Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -136,49 +136,31 @@ button.variant-secondary {
|
|
| 136 |
border-radius: 6px !important;
|
| 137 |
}
|
| 138 |
|
| 139 |
-
/* =========================================
|
| 140 |
-
/*
|
| 141 |
-
/* =========================================
|
| 142 |
-
|
| 143 |
-
/*
|
| 144 |
-
|
| 145 |
-
|
| 146 |
-
|
| 147 |
-
|
| 148 |
-
/* Dropdown container */
|
| 149 |
-
div[id^="headlessui-portal"] div[role="listbox"] {
|
| 150 |
-
background-color: #f3f4f6 !important;
|
| 151 |
-
border: 1px solid #d1d5db !important;
|
| 152 |
-
box-shadow: 0 8px 24px rgba(0,0,0,0.15) !important;
|
| 153 |
-
border-radius: 8px !important;
|
| 154 |
-
overflow: hidden !important;
|
| 155 |
}
|
| 156 |
|
| 157 |
-
/*
|
| 158 |
-
|
| 159 |
-
background-color: #
|
| 160 |
-
color: #
|
| 161 |
-
padding: 10px 14px !important;
|
| 162 |
-
}
|
| 163 |
-
|
| 164 |
-
/* Hover state */
|
| 165 |
-
div[id^="headlessui-portal"] div[role="option"]:hover {
|
| 166 |
-
background-color: #e5e7eb !important;
|
| 167 |
-
color: #111827 !important;
|
| 168 |
}
|
| 169 |
|
| 170 |
-
/*
|
| 171 |
-
|
| 172 |
-
|
| 173 |
-
|
| 174 |
-
background-color: #2563eb !important;
|
| 175 |
-
color: #ffffff !important;
|
| 176 |
}
|
| 177 |
|
| 178 |
-
/*
|
| 179 |
-
|
| 180 |
color: #ffffff !important;
|
| 181 |
-
fill: #ffffff !important;
|
| 182 |
}
|
| 183 |
/* ================================================= */
|
| 184 |
/* METRIC LABEL BLACK BACKGROUND */
|
|
@@ -233,7 +215,7 @@ class PaperSearchPro:
|
|
| 233 |
filters.append(f"publication_year:<{int(year_to)+1}")
|
| 234 |
if open_access:
|
| 235 |
filters.append("open_access.is_oa:true")
|
| 236 |
-
|
| 237 |
# Add author filter if provided
|
| 238 |
if authors and authors.strip():
|
| 239 |
# Clean and format author names for filtering
|
|
@@ -320,7 +302,7 @@ class PaperSearchPro:
|
|
| 320 |
# Sort by relevance score (our calculated score)
|
| 321 |
filtered.sort(key=lambda x: x["Relevance"], reverse=True)
|
| 322 |
all_papers = filtered[:max_papers]
|
| 323 |
-
|
| 324 |
# ===============================
|
| 325 |
# AUTHOR FILTERING (post-API)
|
| 326 |
# ===============================
|
|
@@ -328,24 +310,24 @@ class PaperSearchPro:
|
|
| 328 |
# If we have author filter but no keyword hierarchy
|
| 329 |
author_list = [a.strip().lower() for a in authors.split(",") if a.strip()]
|
| 330 |
filtered = []
|
| 331 |
-
|
| 332 |
for p in all_papers:
|
| 333 |
all_authors_lower = p['All Authors'].lower()
|
| 334 |
match_all = True
|
| 335 |
matched_authors = []
|
| 336 |
-
|
| 337 |
for author in author_list:
|
| 338 |
if author in all_authors_lower:
|
| 339 |
matched_authors.append(author)
|
| 340 |
else:
|
| 341 |
match_all = False
|
| 342 |
break
|
| 343 |
-
|
| 344 |
if match_all:
|
| 345 |
p["Relevance"] = len(matched_authors)
|
| 346 |
p["Matched Keywords"] = f"Authors: {', '.join(matched_authors)}"
|
| 347 |
filtered.append(p)
|
| 348 |
-
|
| 349 |
all_papers = filtered[:max_papers]
|
| 350 |
|
| 351 |
total = len(all_papers)
|
|
