Spaces:
Sleeping
Sleeping
File size: 16,185 Bytes
e82eaee | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 | """Import/Export functionality for CSV and HuggingFace datasets."""
import os
import io
import re
import json
import pandas as pd
import soundfile as sf
import gradio as gr
from datasets import Dataset, Audio, Features, Value
from huggingface_hub import login, HfApi
import utils
from core.state import get_global_results, set_global_results
from core.cache import get_cached_dataset, cache_dataset
from core.comparison import select_best_model_result, get_all_model_comparison, find_best_model_pair
from ui.dashboard import generate_dashboard_outputs
def sanitize_filename(name):
"""Sanitize string to be used as a filename."""
if not name:
return "results"
s = re.sub(r'[^\w\s-]', '_', name).strip().lower()
s = re.sub(r'[-\s]+', '_', s)
return s
def import_csv_analysis(file_obj, similarity_threshold, dataset_name, limit_files):
"""Import analysis results from CSV file."""
global_results = get_global_results()
limit_files = int(float(limit_files)) if limit_files else 0
similarity_threshold = int(float(similarity_threshold)) if similarity_threshold else 90
if file_obj is None:
return generate_dashboard_outputs(similarity_threshold)
try:
df = pd.read_csv(file_obj.name)
dedup_col = 'file_name' if 'file_name' in df.columns else ('path' if 'path' in df.columns else None)
if dedup_col:
df = df.drop_duplicates(subset=[dedup_col])
rename_map = {}
if 'file_name' in df.columns and 'path' not in df.columns:
rename_map['file_name'] = 'path'
if 'idx' in df.columns and 'id' not in df.columns:
rename_map['idx'] = 'id'
if 'score_%' in df.columns and 'score' not in df.columns:
rename_map['score_%'] = 'score'
if 'ref' in df.columns and 'ref_text' not in df.columns:
rename_map['ref'] = 'ref_text'
if 'hyp' in df.columns and 'hyp_text' not in df.columns:
rename_map['hyp'] = 'hyp_text'
if rename_map:
df = df.rename(columns=rename_map)
audio_map = _load_audio_map(dataset_name, df)
results = _process_csv_rows(df, audio_map, similarity_threshold)
_merge_results(results, similarity_threshold)
print(f"Total records after import: {len(get_global_results())}")
return generate_dashboard_outputs(similarity_threshold)
except Exception as e:
print(f"Error importing CSV: {e}")
return "", "", pd.DataFrame()
def _load_audio_map(dataset_name, df):
"""Load audio map from dataset."""
audio_map = {}
limit = None
cached_ds = get_cached_dataset(dataset_name, limit)
if not cached_ds:
try:
target_paths = set()
for _, r_row in df.iterrows():
fname_t = str(r_row.get('path', ''))
if fname_t:
target_paths.add(fname_t)
target_paths.add(os.path.basename(fname_t))
ds = utils.load_hf_dataset(dataset_name, limit=limit, allowed_paths=target_paths)
cached_ds = ds
except Exception as e:
print(f"Warning: Could not load dataset: {e}")
cached_ds = []
if cached_ds:
for item in cached_ds:
path = item['audio']['path']
if path:
audio_map[os.path.basename(path)] = item
audio_map[path] = item
return audio_map
def _process_csv_rows(df, audio_map, similarity_threshold):
"""Process CSV rows into results."""
results = []
def safe_str(val, default=''):
if pd.isna(val): return default
return str(val)
for idx, row in df.iterrows():
fname = safe_str(row.get('path', ''))
ref = safe_str(row.get('ref_text', ''))
hyp = safe_str(row.get('hyp_text', ''))
score, norm_ref, norm_hyp = utils.calculate_similarity(ref, hyp)
model_used = safe_str(row.get('model_used'), 'imported_csv')
verification_status = safe_str(row.get('verification_status'), 'unknown')
if int(round(score)) >= similarity_threshold:
verification_status = 'correct'
row_id = row.get('id', idx)
item = audio_map.get(fname) or audio_map.get(os.path.basename(fname))
audio_array = item['audio']['array'] if item else None
sampling_rate = item['audio']['sampling_rate'] if item else None
model_results = {}
model_results_val = row.get('model_results')
if pd.notnull(model_results_val) and model_results_val:
try:
model_results = json.loads(str(model_results_val))
except:
pass
if hyp and score > 0:
source_name = f"imported_{model_used}" if model_used != 'imported_csv' else 'imported_csv'
model_results[source_name] = {"hyp_text": hyp, "score": score, "norm_ref": norm_ref, "norm_hyp": norm_hyp}
results.append({
"id": int(row_id) if pd.notnull(row_id) else idx,
"path": fname, "score": score, "ref_text": ref, "hyp_text": hyp,
"audio_array": audio_array, "sampling_rate": sampling_rate,
"status": "processed", "verification_status": verification_status,
"model_used": model_used, "norm_ref": norm_ref, "norm_hyp": norm_hyp,
"model_results": model_results
})
return results
def _merge_results(results, similarity_threshold):
"""Merge new results into global_results."""
global_results = get_global_results()
if not global_results:
set_global_results(results)
return
updated = list(global_results)
existing_map = {r.get('path'): i for i, r in enumerate(updated) if r.get('path')}
used_ids = {r.get('id') for r in updated if r.get('id') is not None}
max_id = max(used_ids) if used_ids else -1
for new_item in results:
path = new_item.get('path')
if path and path in existing_map:
idx = existing_map[path]
old_item = updated[idx]
merged = old_item.get('model_results', {}).copy()
merged.update(new_item.get('model_results', {}))
new_item['model_results'] = merged
best_model, best_res = select_best_model_result(merged, similarity_threshold)
if best_res:
new_item.update({
"hyp_text": best_res['hyp_text'], "score": best_res['score'],
"norm_ref": best_res['norm_ref'], "norm_hyp": best_res['norm_hyp'],
"model_used": best_model
})
if old_item.get('model_used') != 'manual':
new_item['verification_status'] = 'correct' if int(round(best_res['score'])) >= similarity_threshold else 'incorrect'
if new_item.get('audio_array') is None and old_item.get('audio_array') is not None:
new_item['audio_array'] = old_item['audio_array']
new_item['sampling_rate'] = old_item['sampling_rate']
new_item['id'] = old_item.get('id', new_item.get('id'))
updated[idx] = new_item
else:
if new_item.get('id') in used_ids:
max_id += 1
new_item['id'] = max_id
updated.append(new_item)
if new_item.get('id') is not None:
used_ids.add(new_item['id'])
max_id = max(max_id, new_item['id'])
set_global_results(updated)
def save_results_csv(dataset_name):
"""Save results to CSV file."""
global_results = get_global_results()
if not global_results:
return None
try:
export_data = []
for result in global_results:
export_row = {k: v for k, v in result.items() if k not in ['audio_array', 'sampling_rate']}
if 'model_results' in export_row and export_row['model_results']:
export_row['model_results'] = json.dumps(export_row['model_results'], ensure_ascii=False)
export_data.append(export_row)
df_export = pd.DataFrame(export_data)
clean_name = sanitize_filename(dataset_name)
filename = f"{clean_name}_results.csv"
abs_path = os.path.abspath(filename)
df_export.to_csv(abs_path, index=False)
print(f"💾 Exporting main CSV: {abs_path}")
detailed_data = []
for result in global_results:
model_results = result.get('model_results', {})
if model_results:
comparison = get_all_model_comparison(result)
for model_name, model_result in model_results.items():
detailed_data.append({
"id": result.get('id'), "path": result.get('path'),
"model_name": model_name, "hyp_text": model_result.get('hyp_text', ''),
"score": model_result.get('score', 0),
"is_best": model_name == comparison.get('best_model', ''),
"ref_text": result.get('ref_text', '')
})
if detailed_data:
df_detailed = pd.DataFrame(detailed_data)
detailed_filename = f"{clean_name}_model_comparison.csv"
df_detailed.to_csv(os.path.abspath(detailed_filename), index=False)
return abs_path
except Exception as e:
print(f"Error creating CSV: {e}")
return None
def _find_index_by_id(record_id: int):
"""Find index by record ID."""
for i, r in enumerate(get_global_results()):
if r.get("id") == record_id:
return i
return None
def verify_action(data_str, similarity_threshold, dataset_name):
"""Handle verification button click."""
