scitex 2.4.2__py3-none-any.whl → 2.4.3__py3-none-any.whl

scitex/scholar/pdf_download/strategies/open_access_download.py

@@ -0,0 +1,186 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+# File: ./src/scitex/scholar/pdf_download/strategies/open_access_download.py
+"""
+Open Access PDF Download Strategy.
+
+Downloads PDFs from known Open Access sources with appropriate handling
+for each source type (arXiv, PubMed Central, OpenAlex OA URLs, etc.).
+"""
+
+from pathlib import Path
+from typing import Optional, Dict, Any
+import aiohttp
+
+from scitex import logging
+
+logger = logging.getLogger(__name__)
+
+
+# Known OA source patterns and their handlers
+OA_SOURCE_PATTERNS = {
+    'arxiv': {
+        'patterns': ['arxiv.org'],
+        'pdf_transform': lambda url: url.replace('/abs/', '/pdf/') + '.pdf' if '/abs/' in url else url,
+    },
+    'pmc': {
+        'patterns': ['ncbi.nlm.nih.gov/pmc', 'europepmc.org'],
+        'pdf_transform': lambda url: url,  # PMC links are usually direct
+    },
+    'biorxiv': {
+        'patterns': ['biorxiv.org', 'medrxiv.org'],
+        'pdf_transform': lambda url: url + '.full.pdf' if not url.endswith('.pdf') else url,
+    },
+    'doaj': {
+        'patterns': ['doaj.org'],
+        'pdf_transform': lambda url: url,
+    },
+    'zenodo': {
+        'patterns': ['zenodo.org'],
+        'pdf_transform': lambda url: url,
+    },
+}
+
+
+def _identify_oa_source(url: str) -> Optional[str]:
+    """Identify which OA source a URL belongs to."""
+    url_lower = url.lower()
+    for source_name, config in OA_SOURCE_PATTERNS.items():
+        for pattern in config['patterns']:
+            if pattern in url_lower:
+                return source_name
+    return None
+
+
+def _transform_to_pdf_url(url: str, source: str) -> str:
+    """Transform URL to direct PDF URL based on source."""
+    if source in OA_SOURCE_PATTERNS:
+        transform_func = OA_SOURCE_PATTERNS[source]['pdf_transform']
+        return transform_func(url)
+    return url
+
+
+async def try_download_open_access_async(
+    oa_url: str,
+    output_path: Path,
+    metadata: Optional[Dict[str, Any]] = None,
+    func_name: str = "try_download_open_access_async",
+    timeout: int = 60,
+) -> Optional[Path]:
+    """
+    Download PDF from an Open Access URL.
+
+    This strategy is simpler than browser-based strategies because OA PDFs
+    are typically directly accessible without authentication.
+
+    Args:
+        oa_url: Open Access URL (from OpenAlex oa_url, arXiv, PMC, etc.)
+        output_path: Path to save the downloaded PDF
+        metadata: Optional paper metadata for logging
+        func_name: Function name for logging
+        timeout: Download timeout in seconds
+
+    Returns:
+        Path to downloaded PDF if successful, None otherwise
+    """
+    if not oa_url:
+        logger.debug(f"{func_name}: No OA URL provided")
+        return None
+
+    # Identify source and transform URL if needed
+    source = _identify_oa_source(oa_url)
+    pdf_url = _transform_to_pdf_url(oa_url, source) if source else oa_url
+
+    logger.info(f"{func_name}: Attempting OA download from {source or 'unknown'}: {pdf_url[:80]}...")
+
+    try:
+        # Create output directory if needed
+        output_path = Path(output_path)
+        output_path.parent.mkdir(parents=True, exist_ok=True)
+
+        # Use aiohttp for async download
+        async with aiohttp.ClientSession() as session:
+            headers = {
+                'User-Agent': 'SciTeX/1.0 (Academic Research Tool; mailto:contact@scitex.io)',
+                'Accept': 'application/pdf,*/*',
+            }
+
+            async with session.get(pdf_url, headers=headers, timeout=aiohttp.ClientTimeout(total=timeout)) as response:
+                if response.status != 200:
+                    logger.warning(f"{func_name}: HTTP {response.status} from {pdf_url}")
+                    return None
+
+                content_type = response.headers.get('Content-Type', '')
+
+                # Verify we're getting a PDF
+                if 'pdf' not in content_type.lower() and not pdf_url.endswith('.pdf'):
+                    # Some servers don't set content-type correctly, check magic bytes
+                    first_bytes = await response.content.read(5)
+                    if first_bytes != b'%PDF-':
+                        logger.warning(f"{func_name}: Response is not a PDF (content-type: {content_type})")
+                        return None
+                    # Reset for full download
+                    content = first_bytes + await response.content.read()
+                else:
+                    content = await response.read()
+
+                # Validate PDF content
+                if len(content) < 1000:  # PDF should be at least 1KB
+                    logger.warning(f"{func_name}: Downloaded content too small ({len(content)} bytes)")
+                    return None
+
+                if not content.startswith(b'%PDF-'):
+                    logger.warning(f"{func_name}: Downloaded content is not a valid PDF")
+                    return None
+
+                # Save to file
+                with open(output_path, 'wb') as f:
+                    f.write(content)
+
+                size_mb = len(content) / 1024 / 1024
+                logger.info(f"{func_name}: Successfully downloaded {size_mb:.2f} MB to {output_path}")
+                return output_path
+
+    except aiohttp.ClientError as e:
+        logger.warning(f"{func_name}: HTTP client error: {e}")
+        return None
+    except TimeoutError:
+        logger.warning(f"{func_name}: Download timed out after {timeout}s")
+        return None
+    except Exception as e:
+        logger.error(f"{func_name}: Download failed: {e}")
+        return None
+
+
+def try_download_open_access_sync(
+    oa_url: str,
+    output_path: Path,
+    metadata: Optional[Dict[str, Any]] = None,
+    timeout: int = 60,
+) -> Optional[Path]:
+    """
+    Synchronous wrapper for try_download_open_access_async.
+
+    Args:
+        oa_url: Open Access URL
+        output_path: Path to save the downloaded PDF
+        metadata: Optional paper metadata
+        timeout: Download timeout in seconds
+
+    Returns:
+        Path to downloaded PDF if successful, None otherwise
+    """
+    import asyncio
+
+    try:
+        loop = asyncio.get_event_loop()
+    except RuntimeError:
+        loop = asyncio.new_event_loop()
+        asyncio.set_event_loop(loop)
+
+    return loop.run_until_complete(
+        try_download_open_access_async(oa_url, output_path, metadata, timeout=timeout)
+    )
+
+
+# EOF

