pyeasyphd 0.3.4__py3-none-any.whl → 0.3.5__py3-none-any.whl

pyeasyphd/scripts/__init__.py

@@ -2,11 +2,12 @@ __all__ = [
     "run_article_md_daily_notes",
     "run_article_tex_submit",
     "run_beamer_tex_weekly_reports",
-    "run_search_for_files",
     "run_search_for_screen",
+    "run_search_for_files",
+    "run_compare_after_search",
 ]
 
 from .run_article_md import run_article_md_daily_notes
 from .run_article_tex import run_article_tex_submit
 from .run_beamer_tex import run_beamer_tex_weekly_reports
-from .run_search_keywords import run_search_for_files, run_search_for_screen
+from .run_search_keywords import run_compare_after_search, run_search_for_files, run_search_for_screen

pyeasyphd/scripts/run_search_keywords.py

@@ -1,6 +1,9 @@
 import os
 from typing import Any, Dict, List
 
+from pyadvtools import transform_to_data_list
+from pybibtexer.tools import compare_bibs_with_zotero
+
 from pyeasyphd.tools import Searchkeywords
 
 
@@ -157,3 +160,103 @@ def _execute_searches(
         path_storage = os.path.join(path_spidering_bibs, je)
         path_output = os.path.join(path_main_output, "Search_spidering_bib", je)
         Searchkeywords(path_storage, path_output, options).run()
+
+
+def run_compare_after_search(
+    zotero_bib: str,
+    keywords_type: str,
+    path_main_output: str,
+    path_conferences_journals_json: str,
+):
+    """
+    Compare search results with Zotero bibliography and generate comparison report.
+
+    Args:
+        zotero_bib: Path to Zotero bibliography file
+        keywords_type: Category name for the search keywords used
+        path_main_output: Main output directory for search results and comparison
+        path_conferences_journals_json: Path to conferences/journals JSON files
+    """
+    # Expand and normalize file paths
+    zotero_bib = _expand_path(zotero_bib)
+    path_main_output = _expand_path(path_main_output)
+    path_conferences_journals_json = _expand_path(path_conferences_journals_json)
+
+    # Configure search options
+    options = _build_search_options(
+        print_on_screen=False,
+        search_year_list=[],
+        include_publisher_list=[],
+        include_abbr_list=[],
+        exclude_publisher_list=["arXiv"],
+        exclude_abbr_list=[],
+        keywords_type=keywords_type,
+        keywords_list_list=[],
+        path_conferences_journals_json=path_conferences_journals_json,
+    )
+
+    # Download bibliography files from local search results
+    download_bib = _download_bib_from_local(path_main_output, keywords_type)
+
+    # Generate comparison output path and run comparison
+    path_output = os.path.join(path_main_output, "comparison_new")
+    compare_bibs_with_zotero(zotero_bib, download_bib, path_output, options)
+
+    return None
+
+
+def _generate_data_list(path_output: str, folder_name: str, keywords_type: str) -> list[str]:
+    """
+    Extract bibliography data content from files in specified folder structure.
+
+    Args:
+        path_output: Base output path for search results
+        folder_name: Specific folder name within the output structure
+        keywords_type: Category name for the search keywords used
+
+    Returns:
+        List of bibliography data content extracted from .bib files in the specified folders
+    """
+    data_list = []
+
+    # Extract data from both title and abstract bibliography folders
+    for bib_type in ["title-bib-zotero", "abstract-bib-zotero"]:
+        folder_path = os.path.join(
+            path_output,
+            f"{folder_name}-Separate",
+            "article",
+            keywords_type,
+            bib_type
+        )
+
+        # Extract bibliography data content if folder exists
+        if os.path.exists(folder_path):
+            data_list.extend(transform_to_data_list(folder_path, ".bib"))
+
+    return data_list
+
+
+def _download_bib_from_local(path_output: str, keywords_type: str) -> list[str]:
+    """
+    Collect bibliography data content from all local search result directories.
+
+    Args:
+        path_output: Base output path containing search results
+        keywords_type: Category name for the search keywords used
+
+    Returns:
+        Combined list of bibliography data content from all .bib files in search results
+    """
+    data_list = []
+
+    # Collect data from spidered bibliographies (Conferences and Journals)
+    for cj in ["Conferences", "Journals"]:
+        folder_name = os.path.join("Search_spidered_bib", cj)
+        data_list.extend(_generate_data_list(path_output, folder_name, keywords_type))
+
+    # Collect data from spidering bibliographies (journal sources)
+    for je in ["spider_j", "spider_j_e"]:
+        folder_name = os.path.join("Search_spidering_bib", je)
+        data_list.extend(_generate_data_list(path_output, folder_name, keywords_type))
+
+    return data_list

