:
+        band type
+    """
+    if info is None:
+        info = gdal_info(file, False)
+
+    bands = info["bands"]
+    return bands[0]["type"]

scripts/gdal/gdal_preset.py

@@ -62,8 +62,21 @@
     "overview_quality=90",
 ]
 
+# Arguments to convert TIFF from 16 bits to 8 bits
+CONVERT_16BITS_TO_8BITS = [
+    "-ot",
+    "Byte",
+    "-scale",
+    # 16 bit --> 2^16 = 65536 values --> 0-65535
+    "0",
+    "65535",
+    # 8 bit --> 2^8 = 256 values --> 0-255
+    "0",
+    "255",
+]
+
 
-def get_gdal_command(preset: str) -> List[str]:
+def get_gdal_command(preset: str, convert_from: Optional[str] = None) -> List[str]:
     get_log().info("gdal_preset", preset=preset)
     gdal_command: List[str] = ["gdal_translate"]
 
@@ -79,6 +92,9 @@ def get_gdal_command(preset: str) -> List[str]:
 
     gdal_command.extend(WEBP_OVERVIEWS)
 
+    if convert_from == "UInt16":
+        gdal_command.extend(CONVERT_16BITS_TO_8BITS)
+
     return gdal_command
 
 

scripts/gdal/tests/gdal_bands_test.py

@@ -1,4 +1,4 @@
-from scripts.gdal.gdal_bands import get_gdal_band_offset
+from scripts.gdal.gdal_bands import get_gdal_band_offset, get_gdal_band_type
 from scripts.gdal.tests.gdalinfo import add_band, fake_gdal_info
 
 
@@ -62,3 +62,14 @@ def test_gdal_default_rgb() -> None:
     bands = get_gdal_band_offset("some_file.tiff", gdalinfo)
 
     assert " ".join(bands) == "-b 1 -b 2 -b 3"
+
+
+def test_get_band_type() -> None:
+    gdalinfo = fake_gdal_info()
+    add_band(gdalinfo, band_type="UInt16")
+    add_band(gdalinfo, band_type="UInt16")
+    add_band(gdalinfo, band_type="UInt16")
+
+    band_type = get_gdal_band_type("some_file.tiff", gdalinfo)
+
+    assert band_type == "UInt16"

scripts/gdal/tests/gdalinfo.py

@@ -7,13 +7,23 @@ def fake_gdal_info() -> GdalInfo:
     return cast(GdalInfo, {})
 
 
-def add_band(gdalinfo: GdalInfo, color_interpretation: Optional[str] = None, no_data_value: Optional[int] = None) -> None:
+def add_band(
+    gdalinfo: GdalInfo,
+    color_interpretation: Optional[str] = None,
+    no_data_value: Optional[int] = None,
+    band_type: Optional[str] = None,
+) -> None:
     if gdalinfo.get("bands", None) is None:
         gdalinfo["bands"] = []
 
     gdalinfo["bands"].append(
         cast(
             GdalInfoBand,
-            {"band": len(gdalinfo["bands"]) + 1, "colorInterpretation": color_interpretation, "noDataValue": no_data_value},
+            {
+                "band": len(gdalinfo["bands"]) + 1,
+                "colorInterpretation": color_interpretation,
+                "noDataValue": no_data_value,
+                "type": band_type,
+            },
         )
     )

scripts/standardising.py

@@ -13,9 +13,10 @@
 from scripts.files.file_tiff import FileTiff
 from scripts.files.files_helper import get_file_name_from_path, is_tiff, is_vrt
 from scripts.files.fs import read, write
-from scripts.gdal.gdal_bands import get_gdal_band_offset
+from scripts.gdal.gdal_bands import get_gdal_band_offset, get_gdal_band_type
 from scripts.gdal.gdal_helper import get_gdal_version, run_gdal
 from scripts.gdal.gdal_preset import get_cutline_command, get_gdal_command
+from scripts.gdal.gdalinfo import gdal_info
 from scripts.logging.time_helper import time_in_ms
 
 
@@ -100,8 +101,10 @@ def standardising(file: str, preset: str, cutline: Optional[str]) -> FileTiff:
             run_gdal(get_cutline_command(input_cutline_path), input_file=input_file, output_file=target_vrt)
             input_file = target_vrt
 
-        command = get_gdal_command(preset)
-        command.extend(get_gdal_band_offset(input_file))
+        # gdalinfo to get band offset and band type
+        info = gdal_info(file, False)
+        command = get_gdal_command(preset, convert_from=get_gdal_band_type(input_file, info))
+        command.extend(get_gdal_band_offset(input_file, info))
 
         run_gdal(command, input_file=input_file, output_file=standardized_file_path)