scitex/scholar/pipelines/ScholarPipelineSearchParallel.py

@@ -32,6 +32,7 @@ from datetime import datetime
 
 from scitex import logging
 from scitex.scholar.core import Paper
+from scitex.scholar.core import normalize_journal_name
 from scitex.scholar.search_engines.individual.PubMedSearchEngine import PubMedSearchEngine
 from scitex.scholar.search_engines.individual.CrossRefSearchEngine import CrossRefSearchEngine
 from scitex.scholar.search_engines.individual.ArXivSearchEngine import ArXivSearchEngine
@@ -331,12 +332,18 @@ class ScholarPipelineSearchParallel:
                 if 'metrics' in result:
                     if result['metrics'].get('citation_count'):
                         paper.metadata.citation_count.total = result['metrics']['citation_count']
-                    # Note: is_open_access not in Paper structure
+                    if 'is_open_access' in result['metrics']:
+                        paper.metadata.access.is_open_access = result['metrics']['is_open_access']
+                        paper.metadata.access.is_open_access_engines = [engine_name]
 
                 if 'urls' in result:
                     if result['urls'].get('pdf'):
                         # pdfs is a list of dicts with url/source keys
                         paper.metadata.url.pdfs = [{'url': result['urls']['pdf'], 'source': 'search'}]
+                        # If this is an open access paper, also store the PDF URL as oa_url
+                        if paper.metadata.access.is_open_access:
+                            paper.metadata.access.oa_url = result['urls']['pdf']
+                            paper.metadata.access.oa_url_engines = [engine_name]
                     if result['urls'].get('publisher'):
                         paper.metadata.url.publisher = result['urls']['publisher']
                     if result['urls'].get('doi_url'):
@@ -733,13 +740,21 @@ class ScholarPipelineSearchParallel:
 
         # Publication info
         if hasattr(meta, 'publication'):
-            result['journal'] = meta.publication.journal or ''
+            journal_raw = meta.publication.journal or ''
+            result['journal'] = normalize_journal_name(journal_raw) if journal_raw else ''
             result['impact_factor'] = meta.publication.impact_factor
 
         # Metrics
         if hasattr(meta, 'citation_count'):
             result['citation_count'] = meta.citation_count.total or 0
-        result['is_open_access'] = False  # Not stored in current Paper structure
+
+        # Access metadata
+        if hasattr(meta, 'access'):
+            result['is_open_access'] = meta.access.is_open_access or False
+            result['oa_status'] = meta.access.oa_status
+            result['oa_url'] = meta.access.oa_url
+        else:
+            result['is_open_access'] = False
 
         # URLs
         if hasattr(meta, 'url'):

scitex/scholar/pipelines/ScholarPipelineSearchSingle.py

@@ -268,12 +268,18 @@ class ScholarPipelineSearchSingle:
                 if 'metrics' in result:
                     if result['metrics'].get('citation_count'):
                         paper.metadata.citation_count.total = result['metrics']['citation_count']
-                    # Note: is_open_access not in Paper structure
+                    if 'is_open_access' in result['metrics']:
+                        paper.metadata.access.is_open_access = result['metrics']['is_open_access']
+                        paper.metadata.access.is_open_access_engines = [engine_name]
 
                 if 'urls' in result:
                     if result['urls'].get('pdf'):
                         # pdfs is a list of dicts with url/source keys
                         paper.metadata.url.pdfs = [{'url': result['urls']['pdf'], 'source': 'search'}]
+                        # If this is an open access paper, also store the PDF URL as oa_url
+                        if paper.metadata.access.is_open_access:
+                            paper.metadata.access.oa_url = result['urls']['pdf']
+                            paper.metadata.access.oa_url_engines = [engine_name]
                     if result['urls'].get('publisher'):
                         paper.metadata.url.publisher = result['urls']['publisher']
                     if result['urls'].get('doi_url'):
@@ -461,7 +467,14 @@ class ScholarPipelineSearchSingle:
         # Metrics
         if hasattr(meta, 'citation_count'):
             result['citation_count'] = meta.citation_count.total or 0
-        result['is_open_access'] = False  # Not stored in current Paper structure
+
+        # Access metadata
+        if hasattr(meta, 'access'):
+            result['is_open_access'] = meta.access.is_open_access or False
+            result['oa_status'] = meta.access.oa_status
+            result['oa_url'] = meta.access.oa_url
+        else:
+            result['is_open_access'] = False
 
         # URLs
         if hasattr(meta, 'url'):

{scitex-2.4.2.dist-info → scitex-2.4.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scitex
-Version: 2.4.2
+Version: 2.4.3
 Summary: A comprehensive Python library for scientific computing and data analysis
 Project-URL: Homepage, https://github.com/ywatanabe1989/scitex-code
 Project-URL: Documentation, https://scitex.readthedocs.io

{scitex-2.4.2.dist-info → scitex-2.4.3.dist-info}/RECORD

@@ -1,7 +1,7 @@
 scitex/.mcp.json,sha256=ge1NDrBZkRIxJgx_tCS9QvcEa_P_ay8AAAafraZVRP4,1259
 scitex/__init__.py,sha256=8MkZBGNQrJj9Iy0qKI5aVUYWXyFW8I2HCIdxG8O9WuY,5778
 scitex/__main__.py,sha256=m0B6tnYHtUnkY7tXQab7qlqfNsLTtHqtWOs1Sg-wnhw,488
-scitex/__version__.py,sha256=lH8tHpaUufmR8xJZbWuMiClgRz_ZaOPAKQsQbIyY6Do,400
+scitex/__version__.py,sha256=9jMqN4MJ8401QbqhffA5qUyQAvmSxbq3WRWDLytDssY,400
 scitex/_optional_deps.py,sha256=oHwPdE6uJ4MmWNgt7JXojNtbwkUjR4c0mCCNMOK4k-w,3696
 scitex/errors.py,sha256=sgl1csU2sfHCINxO9jwPCdNA6hr-CqaT8tQMiuH93sQ,15247
 scitex/units.py,sha256=2T6VCwcjHBT_zPr_DG3JhUIF7a7TU1pUxMQPDBKZNCw,10373
@@ -142,7 +142,7 @@ scitex/benchmark/benchmark.py,sha256=5WEubBdW4jqZAqd1AFrlDeKjTg15DvUc7wTjtNgL8Aw
 scitex/benchmark/monitor.py,sha256=ge7eLEl3mtLprdtIT4liWqim9cquMekgQY4xyaVlTRI,11609
 scitex/benchmark/profiler.py,sha256=aC8Cb_AZdtTS2T-HWj3srUk-unGpZ_y4hz-6K387jcw,8633
 scitex/browser/README.md,sha256=xpYYwib632qOyLCeAIBNtwE2o8frDiLXTyacw2fAz6k,4436
-scitex/browser/__init__.py,sha256=IJmGAT7y263VGawwlLpe4gc5aozIqcEAkr9gdnnA0Ws,1024
+scitex/browser/__init__.py,sha256=piiLZLPHPzeW0QrTPx2aiRS7_6TzDptAomDMn1ffs10,2725
 scitex/browser/auth/__init__.py,sha256=mSHD6yUmDsgADN9U4D2Hm5Pi7y4mWYBCsdRnTcB9BHY,910
 scitex/browser/auth/google.py,sha256=a9TTxAhry6fezLv5JAGKFMgDtuRy5a-1HWpsTSO3Qcw,13054
 scitex/browser/automation/CookieHandler.py,sha256=EAWEOQwdneVHCrN-evtrouYBw7wsT0rG472DpeimBVM,6810
@@ -159,10 +159,14 @@ scitex/browser/collaboration/visual_feedback.py,sha256=0UOLczpMV408AU4zcToOKK3J0
 scitex/browser/core/BrowserMixin.py,sha256=P1CNBj0Rx9op6cONfwZMs575A6J2Bsp1TGaO7Dy51n0,10968
 scitex/browser/core/ChromeProfileManager.py,sha256=vWfGQF05dAmmQBcRVmuUV9axoST7RDIczNDPvQq1KUo,14542
 scitex/browser/core/__init__.py,sha256=bl533RumUuDg4-zrzr5X8BbTn_wIMcc4wkuG5A0QN9U,200
-scitex/browser/debugging/__init__.py,sha256=j32aq6AK7UkCtTguDcksnPc-s-Y4jtukoakpacTP_Zw,443
+scitex/browser/debugging/__init__.py,sha256=q7e1tYecO3nJflNHfur2JkLXZazJW4AZT-Dxz3vnc8k,1971
 scitex/browser/debugging/_browser_logger.py,sha256=ZhQQ5ohZHpeWZ6fRBFjPV6zLWnKP5ZiCxW1PIIDW0SU,20406
+scitex/browser/debugging/_failure_capture.py,sha256=qc5xIkkjYDVge-74bRh60X9MARVe9GVisbOlnt9pjWQ,11662
 scitex/browser/debugging/_highlight_element.py,sha256=EXYInkCGwbKM2NWxX1_aWWSiQNFTwu03zneNsBF9U_w,3925
 scitex/browser/debugging/_show_grid.py,sha256=dyBvAZFAeD5-EgldlC8mfcPDYA9J28K4hVY4YFdidUI,4042
+scitex/browser/debugging/_sync_session.py,sha256=GXJsr20rEVY71bbQpmlBFx13Bi0S0frridLFQdZUs8s,8044
+scitex/browser/debugging/_test_monitor.py,sha256=_1yiDXcPRP6iEA8304lagXs38N_2yAM65OKzf7ydkUE,8227
+scitex/browser/debugging/_visual_cursor.py,sha256=e-bC5Zf4tW8GPpFc4l4aGxvHKFGIZvfFHUKAj1enBYs,13855