@@ -354,24 +336,24 @@ class PaperSearchPro:
|
|
| 354 |
year_range = f"{min(years)} - {max(years)}" if years else "N/A"
|
| 355 |
|
| 356 |
df = pd.DataFrame(all_papers)
|
| 357 |
-
|
| 358 |
# Reorder columns to include BibTeX and remove All Authors from display
|
| 359 |
-
column_order = ['Title', 'Authors', 'Year', 'Citations', 'Journal', 'Abstract',
|
| 360 |
'BibTeX', 'Relevance', 'Matched Keywords']
|
| 361 |
df = df[column_order]
|
| 362 |
|
| 363 |
return df, all_papers, total, avg_rel, year_range
|
| 364 |
-
|
| 365 |
def search_specific_paper(self, search_type, search_value):
|
| 366 |
"""Search for a specific paper by title, author, or DOI"""
|
| 367 |
-
|
| 368 |
all_papers = []
|
| 369 |
-
|
| 370 |
if search_type == "DOI":
|
| 371 |
# Clean DOI (remove URL if present)
|
| 372 |
if "doi.org/" in search_value:
|
| 373 |
search_value = search_value.split("doi.org/")[-1]
|
| 374 |
-
|
| 375 |
# Direct lookup by DOI
|
| 376 |
url = f"https://api.openalex.org/works/doi/{search_value}"
|
| 377 |
try:
|
|
@@ -382,7 +364,7 @@ class PaperSearchPro:
|
|
| 382 |
all_papers.append(paper)
|
| 383 |
except Exception as e:
|
| 384 |
print(f"Error fetching DOI: {e}")
|
| 385 |
-
|
| 386 |
elif search_type == "Exact Title":
|
| 387 |
# Search by exact title
|
| 388 |
params = {
|
|
@@ -394,7 +376,7 @@ class PaperSearchPro:
|
|
| 394 |
response = requests.get(self.base_url, params=params, timeout=15)
|
| 395 |
data = response.json()
|
| 396 |
results = data.get("results", [])
|
| 397 |
-
|
| 398 |
for work in results:
|
| 399 |
# Check for close title match
|
| 400 |
title = work.get("title", "").lower()
|
|
@@ -404,7 +386,7 @@ class PaperSearchPro:
|
|
| 404 |
break # Take the first good match
|
| 405 |
except Exception as e:
|
| 406 |
print(f"Error searching title: {e}")
|
| 407 |
-
|
| 408 |
elif search_type == "Specific Author":
|
| 409 |
# Search for papers by a specific author
|
| 410 |
params = {
|
|
@@ -417,26 +399,26 @@ class PaperSearchPro:
|
|
| 417 |
response = requests.get(self.base_url, params=params, timeout=15)
|
| 418 |
data = response.json()
|
| 419 |
results = data.get("results", [])
|
| 420 |
-
|
| 421 |
for work in results[:20]: # Limit to 20 papers
|
| 422 |
paper = self.extract_info(work)
|
| 423 |
all_papers.append(paper)
|
| 424 |
except Exception as e:
|
| 425 |
print(f"Error searching author: {e}")
|
| 426 |
-
|
| 427 |
if all_papers:
|
| 428 |
for i, p in enumerate(all_papers):
|
| 429 |
p['Rank'] = i + 1
|
| 430 |
-
|
| 431 |
df = pd.DataFrame(all_papers)
|
| 432 |
column_order = ['Title', 'Authors', 'Year', 'Citations', 'Journal', 'Abstract', 'BibTeX']
|
| 433 |
# Only include columns that exist
|
| 434 |
available_cols = [col for col in column_order if col in df.columns]
|
| 435 |
df = df[available_cols]
|
| 436 |
-
|
| 437 |
summary = f"✅ Found **{len(all_papers)}** paper(s) matching your query"
|
| 438 |
return df, all_papers, summary
|
| 439 |
-
|
| 440 |
return pd.DataFrame(), [], "❌ No papers found"
|
| 441 |
|
| 442 |
def extract_info(self, work):
|
|
@@ -454,12 +436,12 @@ class PaperSearchPro:
