sqlglot.trie

 1import typing as t
 2
 3key = t.Sequence[t.Hashable]
 4
 5
 6def new_trie(keywords: t.Iterable[key]) -> t.Dict:
 7    """
 8    Creates a new trie out of a collection of keywords.
 9
10    The trie is represented as a sequence of nested dictionaries keyed by either single character
11    strings, or by 0, which is used to designate that a keyword is in the trie.
12
13    Example:
14        >>> new_trie(["bla", "foo", "blab"])
15        {'b': {'l': {'a': {0: True, 'b': {0: True}}}}, 'f': {'o': {'o': {0: True}}}}
16
17    Args:
18        keywords: the keywords to create the trie from.
19
20    Returns:
21        The trie corresponding to `keywords`.
22    """
23    trie: t.Dict = {}
24
25    for key in keywords:
26        current = trie
27
28        for char in key:
29            current = current.setdefault(char, {})
30        current[0] = True
31
32    return trie
33
34
35def in_trie(trie: t.Dict, key: key) -> t.Tuple[int, t.Dict]:
36    """
37    Checks whether a key is in a trie.
38
39    Examples:
40        >>> in_trie(new_trie(["cat"]), "bob")
41        (0, {'c': {'a': {'t': {0: True}}}})
42
43        >>> in_trie(new_trie(["cat"]), "ca")
44        (1, {'t': {0: True}})
45
46        >>> in_trie(new_trie(["cat"]), "cat")
47        (2, {0: True})
48
49    Args:
50        trie: the trie to be searched.
51        key: the target key.
52
53    Returns:
54        A pair `(value, subtrie)`, where `subtrie` is the sub-trie we get at the point where the search stops, and `value`
55        is either 0 (search was unsuccessful), 1 (`value` is a prefix of a keyword in `trie`) or 2 (`key is in `trie`).
56    """
57    if not key:
58        return (0, trie)
59
60    current = trie
61
62    for char in key:
63        if char not in current:
64            return (0, current)
65        current = current[char]
66
67    if 0 in current:
68        return (2, current)
69    return (1, current)