From d52df58b046d8e364b2d5074dc7934c10da3cb29 Mon Sep 17 00:00:00 2001
From: manonBlanco <blanco@teklia.com>
Date: Wed, 28 Jun 2023 16:30:49 +0200
Subject: [PATCH] Use a random transcription of an element when more than one
found
---
dan/datasets/extract/exceptions.py | 9 ---------
dan/datasets/extract/extract.py | 7 ++-----
2 files changed, 2 insertions(+), 14 deletions(-)
diff --git a/dan/datasets/extract/exceptions.py b/dan/datasets/extract/exceptions.py
index 7d3a8e33..22c47a6c 100644
--- a/dan/datasets/extract/exceptions.py
+++ b/dan/datasets/extract/exceptions.py
@@ -49,15 +49,6 @@ class NoTranscriptionError(ElementProcessingError):
return f"No transcriptions found on element ({self.element_id}) with this config. Skipping."
-class MultipleTranscriptionsError(ElementProcessingError):
- """
- Raised when there are more than one transcription on an element
- """
-
- def __str__(self) -> str:
- return f"More than one transcription found on element ({self.element_id}) with this config."
-
-
class UnknownLabelError(ProcessingError):
"""
Raised when the specified label is not known
diff --git a/dan/datasets/extract/extract.py b/dan/datasets/extract/extract.py
index d35b08d0..b4dd1da9 100644
--- a/dan/datasets/extract/extract.py
+++ b/dan/datasets/extract/extract.py
@@ -18,7 +18,6 @@ from dan.datasets.extract.db import (
get_transcriptions,
)
from dan.datasets.extract.exceptions import (
- MultipleTranscriptionsError,
NoTranscriptionError,
ProcessingError,
UnknownLabelError,
@@ -129,12 +128,10 @@ class ArkindexExtractor:
transcriptions = get_transcriptions(
element.id, self.transcription_worker_version
)
- if len(transcriptions) > 1:
- raise MultipleTranscriptionsError(element.id)
- elif len(transcriptions) == 0:
+ if len(transcriptions) == 0:
raise NoTranscriptionError(element.id)
- transcription = transcriptions.pop()
+ transcription = random.choice(transcriptions)
if self.load_entities:
entities = get_transcription_entities(
--
GitLab