Optimize Array.info and Group.info (#1733)

* Optimize Array.info.

Avoid repeated computes of the same value (getsize)

* Don't have InfoReporter query items twice.

Apparently IPython will run both __repr__ and _repr_html_
so we were calling `getsize` twice.

* Group too

* Apply suggestions from code review

Co-authored-by: Joe Hamman <[email protected]>

---------

Co-authored-by: Joe Hamman <[email protected]>

Author: dcherian

docs/release.rst

@@ -21,6 +21,8 @@ Unreleased
 Enhancements
 ~~~~~~~~~~~~
 
+* Optimize ``Array.info`` so that it calls `getsize` only once.
+  By :user:`Deepak Cherian <dcherian>`.
 * Override IPython ``_repr_*_`` methods to avoid expensive lookups against object stores.
   By :user:`Deepak Cherian <dcherian>` :issue:`1716`.
 

zarr/core.py

@@ -176,7 +176,6 @@ def __init__(
         )
 
         # initialize info reporter
-        self._info_reporter = InfoReporter(self)
 
         # initialize indexing helpers
         self._oindex = OIndex(self)
@@ -2429,7 +2428,7 @@ def info(self):
         Chunks initialized : 0/10
 
         """
-        return self._info_reporter
+        return InfoReporter(self)
 
     def info_items(self):
         return self._synchronized_op(self._info_items_nosync)
@@ -2471,14 +2470,16 @@ def bytestr(n):
             items += [("Synchronizer type", typestr(self._synchronizer))]
 
         # storage info
+        nbytes = self.nbytes
+        nbytes_stored = self.nbytes_stored
         items += [("Store type", typestr(self._store))]
         if self._chunk_store is not None:
             items += [("Chunk store type", typestr(self._chunk_store))]
-        items += [("No. bytes", bytestr(self.nbytes))]
-        if self.nbytes_stored > 0:
+        items += [("No. bytes", bytestr(nbytes))]
+        if nbytes_stored > 0:
             items += [
-                ("No. bytes stored", bytestr(self.nbytes_stored)),
-                ("Storage ratio", f"{self.nbytes / self.nbytes_stored:.1f}"),
+                ("No. bytes stored", bytestr(nbytes_stored)),
+                ("Storage ratio", f"{nbytes / nbytes_stored:.1f}"),
             ]
         items += [("Chunks initialized", f"{self.nchunks_initialized}/{self.nchunks}")]
 

zarr/hierarchy.py

@@ -211,7 +211,6 @@ def __init__(
         )
 
         # setup info
-        self._info = InfoReporter(self)
 
     @property
     def store(self):
@@ -266,7 +265,7 @@ def attrs(self):
     @property
     def info(self):
         """Return diagnostic information about the group."""
-        return self._info
+        return InfoReporter(self)
 
     @property
     def meta_array(self):

zarr/util.py

@@ -408,14 +408,13 @@ def info_html_report(items) -> str:
 class InfoReporter:
     def __init__(self, obj):
         self.obj = obj
+        self.items = self.obj.info_items()
 
     def __repr__(self):
-        items = self.obj.info_items()
-        return info_text_report(items)
+        return info_text_report(self.items)
 
     def _repr_html_(self):
-        items = self.obj.info_items()
-        return info_html_report(items)
+        return info_html_report(self.items)
 
 
 class TreeNode: