feat: connect isolated tokens based on index

src/parallel_corpus/graph.py

@@ -1,5 +1,6 @@
 """Parallel corpus as a graph."""
 
+import copy
 import itertools
 import logging
 import re
@@ -235,6 +236,27 @@ def target_text(g: SourceTarget[List[text_token.Text]]) -> str:
     return text_token.text(g.target)
 
 
+def connect_isolated_tokens_based_on_index(g: Graph) -> Graph:
+    g_new = copy.deepcopy(g)
+    isolated_source_edges = [
+        e for e in g_new.edges.values() if len(e.ids) == 1 and e.ids[0].startswith("s")
+    ]
+    isolated_target_edges = [
+        e for e in g_new.edges.values() if len(e.ids) == 1 and e.ids[0].startswith("t")
+    ]
+    for s_edge in isolated_source_edges:
+        needle = s_edge.ids[0].replace("s", "t")
+        t_edge = next(filter(lambda t: t.ids[0] == needle, isolated_target_edges), None)
+        if t_edge:
+            del g_new.edges[s_edge.id]
+            del g_new.edges[t_edge.id]
+
+            new_edge = merge_edges(s_edge, t_edge)
+            g_new.edges[new_edge.id] = new_edge
+
+    return g_new
+
+
 @dataclass
 class CharIdPair:
     char: str

tests/test_graph.py

@@ -305,3 +305,13 @@ def test_unaligned_rearrange() -> None:
 
 
 # target_text(unaligned_rearrange(init(), 1, 2, 0)) // =>
+
+
+def test_connecting_token_based_on_index() -> None:
+    original_text = "Den 24 maj: JfS 404: pä ansökan af handlanden E. G. Petersson i Mönstcräs. Den 25,maj: M 383: pá ansökan as"  # noqa: E501
+    generated_text = "Den 24 maj: № 404: på ansökan af handlanden E. G. Petersson i Mönsterås. Den 25 maj: № 383: på ansökan af"  # noqa: E501
+
+    g = graph.init_with_source_and_target(original_text, generated_text)
+    print(f"{g=}")
+    g_new = graph.connect_isolated_tokens_based_on_index(g)
+    assert "e-s3-t3" in g_new.edges