GitHub Desktop completely botched things. Should be fixed.

Author: ptth222

src/academic_tracker/__init__.py

@@ -39,8 +39,4 @@
     This module contains general functions for interfacing with the internet.
 
 """
-<<<<<<< Updated upstream:academic_tracker/__init__.py
-__version__ = "1.0.0"
-=======
 __version__ = "1.0.4"
->>>>>>> Stashed changes:src/academic_tracker/__init__.py

src/academic_tracker/__main__.py

@@ -1,4 +1,4 @@
-"""   
+"""
 Usage:
     academic_tracker author_search <config_json_file> [--test --prev_pub=<file-path> --no_GoogleScholar --no_ORCID --no_Crossref --no_PubMed --verbose --silent]
     academic_tracker reference_search <config_json_file> <references_file_or_URL> [--test --prev_pub=<file-path> --PMID_reference --MEDLINE_reference --no_Crossref --no_PubMed --verbose --silent]
@@ -11,7 +11,7 @@
     
 Options:
     -h --help                         Show this screen.
-    --version                         Show version.
+    -v --version                      Show version.
     --verbose                         Print hidden error messages.
     --silent                          Do not print anything to the screen.
     --test                            Generate pubs and email texts, but do not send emails.
@@ -29,18 +29,6 @@
 """
 
 
-## Need help with pycharm UML diagram.
-## add defulat pandas excel engine to required packages xlsxwriter I think
-
-## take docs/_build  out of gitignore and point GitHub pages there for documentation.
-## Add the url of this page to the about section on gitHub.
-## When to merge to master?
-## When to create github pages?
-## Is BSD license okay?
-## Putting on pypi? When?
-## Does the lab have github actions for sphinx documentation? 
-
-
 import re
 import os
 import datetime
@@ -140,11 +128,7 @@ def author_search(config_json_filepath, no_ORCID, no_GoogleScholar, no_Crossref,
         user_input_checking.prev_pubs_file_check(prev_pubs)
 
     ## Query sources and build publication_dict.
-<<<<<<< Updated upstream:academic_tracker/__main__.py
-    publication_dict, prev_pubs = athr_srch_modularized.build_publication_dict(config_dict, prev_pubs, no_ORCID, no_GoogleScholar, no_Crossref)            
-=======
     publication_dict = athr_srch_modularized.build_publication_dict(config_dict, prev_pubs, no_ORCID, no_GoogleScholar, no_Crossref, no_PubMed)            
->>>>>>> Stashed changes:src/academic_tracker/__main__.py
 
     save_dir_name = athr_srch_modularized.save_and_send_reports_and_emails(authors_by_project_dict, publication_dict, config_dict, test)
 
@@ -197,12 +181,8 @@ def PMID_reference(config_json_filepath, ref_path_or_URL, test):
     config_dict = fileio.load_json(config_json_filepath)
 
     ## Get inputs from config file and check them for errors.
-<<<<<<< Updated upstream:academic_tracker/__main__.py
-    user_input_checking.ref_config_file_check(config_dict, True)
-=======
     user_input_checking.ref_config_file_check(config_dict, True, False)
     user_input_checking.config_report_check(config_dict)
->>>>>>> Stashed changes:src/academic_tracker/__main__.py
 
     ## Check the DOI file extension and call the correct read in function.
     extension = os.path.splitext(ref_path_or_URL)[1][1:].lower()
@@ -220,10 +200,8 @@ def PMID_reference(config_json_filepath, ref_path_or_URL, test):
         helper_functions.vprint("Nothing was read from the PMID file. Make sure the file is not empty or is a supported file type.")
         sys.exit()
 
-    if type(file_contents) == str:
-        PMID_list = [line for line in file_contents.split("\n") if line]
-    else:
-        PMID_list = file_contents
+    PMID_list = [line for line in file_contents.split("\n") if line] if type(file_contents) == str else file_contents
+    
     user_input_checking.tracker_validate(PMID_list, tracker_schema.PMID_reference_schema)
 
     helper_functions.vprint("Querying PubMed and building the publication list.")
@@ -232,10 +210,9 @@ def PMID_reference(config_json_filepath, ref_path_or_URL, test):
     ## Build the save directory name.
     if test:
         save_dir_name = "tracker-test-" + re.sub(r"\-| |\:", "", str(datetime.datetime.now())[2:16])
-        os.mkdir(save_dir_name)
     else:
         save_dir_name = "tracker-" + re.sub(r"\-| |\:", "", str(datetime.datetime.now())[2:16])
-        os.mkdir(save_dir_name)
+    os.mkdir(save_dir_name)
 
     ## combine previous and new publications lists and save
     fileio.save_publications_to_file(save_dir_name, publication_dict, {})
@@ -396,6 +373,7 @@ def gen_reports_and_emails_auth(config_json_filepath, publication_json_filepath,
 
     ## Get inputs from config file and check them for errors.
     user_input_checking.tracker_validate(config_dict, tracker_schema.gen_reports_auth_schema)
+    user_input_checking.config_report_check(config_dict)
 
     ## Create an authors_json for each project in the config_dict and update those authors attributes with the project attributes.
     authors_by_project_dict, config_dict = athr_srch_modularized.generate_internal_data_and_check_authors(config_dict)
@@ -429,6 +407,7 @@ def gen_reports_and_emails_ref(config_json_filepath, ref_path_or_URL, publicatio
 