 scitex/browser/docs/ABOUT_PLAYWRIGHT.md,sha256=3ovUznFqms50F1CuRr-3lNC2Xk86TR_ytItNO2mDNAQ,1418
 scitex/browser/interaction/__init__.py,sha256=wT0bf6VlTuuyyazN3SDrKelIaIlUYjpD1aBM3HOcGqY,586
 scitex/browser/interaction/click_center.py,sha256=sXFHnSuSohYlbOyVWzkviKjnBjtqB_JkBU1ut0O8jvo,3514
@@ -1591,7 +1595,7 @@ scitex/scholar/browser/utils/wait_redirects.py,sha256=5FZgp4qKZuwEA1wyFrvpu8EGNI
 scitex/scholar/citation_graph/README.md,sha256=CGEU_W15eLfu3iV33cQT3jwSHmagA3UmSYI5vAaDW2A,3316
 scitex/scholar/citation_graph/__init__.py,sha256=dmA79jVRsVlghbl2SlXS-i2FS-x1QJknCZZDM6Kidog,831
 scitex/scholar/citation_graph/builder.py,sha256=dAXksvgiwyxnEeW9gYuC_vOgjYzVghMQHUR7MN28wkw,6618
-scitex/scholar/citation_graph/database.py,sha256=M2GLVa8O_2sG9tC974ywuSqv3KAfEHBb9m00vk7W6S8,6863
+scitex/scholar/citation_graph/database.py,sha256=NYoLF7VauK8RiEFt8_0ijp5T9fG-ZMl1Q5Lx6CCC0Z8,7067
 scitex/scholar/citation_graph/example.py,sha256=HtvSBnm41cdYycSzwXmRomL7qxxPIOkKQN6A0nvBhKw,2859
 scitex/scholar/citation_graph/models.py,sha256=qz2KgTq85-7QqzdjzTnk6Vho7bclQemxC6SMWzHgQww,2158
 scitex/scholar/cli/README.md,sha256=Qp9UBx4XHFwIhn61rq4pwXxmVwMNyAPHruSi8DowkiM,10946
@@ -1610,9 +1614,9 @@ scitex/scholar/cli/handlers/doi_handler.py,sha256=aEBa6imcAq9aeQRymBWlNgSUhfokyv
 scitex/scholar/cli/handlers/project_handler.py,sha256=lA9IUuHO3C3ZPhHKEkcMTpFgS9UppMCebOwhN5J5FEc,10120
 scitex/scholar/config/PublisherRules.py,sha256=wumxQXIaiFqG5gdSzd1_7bPytXZu2-WBpHZas4koHts,4570
 scitex/scholar/config/README.md,sha256=318vSWix8XiZCNSjahq6Cswww7ZmHPKB3eYi54XyZfM,880
-scitex/scholar/config/ScholarConfig.py,sha256=KbiwbTPH-3S-1fpNbksog3XRSEVyfmyTZ9JhhhiAZCA,4320
+scitex/scholar/config/ScholarConfig.py,sha256=WsmDzLjJhNXYjjMsDs3yIWyco5s4qKHm-4XEz89DL28,5233
 scitex/scholar/config/__init__.py,sha256=xSQfplE2KxqFHxtAS44V-sNabAnK9JIxM68_9GFVIqI,499
-scitex/scholar/config/default.yaml,sha256=Jgyq08_C6sKk-QeFrqEEJiRi9nsmXVYAUBP1-GfQjLI,17064
+scitex/scholar/config/default.yaml,sha256=DLO0pzCn_xFKr08rqOEFndQh_xs-Qkezt9RAENzvi-k,18693
 scitex/scholar/config/_categories/README.md,sha256=swb9Y-fSDK5IBst3bF96xH_syrRWpv7N8jz7gYQCyeQ,4806
 scitex/scholar/config/_categories/api_keys.yaml,sha256=wo2fAxXqQYU6JLkAf4VxvcIiJPIc1qOjx2LMp2hK-k8,739
 scitex/scholar/config/_categories/auth_gateway.yaml,sha256=resIZWaEyCRg0HIk0yJV194I4_Ur_OIQ2hLDswC-PqM,3448
@@ -1628,11 +1632,14 @@ scitex/scholar/config/_categories/url_finder_openurl.yaml,sha256=qIStUtKD1cII6g1
 scitex/scholar/config/_categories/url_finder_selectors.yaml,sha256=Hn_Amb1egy-2_9UxOqSd_g85zLwy0J8pIoztFGANsCo,3446
 scitex/scholar/config/core/_CascadeConfig.py,sha256=DR28mG-dvb4vTgk2bCTZKdT0oOKX3JSxlcmqmp-zAvQ,4855
