feat(whoosh): implement SkipLists for documents searching

Co-authored-by: Matheus Virgolino <matheus.virgolino.abilio.da.silva@ccc.ufcg.edu.br>
Co-authored-by: Manoel Netto <manoel.da.nobrega.eustaqueo.netto@ccc.ufcg.edu.br>
Co-authored-by: Pedro <pedroalmeida1896@gmail.com>
Co-authored-by: Lucaslg7 <lucasmoizinholg7@gmail.com>
Co-authored-by: RailtonDantas <railtondantas.code@gmail.com>
Co-authored-by: João Pereira <joao.pereira.de.oliveira@ccc.ufcg.edu.br>

Author: 6 people

src/whoosh/matching/__init__.py

@@ -48,6 +48,7 @@
     NullMatcher,
     NullMatcherClass,
     ReadTooFar,
+    SkipListMatcher,
 )
 from whoosh.matching.wrappers import (
     ConstantScoreWrapperMatcher,

src/whoosh/matching/mcore.py

@@ -51,7 +51,8 @@
 
 from abc import abstractmethod
 from itertools import repeat
-
+from whoosh.matching.skiplist import SkipList
+from bisect import bisect_left
 # Exceptions
 
 
@@ -572,6 +573,22 @@ def score(self):
         else:
             return self.weight()
 
+class SkipListMatcher(ListMatcher):
+    def __init__(self, ids, **kwargs):
+        super().__init__(ids, **kwargs)
+        self._skiplist = SkipList(ids)
+
+    def skip_to(self, id):
+        if not self.is_active():
+            raise ReadTooFar
+        if id < self.id():
+            return
+
+        node = self._skiplist.skip_to(id)
+        if node is not None:
+           self._i = bisect_left(self._ids, node.doc_id, self._i)
+        else:
+            self._i = len(self._ids)
 
 # Term/vector leaf posting matcher middleware
 

src/whoosh/matching/skiplist.py

@@ -0,0 +1,63 @@
+import random
+
+class SkipNode:
+    __slots__ = ("doc_id", "forward")
+
+    def __init__(self, doc_id, level):
+        self.doc_id = doc_id
+        self.forward = [None] * (level + 1)
+
+class SkipList:
+    def __init__(self, ids, max_level=16, p = 0.5):
+        self.max_level = max_level
+        self.p = p
+        self.header = SkipNode(-1, max_level)
+        self.level = 0
+        self.size = len(ids)
+        self._build(ids)
+
+    def _random_level(self):
+        level = 0
+        while random.random() < self.p and level < self.max_level:
+            level += 1
+        return level
+    
+    def _build(self, ids):
+        update = [self.header] * (self.max_level + 1)
+        for doc_id in ids:
+            lvl = self._random_level()
+            if lvl > self.level:
+                self.level = lvl
+            
+            node = SkipNode(doc_id, lvl)
+            for i in range(lvl + 1):
+                node.forward[i] = update[i].forward[i]
+                update[i].forward[i] = node
+                update[i] = node
+
+    
+    def skip_to(self, target_id):
+        current = self.header
+
+        for i in range(self.level, -1, -1):
+            while (
+                current.forward[i] is not None
+                and current.forward[i].doc_id < target_id
+            ):
+                current = current.forward[i]
+
+        current = current.forward[0]
+        return current
+    
+    def __len__(self):
+        return self.size
+    
+    def __iter__(self):
+        node = self.header.forward[0]
+        while node is not None:
+            yield node.doc_id
+            node = node.forward[0]
+    def __contains__(self, doc_id):
+        node = self.skip_to(doc_id)
+        return node is not None and node.doc_id == doc_id
+