More porting

lib/Docopt.php

@@ -18,15 +18,20 @@ class Docopt {
     protected static function levenshtein_norm(string $source, string $target): float {
         // NOTE: We split the strings into arrays of UTF-8 characters to match Python's behaviour (for UTF-8 input)
         // TODO: How is argv encoded in Windows and macOS? Does the Windows Command Prompt differ from Powershell?
-        $source = preg_split('/(?<!^)(?!$)/suD', $source);
-        $target = preg_split('/(?<!^)(?!$)/suD', $target);
+        $s = @preg_split('//Su', $source, -1, \PREG_SPLIT_NO_EMPTY);
+        $t = @preg_split('//Su', $target, -1, \PREG_SPLIT_NO_EMPTY);
+        // If the input was not UTF-8, we can just treat it as bytes rather than failing, since Python cannot fail here
+        if ($s === false || $t === false) {
+            $s = @preg_split('//S', $source, -1, \PREG_SPLIT_NO_EMPTY);
+            $t = @preg_split('//S', $target, -1, \PREG_SPLIT_NO_EMPTY);
+        }
         # # Compute Levenshtein distance using helper function. The max is always
         # # just the length of the longer string, so this is used to normalize result
         # # before returning it
         # distance = levenshtein(source, target)
         # return float(distance) / max(len(source), len(target))
-        $distance = static::levenshtein($source, $target);
-        return (float) ($distance / max(sizeof($source), sizeof($target)));
+        $distance = static::levenshtein($s, $t);
+        return (float) ($distance / max(sizeof($s), sizeof($t)));
     }
 
     # def levenshtein(source: str, target: str) -> int:
@@ -100,6 +105,45 @@ class Docopt {
         # return matrix[len(source)][len(target)]
         return $matrix[sizeof($source)][sizeof($target)];
     }
+
+    # def transform(pattern: "BranchPattern") -> "Either":
+    #     """Expand pattern into an (almost) equivalent one, but with single Either.
+    #
+    #     Example: ((-a | -b) (-c | -d)) => (-a -c | -a -d | -b -c | -b -d)
+    #     Quirks: [-a] => (-a), (-a...) => (-a -a)
+    #
+    #     """
+    protected static function transform(BranchPattern $pattern): Either {
+        # result = []
+        # groups = [[pattern]]
+        $result = [];
+        $groups = [[$pattern]];
+        # while groups:
+        while ($groups) {
+            # children = groups.pop(0)
+            $children = array_shift($groups);
+            # parents = [Required, NotRequired, OptionsShortcut, Either, OneOrMore]
+            $parents = [Required::class, NotRequired::class, OptionsShortcut::class, Either::class, OneOrMore::class];
+            # if any(t in map(type, children) for t in parents):
+            $anyParentTypeInChildren = array_reduce($children, function($out, $c) use ($parents) {
+                return $out ?: in_array(get_class($c), $parents);
+            }, false);
+            if ($anyParentTypeInChildren) {
+                # child = [c for c in children if type(c) in parents][0]
+                # children.remove(child)
+                # if type(child) is Either:
+                #     for c in child.children:
+                #         groups.append([c] + children)
+                # elif type(child) is OneOrMore:
+                #     groups.append(child.children * 2 + children)
+                # else:
+                #     groups.append(child.children + children)
+            }
+            # else:
+            #     result.append(children)
+        }
+        # return Either(*[Required(*e) for e in result])
+    }
 }
 
 # class DocoptLanguageError(Exception):
@@ -135,46 +179,32 @@ class DocoptExit extends \Exception {
 }
 
 # class Pattern:
-#     def __init__(self, name: Optional[str], value: Optional[Union[List[str], str, int]] = None) -> None:
-#         self._name, self.value = name, value
-
-#     @property
-#     def name(self) -> Optional[str]:
-#         return self._name
-
-#     def __eq__(self, other: Any) -> bool:
-#         return repr(self) == repr(other)
-
-#     def __hash__(self) -> int:
-#         return hash(repr(self))
-
+class Pattern {
+    public $name = null; // DEVIATION: This is a read-only property in Python. As this class is not part of the public API, this was deemed unnecessary complication
+    public $value = null;
+
+    # def __init__(self, name: Optional[str], value: Optional[Union[List[str], str, int]] = None) -> None:
+    public function __construct(?string $name, $value = null) {
+        # self._name, self.value = name, value
+        $this->name = $name;
+        $this->value = $value;
+    }
 
-# def transform(pattern: "BranchPattern") -> "Either":
-#     """Expand pattern into an (almost) equivalent one, but with single Either.
+    # @property
+    # def name(self) -> Optional[str]:
+    #     return self._name
+    // DEVIATION: Not implemented; see property definition above
 
-#     Example: ((-a | -b) (-c | -d)) => (-a -c | -a -d | -b -c | -b -d)
-#     Quirks: [-a] => (-a), (-a...) => (-a -a)
-
-#     """
-#     result = []
-#     groups = [[pattern]]
-#     while groups:
-#         children = groups.pop(0)
-#         parents = [Required, NotRequired, OptionsShortcut, Either, OneOrMore]
-#         if any(t in map(type, children) for t in parents):
-#             child = [c for c in children if type(c) in parents][0]
-#             children.remove(child)
-#             if type(child) is Either:
-#                 for c in child.children:
-#                     groups.append([c] + children)
-#             elif type(child) is OneOrMore:
-#                 groups.append(child.children * 2 + children)
-#             else:
-#                 groups.append(child.children + children)
-#         else:
-#             result.append(children)
-#     return Either(*[Required(*e) for e in result])
+    # def __eq__(self, other: Any) -> bool:
+    #     return repr(self) == repr(other)
+    // DEVIATION: Operators cannot be overloaded in PHP, thus this is not implemented
+    // TODO: Do we need to implement an explicit means oof comparison?
 
+    # def __hash__(self) -> int:
+    #     return hash(repr(self))
+    // DEVIATION: Not implemented as it is only used internally for object comparison
+    // See https://docs.python.org/3/reference/datamodel.html?highlight=__hash__#object.__hash__
+}
 
 # TSingleMatch = Tuple[Union[int, None], Union["LeafPattern", None]]