Spaces:
Sleeping
Sleeping
diogo.rodrigues.silva commited on
Commit ·
3fa9493
1
Parent(s): af5dc19
update app.py
Browse files- __pycache__/app.cpython-310.pyc +0 -0
- app.py +42 -8
__pycache__/app.cpython-310.pyc
DELETED
|
Binary file (13.7 kB)
|
|
|
app.py
CHANGED
|
@@ -18,6 +18,7 @@ from tempfile import mkdtemp
|
|
| 18 |
from urllib.parse import quote
|
| 19 |
|
| 20 |
import gradio as gr
|
|
|
|
| 21 |
import uvicorn
|
| 22 |
from fastapi import Depends, FastAPI, HTTPException, Query, status
|
| 23 |
from fastapi.responses import FileResponse
|
|
@@ -135,6 +136,26 @@ def _download_markdown(file_path: Path, label: str) -> str:
|
|
| 135 |
return f"[{label}]({_download_href(file_path)})"
|
| 136 |
|
| 137 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
def _auth_credentials() -> tuple[str, str]:
|
| 139 |
username = (os.getenv(APP_USERNAME_ENV) or "").strip()
|
| 140 |
password = (os.getenv(APP_PASSWORD_ENV) or "").strip()
|
|
@@ -338,8 +359,19 @@ def screen_excel(
|
|
| 338 |
return
|
| 339 |
|
| 340 |
progress(1, desc="Screening complete.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 341 |
yield (
|
| 342 |
-
|
| 343 |
_download_markdown(screened_output, "Download Screened Excel"),
|
| 344 |
)
|
| 345 |
|
|
@@ -353,9 +385,9 @@ def build_app() -> gr.Blocks:
|
|
| 353 |
)
|
| 354 |
|
| 355 |
with gr.Blocks(title="Reference Parser + Foundry Screener") as demo:
|
| 356 |
-
gr.Markdown("# Reference
|
| 357 |
gr.Markdown(
|
| 358 |
-
"Upload `.txt/.medline/.ris` files, parse into one Excel, then screen
|
| 359 |
)
|
| 360 |
gr.Markdown(f"**Secrets status:** {secrets_note}")
|
| 361 |
|
|
@@ -386,15 +418,17 @@ def build_app() -> gr.Blocks:
|
|
| 386 |
)
|
| 387 |
criteria_inclusion_text = gr.Textbox(
|
| 388 |
label="Inclusion Criteria (one per line)",
|
| 389 |
-
lines=
|
| 390 |
)
|
| 391 |
criteria_exclusion_text = gr.Textbox(
|
| 392 |
label="Exclusion Criteria (one per line)",
|
| 393 |
-
lines=
|
| 394 |
)
|
| 395 |
-
criteria_notes = gr.Textbox(label="Notes (optional)", lines=
|
| 396 |
-
title_column = gr.Textbox(label="Title Column", value="Title")
|
| 397 |
-
abstract_column = gr.Textbox(label="Abstract Column", value="Abstract")
|
|
|
|
|
|
|
| 398 |
|
| 399 |
parse_btn.click(
|
| 400 |
fn=parse_files,
|
|
|
|
| 18 |
from urllib.parse import quote
|
| 19 |
|
| 20 |
import gradio as gr
|
| 21 |
+
import pandas as pd
|
| 22 |
import uvicorn
|
| 23 |
from fastapi import Depends, FastAPI, HTTPException, Query, status
|
| 24 |
from fastapi.responses import FileResponse
|
|
|
|
| 136 |
return f"[{label}]({_download_href(file_path)})"
|
| 137 |
|
| 138 |
|
| 139 |
+
def _screening_verdict_counts(screened_excel_path: Path) -> dict[str, int]:
|
| 140 |
+
df = pd.read_excel(screened_excel_path, engine="openpyxl")
|
| 141 |
+
if "LLM_verdict" not in df.columns:
|
| 142 |
+
raise KeyError("Expected 'LLM_verdict' column was not found in screening output.")
|
| 143 |
+
|
| 144 |
+
verdict_counts = (
|
| 145 |
+
df["LLM_verdict"]
|
| 146 |
+
.astype(str)
|
| 147 |
+
.str.strip()
|
| 148 |
+
.str.lower()
|
| 149 |
+
.value_counts()
|
| 150 |
+
.to_dict()
|
| 151 |
+
)
|
| 152 |
+
return {
|
| 153 |
+
"include": int(verdict_counts.get("include", 0)),
|
| 154 |
+
"exclude": int(verdict_counts.get("exclude", 0)),
|
| 155 |
+
"unclear": int(verdict_counts.get("unclear", 0)),
|
| 156 |
+
}
|
| 157 |
+
|
| 158 |
+
|
| 159 |
def _auth_credentials() -> tuple[str, str]:
|
| 160 |
username = (os.getenv(APP_USERNAME_ENV) or "").strip()
|
| 161 |
password = (os.getenv(APP_PASSWORD_ENV) or "").strip()
|
|
|
|
| 359 |
return
|
| 360 |
|
| 361 |
progress(1, desc="Screening complete.")
|
| 362 |
+
try:
|
| 363 |
+
verdict_counts = _screening_verdict_counts(screened_output)
|
| 364 |
+
completed_status = (
|
| 365 |
+
"Screening complete: "
|
| 366 |
+
f"Included {verdict_counts['include']} | "
|
| 367 |
+
f"Excluded {verdict_counts['exclude']} | "
|
| 368 |
+
f"Unclear {verdict_counts['unclear']}"
|
| 369 |
+
)
|
| 370 |
+
except Exception:
|
| 371 |
+
completed_status = "Screening complete."
|
| 372 |
+
|
| 373 |
yield (
|
| 374 |
+
completed_status,
|
| 375 |
_download_markdown(screened_output, "Download Screened Excel"),
|
| 376 |
)
|
| 377 |
|
|
|
|
| 385 |
)
|
| 386 |
|
| 387 |
with gr.Blocks(title="Reference Parser + Foundry Screener") as demo:
|
| 388 |
+
gr.Markdown("# Reference Parsing, Deduplication and LLM-assisted Screening")
|
| 389 |
gr.Markdown(
|
| 390 |
+
"Upload `.txt/.medline/.ris` files, parse into one Excel, then screen for inclusion/exclusion criteria."
|
| 391 |
)
|
| 392 |
gr.Markdown(f"**Secrets status:** {secrets_note}")
|
| 393 |
|
|
|
|
| 418 |
)
|
| 419 |
criteria_inclusion_text = gr.Textbox(
|
| 420 |
label="Inclusion Criteria (one per line)",
|
| 421 |
+
lines=2,
|
| 422 |
)
|
| 423 |
criteria_exclusion_text = gr.Textbox(
|
| 424 |
label="Exclusion Criteria (one per line)",
|
| 425 |
+
lines=2,
|
| 426 |
)
|
| 427 |
+
criteria_notes = gr.Textbox(label="Notes (optional)", lines=1)
|
| 428 |
+
# title_column = gr.Textbox(label="Title Column", value="Title")
|
| 429 |
+
# abstract_column = gr.Textbox(label="Abstract Column", value="Abstract")
|
| 430 |
+
title_column = "Title"
|
| 431 |
+
abstract_column = "Abstract"
|
| 432 |
|
| 433 |
parse_btn.click(
|
| 434 |
fn=parse_files,
|