From e86ffbb3ae81dfeedd5992125e7bb6c365bc58b5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=C3=A9d=C3=A9ric=20Boquien?= <mederic.boquien@uantof.cl>
Date: Sat, 30 Jan 2016 13:14:25 -0300
Subject: [PATCH] So far no validation was made on the input parameters
 provided in pcigale.ini. This could lead to runtime crashes when an argument
 took an unexpected value. As we are using the ConfigObj module to handle the
 configuration file, we can also use the validate module that comes with it.
 However the standard validation functions are not precise enough in our case.
 This file defines two validation functions for input parameters, and a new
 exception that will be used in a subsequent patch.

---
 pcigale/session/validation.py | 134 ++++++++++++++++++++++++++++++++++
 1 file changed, 134 insertions(+)
 create mode 100644 pcigale/session/validation.py

diff --git a/pcigale/session/validation.py b/pcigale/session/validation.py
new file mode 100644
index 00000000..3d6dd490
--- /dev/null
+++ b/pcigale/session/validation.py
@@ -0,0 +1,134 @@
+# -*- coding: utf-8 -*-
+# Copyright (C) 2016 Universidad de Antofagasta
+# Licensed under the CeCILL-v2 licence - see Licence_CeCILL_V2-en.txt
+# Author: Médéric Boquien
+
+import numpy as np
+import validate as vdt
+
+
+class VdtValueRepeatedError(vdt.VdtValueError):
+    """The value supplied was repeated."""
+
+    def __init__(self, value):
+        """
+        >>> raise VdtValueRepeatedError('jedie')
+        Traceback (most recent call last):
+        VdtValueRepeatedError: the value "jedie" is too long.
+        """
+        vdt.ValidateError.__init__(self, 'the value {} is repeated.'.format(value))
+
+
+def is_cigale_list(inobject, dtype='float', minvalue=None, maxvalue=None,
+                   options=None):
+    """Function that returns a list made of float or int numpy. It also checks
+    if any element is lower/high then a set minimum/maximum value or if any
+    does not belong to a predefined list. If any of the tests fails, an
+    exception is thrown. All the parameters besides "value" are strings because
+    configobj/validate pass arguments as such.
+
+    Parameters
+    ----------
+    inobject: list or string
+        Object that we need to convert and verify
+    dtype: string
+        Desired type of the object
+    minvalue: string
+        Minimum value of all the elements
+    maxvalue: string
+        Maximum value of all the elements
+    options: string
+        All the values that the array can take
+
+    Returns
+    -------
+    outobject: list
+        List of the requested type and respecting the constraints
+    """
+
+    if dtype == 'float':
+        dtype = float
+        conv_type = lambda x: float(x)
+    elif dtype == 'int':
+        dtype = int
+        conv_type = lambda x: int(float(x))
+    else:
+        raise Exception("Unsupported data type. Only float and int are "
+                        "supported.")
+
+    if isinstance(inobject, str):
+        if inobject.startswith('eval '):
+            outobject = eval(inobject[4:])
+            # If the evaluation lead to a single value, we put it in a list.
+            if not isinstance(outobject, Iterable):
+                outobject = [outobject]
+        elif inobject.startswith('range '):
+            start, stop, step = [conv_type(item) for item in inobject[5:].split()]
+            outobject = np.arange(start, stop+step, step)
+        else:
+            # We need to return a list to combine the list of possible values
+            # for each parameter.
+            outobject = [inobject]
+    else:
+        outobject = inobject
+
+    try:  # We convert to an array as it is more convenient for tests
+        outobject = np.asarray(outobject, dtype)
+    except ValueError:
+        raise vdt.VdtValueError(inobject)
+
+    # Test if there are two identical elements
+    unique_outobject, counts = np.unique(outobject, return_counts=True)
+    if outobject.size > unique_outobject.size:
+        raise VdtValueRepeatedError(unique_outobject[counts > 1])
+
+    # Test if any value is smaller than the minimum value
+    if minvalue is not None:
+        minvalue = conv_type(minvalue)
+        if np.any(outobject < minvalue):
+            raise vdt.VdtValueTooSmallError(outobject[outobject < minvalue])
+
+    # Test if any value is larger than the maximum value
+    if maxvalue is not None:
+        maxvalue = conv_type(maxvalue)
+        if np.any(outobject > maxvalue):
+            raise vdt.VdtValueTooBigError(outobject[outobject > maxvalue])
+
+    # Test if any value is not in the list of allowed options
+    if options is not None:
+        options = {conv_type(option) for option in options.split('&')}
+        if not set(outobject).issubset(options):
+            raise vdt.VdtValueError(set(outobject) - options)
+
+    return list(outobject)
+
+
+def is_cigale_string_list(inobject):
+    """Function that returns a list of strings. If inobject is already a list,
+    it is returned as is. Otherwise the string is put into a list. If the
+    string is empty, it returns an empty list.
+
+    Parameters
+    ----------
+    inobject: list or string
+        Object that we need to convert and verify
+
+    Returns
+    -------
+    outobject: list
+        Object transformed into a list
+
+    """
+    if isinstance(inobject, str):
+        if len(inobject) == 0:
+            return []
+        else:
+            return [inobject]
+
+    return inobject
+
+
+# Dictionary of functions that are to be passed to the validator so it knows
+# about extra tests that are not covered with the base functions
+functions = {'cigale_list': is_cigale_list,
+             'cigale_string_list': is_cigale_string_list}
-- 
GitLab