|
|
| 454 |
|
| 455 |
# Get all authors - FULL LIST (no et al.)
|
| 456 |
all_authors = []
|
| 457 |
-
|
| 458 |
for authorship in work.get("authorships", []):
|
| 459 |
# FIRST PRIORITY: Use raw_author_name (full name as published)
|
| 460 |
raw_name = authorship.get("raw_author_name", "")
|
| 461 |
display_name = authorship.get("author", {}).get("display_name", "")
|
| 462 |
-
|
| 463 |
# Choose the best name
|
| 464 |
if raw_name and len(raw_name) > 2:
|
| 465 |
# Raw name is usually more complete (e.g., "Mamoru SHOJI" vs "M. Shoji")
|
|
@@ -468,18 +450,18 @@ class PaperSearchPro:
|
|
| 468 |
author_name = display_name
|
| 469 |
else:
|
| 470 |
author_name = ""
|
| 471 |
-
|
| 472 |
if author_name:
|
| 473 |
all_authors.append(author_name)
|
| 474 |
-
|
| 475 |
# Show ALL authors, no truncation
|
| 476 |
authors_str = ", ".join(all_authors)
|
| 477 |
-
|
| 478 |
# Get DOI for reference
|
| 479 |
doi = work.get("doi", "")
|
| 480 |
if doi:
|
| 481 |
doi = doi.replace("https://doi.org/", "")
|
| 482 |
-
|
| 483 |
# Generate BibTeX entry
|
| 484 |
bibtex = self.generate_bibtex(work, all_authors)
|
| 485 |
|
|
@@ -496,33 +478,33 @@ class PaperSearchPro:
|
|
| 496 |
"Relevance": 0,
|
| 497 |
"Matched Keywords": ""
|
| 498 |
}
|
| 499 |
-
|
| 500 |
def generate_bibtex(self, work, authors):
|
| 501 |
"""Generate BibTeX entry for a paper"""
|
| 502 |
-
|
| 503 |
# Create citation key
|
| 504 |
first_author = authors[0].split()[-1] if authors else "Unknown"
|
| 505 |
year = work.get("publication_year", "0000")
|
| 506 |
title_words = work.get("title", "Untitled").split()[:3]
|
| 507 |
title_short = "".join([w[0] for w in title_words if w]) if title_words else "Paper"
|
| 508 |
citation_key = f"{first_author}{year}{title_short}"
|
| 509 |
-
|
| 510 |
# Clean the citation key (remove special characters)
|
| 511 |
citation_key = ''.join(c for c in citation_key if c.isalnum())
|
| 512 |
-
|
| 513 |
# Get DOI or URL
|
| 514 |
doi = work.get("doi", "")
|
| 515 |
url = work.get("primary_location", {}).get("landing_page_url", "")
|
| 516 |
if not url and work.get("open_access", {}).get("oa_url"):
|
| 517 |
url = work.get("open_access", {}).get("oa_url")
|
| 518 |
-
|
| 519 |
# Determine entry type (article, book, etc.)
|
| 520 |
pub_type = "article"
|
| 521 |
if work.get("type") == "book":
|
| 522 |
pub_type = "book"
|
| 523 |
elif work.get("type") == "book-chapter":
|
| 524 |
pub_type = "incollection"
|
| 525 |
-
|
| 526 |
# Build BibTeX string
|
| 527 |
bibtex_lines = [
|
| 528 |
f"@{pub_type}{{{citation_key},",
|
|
@@ -530,7 +512,7 @@ class PaperSearchPro:
|
|
| 530 |
f" author = {{{' and '.join(authors)}}},",
|
| 531 |
f" year = {{{work.get('publication_year', '0000')}}},"
|
| 532 |
]
|
| 533 |
-
|
| 534 |
# Add journal/booktitle
|
| 535 |
source = (work.get("primary_location", {}).get("source") or {})
|
| 536 |
if pub_type == "article":
|
|
@@ -539,24 +521,24 @@ class PaperSearchPro:
|
|
| 539 |
else:
|
| 540 |
booktitle = source.get("display_name", "Unknown Publication")
|
| 541 |
bibtex_lines.append(f" booktitle = {{{booktitle}}},")
|
| 542 |
-
|
| 543 |
# Add volume, issue if available
|
| 544 |
if work.get("biblio", {}).get("volume"):
|
| 545 |
bibtex_lines.append(f" volume = {{{work['biblio']['volume']}}},")
|
| 546 |
if work.get("biblio", {}).get("issue"):
|
| 547 |
bibtex_lines.append(f" number = {{{work['biblio']['issue']}}},")
|
| 548 |
-
|
| 549 |
# Add pages if available
|
| 550 |
if work.get("biblio", {}).get("first_page") and work.get("biblio", {}).get("last_page"):
|
| 551 |
pages = f"{work['biblio']['first_page']}--{work['biblio']['last_page']}"
|
| 552 |
bibtex_lines.append(f" pages = {{{pages}}},")
|
| 553 |
-
|
| 554 |
# Add DOI/URL
|
| 555 |
if doi:
|
| 556 |
bibtex_lines.append(f" doi = {{{doi}}},")
|
| 557 |
elif url:
|
| 558 |
bibtex_lines.append(f" url = {{{url}}},")
|
| 559 |
-
|
| 560 |
# Add abstract if available
|
| 561 |
if work.get("abstract_inverted_index"):
|
| 562 |
# Rebuild abstract for BibTeX
|
|
@@ -568,10 +550,10 @@ class PaperSearchPro:
|
|
| 568 |
abstract = " ".join([w[1] for w in words])
|
| 569 |
if abstract:
|
| 570 |
bibtex_lines.append(f" abstract = {{{abstract}}},")
|
| 571 |
-
|
| 572 |
# Close the entry
|
| 573 |
bibtex_lines.append("}")
|
| 574 |
-
|
| 575 |
return "\n".join(bibtex_lines)
|
| 576 |
|
| 577 |
|
|
@@ -603,7 +585,7 @@ def create_interface():
|
|
| 603 |
with gr.Column(elem_classes="white-card"):
|
| 604 |
|
| 605 |
query = gr.Textbox(
|
| 606 |
-
label="Search Topic",
|
| 607 |
value="photoacoustic imaging",
|
| 608 |
placeholder="Enter research topic or keywords..."
