fix pagination bug for H5Image

h5pyd/_hl/files.py

@@ -42,7 +42,12 @@ def is_hdf5(domain, **kwargs):
 
 
 class H5Image(io.RawIOBase):
+    """ file-like-object class that treats bytes of an HSDS dataset as an HDF5 file image
+        Can be used as a subsitute for a file path in h5py.File(filepath).  E.g.:
+        f = h5py.File(H5Image("hdf5:/myhsds_domain"))   """
+
     def __init__(self, domain_path, h5path="h5image", logger=None):
+        """ verify dataset can be accessed and set logger if supplied """
         self._cursor = 0
         if domain_path.startswith("hdf5::/"):
             self._domain_path = domain_path
@@ -62,6 +67,7 @@ def __init__(self, domain_path, h5path="h5image", logger=None):
             self._logger.info(f"domain {self._domain_path} opened")
 
     def __repr__(self):
+        """ Just rturn the domain path"""
         return f'<{self._domain_path}>'
 
     def readable(self):
@@ -82,6 +88,7 @@ def tell(self):
         return self._cursor
 
     def seek(self, offset, whence=io.SEEK_SET):
+        """ set the seek pointer """
         if whence == io.SEEK_SET:
             if self._logger:
                 self._logger.debug(f"SEEK_SET({offset})")
@@ -101,6 +108,8 @@ def seek(self, offset, whence=io.SEEK_SET):
         return self._cursor
 
     def _get_page(self, page_number):
+        """ Return bytes for the given page.
+            Read a page from the HSDS dataset if not already in the cache """
         if self._page_cache[page_number] is None:
             if self._logger:
                 self._logger.info(f"reading page {page_number} from server")
@@ -112,6 +121,7 @@ def _get_page(self, page_number):
         return self._page_cache[page_number]
 
     def read(self, size=-1):
+        """ Read size bytes from the cursor """
         start = self._cursor
         if size < 0 or self._cursor + size >= self.size:
             stop = self.size
@@ -125,17 +135,22 @@ def read(self, size=-1):
 
         buffer = bytearray(stop - start)
         offset = start
-        while offset - start < size:
+        while offset < stop:
             page_number = offset // self.page_size
             page_bytes = self._get_page(page_number)
-            num_bytes = ((offset + 1) + self.page_size // self.page_size) + self.page_size
-            if offset + num_bytes - start > size:
-                num_bytes = start + size - offset
-            page_start = offset % self.page_size
-            page_stop = page_start + num_bytes
+            n = offset % self.page_size
+            if stop // self.page_size > page_number:
+                # just read to the end of the page
+                m = self.page_size
+            else:
+                # remaing bytes don't cross page boundry
+                m = n + (stop - offset)
+
+            num_bytes = m - n
             buffer_start = offset - start
             buffer_stop = buffer_start + num_bytes
-            buffer[buffer_start:buffer_stop] = page_bytes[page_start:page_stop]
+            buffer[buffer_start:buffer_stop] = page_bytes[n:m]
+
             offset += num_bytes
 
         if self._logger: