Spaces:
Running
Running
File size: 16,582 Bytes
f7ef7d3 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 |
\
import json
import os
import sys
from datetime import datetime
# Adjust path to import project modules
SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
PROJECT_ROOT = os.path.dirname(SCRIPT_DIR) # e.g. /home/psyborg/Desktop/tts_labeling
# Ensure the project root is at the beginning of sys.path
if PROJECT_ROOT in sys.path and sys.path[0] != PROJECT_ROOT:
sys.path.remove(PROJECT_ROOT) # Remove if it exists but not at index 0
if PROJECT_ROOT not in sys.path: # Add if it doesn't exist at all (it will be added at index 0)
sys.path.insert(0, PROJECT_ROOT)
from utils.database import get_db, SessionLocal # Changed Session to SessionLocal
from sqlalchemy.orm import Session as SQLAlchemySession # Import Session for type hinting
from data.models import TTSData, Annotator, Annotation, AudioTrim, AnnotationInterval # Added AnnotationInterval
from utils.logger import Logger
log = Logger()
ANNOTATIONS_FILE_PATH = os.path.join(PROJECT_ROOT, "annotations.json")
BATCH_SIZE = 100 # Define batch size for commits
def import_annotations(db: SQLAlchemySession, data: dict): # Changed SessionLocal to SQLAlchemySession for type hint
samples = data.get("samples", [])
imported_count = 0
updated_count = 0
skipped_count = 0
samples_processed_in_batch = 0
# Caches to potentially reduce DB lookups within the script run
tts_data_cache = {}
annotator_cache = {}
annotation_ids_for_trim_deletion_in_batch = [] # For batch deletion of trims
# Create a mapping from JSON ID to sample data for efficient lookup
samples_by_id = {s.get("id"): s for s in samples if s.get("id") is not None}
log.info(f"Created a map for {len(samples_by_id)} samples based on their JSON IDs.")
# Load all annotator intervals from the database
db_intervals = db.query(AnnotationInterval).all()
annotator_intervals = {interval.annotator_id: (interval.start_index, interval.end_index) for interval in db_intervals}
log.info(f"Loaded {len(annotator_intervals)} annotator intervals from the database.")
for sample_idx, sample_data in enumerate(samples): # Renamed sample to sample_data for clarity
current_sample_json_id = sample_data.get("id")
if current_sample_json_id is None: # Check for None explicitly
log.warning("Sample missing ID, skipping.")
skipped_count += 1
continue
# Assuming TTSData.id in DB matches JSON 'id' for lookup,
# but interval checks use an adjusted ID.
# The effective ID for checking against DB intervals (which are potentially 1-based for JSON's 0).
effective_id_for_interval_check = current_sample_json_id + 1
# Check if TTSData entry exists
if current_sample_json_id in tts_data_cache:
tts_data_entry = tts_data_cache[current_sample_json_id]
else:
# Query TTSData using the direct ID from JSON
tts_data_entry = db.query(TTSData).filter_by(id=current_sample_json_id).first()
if tts_data_entry:
tts_data_cache[current_sample_json_id] = tts_data_entry
if not tts_data_entry:
log.warning(f"TTSData with JSON ID {current_sample_json_id} not found in database, skipping sample.")
skipped_count += 1
continue
# Use the tts_data_entry.id for foreign keys, which should be the same as current_sample_json_id
db_tts_data_id = tts_data_entry.id
json_annotations = sample_data.get("annotations", [])
if not json_annotations:
continue
objects_to_add_this_sample = []
for json_ann in json_annotations:
json_annotator_name = json_ann.get("annotator")
# Determine the final_annotated_sentence based on the N+1 rule.
# Rule: Use original_subtitle from the (logical) next sample (N+1).
# Fallback 1: If N+1 doesn't exist, or its original_subtitle is None,
# use annotated_subtitle from the current sample's current annotation (json_ann).
# Fallback 2: If that's also None, use original_subtitle from the current sample (sample_data, top-level).
# Fallback 3: If all else fails, use an empty string.
sentence_to_use = None
used_n_plus_1 = False
logical_next_sample_json_id = current_sample_json_id - 1
next_sample_data_for_sentence = samples_by_id.get(logical_next_sample_json_id)
if next_sample_data_for_sentence:
sentence_from_n_plus_1 = next_sample_data_for_sentence.get("original_subtitle")
if sentence_from_n_plus_1 is not None:
sentence_to_use = sentence_from_n_plus_1
used_n_plus_1 = True
# log.debug(f"For sample {current_sample_json_id}, using original_subtitle from next sample {logical_next_sample_json_id}.")
# else: N+1 exists but its original_subtitle is None. Fall through.
# else: N+1 does not exist. Fall through.
if not used_n_plus_1:
# log.debug(f"For sample {current_sample_json_id}, N+1 rule not applied. Using current sample's subtitles.")
sentence_to_use = json_ann.get("annotated_subtitle") # Primary fallback from current annotation
if sentence_to_use is None:
# Secondary fallback to the top-level original_subtitle of the current sample
sentence_to_use = sample_data.get("original_subtitle")
# log.debug(f"For sample {current_sample_json_id}, json_ann.annotated_subtitle is None, falling back to sample_data.original_subtitle.")
final_annotated_sentence = sentence_to_use if sentence_to_use is not None else ""
if not json_annotator_name:
log.warning(f"Annotation for TTSData JSON ID {current_sample_json_id} missing annotator name, skipping.")
skipped_count +=1
continue
# Get initial annotator details from JSON
initial_annotator_entry = annotator_cache.get(json_annotator_name)
if not initial_annotator_entry:
initial_annotator_entry = db.query(Annotator).filter_by(name=json_annotator_name).first()
if not initial_annotator_entry:
log.warning(f"Annotator '{json_annotator_name}' (from JSON) not found in DB for TTSData JSON ID {current_sample_json_id}. Skipping this annotation.")
skipped_count += 1
continue
annotator_cache[json_annotator_name] = initial_annotator_entry
initial_annotator_id = initial_annotator_entry.id
# These will be the annotator details used for saving the annotation.
# They start as the initial annotator and may be reassigned.
save_annotator_id = initial_annotator_id
save_annotator_name = json_annotator_name # For logging
initial_annotator_interval = annotator_intervals.get(initial_annotator_id)
is_within_initial_interval = False
if initial_annotator_interval:
db_start_index, db_end_index = initial_annotator_interval
if db_start_index is not None and db_end_index is not None and \
db_start_index <= effective_id_for_interval_check <= db_end_index:
is_within_initial_interval = True
if not is_within_initial_interval:
log_message_prefix = f"TTSData JSON ID {current_sample_json_id} (effective: {effective_id_for_interval_check})"
if initial_annotator_interval:
log.warning(f"{log_message_prefix} is outside interval [{initial_annotator_interval[0]}, {initial_annotator_interval[1]}] for annotator '{json_annotator_name}'. Attempting to reassign.")
else:
log.warning(f"{log_message_prefix}: Annotator '{json_annotator_name}' (ID: {initial_annotator_id}) has no defined interval. Attempting to reassign to an interval owner.")
reassigned_successfully = False
for potential_owner_id, (owner_start, owner_end) in annotator_intervals.items():
if owner_start is not None and owner_end is not None and \
owner_start <= effective_id_for_interval_check <= owner_end:
save_annotator_id = potential_owner_id
reassigned_annotator_db_entry = db.query(Annotator).filter_by(id=save_annotator_id).first()
if reassigned_annotator_db_entry:
save_annotator_name = reassigned_annotator_db_entry.name
if save_annotator_name not in annotator_cache:
annotator_cache[save_annotator_name] = reassigned_annotator_db_entry
else:
save_annotator_name = f"ID:{save_annotator_id}"
log.error(f"Critical: Could not find Annotator DB entry for reassigned ID {save_annotator_id}, though an interval exists. Check data integrity.")
log.info(f"Reassigning annotation for {log_message_prefix} from '{json_annotator_name}' to '{save_annotator_name}' (ID: {save_annotator_id}) as they own the interval.")
reassigned_successfully = True
break
if not reassigned_successfully:
log.error(f"No annotator found with an interval covering {log_message_prefix}. Skipping this annotation by '{json_annotator_name}'.")
skipped_count += 1
continue
annotator_id = save_annotator_id
current_annotator_name_for_logs = save_annotator_name
annotated_at_str = json_ann.get("update_at") or json_ann.get("create_at")
annotated_at_dt = None
if annotated_at_str:
try:
annotated_at_dt = datetime.fromisoformat(annotated_at_str.replace('Z', '+00:00'))
except ValueError:
try:
annotated_at_dt = datetime.strptime(annotated_at_str.split('.')[0], "%Y-%m-%dT%H:%M:%S")
except ValueError as e_parse:
log.error(f"Could not parse timestamp '{annotated_at_str}' for TTSData JSON ID {current_sample_json_id}, annotator {current_annotator_name_for_logs}: {e_parse}")