|
| 609 |
)
|
|
@@ -615,7 +597,7 @@ def create_interface():
|
|
| 615 |
value="blood, photoacoustics, imaging",
|
| 616 |
placeholder="e.g., blood, photoacoustics, imaging (comma-separated)"
|
| 617 |
)
|
| 618 |
-
|
| 619 |
with gr.Column():
|
| 620 |
authors = gr.Textbox(
|
| 621 |
label="Filter by Authors (optional)",
|
|
@@ -627,7 +609,7 @@ def create_interface():
|
|
| 627 |
# MAX PAPERS SLIDER WITH BLACK REGION COMMENT
|
| 628 |
# ==========================
|
| 629 |
max_papers = gr.Slider(1000, 50000, value=1000, step=100, label="Max Papers")
|
| 630 |
-
|
| 631 |
# Add black region comment below slider
|
| 632 |
gr.HTML("""
|
| 633 |
<div class="slider-comment">
|
|
@@ -666,23 +648,23 @@ def create_interface():
|
|
| 666 |
- **Exact Title** (e.g., "Photoacoustic imaging in biomedicine")
|
| 667 |
- **Author** (find papers by a specific researcher)
|
| 668 |
""")
|
| 669 |
-
|
| 670 |
with gr.Row():
|
| 671 |
specific_search_type = gr.Dropdown(
|
| 672 |
label="Search by",
|
| 673 |
choices=["DOI", "Exact Title", "Specific Author"],
|
| 674 |
value="DOI"
|
| 675 |
)
|
| 676 |
-
|
| 677 |
specific_search_value = gr.Textbox(
|
| 678 |
label="Enter value",
|
| 679 |
placeholder="e.g., 10.1038/nature12373 or exact paper title or author name",
|
| 680 |
scale=2
|
| 681 |
)
|
| 682 |
-
|
| 683 |
specific_search_btn = gr.Button("🔍 FIND PAPER", variant="primary")
|
| 684 |
specific_summary = gr.Markdown("Ready")
|
| 685 |
-
|
| 686 |
# Metrics section (common for both tabs)
|
| 687 |
with gr.Row():
|
| 688 |
total_display = gr.Markdown("**0**", elem_classes="metric-value")
|
|
@@ -714,7 +696,7 @@ def create_interface():
|
|
| 714 |
|
| 715 |
with gr.Row():
|
| 716 |
filename = gr.Textbox(
|
| 717 |
-
label="Filename",
|
| 718 |
value="papers_results",
|
| 719 |
placeholder="Enter base filename without extension"
|
| 720 |
)
|
|
@@ -745,13 +727,13 @@ def create_interface():
|
|
| 745 |
df, papers, summary = searcher.search_specific_paper(search_type, search_value)
|
| 746 |
if not papers:
|
| 747 |
return df, papers, summary, "**0**", "**0.0**", "**N/A**"
|
| 748 |
-
|
| 749 |
# Calculate metrics for display
|
| 750 |
total = len(papers)
|
| 751 |
avg_rel = 0
|
| 752 |
years = [p["Year"] for p in papers if p["Year"] != "N/A"]
|
| 753 |
year_range = f"{min(years)} - {max(years)}" if years else "N/A"
|
| 754 |
-
|
| 755 |
return df, papers, summary, f"**{total}**", f"**{avg_rel:.2f}**", f"**{year_range}**"
|
| 756 |
except Exception as e:
|
| 757 |
empty_df = pd.DataFrame(columns=['Title', 'Authors', 'Year', 'Citations', 'Journal', 'Abstract', 'BibTeX'])
|
|
@@ -763,7 +745,7 @@ def create_interface():
|
|
| 763 |
inputs=[query, keywords, authors, max_papers, year_from, year_to, open_access, sort_by, sort_order],
|
| 764 |
outputs=[results, state, total_display, avg_display, year_display]
|
| 765 |
)
|
| 766 |
-
|
| 767 |
specific_search_btn.click(
|
| 768 |
specific_search_wrapper,
|
| 769 |
inputs=[specific_search_type, specific_search_value],
|
|
@@ -777,7 +759,7 @@ def create_interface():
|
|
| 777 |
|
| 778 |
df = pd.DataFrame(papers)
|
| 779 |
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
