Merge pull request #4 from analysiscenter/remove_samples

Change processing of file samples

Author: SergeyTsimfer

examples/01_Basic_usage.ipynb

@@ -132,7 +132,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "data/A.sgy 2.0 500.0\n"
+      "data/A.sgy 2000\n"
      ]
     }
    ],
@@ -141,7 +141,7 @@
     "\n",
     "# Context manager: recommended\n",
     "with segfast.open(path=path) as file:\n",
-    "    print(file.path, file.sample_interval, file.sample_rate)"
+    "    print(file.path, file.sample_interval)"
    ]
   },
   {
@@ -163,7 +163,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "data/A.sgy 2.0 500.0\n"
+      "data/A.sgy 2000\n"
      ]
     }
    ],
@@ -174,7 +174,7 @@
     "with open('tester.pckl', 'rb') as file:\n",
     "    reloaded = pickle.load(file)\n",
     "\n",
-    "print(reloaded.path, reloaded.sample_interval, reloaded.sample_rate)"
+    "print(reloaded.path, reloaded.sample_interval)"
    ]
   },
   {

segfast/__init__.py

@@ -3,4 +3,4 @@
 from .memmap_loader import MemmapLoader
 from .loader import Loader, File, open #pylint: disable=redefined-builtin
 
-__version__ = '1.0.0'
+__version__ = '1.0.1'

segfast/memmap_loader.py

@@ -223,7 +223,7 @@ def _make_mmap_headers_dtype(headers, endian_symbol='>'):
 
 
     # Data loading
-    def load_traces(self, indices, limits=None, buffer=None, return_samples=False):
+    def load_traces(self, indices, limits=None, buffer=None):
         """ Load traces by their indices.
         Under the hood, we use a pre-made memory mapping over the file, where trace data is viewed with a special dtype.
         Regardless of the numerical dtype of SEG-Y file, we output IEEE float32:
@@ -237,8 +237,6 @@ def load_traces(self, indices, limits=None, buffer=None, return_samples=False):
             Slice of the data along the depth axis.
         buffer : np.ndarray, optional
             Buffer to read the data into. If possible, avoids copies.
-        return_samples : bool
-            Whether to return samples of loaded traces in accordance to `limits`.
         """
         limits = self.process_limits(limits)
 
@@ -251,11 +249,10 @@ def load_traces(self, indices, limits=None, buffer=None, return_samples=False):
             traces = self._ibm_to_ieee(traces)
 
         if buffer is None:
-            buffer = np.require(traces, dtype=self.dtype, requirements='C')
-            return buffer if return_samples is False else (buffer, self.samples[limits])
+            return np.require(traces, dtype=self.dtype, requirements='C')
 
         buffer[:len(indices)] = traces
-        return buffer if return_samples is False else (buffer, self.samples[limits])
+        return buffer
 
     def load_depth_slices(self, indices, buffer=None):
         """ Load horizontal (depth) slices of the data.

segfast/segyio_loader.py

@@ -64,19 +64,13 @@ def __init__(self, path, endian='big', strict=False, ignore_geometry=True):
         self.n_traces = self.file_handler.trace.length
         self.dtype = self.file_handler.dtype
 
-        # Sample interval, rate and delay
-        self.sample_interval = self._infer_sample_interval() # ms
-        self.sample_rate = 1000 / self.sample_interval       # Hz
-        self.samples = np.arange(self.n_samples) * self.sample_interval
-        self.delay = self.file_handler.header[0].get(segyio.TraceField.DelayRecordingTime)
-
         # Misc
         self.metrics = self.file_handler.xfd.metrics()
         self.text = [self.file_handler.text[i] for i in range(1 + self.file_handler.ext_headers)]
 
-
-    def _infer_sample_interval(self):
-        """ Get sample interval from file headers. """
+    @property
+    def sample_interval(self):
+        """ Sample interval of seismic traces. """
         bin_sample_interval = self.file_handler.bin[segyio.BinField.Interval]
         trace_sample_interval = self.file_handler.header[0][segyio.TraceField.TRACE_SAMPLE_INTERVAL]
         # 0 means undefined sample interval, so it is removed from the set
@@ -87,7 +81,12 @@ def _infer_sample_interval(self):
                              "either both `Interval` (bytes 3217-3218 in the binary header) "
                              "and `TRACE_SAMPLE_INTERVAL` (bytes 117-118 in the header of the first trace) "
                              "are undefined or they have different values.")
-        return union_sample_interval.pop() / 1000  # convert from microseconds to milliseconds
+        return union_sample_interval.pop()
+
+    @property
+    def delay(self):
+        """ Delay recording time of seismic traces. """
+        return self.file_handler.header[0].get(segyio.TraceField.DelayRecordingTime)
 
 
     # Headers
@@ -160,7 +159,7 @@ def make_tsf_header(self):
 
 
     # Data loading: traces
-    def load_traces(self, indices, limits=None, buffer=None, return_samples=False):
+    def load_traces(self, indices, limits=None, buffer=None):
         """ Load traces by their indices.
         By pre-allocating memory for all of the requested traces, we significantly speed up the process.
 
@@ -172,19 +171,16 @@ def load_traces(self, indices, limits=None, buffer=None, return_samples=False):
             Slice of the data along the depth axis.
         buffer : np.ndarray, optional
             Buffer to read the data into. If possible, avoids copies.
-        return_samples : bool
-            Whether to return samples of loaded traces in accordance to `limits`.
         """
         limits = self.process_limits(limits)
-        samples = self.samples[limits]
-        n_samples = len(samples)
+        n_samples = len(range(*limits.indices(self.n_samples)))
 
         if buffer is None:
             buffer = np.empty((len(indices), n_samples), dtype=self.dtype)
 
         for i, index in enumerate(indices):
             self.load_trace(index=index, buffer=buffer[i], limits=limits)
-        return buffer if return_samples is False else (buffer, samples)
+        return buffer
 
     def process_limits(self, limits):
         """ Convert given `limits` to a `slice`. """
@@ -209,6 +205,7 @@ def load_trace(self, index, buffer, limits):
                                     limits.start, limits.stop, limits.step,
                                     buffer.size)
 
+
     # Data loading: depth slices
     def load_depth_slices(self, indices, buffer=None):
         """ Load horizontal (depth) slices of the data.