Fix CorpusReplaceOrthFromReferenceCorpus (#290)

* followup fix

* first draft test

* tests

* updates

* dont change search

corpus/convert.py

@@ -50,7 +50,7 @@ def run(self):
         c.load(self.bliss_corpus.get_path())
 
         for s in c.segments():
-            tag in s.fullname()
+            tag = s.fullname()
             assert tag in orths.keys(), "Segment %s not found in reference corpus" % tag
             s.orth = orths[tag]
 

tests/job_tests/returnn/files/test_replace.corpus.xml

@@ -0,0 +1,12 @@
+<?xml version="1.0" encoding="utf-8"?>
+<corpus name="job-testing">
+  <speaker-description name="116">
+    <gender>male</gender>
+  </speaker-description>
+  <recording name="116-288045-0000" audio="sample.wav">
+    <speaker name="116"/>
+    <segment name="116-288045-0000" start="0.0000" end="10.6500">
+      <orth> AS I APPROACHED [non-speech] THE CITY, I HEARD BELLS RINGING [noise] [noise] AND A LITTLE LATER I FOUND THE THIS DOES NOT BELONG HERE! BUT I WANT IT HERE! STREETS ASTIR WITH THRONGS OF WELL DRESSED PEOPLE IN FAMILY GROUPS WENDING THEIR WAY HITHER AND THITHER!!! </orth>
+    </segment>
+  </recording>
+</corpus>

tests/job_tests/returnn/files/test_replace.corrupt.corpus.xml

@@ -0,0 +1,12 @@
+<?xml version="1.0" encoding="utf-8"?>
+<corpus name="job-testing">
+  <speaker-description name="116">
+    <gender>male</gender>
+  </speaker-description>
+  <recording name="116-288045-0000" audio="sample.wav">
+    <speaker name="116"/>
+    <segment name="116-288045-0000" start="0.0000" end="10.6500">
+      <orth> AS I APPROACHED [non-speech] THE CITY, I HEARD BELLS RINGING [noise] [noise] AND A LITTLE LATER THIS DOES NOT BELONG HERE I FOUND THE THIS DOES NOT BELONG HERE! BUT I WANT IT HERE! STREETS ASTIR WITH THRONGS OF WELL DRESSED PEOPLE IN FAMILY GROUPS WENDING THEIR WAY HITHER AND THITHER!!!</orth>
+    </segment>
+  </recording>
+</corpus>

tests/job_tests/returnn/test_convert.py

@@ -0,0 +1,23 @@
+import os
+import tempfile
+import filecmp
+from sisyphus import setup_path
+
+from i6_core.corpus.convert import CorpusReplaceOrthFromReferenceCorpus
+
+Path = setup_path(__package__)
+
+
+def test_corpus_replace_orth_from_reference_corpus():
+
+    with tempfile.TemporaryDirectory() as tmpdir:
+        reference_corpus = Path("files/test_replace.corpus.xml")
+        bliss_corpus_corrupt = Path("files/test_replace.corrupt.corpus.xml")
+
+        replace_job = CorpusReplaceOrthFromReferenceCorpus(
+            bliss_corpus=bliss_corpus_corrupt, reference_bliss_corpus=reference_corpus
+        )
+        replace_job.out_corpus = Path(os.path.join(tmpdir, "replaced.corpus.xml"))
+        replace_job.run()
+
+        assert filecmp.cmp(replace_job.out_corpus, reference_corpus, shallow=False)