Spaces:
Paused
Paused
revert html experiments
Browse files
app.py
CHANGED
|
@@ -88,7 +88,6 @@ def flag(query, language, num_results, issue_description):
|
|
| 88 |
|
| 89 |
|
| 90 |
def format_result(result, highlight_terms, exact_search, datasets_filter=None):
|
| 91 |
-
# print("result", result)
|
| 92 |
text, url, docid = result
|
| 93 |
if datasets_filter is not None:
|
| 94 |
datasets_filter = set(datasets_filter)
|
|
@@ -134,7 +133,8 @@ def format_result(result, highlight_terms, exact_search, datasets_filter=None):
|
|
| 134 |
language = "FIXME"
|
| 135 |
result_html = """{}
|
| 136 |
<span style='font-size:14px; font-family: Arial; color:#7978FF; text-align: left;'>Document ID: {}</span>
|
| 137 |
-
<
|
|
|
|
| 138 |
<!-- <span style='font-size:12px; font-family: Arial; color:MediumAquaMarine'>Language: {}</span><br> -->
|
| 139 |
<span style='font-family: Arial;'>{}</span><br>
|
| 140 |
<br>
|
|
@@ -144,6 +144,58 @@ def format_result(result, highlight_terms, exact_search, datasets_filter=None):
|
|
| 144 |
return "<p>" + result_html + "</p>"
|
| 145 |
|
| 146 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 147 |
def format_result_page(
|
| 148 |
results, highlight_terms, num_results, exact_search, datasets_filter=None
|
| 149 |
):
|
|
@@ -160,34 +212,32 @@ def format_result_page(
|
|
| 160 |
def extract_results_from_payload(query, language, payload, exact_search):
|
| 161 |
results = payload["results"]
|
| 162 |
|
| 163 |
-
processed_results =
|
| 164 |
datasets = set()
|
| 165 |
highlight_terms = None
|
| 166 |
num_results = None
|
| 167 |
-
|
| 168 |
if exact_search:
|
| 169 |
highlight_terms = query
|
| 170 |
num_results = payload["num_results"]
|
|
|
|
| 171 |
else:
|
| 172 |
highlight_terms = payload["highlight_terms"]
|
| 173 |
-
results = []
|
| 174 |
-
for lang, res_for_lang in payload["results"].items():
|
| 175 |
-
for result in res_for_lang:
|
| 176 |
-
results.append(result)
|
| 177 |
|
| 178 |
-
for
|
| 179 |
-
|
| 180 |
-
|
| 181 |
-
result["
|
| 182 |
-
|
| 183 |
-
|
| 184 |
-
|
| 185 |
-
|
| 186 |
-
|
| 187 |
-
|
| 188 |
-
|
| 189 |
-
|
| 190 |
-
|
|
|
|
|
|
|
| 191 |
|
| 192 |
return processed_results, highlight_terms, num_results, list(datasets)
|
| 193 |
|
|
@@ -277,6 +327,7 @@ if __name__ == "__main__":
|
|
| 277 |
"vi",
|
| 278 |
"zh",
|
| 279 |
"detect_language",
|
|
|
|
| 280 |
],
|
| 281 |
value="en",
|
| 282 |
label="Language",
|
|
@@ -293,10 +344,8 @@ if __name__ == "__main__":
|
|
| 293 |
multiselect=True,
|
| 294 |
)
|
| 295 |
with gr.Row():
|
| 296 |
-
|
| 297 |
-
|
| 298 |
-
for i in range(100):
|
| 299 |
-
results_html.append(gr.HTML(label="Results"))
|
| 300 |
with gr.Row(visible=False) as pagination:
|
| 301 |
next_page_btn = gr.Button("Next Page")
|
| 302 |
|
|
@@ -344,8 +393,8 @@ if __name__ == "__main__":
|
|
| 344 |
num_results
|
| 345 |
)
|
| 346 |
# print("processed_results", processed_results)
|
| 347 |
-
results_html =
|
| 348 |
-
processed_results, highlight_terms, num_results, exact_search
|
| 349 |
)
|
| 350 |
return (
|
| 351 |
processed_results,
|
|
@@ -366,20 +415,17 @@ if __name__ == "__main__":
|
|
| 366 |
datasets,
|
| 367 |
) = run_query(query, lang, k, dropdown_input, max_page_size, 0)
|
| 368 |
has_more_results = exact_search and (num_results > max_page_size)
|
| 369 |
-
return
