fix: identifier not always set for generated instances of QPImage

CHANGELOG

@@ -1,3 +1,5 @@
+0.4.3
+ - fix: "identifier" not always set for generated instances of QPImage
 0.4.2
  - fix: implement `get_name` method for SeriesFolder format
  - enh: start identifier/name indexing at 1 instead of 0

qpformat/file_formats/__init__.py

@@ -147,7 +147,9 @@ def get_name(self, idx):
     def get_qpimage_raw(self, idx):
         """Return QPImage without background correction"""
         ds = self._get_dataset(idx)
-        return ds.get_qpimage_raw()
+        qpi = ds.get_qpimage_raw()
+        qpi["identifier"] = self.get_identifier(idx)
+        return qpi
 
     def get_time(self, idx):
         ds = self._get_dataset(idx)

qpformat/file_formats/dataset.py

@@ -2,6 +2,7 @@
 import copy
 import functools
 import hashlib
+import io
 import pathlib
 
 import numpy as np
@@ -31,12 +32,13 @@ class SeriesData(object):
     def __init__(self, path, meta_data={}, holo_kw={}, as_type="float32"):
         #: Enforced dtype via keyword arguments
         self.as_type = as_type
-        if isinstance(path, (str, pathlib.Path)):
-            #: pathlib.Path to data file or BytesIO
-            self.path = pathlib.Path(path).resolve()
-        else:
-            # _io.BytesIO
+        if isinstance(path, io.IOBase):
+            # io.IOBase
             self.path = path
+        else:
+            #: pathlib.Path to data file or io.IOBase
+            self.path = pathlib.Path(path).resolve()
+
         # check for valid metadata keys
         for key in meta_data:
             if key not in qpimage.meta.DATA_KEYS:
@@ -113,8 +115,11 @@ def _compute_bgid(self, bg=None):
     def _identifier_data(self):
         data = []
         # data
-        with self.path.open("rb") as fd:
-            data.append(fd.read(50 * 1024))
+        if isinstance(self.path, io.IOBase):
+            data.append(self.path.read(50 * 1024))
+        else:
+            with self.path.open("rb") as fd:
+                data.append(fd.read(50 * 1024))
         data += self._identifier_meta()
         return hash_obj(data)
 
@@ -161,6 +166,10 @@ def get_qpimage(self, idx):
         """Return background-corrected QPImage of data at index `idx`"""
         # raw data
         qpi = self.get_qpimage_raw(idx)
+        if "identifier" not in qpi:
+            msg = "`get_qpimage_raw` does not set 'identifier' " \
+                  + "in class '{}'!".format(self.__class__)
+            raise KeyError(msg)
         # bg data
         if self._bgdata:
             if len(self._bgdata) == 1:
@@ -176,13 +185,15 @@ def get_qpimage(self, idx):
                 # `self._bgdata` is a QPImage
                 bg = self._bgdata[bgidx]
             qpi.set_bg_data(bg_data=bg)
-        # set identifier
-        qpi["identifier"] = self.get_identifier(idx)
         return qpi
 
     @abc.abstractmethod
     def get_qpimage_raw(self, idx):
-        """Return QPImage without background correction"""
+        """Return QPImage without background correction
+
+        Note that this method must always return a QPImage instance with
+        the "identifier" metadata key set!
+        """
 
     def get_time(self, idx):
         """Return time of data at index `idx`

qpformat/file_formats/series_hdf5_hyperspy.py

@@ -93,6 +93,8 @@ def get_qpimage_raw(self, idx=0):
                               meta_data=meta_data,
                               holo_kw=self.holo_kw,
                               h5dtype=self.as_type)
+        # set identifier
+        qpi["identifier"] = self.get_identifier(idx)
         return qpi
 
     @staticmethod

qpformat/file_formats/series_zip_tif_holo.py

@@ -60,7 +60,10 @@ def files(self):
     def get_qpimage_raw(self, idx):
         """Return QPImage without background correction"""
         ds = self._get_dataset(idx)
-        return ds.get_qpimage_raw()
+        qpi = ds.get_qpimage_raw()
+        # set identifier
+        qpi["identifier"] = self.get_identifier(idx)
+        return qpi
 
     @staticmethod
     def verify(path):

qpformat/file_formats/series_zip_tif_phasics.py

@@ -61,7 +61,10 @@ def files(self):
     def get_qpimage_raw(self, idx):
         """Return QPImage without background correction"""
         ds = self._get_dataset(idx)
-        return ds.get_qpimage_raw()
+        qpi = ds.get_qpimage_raw()
+        # get identifier
+        qpi["identifier"] = self.get_identifier(idx)
+        return qpi
 
     def get_time(self, idx):
         # Obtain the time from the tif meta data

qpformat/file_formats/single_hdf5_qpimage.py

@@ -52,6 +52,8 @@ def get_qpimage_raw(self, idx=0):
         # Force meta data
         for key in self.meta_data:
             qpi[key] = self.meta_data[key]
+        # set identifier
+        qpi["identifier"] = self.get_identifier(idx)
         return qpi
 
