Merge pull request #317 from MDAnalysis/feature-unified_iter

Cleaned up Reader classes

package/CHANGELOG

@@ -33,6 +33,8 @@ The rules for this file:
   * Removed FormatError, now replaced by ValueError
   * distances.contact_matrix(): changed keyword 'suppress_progmet' to
     'quiet'
+  * base.Reader now defines __iter__
+  * __iter__ removed from many Readers (now use base.Reader implementation)
 
   Fixes
   * analysis.hbonds.HydrogenBondAnalysis performs a sanity check for

package/MDAnalysis/coordinates/DCD.py

@@ -462,20 +462,9 @@ def _dcd_header(self):
             'charmm', 'first', 'with_unitcell']
         return dict(zip(desc, struct.unpack("LLiiiiidiPPiiii", self._dcd_C_str)))
 
-    def __iter__(self):
-        # Reset the trajectory file, read from the start
-        # usage is "from ts in dcd:" where dcd does not have indexes
+    def _reopen(self):
         self._reset_dcd_read()
 
-        def iterDCD():
-            for i in xrange(0, self.numframes, self.skip):  # FIXME: skip is not working!!!
-                try:
-                    yield self._read_next_timestep()
-                except IOError:
-                    raise StopIteration
-
-        return iterDCD()
-
     def _read_next_timestep(self, ts=None):
         if ts is None:
             ts = self.ts

package/MDAnalysis/coordinates/DLPoly.py

@@ -260,15 +260,6 @@ def _read_numframes(self):
 
         return numframes
 
-    def __iter__(self):
-        self._reopen()
-        while True:
-            try:
-                yield self._read_next_timestep()
-            except IOError:
-                self.rewind()
-                raise StopIteration
-
     def rewind(self):
         self._reopen()
         self.next()

package/MDAnalysis/coordinates/GMS.py

@@ -165,16 +165,6 @@ def _read_out_numframes(self, filename):
         self.close()
         return int(counter)
 
-    def __iter__(self):
-        self.ts.frame = 0  # start at 0 so that the first frame becomes 1
-        self._reopen()
-        while True:
-            try:
-                yield self._read_next_timestep()
-            except EOFError:
-                self.close()
-                raise StopIteration
-
     def _read_next_timestep(self, ts=None):
         # check that the timestep object exists
         if ts is None:

package/MDAnalysis/coordinates/MOL2.py

@@ -96,13 +96,6 @@ def __init__(self, filename, convert_units=None, **kwargs):
         self.delta = 0
         self.skip_timestep = 1
 
-    def __iter__(self):
-        for i in xrange(0, self.numframes):
-            try:
-                yield self._read_frame(i)
-            except IOError:
-                raise StopIteration
-
     def parse_block(self, block):
         sections = {}
         cursor = None

package/MDAnalysis/coordinates/PDB.py

@@ -551,12 +551,13 @@ def Writer(self, filename, **kwargs):
         kwargs.setdefault('multiframe', self.numframes > 1)
         return PrimitivePDBWriter(filename, **kwargs)
 
-    def __iter__(self):
-        for i in xrange(0, self.numframes):
-            try:
-                yield self._read_frame(i)
-            except IOError:
-                raise StopIteration
+    def rewind(self):
+        self._read_frame(0)
+
+    def _reopen(self):
+        # Pretend the current TS is 0 (in 1 based) so "next" is the
+        # first frame
+        self.ts.frame = 0
 
     def _read_next_timestep(self, ts=None):
         if ts is None:
@@ -571,8 +572,6 @@ def _read_next_timestep(self, ts=None):
         return self._read_frame(frame)
 
     def _read_frame(self, frame):
-        if numpy.dtype(type(frame)) != numpy.dtype(int):
-            raise TypeError("frame must be a integer")
         # Assume _read_frame is passed a 0-based frame number. Convert this to
         # 1-based.
         frame += 1

package/MDAnalysis/coordinates/TRJ.py

@@ -109,9 +109,6 @@
 .. autoclass:: NCDFReader
    :members:
 
-   .. automethod:: __getitem__
-   .. automethod:: __iter__
-
 .. autoclass:: NCDFWriter
    :members:
 
@@ -369,15 +366,6 @@ def rewind(self):
         self._reopen()
         self.next()
 
-    def __iter__(self):
-        self._reopen()
-        while True:
-            try:
-                yield self._read_next_timestep()
-            except EOFError:
-                self.rewind()
-                raise StopIteration
-
 
 class NCDFReader(base.Reader):
     """Reader for `AMBER NETCDF format`_ (version 1.0).
@@ -536,6 +524,9 @@ def _read_frame(self, frame):
         self._current_frame = frame
         return ts
 