|
| 370 |
-
|
| 371 |
-
|
| 372 |
-
|
| 373 |
-
|
| 374 |
-
|
| 375 |
-
|
| 376 |
-
|
| 377 |
-
|
| 378 |
-
|
| 379 |
-
|
| 380 |
-
+ results_html
|
| 381 |
-
+ [gr.update(visible=False)] * (100 - len(results_html))
|
| 382 |
-
)
|
| 383 |
|
| 384 |
def next_page(
|
| 385 |
query,
|
|
@@ -405,23 +451,19 @@ if __name__ == "__main__":
|
|
| 405 |
print("num_processed_results", num_processed_results)
|
| 406 |
print("has_more_results", has_more_results)
|
| 407 |
print("received_results", received_results)
|
| 408 |
-
return
|
| 409 |
-
|
| 410 |
-
|
| 411 |
-
|
| 412 |
-
|
| 413 |
-
|
| 414 |
-
|
| 415 |
-
|
| 416 |
-
|
| 417 |
-
|
| 418 |
-
|
| 419 |
-
|
| 420 |
-
|
| 421 |
-
]
|
| 422 |
-
+ results_html
|
| 423 |
-
+ [gr.update(visible=False)] * (100 - len(results_html))
|
| 424 |
-
)
|
| 425 |
|
| 426 |
def filter_datasets(
|
| 427 |
lang,
|
|
@@ -431,14 +473,14 @@ if __name__ == "__main__":
|
|
| 431 |
exact_search,
|
| 432 |
datasets_filter,
|
| 433 |
):
|
| 434 |
-
results_html =
|
| 435 |
processed_results,
|
| 436 |
highlight_terms,
|
| 437 |
num_results,
|
| 438 |
exact_search,
|
| 439 |
datasets_filter,
|
| 440 |
)
|
| 441 |
-
return results_html
|
| 442 |
|
| 443 |
query.submit(
|
| 444 |
fn=submit,
|
|
@@ -452,8 +494,8 @@ if __name__ == "__main__":
|
|
| 452 |
available_datasets,
|
| 453 |
pagination,
|
| 454 |
received_results_state,
|
| 455 |
-
|
| 456 |
-
|
| 457 |
)
|
| 458 |
submit_btn.click(
|
| 459 |
submit,
|
|
@@ -467,8 +509,8 @@ if __name__ == "__main__":
|
|
| 467 |
available_datasets,
|
| 468 |
pagination,
|
| 469 |
received_results_state,
|
| 470 |
-
|
| 471 |
-
|
| 472 |
)
|
| 473 |
|
| 474 |
next_page_btn.click(
|
|
@@ -491,8 +533,8 @@ if __name__ == "__main__":
|
|
| 491 |
available_datasets,
|
| 492 |
pagination,
|
| 493 |
received_results_state,
|
| 494 |
-
|
| 495 |
-
|
| 496 |
)
|
| 497 |
|
| 498 |
available_datasets.change(
|
|
@@ -507,16 +549,4 @@ if __name__ == "__main__":
|
|
| 507 |
],
|
| 508 |
outputs=results_html,
|
| 509 |
)
|
| 510 |
-
results_html[0].change(
|
| 511 |
-
filter_datasets,
|
| 512 |
-
inputs=[
|
| 513 |
-
lang,
|
| 514 |
-
processed_results_state,
|
| 515 |
-
highlight_terms_state,
|
| 516 |
-
num_results_state,
|
| 517 |
-
exact_search_state,
|
| 518 |
-
available_datasets,
|
| 519 |
-
],
|
| 520 |
-
outputs=results_html,
|
| 521 |
-
)
|
| 522 |
demo.launch(enable_queue=True, debug=True)
|
|
|
|
| 88 |
|
| 89 |
|
| 90 |
def format_result(result, highlight_terms, exact_search, datasets_filter=None):
|
|
|
|
| 91 |
text, url, docid = result
|
| 92 |
if datasets_filter is not None:
|
| 93 |
datasets_filter = set(datasets_filter)
|
|
|
|
| 133 |
language = "FIXME"
|
| 134 |
result_html = """{}
|
| 135 |
<span style='font-size:14px; font-family: Arial; color:#7978FF; text-align: left;'>Document ID: {}</span>
|
| 136 |
+
<a href="https://forms.gle/AdBLLwRApqcLkHYA8" target="_blank"><button>🏴☠️ Flag result 🏴☠️</button></a><br>
|
| 137 |
+
|
| 138 |
<!-- <span style='font-size:12px; font-family: Arial; color:MediumAquaMarine'>Language: {}</span><br> -->
|
| 139 |
<span style='font-family: Arial;'>{}</span><br>
|
| 140 |
<br>
|
|
|
|
| 144 |
return "<p>" + result_html + "</p>"
|
| 145 |
|
| 146 |
|
| 147 |
+
def format_result_page_old(
|
| 148 |
+
language, results, highlight_terms, num_results, exact_search, datasets_filter=None