|
| 780 |
-
|
| 781 |
# Remove All Authors column from export if it exists
|
| 782 |
if 'All Authors' in df.columns:
|
| 783 |
df = df.drop(columns=['All Authors'])
|
|
@@ -805,7 +787,7 @@ def create_interface():
|
|
| 805 |
f.write(f"Generated: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n")
|
| 806 |
f.write(f"Total Papers: {len(df)}\n")
|
| 807 |
f.write("="*80 + "\n\n")
|
| 808 |
-
|
| 809 |
for i, row in df.iterrows():
|
| 810 |
f.write(f"PAPER #{i+1}\n")
|
| 811 |
f.write("-"*40 + "\n")
|
|
@@ -834,7 +816,7 @@ def create_interface():
|
|
| 834 |
if __name__ == "__main__":
|
| 835 |
demo = create_interface()
|
| 836 |
demo.launch(
|
| 837 |
-
inbrowser=True,
|
| 838 |
height=1000,
|
| 839 |
server_name="0.0.0.0", # Makes it accessible on network
|
| 840 |
share=False # Set to True for temporary public link
|
|
|
|
| 136 |
border-radius: 6px !important;
|
| 137 |
}
|
| 138 |
|
| 139 |
+
/* ========================================= */
|
| 140 |
+
/* FORCE BUTTON TEXT TO ALWAYS BE WHITE */
|
| 141 |
+
/* ========================================= */
|
| 142 |
+
|
| 143 |
+
/* All buttons */
|
| 144 |
+
button,
|
| 145 |
+
.gr-button {
|
| 146 |
+
color: #ffffff !important; /* TEXT WHITE */
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 147 |
}
|
| 148 |
|
| 149 |
+
/* Primary button */
|
| 150 |
+
button.variant-primary {
|
| 151 |
+
background-color: #2563eb !important; /* keep blue */
|
| 152 |
+
color: #ffffff !important; /* TEXT WHITE */
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 153 |
}
|
| 154 |
|
| 155 |
+
/* Secondary button */
|
| 156 |
+
button.variant-secondary {
|
| 157 |
+
background-color: #6b7280 !important; /* keep gray */
|
| 158 |
+
color: #ffffff !important; /* TEXT WHITE */
|
|
|
|
|
|
|
| 159 |
}
|
| 160 |
|
| 161 |
+
/* Keep text white even on hover */
|
| 162 |
+
button:hover {
|
| 163 |
color: #ffffff !important;
|
|
|
|
| 164 |
}
|
| 165 |
/* ================================================= */
|
| 166 |
/* METRIC LABEL BLACK BACKGROUND */
|
|
|
|
| 215 |
filters.append(f"publication_year:<{int(year_to)+1}")
|
| 216 |
if open_access:
|
| 217 |
filters.append("open_access.is_oa:true")
|
| 218 |
+
|
| 219 |
# Add author filter if provided
|
| 220 |
if authors and authors.strip():
|
| 221 |
# Clean and format author names for filtering
|
|
|
|
| 302 |
# Sort by relevance score (our calculated score)
|
| 303 |
filtered.sort(key=lambda x: x["Relevance"], reverse=True)
|
| 304 |
all_papers = filtered[:max_papers]
|
| 305 |
+
|
| 306 |
# ===============================
|
| 307 |
# AUTHOR FILTERING (post-API)
|
| 308 |
# ===============================
|
|
|
|
| 310 |
# If we have author filter but no keyword hierarchy
|
| 311 |
author_list = [a.strip().lower() for a in authors.split(",") if a.strip()]
|
| 312 |
filtered = []
|
| 313 |
+
|
| 314 |
for p in all_papers:
|
| 315 |
all_authors_lower = p['All Authors'].lower()
|
| 316 |
match_all = True
|
| 317 |
matched_authors = []
|
| 318 |
+
|
| 319 |
for author in author_list:
|
| 320 |
if author in all_authors_lower:
|
| 321 |
matched_authors.append(author)
|
| 322 |
else:
|
| 323 |
