Spaces:
Runtime error
Runtime error
ugmSorcero
commited on
Commit
·
42468fb
1
Parent(s):
6bb1fd5
Fixes linting
Browse files- core/pipelines.py +7 -4
- core/search_index.py +4 -6
- interface/components.py +3 -3
core/pipelines.py
CHANGED
|
@@ -11,9 +11,10 @@ from haystack.nodes.ranker import SentenceTransformersRanker
|
|
| 11 |
from haystack.nodes.audio.document_to_speech import DocumentToSpeech
|
| 12 |
import os
|
| 13 |
|
| 14 |
-
data_path =
|
| 15 |
os.makedirs(data_path, exist_ok=True)
|
| 16 |
|
|
|
|
| 17 |
def keyword_search(index="documents", split_word_length=100, audio_output=False):
|
| 18 |
"""
|
| 19 |
**Keyword Search Pipeline**
|
|
@@ -49,13 +50,15 @@ def keyword_search(index="documents", split_word_length=100, audio_output=False)
|
|
| 49 |
index_pipeline.add_node(
|
| 50 |
document_store, name="DocumentStore", inputs=["TfidfRetriever"]
|
| 51 |
)
|
| 52 |
-
|
| 53 |
if audio_output:
|
| 54 |
doc2speech = DocumentToSpeech(
|
| 55 |
model_name_or_path="espnet/kan-bayashi_ljspeech_vits",
|
| 56 |
-
generated_audio_dir=Path(data_path +
|
|
|
|
|
|
|
|
|
|
| 57 |
)
|
| 58 |
-
search_pipeline.add_node(doc2speech, name='DocumentToSpeech', inputs=['TfidfRetriever'])
|
| 59 |
|
| 60 |
return search_pipeline, index_pipeline
|
| 61 |
|
|
|
|
| 11 |
from haystack.nodes.audio.document_to_speech import DocumentToSpeech
|
| 12 |
import os
|
| 13 |
|
| 14 |
+
data_path = "data/"
|
| 15 |
os.makedirs(data_path, exist_ok=True)
|
| 16 |
|
| 17 |
+
|
| 18 |
def keyword_search(index="documents", split_word_length=100, audio_output=False):
|
| 19 |
"""
|
| 20 |
**Keyword Search Pipeline**
|
|
|
|
| 50 |
index_pipeline.add_node(
|
| 51 |
document_store, name="DocumentStore", inputs=["TfidfRetriever"]
|
| 52 |
)
|
| 53 |
+
|
| 54 |
if audio_output:
|
| 55 |
doc2speech = DocumentToSpeech(
|
| 56 |
model_name_or_path="espnet/kan-bayashi_ljspeech_vits",
|
| 57 |
+
generated_audio_dir=Path(data_path + "audio"),
|
| 58 |
+
)
|
| 59 |
+
search_pipeline.add_node(
|
| 60 |
+
doc2speech, name="DocumentToSpeech", inputs=["TfidfRetriever"]
|
| 61 |
)
|
|
|
|
| 62 |
|
| 63 |
return search_pipeline, index_pipeline
|
| 64 |
|
core/search_index.py
CHANGED
|
@@ -38,12 +38,10 @@ def search(queries, pipeline):
|
|
| 38 |
"fragment_id": res.id,
|
| 39 |
}
|
| 40 |
if not score_is_empty:
|
| 41 |
-
match.update({
|
| 42 |
-
if hasattr(res,
|
| 43 |
-
match.update({
|
| 44 |
-
query_results.append(
|
| 45 |
-
match
|
| 46 |
-
)
|
| 47 |
if not score_is_empty:
|
| 48 |
query_results = sorted(
|
| 49 |
query_results, key=lambda x: x["score"], reverse=True
|
|
|
|
| 38 |
"fragment_id": res.id,
|
| 39 |
}
|
| 40 |
if not score_is_empty:
|
| 41 |
+
match.update({"score": res.score})
|
| 42 |
+
if hasattr(res, "content_audio"):
|
| 43 |
+
match.update({"content_audio": res.content_audio})
|
| 44 |
+
query_results.append(match)
|
|
|
|
|
|
|
| 45 |
if not score_is_empty:
|
| 46 |
query_results = sorted(
|
| 47 |
query_results, key=lambda x: x["score"], reverse=True
|
interface/components.py
CHANGED
|
@@ -59,10 +59,10 @@ def component_show_search_result(container, results):
|
|
| 59 |
st.markdown(f"### Match {idx+1}")
|
| 60 |
st.markdown(f"**Text**: {document['text']}")
|
| 61 |
st.markdown(f"**Document**: {document['id']}")
|
| 62 |
-
if
|
| 63 |
st.markdown(f"**Score**: {document['score']:.3f}")
|
| 64 |
-
if
|
| 65 |
-
st.audio(str(document[
|
| 66 |
st.markdown("---")
|
| 67 |
|
| 68 |
|
|
|
|
| 59 |
st.markdown(f"### Match {idx+1}")
|
| 60 |
st.markdown(f"**Text**: {document['text']}")
|
| 61 |
st.markdown(f"**Document**: {document['id']}")
|
| 62 |
+
if "score" in document:
|
| 63 |
st.markdown(f"**Score**: {document['score']:.3f}")
|
| 64 |
+
if "content_audio" in document:
|
| 65 |
+
st.audio(str(document["content_audio"]))
|
| 66 |
st.markdown("---")
|
| 67 |
|
| 68 |
|