Segment boundaries (#61)

* handle repeat words at segment boundaries

* correct overlapping logic

* increment version

* cover case of empy list

Author: joeferraratonic

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "tonic-textual"
-version = "3.10.5"
+version = "3.10.6"
 description = "Wrappers around the Tonic Textual API"
 authors = ["Adam Kamor <[email protected]>", "Joe Ferrara <[email protected]>", "Ander Steele <[email protected]>", "Ethan Philpott <[email protected]>", "Lyon Van Voorhis <[email protected]>", "Kirill Medvedev <[email protected]>", "Travis Matthews <[email protected]>"]
 license = "MIT"

tonic_textual/__init__.py

@@ -1 +1 @@
-__version__ = "3.10.5"
+__version__ = "3.10.6"

tonic_textual/helpers/redact_audio_file_helper.py

@@ -113,7 +113,12 @@ def get_intervals_to_redact(
     """
     transcript_words = []
     for segment in transcript_segments:
-        transcript_words.extend(segment.words)
+        # sometimes the last word of the previous segmant matches the first word of the new segment
+        # this causes issues later on if repeats are not removed
+        if len(transcript_words) > 0 and len(segment.words[0]) > 0 and segment.words[0] == transcript_words[-1]:
+            transcript_words.extend(segment.words[1:])
+        else:
+            transcript_words.extend(segment.words)
     enriched_transcript_words = add_character_indices_to_words(
         transcript_text, transcript_words
     )
@@ -129,7 +134,7 @@ def get_intervals_to_redact(
             # this beeps entire word when span is part of a word
             if word_start < span_end and word_start >= span_start:
                 intersecting_words.append(word_obj)
-            elif word_end >= span_start and word_end < span_end:
+            elif word_end > span_start and word_end <= span_end:
                 intersecting_words.append(word_obj)
             elif word_start > span_end: # done
                 break