|
| 149 |
+
) -> gr.HTML:
|
| 150 |
+
|
| 151 |
+
filtered_num_results = 0
|
| 152 |
+
header_html = ""
|
| 153 |
+
|
| 154 |
+
if language == "detect_language" and not exact_search:
|
| 155 |
+
header_html += """<div style='font-family: Arial; color:MediumAquaMarine; text-align: center; line-height: 3em'>
|
| 156 |
+
Detected language: <b style='color:MediumAquaMarine'>{}</b></div>""".format(
|
| 157 |
+
list(results.keys())[0]
|
| 158 |
+
)
|
| 159 |
+
|
| 160 |
+
results_html = ""
|
| 161 |
+
for lang, results_for_lang in results.items():
|
| 162 |
+
if len(results_for_lang) == 0:
|
| 163 |
+
if exact_search:
|
| 164 |
+
results_html += """<div style='font-family: Arial; color:Silver; text-align: left; line-height: 3em'>
|
| 165 |
+
No results found.</div>"""
|
| 166 |
+
else:
|
| 167 |
+
results_html += """<div style='font-family: Arial; color:Silver; text-align: left; line-height: 3em'>
|
| 168 |
+
No results for language: <b>{}</b></div>""".format(
|
| 169 |
+
lang
|
| 170 |
+
)
|
| 171 |
+
continue
|
| 172 |
+
results_for_lang_html = ""
|
| 173 |
+
for result in results_for_lang:
|
| 174 |
+
result_html = format_result(
|
| 175 |
+
result, highlight_terms, exact_search, datasets_filter
|
| 176 |
+
)
|
| 177 |
+
if result_html != "":
|
| 178 |
+
filtered_num_results += 1
|
| 179 |
+
results_for_lang_html += result_html
|
| 180 |
+
if language == "all" and not exact_search:
|
| 181 |
+
results_for_lang_html = f"""
|
| 182 |
+
<details>
|
| 183 |
+
<summary style='font-family: Arial; color:MediumAquaMarine; text-align: left; line-height: 3em'>
|
| 184 |
+
Results for language: <b>{lang}</b>
|
| 185 |
+
</summary>
|
| 186 |
+
{results_for_lang_html}
|
| 187 |
+
</details>"""
|
| 188 |
+
results_html += results_for_lang_html
|
| 189 |
+
|
| 190 |
+
if num_results is not None:
|
| 191 |
+
header_html += """<div style='font-family: Arial; color:MediumAquaMarine; text-align: center; line-height: 3em'>
|
| 192 |
+
Total number of matches: <b style='color:MediumAquaMarine'>{}</b></div>""".format(
|
| 193 |
+
num_results
|
| 194 |
+
)
|
| 195 |
+
|
| 196 |
+
return header_html + results_html
|
| 197 |
+
|
| 198 |
+
|
| 199 |
def format_result_page(
|
| 200 |
results, highlight_terms, num_results, exact_search, datasets_filter=None
|
| 201 |
):
|
|
|
|
| 212 |
def extract_results_from_payload(query, language, payload, exact_search):
|
| 213 |
results = payload["results"]
|
| 214 |
|
| 215 |
+
processed_results = dict()
|
| 216 |
datasets = set()
|
| 217 |
highlight_terms = None
|
| 218 |
num_results = None
|
|
|
|
| 219 |
if exact_search:
|
| 220 |
highlight_terms = query
|
| 221 |
num_results = payload["num_results"]
|
| 222 |
+
results = {"dummy": results}
|
| 223 |
else:
|
| 224 |
highlight_terms = payload["highlight_terms"]
|
|
|
|
|
|
|
|
|
|
|
|
|
| 225 |
|
| 226 |
+
for lang, results_for_lang in results.items():
|
| 227 |
+
processed_results[lang] = list()
|
| 228 |
+
for result in results_for_lang:
|
| 229 |
+
text = result["text"]
|
| 230 |
+
url = (
|
| 231 |
+
result["meta"]["url"]
|
| 232 |
+
if "meta" in result
|
| 233 |
+
and result["meta"] is not None
|
| 234 |
+
and "url" in result["meta"]
|
| 235 |
+
else None
|
| 236 |
+
)
|
| 237 |
+
docid = result["docid"]
|
| 238 |
+
_, dataset, _ = docid.split("/")
