src/Alignment.bend

bend-edit-distance@0.2.1.0 · 110 lines · 0 definitions

1import Base
2import bend-edit-distance@0.2.1.0/EdlibTypes.bend as E
3import bend-edit-distance@0.2.1.0/src/Reconstruction.bend as R
4
5# Myers owns nonempty distance and location searches. This module builds
6# ordered results, preserves empty-input returns and validates reconstructed PATH.
7type Location is Data:
8  Location{end: E.Coordinate, start: Nat}
9
10type Best is Data:
11  Best{score: U32, locations: List<&2, Location>}
12
13def update_best(best: Best, score: U32, location: Location, comparison: Cmp) -> Best:
14  match best comparison:
15    case Best{_, _} LT{}:
16      Best{score, [location]}
17    case Best{old, locations} EQ{}:
18      Best{old, location <> locations}
19    case _ GT{}:
20      best
21
22def ordered(best: Best) -> Best:
23  match best:
24    case Best{score, locations}:
25      Best{score, List.reverse(&2, Location, locations)}
26
27def ends(locations: List<&2, Location>) -> List<&2, E.Coordinate>:
28  match locations:
29    case Nil{}:
30      Nil{}
31    case Location{end, _} <> tail:
32      end <> ends(tail)
33
34def starts(locations: List<&2, Location>) -> List<&2, E.Coordinate>:
35  match locations:
36    case Nil{}:
37      Nil{}
38    case Location{_, start} <> tail:
39      E.Position{start} <> starts(tail)
40
41def span(target: String, location: Location) -> String:
42  match location:
43    case Location{E.BeforeStart{}, _}:
44      ""
45    case Location{E.Position{index}, start}:
46      String.drop(String.take(target, 1n+index), start)
47
48def path_result(score: U32, +locations: List<&2, Location>,
49                +operations: List<&2, E.Operation>, format: E.CigarFormat) -> E.AlignResult:
50  E.AlignResult{E.Ok{}, E.Found{U32.to_nat(score)}, ends(locations),
51    Some{starts(locations)}, Some{operations}, Some{E.alignment_to_cigar(operations, format)}}
52
53def location_valid(target: String, location: Location) -> Bool:
54  match location:
55    case Location{E.BeforeStart{}, start}:
56      Nat.is_eq(start, 0n)
57    case Location{E.Position{+end}, start}:
58      Bool.and(Nat.is_lt(end, String.length(target)), Nat.is_le(start, 1n+end))
59
60def path_checked(score: U32, locations: List<&2, Location>, format: E.CigarFormat,
61                  operations: List<&2, E.Operation>, valid: Bool) -> E.AlignResult:
62  match valid:
63    case True{}:
64      path_result(score, locations, operations, format)
65    case False{}:
66      E.AlignResult{E.Error{}, E.NotFound{}, Nil{}, None{}, None{}, None{}}
67
68def path_validate(query: String, segment: String, pairs: List<&2, E.EqualityPair>,
69                   +score: U32, locations: List<&2, Location>, format: E.CigarFormat,
70                   +operations: List<&2, E.Operation>, valid_location: Bool) -> E.AlignResult:
71  path_checked(score, locations, format, operations,
72    Bool.and(valid_location, R.valid_path(query, segment, operations, pairs, U32.to_nat(score))))
73
74def path_reconstruct(+query: String, +segment: String, +pairs: List<&2, E.EqualityPair>,
75                      score: U32, locations: List<&2, Location>, format: E.CigarFormat,
76                      valid_location: Bool) -> E.AlignResult:
77  path_validate(query, segment, pairs, score, locations, format,
78    R.reconstruct(query, segment, pairs), valid_location)
79
80def result(query: String, target: String, task: E.Task,
81           pairs: List<&2, E.EqualityPair>, format: E.CigarFormat,
82           best: Best, accepted: Bool) -> E.AlignResult:
83  match task best accepted:
84    case _ _ False{}:
85      E.not_found()
86    case E.Distance{} Best{score, locations} True{}:
87      E.AlignResult{E.Ok{}, E.Found{U32.to_nat(score)}, ends(locations), None{}, None{}, None{}}
88    case E.Locations{} Best{score, +locations} True{}:
89      E.AlignResult{E.Ok{}, E.Found{U32.to_nat(score)}, ends(locations), Some{starts(locations)}, None{}, None{}}
90    case E.Path{} Best{_, Nil{}} True{}:
91      E.AlignResult{E.Error{}, E.NotFound{}, Nil{}, None{}, None{}, None{}}
92    case E.Path{} Best{score, +location <> tail} True{}:
93      +sequence = target
94      path_reconstruct(query, span(sequence, location), pairs, score,
95        location <> tail, format, location_valid(sequence, location))
96
97# Pinned Edlib returns before applying k or calculating starts/path on empty
98# inputs, including PATH. Preserve that observable oracle behavior.
99def empty_result(query: String, target: String, config: E.AlignConfig) -> E.AlignResult:
100  match query target config:
101    case SNil{} +t E.AlignConfig{E.NW{}, _, _, _, _}:
102      +length = String.length(t)
103      E.distance_result(length, E.end_coordinate(length), True{})
104    case SNil{} _ _:
105      E.distance_result(0n, E.BeforeStart{}, True{})
106    case SCon{_, qt} SNil{} _:
107      E.distance_result(1n+String.length(qt), E.BeforeStart{}, True{})
108    # Internal precondition: Myers calls this only when one input is empty.
109    case _ _ _:
110      E.AlignResult{E.Error{}, E.NotFound{}, Nil{}, None{}, None{}, None{}}