match_all = False
|
| 324 |
break
|
| 325 |
+
|
| 326 |
if match_all:
|
| 327 |
p["Relevance"] = len(matched_authors)
|
| 328 |
p["Matched Keywords"] = f"Authors: {', '.join(matched_authors)}"
|
| 329 |
filtered.append(p)
|
| 330 |
+
|
| 331 |
all_papers = filtered[:max_papers]
|
| 332 |
|
| 333 |
total = len(all_papers)
|
|
|
|
| 336 |
year_range = f"{min(years)} - {max(years)}" if years else "N/A"
|
| 337 |
|
| 338 |
df = pd.DataFrame(all_papers)
|
| 339 |
+
|
| 340 |
# Reorder columns to include BibTeX and remove All Authors from display
|
| 341 |
+
column_order = ['Title', 'Authors', 'Year', 'Citations', 'Journal', 'Abstract',
|
| 342 |
'BibTeX', 'Relevance', 'Matched Keywords']
|
| 343 |
df = df[column_order]
|
| 344 |
|
| 345 |
return df, all_papers, total, avg_rel, year_range
|
| 346 |
+
|
| 347 |
def search_specific_paper(self, search_type, search_value):
|
| 348 |
"""Search for a specific paper by title, author, or DOI"""
|
| 349 |
+
|
| 350 |
all_papers = []
|
| 351 |
+
|
| 352 |
if search_type == "DOI":
|
| 353 |
# Clean DOI (remove URL if present)
|
| 354 |
if "doi.org/" in search_value:
|
| 355 |
search_value = search_value.split("doi.org/")[-1]
|
| 356 |
+
|
| 357 |
# Direct lookup by DOI
|
| 358 |
url = f"https://api.openalex.org/works/doi/{search_value}"
|
| 359 |
try:
|
|
|
|
| 364 |
all_papers.append(paper)
|
| 365 |
except Exception as e:
|
| 366 |
print(f"Error fetching DOI: {e}")
|
| 367 |
+
|
| 368 |
elif search_type == "Exact Title":
|
| 369 |
# Search by exact title
|
| 370 |
params = {
|
|
|
|
| 376 |
response = requests.get(self.base_url, params=params, timeout=15)
|
| 377 |
data = response.json()
|
| 378 |
results = data.get("results", [])
|
| 379 |
+
|
| 380 |
for work in results:
|
| 381 |
# Check for close title match
|
| 382 |
title = work.get("title", "").lower()
|
|
|
|
| 386 |
break # Take the first good match
|
| 387 |
except Exception as e:
|
| 388 |
print(f"Error searching title: {e}")
|
| 389 |
+
|
| 390 |
elif search_type == "Specific Author":
|
| 391 |
# Search for papers by a specific author
|
| 392 |
params = {
|
|
|
|
| 399 |
response = requests.get(self.base_url, params=params, timeout=15)
|
| 400 |
data = response.json()
|
| 401 |
results = data.get("results", [])
|
| 402 |
+
|
| 403 |
for work in results[:20]: # Limit to 20 papers
|
| 404 |
paper = self.extract_info(work)
|
| 405 |
all_papers.append(paper)
|
| 406 |
except Exception as e:
|
| 407 |
print(f"Error searching author: {e}")
|
| 408 |
+
|
| 409 |
if all_papers:
|
| 410 |
for i, p in enumerate(all_papers):
|
| 411 |
p['Rank'] = i + 1
|
| 412 |
+
|
| 413 |
df = pd.DataFrame(all_papers)
|
| 414 |
column_order = ['Title', 'Authors', 'Year', 'Citations', 'Journal', 'Abstract', 'BibTeX']
|
| 415 |
# Only include columns that exist
|
| 416 |
available_cols = [col for col in column_order if col in df.columns]
|
| 417 |
df = df[available_cols]
|
| 418 |
+
|
| 419 |
summary = f"✅ Found **{len(all_papers)}** paper(s) matching your query"
|
| 420 |
return df, all_papers, summary
|
| 421 |
+
|
| 422 |
return pd.DataFrame(), [], "❌ No papers found"
|
| 423 |
|
| 424 |
def extract_info(self, work):
