src/Alignment.bend

bend-edit-distance@0.2.2.0 · 111 lines · 14 definitions

1import Base
2import bend-edit-distance@0.2.2.0/src/Sequence.bend as S
3import bend-edit-distance@0.2.2.0/EdlibTypes.bend as E
4import bend-edit-distance@0.2.2.0/src/Reconstruction.bend as R
5
6# Myers owns nonempty distance and location searches. This module builds
7# ordered results, preserves empty-input returns and validates reconstructed PATH.
8type Location is Data:
9  Location{end: E.Coordinate, start: Nat}
10
11type Best is Data:
12  Best{score: U32, locations: List<&2, Location>}
13
14def update_best(best: Best, score: U32, location: Location, comparison: Cmp) -> Best:
15  match best comparison:
16    case Best{_, _} LT{}:
17      Best{score, [location]}
18    case Best{old, locations} EQ{}:
19      Best{old, location <> locations}
20    case _ GT{}:
21      best
22
23def ordered(best: Best) -> Best:
24  match best:
25    case Best{score, locations}:
26      Best{score, List.reverse(&2, Location, locations)}
27
28def ends(locations: List<&2, Location>) -> List<&2, E.Coordinate>:
29  match locations:
30    case Nil{}:
31      Nil{}
32    case Location{end, _} <> tail:
33      end <> ends(tail)
34
35def starts(locations: List<&2, Location>) -> List<&2, E.Coordinate>:
36  match locations:
37    case Nil{}:
38      Nil{}
39    case Location{_, start} <> tail:
40      E.Position{start} <> starts(tail)
41
42def span(target: String, location: Location) -> String:
43  match location:
44    case Location{E.BeforeStart{}, _}:
45      ""
46    case Location{E.Position{index}, start}:
47      String.drop(String.take(target, 1n+index), start)
48
49def path_result(score: U32, +locations: List<&2, Location>,
50                +operations: List<&2, E.Operation>, format: E.CigarFormat) -> E.AlignResult:
51  E.AlignResult{E.Ok{}, E.Found{U32.to_nat(score)}, ends(locations),
52    Some{starts(locations)}, Some{operations}, Some{E.alignment_to_cigar(operations, format)}}
53
54def location_valid(target: String, location: Location) -> Bool:
55  match location:
56    case Location{E.BeforeStart{}, start}:
57      Nat.is_eq(start, 0n)
58    case Location{E.Position{+end}, start}:
59      Bool.and(Nat.is_lt(end, S.length(target)), Nat.is_le(start, 1n+end))
60
61def path_checked(score: U32, locations: List<&2, Location>, format: E.CigarFormat,
62                  operations: List<&2, E.Operation>, valid: Bool) -> E.AlignResult:
63  match valid:
64    case True{}:
65      path_result(score, locations, operations, format)
66    case False{}:
67      E.AlignResult{E.Error{}, E.NotFound{}, Nil{}, None{}, None{}, None{}}
68
69def path_validate(query: String, segment: String, pairs: List<&2, E.EqualityPair>,
70                   +score: U32, locations: List<&2, Location>, format: E.CigarFormat,
71                   +operations: List<&2, E.Operation>, valid_location: Bool) -> E.AlignResult:
72  path_checked(score, locations, format, operations,
73    Bool.and(valid_location, R.valid_path(query, segment, operations, pairs, U32.to_nat(score))))
74
75def path_reconstruct(+query: String, +segment: String, +pairs: List<&2, E.EqualityPair>,
76                      score: U32, locations: List<&2, Location>, format: E.CigarFormat,
77                      valid_location: Bool) -> E.AlignResult:
78  path_validate(query, segment, pairs, score, locations, format,
79    R.reconstruct(query, segment, pairs), valid_location)
80
81def result(query: String, target: String, task: E.Task,
82           pairs: List<&2, E.EqualityPair>, format: E.CigarFormat,
83           best: Best, accepted: Bool) -> E.AlignResult:
84  match task best accepted:
85    case _ _ False{}:
86      E.not_found()
87    case E.Distance{} Best{score, locations} True{}:
88      E.AlignResult{E.Ok{}, E.Found{U32.to_nat(score)}, ends(locations), None{}, None{}, None{}}
89    case E.Locations{} Best{score, +locations} True{}:
90      E.AlignResult{E.Ok{}, E.Found{U32.to_nat(score)}, ends(locations), Some{starts(locations)}, None{}, None{}}
91    case E.Path{} Best{_, Nil{}} True{}:
92      E.AlignResult{E.Error{}, E.NotFound{}, Nil{}, None{}, None{}, None{}}
93    case E.Path{} Best{score, +location <> tail} True{}:
94      +sequence = target
95      path_reconstruct(query, span(sequence, location), pairs, score,
96        location <> tail, format, location_valid(sequence, location))
97
98# Pinned Edlib returns before applying k or calculating starts/path on empty
99# inputs, including PATH. Preserve that observable oracle behavior.
100def empty_result(query: String, target: String, config: E.AlignConfig) -> E.AlignResult:
101  match query target config:
102    case SNil{} +t E.AlignConfig{E.NW{}, _, _, _, _}:
103      +length = S.length(t)
104      E.distance_result(length, E.end_coordinate(length), True{})
105    case SNil{} _ _:
106      E.distance_result(0n, E.BeforeStart{}, True{})
107    case SCon{_, qt} SNil{} _:
108      E.distance_result(1n+S.length(qt), E.BeforeStart{}, True{})
109    # Internal precondition: Myers calls this only when one input is empty.
110    case _ _ _:
111      E.AlignResult{E.Error{}, E.NotFound{}, Nil{}, None{}, None{}, None{}}