|
| 239 |
+
datasets.add(dataset)
|
| 240 |
+
processed_results[lang].append((text, url, docid))
|
| 241 |
|
| 242 |
return processed_results, highlight_terms, num_results, list(datasets)
|
| 243 |
|
|
|
|
| 327 |
"vi",
|
| 328 |
"zh",
|
| 329 |
"detect_language",
|
| 330 |
+
"all",
|
| 331 |
],
|
| 332 |
value="en",
|
| 333 |
label="Language",
|
|
|
|
| 344 |
multiselect=True,
|
| 345 |
)
|
| 346 |
with gr.Row():
|
| 347 |
+
results_html = gr.HTML(label="Results")
|
| 348 |
+
|
|
|
|
|
|
|
| 349 |
with gr.Row(visible=False) as pagination:
|
| 350 |
next_page_btn = gr.Button("Next Page")
|
| 351 |
|
|
|
|
| 393 |
num_results
|
| 394 |
)
|
| 395 |
# print("processed_results", processed_results)
|
| 396 |
+
results_html = format_result_page_old(
|
| 397 |
+
lang, processed_results, highlight_terms, num_results, exact_search
|
| 398 |
)
|
| 399 |
return (
|
| 400 |
processed_results,
|
|
|
|
| 415 |
datasets,
|
| 416 |
) = run_query(query, lang, k, dropdown_input, max_page_size, 0)
|
| 417 |
has_more_results = exact_search and (num_results > max_page_size)
|
| 418 |
+
return [
|
| 419 |
+
processed_results,
|
| 420 |
+
highlight_terms,
|
| 421 |
+
num_results,
|
| 422 |
+
exact_search,
|
| 423 |
+
gr.update(visible=True),
|
| 424 |
+
gr.Dropdown.update(choices=datasets, value=datasets),
|
| 425 |
+
gr.update(visible=has_more_results),
|
| 426 |
+
len(processed_results),
|
| 427 |
+
results_html,
|
| 428 |
+
]
|
|
|
|
|
|
|
|
|
|
| 429 |
|
| 430 |
def next_page(
|
| 431 |
query,
|
|
|
|
| 451 |
print("num_processed_results", num_processed_results)
|
| 452 |
print("has_more_results", has_more_results)
|
| 453 |
print("received_results", received_results)
|
| 454 |
+
return [
|
| 455 |
+
processed_results,
|
| 456 |
+
highlight_terms,
|
| 457 |
+
num_results,
|
| 458 |
+
exact_search,
|
| 459 |
+
gr.update(visible=True),
|
| 460 |
+
gr.Dropdown.update(choices=datasets, value=datasets),
|
| 461 |
+
gr.update(
|
| 462 |
+
visible=num_processed_results >= max_page_size and has_more_results
|
| 463 |
+
),
|
| 464 |
+
received_results + num_processed_results,
|
| 465 |
+
results_html,
|
| 466 |
+
]
|
|
|
|
|
|
|
|
|
|
|
|
|
| 467 |
|
| 468 |
def filter_datasets(
|
| 469 |
lang,
|
|
|
|
| 473 |
exact_search,
|
| 474 |
datasets_filter,
|
| 475 |
):
|
| 476 |
+
results_html = format_result_page_old(
|
| 477 |
processed_results,
|
| 478 |
highlight_terms,
|
| 479 |
num_results,
|
| 480 |
exact_search,
|
| 481 |
datasets_filter,
|
| 482 |
)
|
| 483 |
+
return results_html
|
| 484 |
|
| 485 |
query.submit(
|
| 486 |
fn=submit,
|
|
|
|
| 494 |
available_datasets,
|
| 495 |
pagination,
|
| 496 |
received_results_state,
|
| 497 |
+
results_html,
|
| 498 |
+
],
|
| 499 |
)
|
| 500 |
submit_btn.click(
|
| 501 |
submit,
|
|
|
|
| 509 |
available_datasets,
|
| 510 |
pagination,
|
| 511 |
received_results_state,
|
| 512 |
+
results_html,
|
| 513 |
+
],
|
| 514 |
)
|
| 515 |
|
| 516 |
next_page_btn.click(
|
|
|
|
| 533 |
available_datasets,
|
| 534 |
pagination,
|
| 535 |
received_results_state,
|
| 536 |
+
results_html,
|
| 537 |
+
],
|
| 538 |
)
|
| 539 |
|
| 540 |
available_datasets.change(
|
|
|
|
| 549 |
],
|
| 550 |
outputs=results_html,
|
| 551 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 552 |
demo.launch(enable_queue=True, debug=True)
|