Support for multi formatting elements in savetxt (Trac #663)

[numpy-gitbot]

Original ticket http://projects.scipy.org/numpy/ticket/663 on 2008-02-21 by @huard, assigned to unknown.

Here is a patch to add support for multiple formats in savetxt.

Index: numpy/lib/io.py
===================================================================
--- numpy/lib/io.py     (revision 4815)
+++ numpy/lib/io.py     (working copy)
@@ -326,6 +326,10 @@
     the file is automatically saved in compressed gzip format.  The load()
     command understands gzipped files transparently.

+    fmt can be a single format (%10.5f), a sequence of formats, or a 
+    single multi-format string, e.g. 'Iteration %d -- %10.5f', in which
+    case delimiter is ignored.
+    
     Example usage:

     save('test.out', X)         # X is an array
@@ -354,8 +358,23 @@
     if len(X.shape)==1:
         origShape = X.shape
         X.shape = len(X), 1
+    
+    # Fmt can be a string with multiple insertion points or a list of formats.
+    # E.g. '%10.5f\t%10d' or ('%10.5f', '$10d')
+    if type(fmt) in (list, tuple):
+        if len(fmt) != X.shape[1]:
+            raise AttributeError, 'fmt has wrong shape.  '+ str(fmt)
+        format = delimiter.join(fmt)
+    elif type(fmt) is str:
+        if fmt.count('%') == 1:
+            fmt = [fmt,]*X.shape[1]
+            format = delimiter.join(fmt)
+        elif fmt.count('%') != X.shape[1]:
+            raise AttributeError, 'fmt has wrong number of % formats.  ' + fmt
+        else:
+            format = fmt
     for row in X:
-        fh.write(delimiter.join([fmt%val for val in row]) + '\n')
+        fh.write(format%tuple(row) + '\n')

     if origShape is not None:
         X.shape = origShape

[numpy-gitbot]

@huard wrote on 2008-04-09

Let me add some tests, it's going to be easier to review.

[numpy-gitbot]

Attachment added by @huard on 2008-04-09: io.patch

[numpy-gitbot]

@huard wrote on 2008-04-16

To increase the probability of someone reviewing this, I tried to simplify the process. I committed the part of the patch related to docstring enhancements as well as fixes and additions to the tests (r5037). The test for multi-format is simply commented out in the test_io.py file. Here is the slimmed version of the patch for multi-format support in io.py. Simply uncomment the tests related to format in test_io.py to make sure everything works as advertised.

Index: io.py
===================================================================
--- io.py       (revision 5037)
+++ io.py       (working copy)
@@ -316,7 +316,6 @@
     else:  return X


-# adjust so that fmt can change across columns if desired.

 def savetxt(fname, X, fmt='%.18e',delimiter=' '):
     """
@@ -331,9 +330,9 @@
       transparently.
     X : array or sequence
       Data to write to file.
-    fmt : string 
-      A format string %[flags][width][.precision]specifier. See notes below for 
-      a description of some common flags and specifiers.
+    fmt : format string or sequence of format strings 
+      A single format (%10.5f), a sequence of formats, or a multi-format string, 
+      e.g. 'Iteration %d -- %10.5f', in which case delimiter is ignored.
     delimiter : str
       Character separating columns.

@@ -382,18 +381,34 @@
     else:
         raise ValueError('fname must be a string or file handle')

+    X = np.asarray(X)
+    if X.ndim == 1:
+        if X.dtype.names is None:
+            X = np.atleast_2d(X).T
+            ncol = 1
+        else:
+            ncol = len(X.dtype.descr)
+    else:
+        ncol = X.shape[1]
+        
+    # Fmt can be a string with multiple insertion points or a list of formats.
+    # E.g. '%10.5f\t%10d' or ('%10.5f', '$10d')
+    if type(fmt) in (list, tuple):
+        if len(fmt) != ncol:
+            raise AttributeError, 'fmt has wrong shape.  '+ str(fmt)
+        format = delimiter.join(fmt)
+    elif type(fmt) is str:
+        if fmt.count('%') == 1:
+            fmt = [fmt,]*ncol
+            format = delimiter.join(fmt)
+        elif fmt.count('%') != ncol:
+            raise AttributeError, 'fmt has wrong number of % formats.  ' + fmt
+        else:
+            format = fmt

-    X = np.asarray(X)
-    origShape = None
-    if len(X.shape)==1 and X.dtype.names is None:
-        origShape = X.shape
-        X.shape = len(X), 1
     for row in X:
-        fh.write(delimiter.join([fmt%val for val in row]) + '\n')
+        fh.write(format%tuple(row) + '\n')

-    if origShape is not None:
-        X.shape = origShape
-
 import re
 def fromregex(file, regexp, dtype):
     """Construct a record array from a text file, using regular-expressions parsing.

[numpy-gitbot]

@stefanv wrote on 2008-04-28

Thank you, David. Applied in r5108.