lccs now retrieving the sequence

README.md

@@ -35,6 +35,7 @@ lccs_len = lccs_length(seq1, seq2)  # 5, [2, 3, 4, 5, 6]
 
 ## TODOs
 
+* implement lccs with suffix tree;
 * batch methods, i.e. supporting 2D arrays;
 * batch methods with any number of dimensions (nD array) and add a `dim` argument;
 * make it work with an unlimited number of sequences, and set `dim` and `pad_token` as kwargs only;

src/cpu/lccs_cpu_dyn.cpp

@@ -6,8 +6,8 @@ namespace nb = nanobind;
 using namespace nb::literals;
 
 
-// Calculating the length of the longest common contiguous subsequence with dynamic programming
-int lccs_length(
+// Returns the length of the longest common subsequence and the idx of its end in s1
+std::vector<int> lccs_length_idx(
     const nb::ndarray<double, nb::ndim<1>>& s1,
     const nb::ndarray<double, nb::ndim<1>>& s2
 ) {
@@ -21,16 +21,45 @@ int lccs_length(
 
     std::vector<std::vector<int>> table(s1Len + 1, std::vector<int>(s2Len + 1, 0));
     int max_length = 0;
+    int imax = 0;  // ending idx of the lccs
     for (int i = 0; i < s1Len; ++i) {
         for (int j = 0; j < s2Len; ++j) {
             if (v1(i) == v2(j)) {
                 table[i + 1][j + 1] = table[i][j] + 1;
                 if (table[i + 1][j + 1] > max_length) {
+                    imax = i;
                     max_length = table[i + 1][j + 1];
                 }
             }
         }
     }
+    std::vector<int> lccs_len_idx = {max_length, imax};
+    return lccs_len_idx;
+}
+
+
+// Calculating the length of the longest common contiguous subsequence with dynamic programming
+int lccs_length(
+    const nb::ndarray<double, nb::ndim<1>>& s1,
+    const nb::ndarray<double, nb::ndim<1>>& s2
+) {
+    std::vector<int> lccs_len_idx = lccs_length_idx(s1, s2);
+    return lccs_len_idx[0];
+}
+
+
+// Calculating the longest common contiguous subsequence with dynamic programming
+std::vector<int> lccs(
+    const nb::ndarray<double, nb::ndim<1>>& s1,
+    const nb::ndarray<double, nb::ndim<1>>& s2
+) {
+    std::vector<int> lccs_len_idx = lccs_length_idx(s1, s2);
+
+    // Extract the longest common substring from s1
+    std::vector<int> longestSubseq(lccs_len_idx[0]);
+    int idx = 0;
+    for (int i = lccs_len_idx[1] - lccs_len_idx[0] + 1; i <= lccs_len_idx[1]; ++i)
+        longestSubseq[idx++] = s1(i);
 
-    return max_length;
+    return longestSubseq;
 }

src/lcs.cpp

@@ -13,6 +13,8 @@ NB_MODULE(lcsvec_ext, m) {
     m.def("lcs_table", &createLCSTable, "seq1"_a, "seq2"_a,
           "Returns the longest common subsequence (lcs) table from `seq1` and `seq2`.");
 
+    m.def("lccs", &lccs, "seq1"_a, "seq2"_a,
+          "Returns the longest common contiguous subsequence (lccs) from `seq1` and `seq2`.");
     m.def("lccs_length", &lccs_length, "seq1"_a, "seq2"_a,
           "Returns the length of the longest common contiguous subsequence (lccs) from `seq1` and `seq2`.");
 }

src/lcsvec/__init__.py

@@ -1,7 +1,7 @@
 """Main LCSvec module."""
 
