dev

Author: davidhassell

cf/data/array/umarray.py

@@ -8,7 +8,7 @@
     load_stash2standard_name,
     parse_indices,
 )
-from ...umread_lib.umfile import File
+from ...umread_lib.umfile import File, Rec
 from .abstract import Array
 from .mixin import FileArrayMixin
 
@@ -272,14 +272,12 @@ def _get_rec(self, f, header_offset):
                 The record container.
 
         """
-        # TODOCFA: This method doesn't require data_offset and disk_length,
-        # so plays nicely with CFA. Is it fast enough that we can
-        # use this method always?
+        return Rec.from_file_and_offsets(f, header_offset)
         for v in f.vars:
             for r in v.recs:
                 if r.hdr_offset == header_offset:
+                    print (r.__dict__)
                     return r
-
     def _set_units(self, int_hdr):
         """The units and calendar properties.
 
@@ -683,4 +681,5 @@ def open(self):
             byte_ordering=self.get_byte_ordering(),
             word_size=self.get_word_size(),
             fmt=self.get_fmt(),
+            parse=False,
         )

cf/umread_lib/c-lib/bits/type_dep_protos.h

@@ -12,6 +12,12 @@ int get_var_gridcode(const INTEGER *int_hdr);
 int get_var_packing(const INTEGER *int_hdr);
 REAL get_var_real_fill_value(const REAL *int_hdr);
 
+/* int my_get_extra_data_length(const INTEGER *int_hdr);
+int my_get_num_data_words(const INTEGER *int_hdr);
+int my_get_ff_disk_length(const INTEGER *int_hdr); */
+
+
+
 /* read.c */
 
 size_t read_words(int fd, 

cf/umread_lib/c-lib/type-dep/interpret_header.c

@@ -139,3 +139,4 @@ REAL get_var_real_fill_value(const REAL *real_hdr)
 {
   return real_hdr[INDEX_BMDI];
 }
+

cf/umread_lib/c-lib/umfile.c

@@ -18,6 +18,7 @@ int get_type_and_num_words(int word_size,
     default:
       return -1;
     }
+  
 }
 
 int get_extra_data_offset_and_length(int word_size, 
@@ -163,3 +164,10 @@ int read_record_data(int fd,
   /* invalid word size falls through to error return */
   ERRBLKI;
 }
+
+
+/* Get the word address of the current file position. */
+int file_pos(int fd)
+{
+  return lseek(fd, 0, SEEK_CUR);
+}

cf/umread_lib/cInterface.py

@@ -377,6 +377,7 @@ def c_rec_to_py_rec(self, c_rec_p):
         header_offset = c_rec.header_offset
         data_offset = c_rec.data_offset
         disk_length = c_rec.disk_length
+
         return umfile.Rec(
             int_hdr, real_hdr, header_offset, data_offset, disk_length
         )

cf/umread_lib/umfile.py

@@ -11,6 +11,15 @@ class UMFileException(Exception):
     pass
 
 
+LBLREC = 14
+LBROW = 17
+LBNPT = 18
+LBEXT = 19
+LBPACK = 20
+LBEGIN = 28
+LBNREC = 29
+
+
 class File:
     """A class for a UM file that gives a view of the file including
     sets of PP records combined into variables."""
@@ -34,7 +43,7 @@ def __init__(
                 'little_endian' or 'big_endian'
 
             word_size: `int`, optional
-                4 or 8
+                The size in bytes of one word. Either ``4`` or ``8``.
 
             fmt: `str`, optional
                 'FF' or 'PP'
@@ -280,8 +289,13 @@ def __init__(
         if file:
             self.file = file
 
+    # import cf; pp  = cf.umread_lib.umfile.File('/home/david/test2.pp', fmt='PP', byte_ordering= 'little_endian' , word_size=4, parse=False)
+    # cf.umread_lib.umfile.Rec.from_file_and_offsets(pp, 4,  268, 46640)
+
     @classmethod
-    def from_file_and_offsets(cls, file, hdr_offset, data_offset, disk_length):
+    def from_file_and_offsets(
+        cls, file, hdr_offset, data_offset=None, disk_length=None
+    ):
         """Instantiate a `Rec` object from the `File` object and the
         header and data offsets.
 
@@ -295,24 +309,59 @@ def from_file_and_offsets(cls, file, hdr_offset, data_offset, disk_length):
                 into variables.
 
             hdr_offset: `int`
-                The start word in the file of the header.
+                The file start address of the header, in bytes.
 
-            data_offset: `int`
-                The start word in the file of the data.
+            data_offset: `int`, optional
+                The file start address of the data, in bytes. If
+                `None`, the default, then the data offset will be
+                calculated from the integer header.
 
             disk_length: `int`
-                The length in words of the data in the file.
+                The length in bytes of the data in the file. If
+                `None`, the default, then the disk length will be
+                calculated from the integer header.
 
         :Returns:
 
              `Rec`
 
         """
         c = file._c_interface
+        word_size = file.word_size
         int_hdr, real_hdr = c.read_header(
-            file.fd, hdr_offset, file.byte_ordering, file.word_size
+            file.fd, hdr_offset, file.byte_ordering, word_size
         )
-
+        PP =  file.fmt == "PP"
+        print (c.lib.arse)
+        if data_offset is None:
+            # Calculate the data offset from the integer header
+            if PP:
+                # We only support 64-word headers, so the data starts
+                # 66 words after the header, i.e. 64 words of the
+                # header plus two block control words.
+                data_offset = hdr_offset + 66 * word_size
+            else:
+                # Fields file
+                data_offset = int_hdr[LBEGIN] * word_size
+
+        if disk_length is None:
+            # Calculate the disk length from the integer header
+            lbpack = int_hdr[LBPACK]
+            lbnrec = int_hdr[LBNREC]
+            lblrec = int_hdr[LBLREC]
+            
+            if not lbpack:
+                disk_length = (lblrec - int_hdr[LBEXT]) * word_size
+            elif lbpack != 0 and lblrec * word_size:
+                disk_length = lblrec * word_size
+            elif lbpack % 10 == 2:
+                lbnpt = int_hdr[LBNPT]
+                lbrow = int_hdr[LBROW]
+                if lbnpt > 0 and lbrow > 0:
+                    disk_length = lbnpt * lbrow  * 4
+                else:
+                    disk_length = (lblrec - int_hdr[LBEXT]) * 4
+        
         return cls(
             int_hdr, real_hdr, hdr_offset, data_offset, disk_length, file=file
         )