     @staticmethod

qpformat/file_formats/single_npy_numpy.py

@@ -38,6 +38,8 @@ def get_qpimage_raw(self, idx=0):
                               which_data=self.storage_type,
                               meta_data=meta_data,
                               h5dtype=self.as_type)
+        # get identifier
+        qpi["identifier"] = self.get_identifier(idx)
         return qpi
 
     @staticmethod

qpformat/file_formats/single_tif_holo.py

@@ -32,6 +32,8 @@ def get_qpimage_raw(self, idx=0):
                               meta_data=meta_data,
                               holo_kw=self.holo_kw,
                               h5dtype=self.as_type)
+        # get identifier
+        qpi["identifier"] = self.get_identifier(idx)
         return qpi
 
     @staticmethod

qpformat/file_formats/single_tif_phasics.py

@@ -153,6 +153,8 @@ def get_qpimage_raw(self, idx=0):
                               which_data="phase,intensity",
                               meta_data=meta_data,
                               h5dtype=self.as_type)
+        # set identifier
+        qpi["identifier"] = self.get_identifier()
         return qpi
 
     def get_time(self, idx=0):

setup.py

@@ -26,7 +26,7 @@
     install_requires=["h5py>=2.7.0",
                       "nrefocus>=0.1.5",
                       "numpy>=1.12.0",
-                      "qpimage>=0.4.0",
+                      "qpimage>=0.4.3",
                       "scikit-image>=0.11.0",
                       "scipy>=0.18.0",
                       ],

tests/test_series_folder.py

@@ -60,10 +60,8 @@ def test_load_data():
     # data should be the same
     qpi0 = ds.get_qpimage(0)
     qpi1 = ds.get_qpimage(1)
-    assert qpi0 != qpi1
+    assert qpi0 == qpi1
     assert qpi0["identifier"] != qpi1["identifier"]
-    qpi0["identifier"] = ""
-    qpi1["identifier"] = ""
     assert qpi0 == qpi1
     assert ds.get_qpimage_raw(0) == ds.get_qpimage_raw(1)
     assert ds.get_qpimage(0).shape == (50, 50)

tests/test_series_hdf5_hyperspy.py

@@ -49,6 +49,16 @@ def test_basic():
     shutil.rmtree(path=tdir, ignore_errors=True)
 
 
+def test_returned_identifier():
+    tdir, hspyf = make_hyperspy()
+    ds = qpformat.load_data(hspyf)
+    qpi = ds.get_qpimage(0)
+    assert "identifier" in qpi
+    qpiraw = ds.get_qpimage_raw(0)
+    assert "identifier" in qpiraw
+    shutil.rmtree(path=tdir, ignore_errors=True)
+
+
 def test_wrong_format():
     path = datapath / "single_qpimage.h5"
     ds = qpformat.load_data(path, fmt="SeriesHdf5HyperSpy")

tests/test_series_hdf5_qpimage.py

@@ -40,9 +40,12 @@ def test_identifier():
 def test_load_data():
     path = datapath / "single_qpimage.h5"
     tf = tempfile.mktemp(suffix=".h5", prefix="qpformat_test_")
-    qpi = qpimage.QPImage(h5file=path, h5mode="r")
+    qpi1 = qpimage.QPImage(h5file=path, h5mode="r").copy()
+    qpi2 = qpi1.copy()
+    qpi1["identifier"] = "test100"
+    qpi2["identifier"] = "test200"
     # generate qpseries hdf5 file
-    with qpimage.QPSeries(qpimage_list=[qpi, qpi],
+    with qpimage.QPSeries(qpimage_list=[qpi1, qpi2],
                           h5file=tf,
                           h5mode="a"):
         pass
@@ -52,13 +55,12 @@ def test_load_data():
     assert ds.path == pathlib.Path(tf)
     assert ds.get_time(1) == 0
     assert "SeriesHdf5Qpimage" in ds.__repr__()
-    # individual identifiers are not extracted anymore as of
-    # qpformat version 0.3.4
     qpd = ds.get_qpimage(1)
-    assert qpd != qpi
-    assert qpd.shape == qpi.shape
-    assert np.allclose(qpd.amp, qpi.amp)
-    assert np.allclose(qpd.pha, qpi.pha)
+    assert qpd["identifier"] != qpi1["identifier"]
+    assert qpd == qpi1
+    assert qpd.shape == qpi1.shape
+    assert np.allclose(qpd.amp, qpi1.amp)
+    assert np.allclose(qpd.pha, qpi1.pha)
 