|
|
|
|
| 436 |
|
| 437 |
# Get all authors - FULL LIST (no et al.)
|
| 438 |
all_authors = []
|
| 439 |
+
|
| 440 |
for authorship in work.get("authorships", []):
|
| 441 |
# FIRST PRIORITY: Use raw_author_name (full name as published)
|
| 442 |
raw_name = authorship.get("raw_author_name", "")
|
| 443 |
display_name = authorship.get("author", {}).get("display_name", "")
|
| 444 |
+
|
| 445 |
# Choose the best name
|
| 446 |
if raw_name and len(raw_name) > 2:
|
| 447 |
# Raw name is usually more complete (e.g., "Mamoru SHOJI" vs "M. Shoji")
|
|
|
|
| 450 |
author_name = display_name
|
| 451 |
else:
|
| 452 |
author_name = ""
|
| 453 |
+
|
| 454 |
if author_name:
|
| 455 |
all_authors.append(author_name)
|
| 456 |
+
|
| 457 |
# Show ALL authors, no truncation
|
| 458 |
authors_str = ", ".join(all_authors)
|
| 459 |
+
|
| 460 |
# Get DOI for reference
|
| 461 |
doi = work.get("doi", "")
|
| 462 |
if doi:
|
| 463 |
doi = doi.replace("https://doi.org/", "")
|
| 464 |
+
|
| 465 |
# Generate BibTeX entry
|
| 466 |
bibtex = self.generate_bibtex(work, all_authors)
|
| 467 |
|
|
|
|
| 478 |
"Relevance": 0,
|
| 479 |
"Matched Keywords": ""
|
| 480 |
}
|
| 481 |
+
|
| 482 |
def generate_bibtex(self, work, authors):
|
| 483 |
"""Generate BibTeX entry for a paper"""
|
| 484 |
+
|
| 485 |
# Create citation key
|
| 486 |
first_author = authors[0].split()[-1] if authors else "Unknown"
|
| 487 |
year = work.get("publication_year", "0000")
|
| 488 |
title_words = work.get("title", "Untitled").split()[:3]
|
| 489 |
title_short = "".join([w[0] for w in title_words if w]) if title_words else "Paper"
|
| 490 |
citation_key = f"{first_author}{year}{title_short}"
|
| 491 |
+
|
| 492 |
# Clean the citation key (remove special characters)
|
| 493 |
citation_key = ''.join(c for c in citation_key if c.isalnum())
|
| 494 |
+
|
| 495 |
# Get DOI or URL
|
| 496 |
doi = work.get("doi", "")
|
| 497 |
url = work.get("primary_location", {}).get("landing_page_url", "")
|
| 498 |
if not url and work.get("open_access", {}).get("oa_url"):
|
| 499 |
url = work.get("open_access", {}).get("oa_url")
|
| 500 |
+
|
| 501 |
# Determine entry type (article, book, etc.)
|
| 502 |
pub_type = "article"
|
| 503 |
if work.get("type") == "book":
|
| 504 |
pub_type = "book"
|
| 505 |
elif work.get("type") == "book-chapter":
|
| 506 |
pub_type = "incollection"
|
| 507 |
+
|
| 508 |
# Build BibTeX string
|
| 509 |
bibtex_lines = [
|
| 510 |
f"@{pub_type}{{{citation_key},",
|
|
|
|
| 512 |
f" author = {{{' and '.join(authors)}}},",
|
| 513 |
f" year = {{{work.get('publication_year', '0000')}}},"
|
| 514 |
]
|
| 515 |
+
|
| 516 |
# Add journal/booktitle
|
| 517 |
source = (work.get("primary_location", {}).get("source") or {})
|
| 518 |
if pub_type == "article":
|
|
|
|
| 521 |
else:
|
| 522 |
booktitle = source.get("display_name", "Unknown Publication")
|
| 523 |
bibtex_lines.append(f" booktitle = {{{booktitle}}},")
|
| 524 |
+
|
| 525 |
# Add volume, issue if available
|
| 526 |
if work.get("biblio", {}).get("volume"):
|
| 527 |
bibtex_lines.append(f" volume = {{{work['biblio']['volume']}}},")
|
| 528 |
if work.get("biblio", {}).get("issue"):
|
| 529 |
bibtex_lines.append(f" number = {{{work['biblio']['issue']}}},")
|
| 530 |
+
|
| 531 |
# Add pages if available
|
| 532 |
if work.get("biblio", {}).get("first_page") and work.get("biblio", {}).get("last_page"):
|
| 533 |
pages = f"{work['biblio']['first_page']}--{work['biblio']['last_page']}"
|
| 534 |
bibtex_lines.append(f" pages = {{{pages}}},")
|
| 535 |
+
|
| 536 |
# Add DOI/URL
|
| 537 |
if doi:
|
| 538 |
bibtex_lines.append(f" doi = {{{doi}}},")
|
| 539 |
elif url:
|
| 540 |
bibtex_lines.append(f" url = {{{url}}},")
|
| 541 |
+
|
| 542 |
# Add abstract if available
|
| 543 |
if work.get("abstract_inverted_index"):
|
| 544 |
# Rebuild abstract for BibTeX
|
|
|
|
| 550 |
abstract = " ".join([w[1] for w in words])
|
| 551 |
if abstract:
|
| 552 |
bibtex_lines.append(f" abstract = {{{abstract}}},")
|
| 553 |
+
|
| 554 |
# Close the entry
|
| 555 |
bibtex_lines.append("}")
|
| 556 |
+
|
| 557 |
return "\n".join(bibtex_lines)
|
| 558 |
|
| 559 |
|
|
|
|
| 585 |
with gr.Column(elem_classes="white-card"):
|
| 586 |
|
| 587 |
query = gr.Textbox(
|
| 588 |
+
label="Search Topic",
|
| 589 |
value="photoacoustic imaging",
|
| 590 |
placeholder="Enter research topic or keywords..."