 scitex/scholar/config/core/_PathManager.py,sha256=ddH2Ik2jeHGA6DKEaDZmrLO3AJz5bUNmhGojjYriISE,20899
-scitex/scholar/core/Paper.py,sha256=sSQ9J02fw2JN4rrfdqYBxoITgMgEZAA6gJYBayP1LvQ,23044
+scitex/scholar/core/Paper.py,sha256=qhe_HKN6MI0CnY08uWTHReNhBA-pxcw6KMrvLpCeXbc,26806
 scitex/scholar/core/Papers.py,sha256=qrlXQIHuISCSEeKnvaSFQw8VgH6ujP946TH1dcu70mM,25123
 scitex/scholar/core/README.md,sha256=lzkBX327j4yo_R2cdAmLczzYqn9-tbewEWB91fUBrmo,3197
 scitex/scholar/core/Scholar.py,sha256=TVXJP0lpVCLR1NIB1gCi_R8kqPRf5IzLXJ4WsoYz2_k,73367
-scitex/scholar/core/__init__.py,sha256=mbBTPyuv6Rnva7NXLB0-Y9H92FOY7DKXKUH6Hdims-A,138
+scitex/scholar/core/__init__.py,sha256=ZkI7H7n_VXHc_GW_oduRh8OERStv03iLyoM4hfvc4A8,1139
+scitex/scholar/core/journal_normalizer.py,sha256=dGkAA1fao-6h5Ofcm7EIudRY8EruRoZv6pQ2Up0Opbc,17497
+scitex/scholar/core/oa_cache.py,sha256=foPNOgdNkm2olSKDQp7Df_yuTdlJf3PdQdAYFyMrx7g,9826
+scitex/scholar/core/open_access.py,sha256=rVm_T1dvcYClnZtjUqI2dlTRl4RyFgz_kPnofOubAuw,14263
 scitex/scholar/docs/DETAILS_FOR_DEVELOPERS.md,sha256=E-Tv_xjaUlwzC8tGNEgQQCQTYxXhSCkccaCwMigxFAM,12878
 scitex/scholar/docs/IMPACT_FACTOR_CITATION_INTEGRATION.md,sha256=g2UPSSOzkt7V_kr3GqhFcrDWfBOFfTGHy22Slycziks,8986
 scitex/scholar/docs/PIPELINE_IMPLEMENTATION_COMPLETE.md,sha256=jG9MnwMjIbU8LFXrlaWhbTVeVudDpf7rrykivhMbsb8,9056
@@ -1875,21 +1882,22 @@ scitex/scholar/metadata_engines/utils/__init__.py,sha256=itAfU-rDnO51wPH1HvjKIwE
 scitex/scholar/metadata_engines/utils/_metadata2bibtex.py,sha256=3yDA29NYtrsyiDWAc2ARPflvrdHvcSZPE-D_GAPyWi8,3092
 scitex/scholar/metadata_engines/utils/_standardize_metadata.py,sha256=32j_GlqkyRXlM9tIXOojlWrvIgmwoDezW78vSDiVXRw,12319
 scitex/scholar/pdf_download/README.md,sha256=W-Xz3CcSPnZU9qLnUiA7s3nd-yH8SReG15sBttp7Gt0,1429
-scitex/scholar/pdf_download/ScholarPDFDownloader.py,sha256=Isb4FOFbPZLXNYzhejLrtcr-BUa7kL6rraIJF_BGe68,18985
+scitex/scholar/pdf_download/ScholarPDFDownloader.py,sha256=Obam1sw8mqXydHmO_l9iAjNb68W-CbverWu8T6vWrrw,24468
 scitex/scholar/pdf_download/__init__.py,sha256=i5mmHd1wUKmTYRu3yUKxPrnRpWVW1OGDwAK-7r2Kma4,98
-scitex/scholar/pdf_download/strategies/__init__.py,sha256=f5x3V3TJzpwrmYTGYXXZiea_PDsawD-9kRgYa2C26oI,1165
+scitex/scholar/pdf_download/strategies/__init__.py,sha256=7Mi9YWDXy5jd0WyDARvCwRHSA7iyd4zxYeRcVttqyC8,1349
 scitex/scholar/pdf_download/strategies/chrome_pdf_viewer.py,sha256=kq7nuIdrrKvvsPIApQcwVtFm8kyRgHaTYTLasY__a5c,14399
 scitex/scholar/pdf_download/strategies/direct_download.py,sha256=48cqMB4g85BKSoBNXTKdpiyv7Gbx3XZOTTRuzVyzruE,4419
 scitex/scholar/pdf_download/strategies/manual_download_fallback.py,sha256=ep8FnhRr7neihaSILZ_ux1I_oWaTXX9tMxWwkiXtdmA,5286
 scitex/scholar/pdf_download/strategies/manual_download_utils.py,sha256=j5holl_qjIGqN0WTyOVwmyv2sFOZAxcAdTglTDUhnFM,35759
