是否可以使用pyDatalog来检查是否满足所有数据依赖？

Question

我试图使用pyDatalog来确定各种特性的依赖关系是否得到满足。一些图书馆(lA，lB，.)提供产出(1,2，.)这是特性(fX，fY，.)需要的。

例如：

+has("lA", 1)   #Library A has output 1
+has("lA", 2)   #Library A has output 2
+has("lB", 2)   #Library B has output 2
+needs("fX", 1) #Feature X needs output 1
+needs("fX", 2) #Feature X needs output 2
+needs("fY", 2) #Feature Y needs output 2

使用pyDatalog图形教程，我可以找到至少提供一个特性所需输出的库：

lib_supports_feature(X, Z) <= has(X, Y) & needs(Z, Y)
lib_supports_feature(X,"fX")

这会返回：('lA'，)，('lB‘)，因为它只是找到至少有一个路径的任何库。

是否有一种使用pyDatalog只返回满足该特性所有需求的库的好方法？

Answer 1

您需要使用双重否定：

missing(X,Y,Z) <= ( # Y is a missing output of X, and is needed by Z
               needs(Z, Y) 
               & has(X, Y1) # needed only if X is not bound
               & ~ has(X, Y))

lib_full_supports_feature(X,Z) <= ( # like supports, but there is no missing output
               has(X,Y) 
               & needs(Z,Y) 
               & ~ missing(X, Y1, Z))

Answer 2

您可以计算supported_and_needed特性的数量，并将它们与所需特性的数量进行比较。如果它们是平等的，那么所有的功能需求都会得到满足。

(lib_num_supported_and_needed_features[X, Z] == len_(Y)) <= (has(X, Y) & needs(Z, Y))
(num_needed_features[Z] == len_(Y)) <= (needs(Z, Y))
lib_full_supports_features(X, Z) <= (num_needed_features[Z] == lib_num_supported_and_needed_features[X, Z])