-from .lcsvec_ext import lccs_length, lcs, lcs_length, lcs_table
+from .lcsvec_ext import lccs, lccs_length, lcs, lcs_length, lcs_table
 
 __version__ = "0.0.1"
 
-__all__ = ["lccs_length", "lcs", "lcs_length", "lcs_table"]
+__all__ = ["lccs", "lccs_length", "lcs", "lcs_length", "lcs_table"]

tests/test_lccs_cpu.py

@@ -5,37 +5,73 @@
 from typing import TYPE_CHECKING
 
 import numpy as np
-from lcsvec import lccs_length
+import pytest
+from lcsvec import lccs, lccs_length
 from torch import IntTensor, LongTensor, arange
 
 if TYPE_CHECKING:
     from numpy.typing import NDArray
 
 
+TEST_CASES = [
+    (range(12), [8, 0, 1, 2, 8, 2, 3, 8, 4, 0], range(3)),
+    (range(12), [8, 0, 9, 2, 8, 2, 7, 3, 4, 5], range(3, 6)),
+    (range(12), [0, 1, 2, 3, 8, 9, 2, 3, 4, 5], range(4)),
+    (range(-2, 10), [9, -1, 0, 1, 2, 9, 2, 4, 4, 5], range(-1, 3)),
+]
+
+
 def _test_lccs(
     seq1: NDArray | IntTensor | LongTensor,
     seq2: NDArray | IntTensor | LongTensor,
     ref: list[int],
-) -> None:
-    lcs_len = lccs_length(seq1, seq2)
-    assert lcs_len == len(ref)
+) -> True:
+    lccs_ = lccs(seq1, seq2)
+    lccs_len = lccs_length(seq1, seq2)
 
+    assert lccs_len == len(ref)
+    assert lccs_ == ref
+    return True
 
-def test_lccs_numpy() -> None:
+
+@pytest.mark.parametrize("sequences", TEST_CASES)
+def test_lccs_numpy(
+    sequences: tuple[list[int] | range, list[int] | range, list[int] | range],
+) -> None:
     r"""Test the LCCS methods with numpy."""
-    seq1 = np.arange(0, 12)
-    seq2 = np.array([8, 0, 1, 2, 8, 2, 3, 8, 4, 0], dtype=np.int64)
-    ref = np.arange(0, 3).tolist()
+    seq1, seq2, ref = sequences
+    seq1 = (
+        np.arange(seq1.start, seq1.stop)
+        if isinstance(seq1, range)
+        else np.array(seq1, dtype=np.int64)
+    )
+    seq2 = (
+        np.arange(seq2.start, seq2.stop)
+        if isinstance(seq2, range)
+        else np.array(seq2, dtype=np.int64)
+    )
+    ref = (
+        np.arange(ref.start, ref.stop).tolist()
+        if isinstance(ref, range)
+        else np.array(ref, dtype=np.int64).tolist()
+    )
 
-    lcs_len = lccs_length(seq1, seq2)
-    assert lcs_len == len(ref)
+    assert _test_lccs(seq1, seq2, ref)
 
 
-def test_lccs_torch() -> None:
+@pytest.mark.parametrize("sequences", TEST_CASES)
+def test_lccs_torch(
+    sequences: tuple[list[int] | range, list[int] | range, list[int] | range],
+) -> None:
     r"""Test the LCCS methods with pytorch."""
-    seq1 = arange(0, 12)
-    seq2 = LongTensor([8, 0, 1, 2, 8, 2, 3, 8, 4, 0])
-    ref = arange(0, 3).tolist()
+    seq1, seq2, ref = sequences
+    seq1 = (
+        arange(seq1.start, seq1.stop) if isinstance(seq1, range) else LongTensor(seq1)
+    )
+    seq2 = (
+        arange(seq2.start, seq2.stop) if isinstance(seq2, range) else LongTensor(seq2)
+    )
+    ref = arange(ref.start, ref.stop) if isinstance(ref, range) else LongTensor(ref)
+    ref = ref.tolist()
 
-    lcs_len = lccs_length(seq1, seq2)
-    assert lcs_len == len(ref)
+    assert _test_lccs(seq1, seq2, ref)