{pyeasyphd-0.3.4.dist-info → pyeasyphd-0.3.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pyeasyphd
-Version: 0.3.4
+Version: 0.3.5
 Summary: pyeasyphd
 License: GPL-3.0-or-later
 License-File: LICENSE

{pyeasyphd-0.3.4.dist-info → pyeasyphd-0.3.5.dist-info}/RECORD

@@ -26,11 +26,11 @@ pyeasyphd/main/python_run_tex.py,sha256=9Syu8qRjPXN3gEabfRUWxwTFBm_izIcB4yFhsz3Q
 pyeasyphd/pyeasyphd.py,sha256=OAwbwq2rSXLSk2AoTAF8hmlOMRSRfvDn1Uqk-zkuqH8,3470
 pyeasyphd/pyeasyphd.sublime-settings,sha256=KcXx3DjyVf8UfnB4FP4u-jaTQU3Cuj24OvxGCZvXAsw,3135
 pyeasyphd/pyeasyphd.sublime-syntax,sha256=pXylbA-tye-K5dCTjEJLFVRqtY1T7AgWZ4laxo-dnaE,73
-pyeasyphd/scripts/__init__.py,sha256=R-gQqvYxkiQFZXjzxIWSNheWRolklXb1QY8b2nrKKZk,415
+pyeasyphd/scripts/__init__.py,sha256=0UW0H3Ht37dI1Kn3Pfdov3T7BvRLYDtRS40euB_t4Ls,473
 pyeasyphd/scripts/run_article_md.py,sha256=ZjdO03YRDokSqo-Rffby-1p6_P35N4amERKHe4yS6_0,4127
 pyeasyphd/scripts/run_article_tex.py,sha256=TuTEQk7y-Ykos3a_mlEJzyFnT9RKsAYjGXnwHJa5cjY,4672
 pyeasyphd/scripts/run_beamer_tex.py,sha256=UUOadRfmyWGfK__9NzJIUo1lAiDRB5edxRsfhrS3Ejo,4209
-pyeasyphd/scripts/run_search_keywords.py,sha256=HAbELDQy2buf2pxZvmdAM6GrS8W9dGSsuf8cOB5GyFA,6178
+pyeasyphd/scripts/run_search_keywords.py,sha256=jvBgYYNvOmiQYXCBPEP6dPo0SxvFRHQE2sPIOVvEmHc,9831
 pyeasyphd/tools/__init__.py,sha256=u1MZu_JjVac3HhEmcSTwroS83UVu0W5Vspy3Wu_-GH8,496
 pyeasyphd/tools/generate/generate_from_bibs.py,sha256=Dp1MyADwIRb9qFTFOkMPJLaeeh7NBjuiSLBN7smP2eo,7640
 pyeasyphd/tools/generate/generate_html.py,sha256=JzUEqgTVCaFzd4hXTYUEf0cVSO1QRe0nVUS72W6oyyU,5349
@@ -44,7 +44,7 @@ pyeasyphd/tools/search/search_keywords.py,sha256=YCurXuoYeU1ftve0cb8Hcn_g2FXCXf7
 pyeasyphd/tools/search/search_writers.py,sha256=Dz6D8m17R7x8NT7_PCjwmzlq29AfUz-N6sjyCguDTo4,15702
 pyeasyphd/tools/search/utils.py,sha256=bo7xtIZu31dQvjol1lwyWq1t6ldbw28oondwK8VbAqk,7562
 pyeasyphd/utils/utils.py,sha256=kWxzzgNwz77K9Q7j-RKTaoPpxqiVLVtaBMMhLuEenwE,3128
-pyeasyphd-0.3.4.dist-info/METADATA,sha256=yTHmsfDu8UqmssQFuVtZhpMeEc8PI8azisZEwGHoyDU,985
-pyeasyphd-0.3.4.dist-info/WHEEL,sha256=zp0Cn7JsFoX2ATtOhtaFYIiE2rmFAD4OcMhtUki8W3U,88
-pyeasyphd-0.3.4.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
-pyeasyphd-0.3.4.dist-info/RECORD,,
+pyeasyphd-0.3.5.dist-info/METADATA,sha256=T1k7yh4ZsdlWxiW1XbmPBMQx1VtjfS8gu-Agnl26Y8g,985
+pyeasyphd-0.3.5.dist-info/WHEEL,sha256=zp0Cn7JsFoX2ATtOhtaFYIiE2rmFAD4OcMhtUki8W3U,88
+pyeasyphd-0.3.5.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
+pyeasyphd-0.3.5.dist-info/RECORD,,