Merge pull request #71 from philipstarkey/philipstarkey/issue42

philipstarkey · web-flow · commit 20cb9afbcf9d · 2020-06-25T12:56:40.000+10:00
diff --git a/lyse/__init__.py b/lyse/__init__.py
@@ -104,11 +104,10 @@ def globals_diff(run1, run2, group=None):
 class Run(object):
     def __init__(self,h5_path,no_write=False):
         self.no_write = no_write
+        self._no_group = None
         self.h5_path = h5_path
         if not self.no_write:
-            with h5py.File(h5_path) as h5_file:
-                if not 'results' in h5_file:
-                     h5_file.create_group('results')
+            self._create_group_if_not_exists(h5_path, '/', 'results')
                      
         try:
             if not self.no_write:
@@ -118,48 +117,65 @@ def __init__(self,h5_path,no_write=False):
                 frame = inspect.currentframe()
                 __file__ = frame.f_back.f_globals['__file__']
                 self.group = os.path.basename(__file__).split('.py')[0]
-                with h5py.File(h5_path) as h5_file:
-                    if not self.group in h5_file['results']:
-                         h5_file['results'].create_group(self.group)
+                self._create_group_if_not_exists(h5_path, 'results', self.group)
         except KeyError:
             # sys.stderr.write('Warning: to write results, call '
             # 'Run.set_group(groupname), specifying the name of the group '
             # 'you would like to save results to. This normally comes from '
             # 'the filename of your script, but since you\'re in interactive '
             # 'mode, there is no scipt name. Opening in read only mode for '
             # 'the moment.\n')
+            
+            # Backup the value of self.no_write for restoration once the group
+            # is set
+            self._no_group = (True, self.no_write)
             self.no_write = True
             
+    def _create_group_if_not_exists(self, h5_path, location, groupname):
+        """Creates a group in the HDF5 file at `location` if it does not exist.
+        
+        Only opens the h5 file in write mode if a group must be created.
+        This ensures the last modified time of the file is only updated if
+        the file is actually written to."""
+        create_group = False
+        with h5py.File(h5_path, 'r') as h5_file:
+            if not groupname in h5_file[location]:
+                create_group = True
+        if create_group:
+            with h5py.File(h5_path, 'r+') as h5_file:
+                h5_file[location].create_group(groupname)
+
     def set_group(self, groupname):
         self.group = groupname
-        with h5py.File(self.h5_path) as h5_file:
-            if not self.group in h5_file['results']:
-                 h5_file['results'].create_group(self.group)
-        self.no_write = False
+        self._create_group_if_not_exists(self.h5_path, '/', 'results')
+        # restore no_write attribute now we have set the group
+        if self._no_group is not None and self._no_group[0]:
+            self.no_write = self._no_group[1]
+            self._no_group = None
 
     def trace_names(self):
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             try:
                 return list(h5_file['data']['traces'].keys())
             except KeyError:
                 return []
                 
     def get_attrs(self, group):
         """Returns all attributes of the specified group as a dictionary."""
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             if not group in h5_file:
                 raise Exception('The group \'%s\' does not exist'%group)
             return get_attributes(h5_file[group])
 
     def get_trace(self,name):
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             if not name in h5_file['data']['traces']:
                 raise Exception('The trace \'%s\' doesn not exist'%name)
             trace = h5_file['data']['traces'][name]
             return array(trace['t'],dtype=float),array(trace['values'],dtype=float)         
 
     def get_result_array(self,group,name):
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             if not group in h5_file['results']:
                 raise Exception('The result group \'%s\' doesn not exist'%group)
             if not name in h5_file['results'][group]:
@@ -169,7 +185,7 @@ def get_result_array(self,group,name):
     def get_result(self, group, name):
         """Return 'result' in 'results/group' that was saved by 
         the save_result() method."""
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             if not group in h5_file['results']:
                 raise Exception('The result group \'%s\' does not exist'%group)
             if not name in h5_file['results'][group].attrs.keys():
@@ -288,7 +304,7 @@ def save_result_arrays(self, *args, **kwargs):
             self.save_result_array(name, value, **kwargs)
     
     def get_image(self,orientation,label,image):
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             if not 'images' in h5_file:
                 raise Exception('File does not contain any images')
             if not orientation in h5_file['images']:
@@ -307,13 +323,13 @@ def get_images(self,orientation,label, *images):
         
     def get_all_image_labels(self):
         images_list = {}
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             for orientation in h5_file['/images'].keys():
                 images_list[orientation] = list(h5_file['/images'][orientation].keys())               
         return images_list                
     
     def get_image_attributes(self, orientation):
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             if not 'images' in h5_file:
                 raise Exception('File does not contain any images')
             if not orientation in h5_file['images']:
@@ -322,18 +338,18 @@ def get_image_attributes(self, orientation):
 
     def get_globals(self,group=None):
         if not group:
-            with h5py.File(self.h5_path) as h5_file:
+            with h5py.File(self.h5_path, 'r') as h5_file:
                 return dict(h5_file['globals'].attrs)
         else:
             try:
-                with h5py.File(self.h5_path) as h5_file:
+                with h5py.File(self.h5_path, 'r') as h5_file:
                     return dict(h5_file['globals'][group].attrs)
             except KeyError:
                 return {}
 
     def get_globals_raw(self, group=None):
         globals_dict = {}
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             if group == None:
                 for obj in h5_file['globals'].values():
                     temp_dict = dict(obj.attrs)
@@ -374,7 +390,7 @@ def append_expansion(name, obj):
                 for key, val in temp_dict.items():
                     if val:
                         expansion_dict[key] = val
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             h5_file['globals'].visititems(append_expansion)
         return expansion_dict
                    
@@ -385,12 +401,12 @@ def append_units(name, obj):
                 temp_dict = dict(obj.attrs)
                 for key, val in temp_dict.items():
                     units_dict[key] = val
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             h5_file['globals'].visititems(append_units)
         return units_dict
 
     def globals_groups(self):
-        with h5py.File(self.h5_path) as h5_file:
+        with h5py.File(self.h5_path, 'r') as h5_file:
             try:
                 return list(h5_file['globals'].keys())
             except KeyError:
@@ -401,14 +417,13 @@ def globals_diff(self, other_run, group=None):
     
         
 class Sequence(Run):
-    def __init__(self,h5_path,run_paths):
+    def __init__(self, h5_path, run_paths, no_write=False):
         if isinstance(run_paths, pandas.DataFrame):
             run_paths = run_paths['filepath']
         self.h5_path = h5_path
-        self.no_write = False
-        with h5py.File(h5_path) as h5_file:
-            if not 'results' in h5_file:
-                 h5_file.create_group('results')
+        self.no_write = no_write
+        if not self.no_write:
+            self._create_group_if_not_exists(h5_path, '/', 'results')
                  
         self.runs = {path: Run(path,no_write=True) for path in run_paths}
         
@@ -419,9 +434,8 @@ def __init__(self,h5_path,run_paths):
         try:
             __file__ = frame.f_back.f_locals['__file__']
             self.group = os.path.basename(__file__).split('.py')[0]
-            with h5py.File(h5_path) as h5_file:
-                if not self.group in h5_file['results']:
-                     h5_file['results'].create_group(self.group)
+            if not self.no_write:
+                self._create_group_if_not_exists(h5_path, 'results', self.group)
         except KeyError:
             sys.stderr.write('Warning: to write results, call '
             'Sequence.set_group(groupname), specifying the name of the group '