+scitex/scholar/pdf_download/strategies/open_access_download.py,sha256=ISPPMsER66ChTvbjwLF-48idFAWLvam3Vf1UAxPacs4,6319
 scitex/scholar/pdf_download/strategies/response_body.py,sha256=fpk0GGiOLQk2Yll76G5YPE5S0a8hde4NQLk8OIrghck,6928
 scitex/scholar/pipelines/README.md,sha256=nvTlfY7qP084iZo93-qXfic8jjmbR4qqUZOtmuH391U,4478
 scitex/scholar/pipelines/ScholarPipelineBibTeX.py,sha256=OQ_CDAj9x5ZPK8gHdLUlBe9UvYCJXX7IJnp555X69CU,11017
 scitex/scholar/pipelines/ScholarPipelineMetadataParallel.py,sha256=IM97YAXHu05uXh3hsa_TuWwBaf2Trjx-kjsEf1qV5t0,9306
 scitex/scholar/pipelines/ScholarPipelineMetadataSingle.py,sha256=Cjo-A1pTEegmYm-EhkvkUDIdiSSCI_ke3BBw1JhSfI0,15550
 scitex/scholar/pipelines/ScholarPipelineParallel.py,sha256=qmMIaocwP1QIvckQKiPjvnOKzvbO5TnFFiLly5ncBNk,14854
-scitex/scholar/pipelines/ScholarPipelineSearchParallel.py,sha256=3-38SC2iEvY6PnVc5BqDxXpWuHp9sSAxKut7Qm1QErA,30411
-scitex/scholar/pipelines/ScholarPipelineSearchSingle.py,sha256=tibeMD5hf_J3_gqUFE7R8dYYtzy0f2IlkjO-iVhScok,18379
+scitex/scholar/pipelines/ScholarPipelineSearchParallel.py,sha256=In8x3hgzlwIyegD4LOJD-I9NZ0k7_8aZWJZ99cXDjT4,31271
+scitex/scholar/pipelines/ScholarPipelineSearchSingle.py,sha256=Qd5SloT5vhXEiOGCPFeBLwzzIjxobrDpQH5C920SoHs,19099
 scitex/scholar/pipelines/ScholarPipelineSingle.py,sha256=M0iFVCtgDWV5yzd-vUBedmxIdc1hoB1PbCNjzZUj62Q,27295
 scitex/scholar/pipelines/SearchQueryParser.py,sha256=7FGjO4Lj7zp-8o8JJjc8qXudY3-pYIjrQarcYFI_lBQ,10018
 scitex/scholar/pipelines/__init__.py,sha256=2739ioL2pNmtGCLmDr0mVd2M5c-N_9z27IfG8xVyKKY,2079
@@ -2925,8 +2933,8 @@ scitex/writer/utils/__init__.py,sha256=wizvQZbOWHsNnkdDsB8J4-lPInRM3gDdwOCRg1fLI
 scitex/writer/utils/_parse_latex_logs.py,sha256=_KfTzSw2IYhUzdZBJsZreAvQZk_BJHUBe2qIUNJfd-s,3154
 scitex/writer/utils/_parse_script_args.py,sha256=vVMQE-AHCs2Q2uVQDuZVN8N3Eft0sxuPtNmnyPXVgnc,4625
 scitex/writer/utils/_watch.py,sha256=qSBbwbeCPmXEWXn-ozCrar43rp664Wo65JzwIMWx7wE,2575
-scitex-2.4.2.dist-info/METADATA,sha256=81TQLW57ldzvO0fvMM1bapwESLdYisGzh6QtaNaQZ7s,29660
-scitex-2.4.2.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-scitex-2.4.2.dist-info/entry_points.txt,sha256=jmgM0XEEIfCoMvwDSUNwRHBHaX_cfcJWQgi-lFc-BNU,48
-scitex-2.4.2.dist-info/licenses/LICENSE,sha256=TfPDBt3ar0uv_f9cqCDMZ5rIzW3CY8anRRd4PkL6ejs,34522
-scitex-2.4.2.dist-info/RECORD,,
+scitex-2.4.3.dist-info/METADATA,sha256=2Cl9m8PVii8rfVZqZceefYFQeYZHW5XTpLiXtmgWA88,29660
+scitex-2.4.3.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+scitex-2.4.3.dist-info/entry_points.txt,sha256=jmgM0XEEIfCoMvwDSUNwRHBHaX_cfcJWQgi-lFc-BNU,48
+scitex-2.4.3.dist-info/licenses/LICENSE,sha256=TfPDBt3ar0uv_f9cqCDMZ5rIzW3CY8anRRd4PkL6ejs,34522
+scitex-2.4.3.dist-info/RECORD,,