|
| 591 |
)
|
|
|
|
| 597 |
value="blood, photoacoustics, imaging",
|
| 598 |
placeholder="e.g., blood, photoacoustics, imaging (comma-separated)"
|
| 599 |
)
|
| 600 |
+
|
| 601 |
with gr.Column():
|
| 602 |
authors = gr.Textbox(
|
| 603 |
label="Filter by Authors (optional)",
|
|
|
|
| 609 |
# MAX PAPERS SLIDER WITH BLACK REGION COMMENT
|
| 610 |
# ==========================
|
| 611 |
max_papers = gr.Slider(1000, 50000, value=1000, step=100, label="Max Papers")
|
| 612 |
+
|
| 613 |
# Add black region comment below slider
|
| 614 |
gr.HTML("""
|
| 615 |
<div class="slider-comment">
|
|
|
|
| 648 |
- **Exact Title** (e.g., "Photoacoustic imaging in biomedicine")
|
| 649 |
- **Author** (find papers by a specific researcher)
|
| 650 |
""")
|
| 651 |
+
|
| 652 |
with gr.Row():
|
| 653 |
specific_search_type = gr.Dropdown(
|
| 654 |
label="Search by",
|
| 655 |
choices=["DOI", "Exact Title", "Specific Author"],
|
| 656 |
value="DOI"
|
| 657 |
)
|
| 658 |
+
|
| 659 |
specific_search_value = gr.Textbox(
|
| 660 |
label="Enter value",
|
| 661 |
placeholder="e.g., 10.1038/nature12373 or exact paper title or author name",
|
| 662 |
scale=2
|
| 663 |
)
|
| 664 |
+
|
| 665 |
specific_search_btn = gr.Button("🔍 FIND PAPER", variant="primary")
|
| 666 |
specific_summary = gr.Markdown("Ready")
|
| 667 |
+
|
| 668 |
# Metrics section (common for both tabs)
|
| 669 |
with gr.Row():
|
| 670 |
total_display = gr.Markdown("**0**", elem_classes="metric-value")
|
|
|
|
| 696 |
|
| 697 |
with gr.Row():
|
| 698 |
filename = gr.Textbox(
|
| 699 |
+
label="Filename",
|
| 700 |
value="papers_results",
|
| 701 |
placeholder="Enter base filename without extension"
|
| 702 |
)
|
|
|
|
| 727 |
df, papers, summary = searcher.search_specific_paper(search_type, search_value)
|
| 728 |
if not papers:
|
| 729 |
return df, papers, summary, "**0**", "**0.0**", "**N/A**"
|
| 730 |
+
|
| 731 |
# Calculate metrics for display
|
| 732 |
total = len(papers)
|
| 733 |
avg_rel = 0
|
| 734 |
years = [p["Year"] for p in papers if p["Year"] != "N/A"]
|
| 735 |
year_range = f"{min(years)} - {max(years)}" if years else "N/A"
|
| 736 |
+
|
| 737 |
return df, papers, summary, f"**{total}**", f"**{avg_rel:.2f}**", f"**{year_range}**"
|
| 738 |
except Exception as e:
|
| 739 |
empty_df = pd.DataFrame(columns=['Title', 'Authors', 'Year', 'Citations', 'Journal', 'Abstract', 'BibTeX'])
|
|
|
|
| 745 |
inputs=[query, keywords, authors, max_papers, year_from, year_to, open_access, sort_by, sort_order],
|
| 746 |
outputs=[results, state, total_display, avg_display, year_display]
|
| 747 |
)
|
| 748 |
+
|
| 749 |
specific_search_btn.click(
|
| 750 |
specific_search_wrapper,
|
| 751 |
inputs=[specific_search_type, specific_search_value],
|
|
|
|
| 759 |
|
| 760 |
df = pd.DataFrame(papers)
|
| 761 |
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
|
| 762 |
+
|
| 763 |
# Remove All Authors column from export if it exists
|
| 764 |
if 'All Authors' in df.columns:
|
| 765 |
df = df.drop(columns=['All Authors'])
|
|
|
|
| 787 |
f.write(f"Generated: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n")
|
| 788 |
f.write(f"Total Papers: {len(df)}\n")
|
| 789 |
f.write("="*80 + "\n\n")
|
| 790 |
+
|
| 791 |
for i, row in df.iterrows():
|
| 792 |
f.write(f"PAPER #{i+1}\n")
|
| 793 |
f.write("-"*40 + "\n")
|
|
|
|
| 816 |
if __name__ == "__main__":
|
| 817 |
demo = create_interface()
|
| 818 |
demo.launch(
|
| 819 |
+
inbrowser=True,
|
| 820 |
height=1000,
|
| 821 |
server_name="0.0.0.0", # Makes it accessible on network
|
| 822 |
share=False # Set to True for temporary public link
|