final_annotated_at = annotated_at_dt
# Previous N+1 logic and interval checks that led to skipping are removed/replaced by the above.
annotation_obj = db.query(Annotation).filter_by(
tts_data_id=db_tts_data_id,
annotator_id=annotator_id
).first()
if annotation_obj:
annotation_obj.annotated_sentence = final_annotated_sentence
annotation_obj.annotated_at = final_annotated_at
updated_count +=1
else:
annotation_obj = Annotation(
tts_data_id=db_tts_data_id,
annotator_id=annotator_id,
annotated_sentence=final_annotated_sentence,
annotated_at=final_annotated_at
)
db.add(annotation_obj)
try:
db.flush()
imported_count +=1
except Exception as e_flush:
log.error(f"Error flushing new annotation for TTSData JSON ID {current_sample_json_id}, Annotator {current_annotator_name_for_logs}: {e_flush}")
db.rollback()
skipped_count +=1
continue
if annotation_obj.id:
if annotation_obj.id not in annotation_ids_for_trim_deletion_in_batch:
annotation_ids_for_trim_deletion_in_batch.append(annotation_obj.id)
json_audio_trims = json_ann.get("audio_trims", [])
if json_audio_trims:
# log.info(f"Preparing to add {len(json_audio_trims)} new trims for Annotation ID {annotation_obj.id}.")
for trim_info in json_audio_trims:
start_sec = trim_info.get("start")
end_sec = trim_info.get("end")
if start_sec is not None and end_sec is not None:
try:
start_ms = int(float(start_sec) * 1000.0)
end_ms = int(float(end_sec) * 1000.0)
if start_ms < 0 or end_ms < 0 or end_ms < start_ms:
log.warning(f"Invalid trim values (start_ms={start_ms}, end_ms={end_ms}) for annotation ID {annotation_obj.id}, TTSData JSON ID {current_sample_json_id}. Skipping.")
continue
new_trim_db_obj = AudioTrim(
annotation_id=annotation_obj.id,
original_tts_data_id=db_tts_data_id,
start=start_ms,
end=end_ms
)
objects_to_add_this_sample.append(new_trim_db_obj)
except ValueError:
log.warning(f"Invalid start/end format in audio trim for annotation ID {annotation_obj.id}, TTSData JSON ID {current_sample_json_id}. Skipping: {trim_info}")
continue
else:
log.warning(f"Skipping trim with missing start/end for Annotation ID {annotation_obj.id}, TTSData JSON ID {current_sample_json_id}: {trim_info}")
else:
log.warning(f"Annotation ID not available for TTSData JSON ID {current_sample_json_id}, Annotator {current_annotator_name_for_logs}. Cannot process audio trims.")
if objects_to_add_this_sample:
db.add_all(objects_to_add_this_sample)
samples_processed_in_batch += 1
if samples_processed_in_batch >= BATCH_SIZE or (sample_idx == len(samples) - 1):
if annotation_ids_for_trim_deletion_in_batch:
log.info(f"Batch deleting trims for {len(annotation_ids_for_trim_deletion_in_batch)} annotations in current batch.")
db.query(AudioTrim).filter(AudioTrim.annotation_id.in_(annotation_ids_for_trim_deletion_in_batch)).delete(synchronize_session=False)
annotation_ids_for_trim_deletion_in_batch.clear()
try:
db.commit()
log.info(f"Committed batch. Total samples processed so far: {sample_idx + 1} out of {len(samples)}")
except Exception as e_commit:
db.rollback()
log.error(f"Failed to commit batch after sample index {sample_idx} (TTSData JSON ID {current_sample_json_id}): {e_commit}. Rolling back this batch.")
annotation_ids_for_trim_deletion_in_batch.clear()
finally:
samples_processed_in_batch = 0 # Reset for next batch or end
log.info(f"Finished import attempt. Final counts - New: {imported_count}, Updated: {updated_count}, Skipped: {skipped_count}")
def main():
log.info("Starting annotation import script...")
if not os.path.exists(ANNOTATIONS_FILE_PATH):
log.error(f"Annotations file not found at: {ANNOTATIONS_FILE_PATH}")
return
try:
with open(ANNOTATIONS_FILE_PATH, 'r', encoding='utf-8') as f:
data = json.load(f)
except json.JSONDecodeError as e:
log.error(f"Error decoding JSON from {ANNOTATIONS_FILE_PATH}: {e}")
return
except Exception as e:
log.error(f"Error reading file {ANNOTATIONS_FILE_PATH}: {e}")
return
try:
with get_db() as db_session:
import_annotations(db_session, data)
except Exception as e:
log.error(f"An error occurred during the import process: {e}")
finally:
log.info("Annotation import script finished.")
if __name__ == "__main__":
main()
|