global_results = get_global_results()
similarity_threshold = int(float(similarity_threshold)) if similarity_threshold else 90
if not data_str:
return generate_dashboard_outputs(similarity_threshold)
try:
data = json.loads(data_str)
record_id = data.get('id')
status = data.get('status')
if record_id is None or status not in ("correct", "incorrect", "update_match"):
return generate_dashboard_outputs(similarity_threshold)
idx = _find_index_by_id(int(record_id))
if idx is None:
return generate_dashboard_outputs(similarity_threshold)
if status == 'update_match':
record = global_results[idx]
model_results = record.get('model_results', {})
ref_text = record.get('ref_text', '')
best_text = ""
if model_results:
if len(model_results) >= 2:
best_pair = find_best_model_pair(record, ref_text)
best_text = best_pair.get('best_hyp', '') if best_pair else ""
if not best_text:
_, best_res = select_best_model_result(model_results)
best_text = best_res.get('hyp_text', '') if best_res else ""
if best_text:
global_results[idx]['ref_text'] = best_text
global_results[idx]['verification_status'] = 'correct'
global_results[idx]['model_used'] = 'manual'
for m_name in model_results:
hyp = model_results[m_name].get('hyp_text', '')
new_score, _, _ = utils.calculate_similarity(best_text, hyp)
global_results[idx]['model_results'][m_name]['score'] = new_score
_, best_res_new = select_best_model_result(global_results[idx]['model_results'])
if best_res_new:
global_results[idx]['score'] = best_res_new['score']
global_results[idx]['hyp_text'] = best_res_new['hyp_text']
else:
global_results[idx]['verification_status'] = status
global_results[idx]['model_used'] = 'manual'
try:
save_df = pd.DataFrame(global_results)
clean_name = sanitize_filename(dataset_name)
save_df.to_csv(f"{clean_name}_results.csv", index=False)
except Exception as e:
print(f"Error saving: {e}")
return generate_dashboard_outputs(similarity_threshold)
except Exception as e:
print(f"Error in verify_action: {e}")
return generate_dashboard_outputs(similarity_threshold)
def create_verified_dataset(hf_token, dataset_name, progress=gr.Progress()):
"""Create a new dataset on HuggingFace with only verified records."""
global_results = get_global_results()
if not hf_token:
raise gr.Error("Калі ласка, увядзіце Hugging Face Token.")
if not global_results:
raise gr.Error("Няма даных для стварэння датасэта.")
verified_data = [r for r in global_results if r.get('verification_status') == 'correct']
if not verified_data:
raise gr.Error("Няма правераных (correct) запісаў.")
try:
login(token=hf_token)
api = HfApi(token=hf_token)
username = api.whoami()['name']
original_slug = dataset_name.split("/")[-1] if "/" in dataset_name else dataset_name
new_repo_id = f"{username}/{original_slug}Checked"
def gen():
ds_ref = None
for row in verified_data:
audio_array = row.get('audio_array')
sr = row.get('sampling_rate')
if audio_array is None or len(audio_array) == 0:
if ds_ref is None:
try:
needed = {r.get('path') for r in verified_data if not r.get('audio_array')}
needed.update(os.path.basename(p) for p in needed if p)
items = utils.load_hf_dataset(dataset_name, allowed_paths=needed)
ds_ref = {item['audio']['path']: item for item in items}
ds_ref.update({os.path.basename(k): v for k, v in ds_ref.items()})
except:
ds_ref = {}
item = ds_ref.get(row.get('path')) or ds_ref.get(os.path.basename(row.get('path', '')))
if item:
audio_array = item['audio']['array']
sr = item['audio']['sampling_rate']
if audio_array is not None and len(audio_array) > 0:
buffer = io.BytesIO()
sf.write(buffer, audio_array, int(float(sr or 16000)), format='WAV')
yield {"audio": {"bytes": buffer.getvalue(), "path": None}, "text": row.get('ref_text', ''), "original_path": row.get('path', '')}
features = Features({"audio": {"bytes": Value("binary"), "path": Value("string")}, "text": Value("string"), "original_path": Value("string")})
new_ds = Dataset.from_generator(gen, features=features)
if "audio" in new_ds.features:
new_ds.info.features["audio"] = Audio(sampling_rate=None)
if len(new_ds) == 0:
raise gr.Error("Не ўдалося сабраць аўдыяданыя.")
progress(0.9, desc=f"Загрузка на Hugging Face...")
new_ds.push_to_hub(new_repo_id, token=hf_token)
return f"✅ Датасэт створаны: https://huggingface.co/datasets/{new_repo_id}"
except Exception as e:
raise gr.Error(f"Памылка: {e}")
|