     # cleanup
     try:
@@ -126,6 +128,30 @@ def test_meta_override():
         pass
 
 
+def test_returned_identifier():
+    path = datapath / "single_qpimage.h5"
+    tf = tempfile.mktemp(suffix=".h5", prefix="qpformat_test_")
+    qpi = qpimage.QPImage(h5file=path, h5mode="r")
+    # generate qpseries hdf5 file
+    with qpimage.QPSeries(qpimage_list=[qpi, qpi],
+                          h5file=tf,
+                          h5mode="a"):
+        pass
+
+    ds = qpformat.load_data(tf)
+
+    qpi = ds.get_qpimage(0)
+    assert "identifier" in qpi
+    qpiraw = ds.get_qpimage_raw(0)
+    assert "identifier" in qpiraw
+
+    # cleanup
+    try:
+        os.remove(tf)
+    except OSError:
+        pass
+
+
 if __name__ == "__main__":
     # Run all tests
     loc = locals()

tests/test_series_zip_tif_holo.py

@@ -48,6 +48,20 @@ def test_load_data():
         pass
 
 
+def test_returned_identifier():
+    path = setup_test_zip()
+    ds = qpformat.load_data(path)
+    qpi = ds.get_qpimage(0)
+    assert "identifier" in qpi
+    qpiraw = ds.get_qpimage_raw(0)
+    assert "identifier" in qpiraw
+
+    try:
+        path.unlink()
+    except OSError:
+        pass
+
+
 if __name__ == "__main__":
     # Run all tests
     loc = locals()

tests/test_series_zip_tif_phasics.py

@@ -36,6 +36,15 @@ def test_data_content():
     assert np.allclose(qpi.pha.max() - qpi.pha.min(), 0.18902349472045898)
 
 
+def test_returned_identifier():
+    path = datapath / "series_phasics.zip"
+    ds = qpformat.load_data(path)
+    qpi = ds.get_qpimage(0)
+    assert "identifier" in qpi
+    qpiraw = ds.get_qpimage_raw(0)
+    assert "identifier" in qpiraw
+
+
 if __name__ == "__main__":
     # Run all tests
     loc = locals()

tests/test_single_hdf5_qpimage.py

@@ -41,11 +41,11 @@ def test_load_data():
     assert ds.path == path.resolve()
     assert ds.get_time() == 0
     assert "SingleHdf5Qpimage" in ds.__repr__()
-    # individual identifiers are not extracted anymore as of
-    # qpformat version 0.3.4
     qpd = ds.get_qpimage()
-    qpi = qpimage.QPImage(h5file=path, h5mode="r")
-    assert qpd != qpi
+    qpi = qpimage.QPImage(h5file=path, h5mode="r").copy()
+    qpi["identifier"] = "test"
+    assert qpd["identifier"] != qpi["identifier"]
+    assert qpd == qpi
     assert qpd.shape == qpi.shape
     assert np.allclose(qpd.amp, qpi.amp)
     assert np.allclose(qpd.pha, qpi.pha)
@@ -98,6 +98,15 @@ def test_meta_override():
         pass
 
 
+def test_returned_identifier():
+    path = datapath / "single_qpimage.h5"
+    ds = qpformat.load_data(path)
+    qpi = ds.get_qpimage(0)
+    assert "identifier" in qpi
+    qpiraw = ds.get_qpimage_raw(0)
+    assert "identifier" in qpiraw
+
+
 if __name__ == "__main__":
     # Run all tests
     loc = locals()

tests/test_single_npy_numpy.py

@@ -44,6 +44,25 @@ def test_load_field():
         pass
 
 
+def test_returned_identifier():
+    # generate test data
+    tf = pathlib.Path(tempfile.mktemp(suffix=".npy", prefix="qpformat_test_"))
+    phase = np.ones((20, 20), dtype=float)
+    phase *= np.linspace(0, .3, 20).reshape(-1, 1)
+    np.save(tf, phase)
+
+    ds = qpformat.load_data(tf, as_type="float64")
+    qpi = ds.get_qpimage(0)
+    assert "identifier" in qpi
+    qpiraw = ds.get_qpimage_raw(0)
+    assert "identifier" in qpiraw
+
+    try:
+        tf.unlink()
+    except OSError:
+        pass
+
+
 if __name__ == "__main__":
     # Run all tests
     loc = locals()

tests/test_single_tif_holo.py

@@ -30,6 +30,15 @@ def test_load_data():
     assert qpi.shape == (238, 267)
 
 
+def test_returned_identifier():
+    path = datapath / "single_holo.tif"
+    ds = qpformat.load_data(path)
+    qpi = ds.get_qpimage(0)
+    assert "identifier" in qpi
+    qpiraw = ds.get_qpimage_raw(0)
+    assert "identifier" in qpiraw
+
+
 if __name__ == "__main__":
     # Run all tests
     loc = locals()

tests/test_single_tif_phasics.py

@@ -65,6 +65,15 @@ def test_data_content():
     assert np.allclose(qpi.pha.max() - qpi.pha.min(), 4.168394088745117)
 
 
+def test_returned_identifier():
+    path = datapath / "single_phasics.tif"
+    ds = qpformat.load_data(path)
+    qpi = ds.get_qpimage(0)
+    assert "identifier" in qpi
+    qpiraw = ds.get_qpimage_raw(0)
+    assert "identifier" in qpiraw
+
+
 if __name__ == "__main__":
     # Run all tests
     loc = locals()