+    def _reopen(self):
+        self._current_frame = -1
+
     def _read_next_timestep(self, ts=None):
         if ts is None:
             ts = self.ts
@@ -544,21 +535,6 @@ def _read_next_timestep(self, ts=None):
         except IndexError:
             raise IOError
 
-    def _sliced_iter(self, start, stop, step):
-        def iterNETCDF(start=start, stop=stop, step=step):
-            for i in xrange(start, stop, step):
-                yield self._read_frame(i)
-
-        return iterNETCDF() 
-
-    def __iter__(self):
-        """Iterate over the whole trajectory"""
-        for i in xrange(0, self.numframes):
-            try:
-                yield self._read_frame(i)
-            except IndexError:
-                raise StopIteration
-
     def close(self):
         """Close trajectory; any further access will raise an :exc:`IOError`"""
         if not self.trjfile is None:

package/MDAnalysis/coordinates/TRZ.py

@@ -341,15 +341,6 @@ def skip_timestep(self):
             self.rewind()
         return self._skip_timestep
 
-    def __iter__(self):
-        self._reopen()
-        while True:
-            try:
-                yield self._read_next_timestep()
-            except IOError:
-                self.rewind()
-                raise StopIteration
-
     # can use base.Reader __getitem__ implementation
     def _read_frame(self, frame):
         """Move to *frame* and fill timestep with data."""

package/MDAnalysis/coordinates/XYZ.py

@@ -315,16 +315,6 @@ def _read_xyz_numframes(self, filename):
         numframes = int(counter / linesPerFrame)
         return numframes
 
-    def __iter__(self):
-        self.ts.frame = 0  # start at 0 so that the first frame becomes 1
-        self._reopen()
-        while True:
-            try:
-                yield self._read_next_timestep()
-            except EOFError:
-                self.close()
-                raise StopIteration
-
     def _read_next_timestep(self, ts=None):
         # check that the timestep object exists
         if ts is None:

package/MDAnalysis/coordinates/base.py

@@ -622,7 +622,8 @@ def next(self):
 
     def rewind(self):
         """Position at beginning of trajectory"""
-        self[0]
+        self._reopen()
+        self.next()
 
     @property
     def dt(self):
@@ -689,7 +690,20 @@ def __del__(self):
         self.close()
 
     def __iter__(self):
-        raise NotImplementedError("BUG: Override __iter__() in the trajectory reader (%s)!" % type(self))
+        self._reopen()
+        while True:
+            try:
+                yield self._read_next_timestep()
+            except (EOFError, IOError):
+                self.rewind()
+                raise StopIteration
+
+    def _reopen(self):
+        """Should position Reader to just before first frame
+        
+        Calling next after this should return the first frame
+        """
+        pass
 
     def __getitem__(self, frame):
         """Return the Timestep corresponding to *frame*.
@@ -742,7 +756,7 @@ def _sliced_iter(self, start, stop, step):
         def _iter(start=start, stop=stop, step=step):
             try:
                 for i in xrange(start, stop, step):
-                    yield self[i]
+                    yield self._read_frame(i)
             except TypeError:  # if _read_frame not implemented
                 raise TypeError("{0} does not support slicing."
                                 "".format(self.__class__.__name__))
@@ -1172,6 +1186,12 @@ def _read_first_frame(self):
         # Override this in subclasses to create and fill a Timestep
         pass
 
+    def rewind(self):
+        pass
+
+    def _reopen(self):
+        pass
+
     def next(self):
         raise IOError(self._err.format(self.__class__.__name__))
 

package/MDAnalysis/coordinates/xdrfile/core.py

@@ -718,11 +718,6 @@ def __iter__(self):
             else:
                 yield ts
 
-    def rewind(self):
-        """Position at beginning of trajectory"""
-        self._reopen()
-        self.next()  # read first frame
-
     def _reopen(self):
         self.close()
         self.open_trajectory()
@@ -744,19 +739,6 @@ def _read_frame(self, frame):
     # Renamed this once upon a time.
     _goto_frame = _read_frame
 
-    def _sliced_iter(self, start, stop, step):
-        def _iter(start=start, stop=stop, step=step):
-            if step == 1:
-                rng = xrange(start + 1, stop, step)
-                yield self._read_frame(start)
-                for framenr in rng:
-                    yield self._read_frame(framenr)
-            else:
-                rng = xrange(start, stop, step)
-                for framenr in rng:
-                    yield self._read_frame(framenr)
-        return _iter()
-
     def timeseries(self, asel, start=0, stop=-1, skip=1, format='afc'):
         raise NotImplementedError("timeseries not available for Gromacs trajectories")