     ## Get inputs from config file and check them for errors.
     user_input_checking.tracker_validate(config_dict, tracker_schema.gen_reports_ref_schema)
+    user_input_checking.config_report_check(config_dict)
 
     if not prev_pub_filepath or prev_pub_filepath.lower() == "ignore":
         prev_pubs = {}

src/academic_tracker/athr_srch_modularized.py

@@ -10,6 +10,8 @@
 import datetime
 import os
 
+import deepdiff
+
 from . import user_input_checking
 from . import fileio
 from . import helper_functions
@@ -35,10 +37,6 @@ def input_reading_and_checking(config_json_filepath, no_ORCID, no_GoogleScholar,
     ## read in config file
     config_dict = fileio.load_json(config_json_filepath)
 
-<<<<<<< Updated upstream:academic_tracker/athr_srch_modularized.py
-    ## Get inputs from config file and check them for errors.
-    user_input_checking.config_file_check(config_dict, no_ORCID, no_GoogleScholar, no_Crossref)
-=======
     if not "ORCID_search" in config_dict:
         no_ORCID = True
 
@@ -51,7 +49,6 @@ def input_reading_and_checking(config_json_filepath, no_ORCID, no_GoogleScholar,
     ## Get inputs from config file and check them for errors.
     user_input_checking.config_file_check(config_dict, no_ORCID, no_GoogleScholar, no_Crossref, no_PubMed)
     user_input_checking.config_report_check(config_dict)
->>>>>>> Stashed changes:src/academic_tracker/athr_srch_modularized.py
 
     return config_dict
 
@@ -107,29 +104,23 @@ def build_publication_dict(config_dict, prev_pubs, no_ORCID, no_GoogleScholar, n
 
     ## Get publications from PubMed 
     helper_functions.vprint("Finding author's publications. This could take a while.")
-<<<<<<< Updated upstream:academic_tracker/athr_srch_modularized.py
-    helper_functions.vprint("Searching PubMed.")
-    PubMed_publication_dict = athr_srch_webio.search_PubMed_for_pubs(prev_pubs, config_dict["Authors"], config_dict["PubMed_search"]["PubMed_email"])
-    prev_pubs.update(PubMed_publication_dict)
-=======
     current_pubs = {}
     if not no_PubMed:
         helper_functions.vprint("Searching PubMed.")
         PubMed_publication_dict = athr_srch_webio.search_PubMed_for_pubs(current_pubs, config_dict["Authors"], config_dict["PubMed_search"]["PubMed_email"])
         current_pubs.update(PubMed_publication_dict)
->>>>>>> Stashed changes:src/academic_tracker/athr_srch_modularized.py
     if not no_ORCID:
         helper_functions.vprint("Searching ORCID.")
-        ORCID_publication_dict = athr_srch_webio.search_ORCID_for_pubs(prev_pubs, config_dict["ORCID_search"]["ORCID_key"], config_dict["ORCID_search"]["ORCID_secret"], config_dict["Authors"])
-        prev_pubs.update(ORCID_publication_dict)
+        ORCID_publication_dict = athr_srch_webio.search_ORCID_for_pubs(current_pubs, config_dict["ORCID_search"]["ORCID_key"], config_dict["ORCID_search"]["ORCID_secret"], config_dict["Authors"])
+        current_pubs.update(ORCID_publication_dict)
     if not no_GoogleScholar:
         helper_functions.vprint("Searching Google Scholar.")
-        Google_Scholar_publication_dict = athr_srch_webio.search_Google_Scholar_for_pubs(prev_pubs, config_dict["Authors"], config_dict["Crossref_search"]["mailto_email"])
-        prev_pubs.update(Google_Scholar_publication_dict)
+        Google_Scholar_publication_dict = athr_srch_webio.search_Google_Scholar_for_pubs(current_pubs, config_dict["Authors"], config_dict["Crossref_search"]["mailto_email"])
+        current_pubs.update(Google_Scholar_publication_dict)
     if not no_Crossref:
         helper_functions.vprint("Searching Crossref.")
-        Crossref_publication_dict = athr_srch_webio.search_Crossref_for_pubs(prev_pubs, config_dict["Authors"], config_dict["Crossref_search"]["mailto_email"])
-        prev_pubs.update(Crossref_publication_dict)
+        Crossref_publication_dict = athr_srch_webio.search_Crossref_for_pubs(current_pubs, config_dict["Authors"], config_dict["Crossref_search"]["mailto_email"])
+        current_pubs.update(Crossref_publication_dict)
 
     publication_dict = {}
     if not no_PubMed:
@@ -148,13 +139,18 @@ def build_publication_dict(config_dict, prev_pubs, no_ORCID, no_GoogleScholar, n
         for key, value in Crossref_publication_dict.items():
             if not key in publication_dict:
                 publication_dict[key] = value
+    
+    ## Compare current pubs with previous and only keep those that are new or updated.
+    for pub_id, pub_values in prev_pubs.items():
+        if pub_id in publication_dict and not deepdiff.DeepDiff(publication_dict[pub_id], pub_values, ignore_order=True, report_repetition=True):
+            del publication_dict[pub_id]
 
 
     if len(publication_dict) == 0:
         helper_functions.vprint("No new publications found.")
         sys.exit()
 
-    return publication_dict, prev_pubs
+    return publication_dict
 
 
 
@@ -174,10 +170,9 @@ def save_and_send_reports_and_emails(authors_by_project_dict, publication_dict,
     ## Build the save directory name.
     if test:
         save_dir_name = "tracker-test-" + re.sub(r"\-| |\:", "", str(datetime.datetime.now())[2:16])
-        os.mkdir(save_dir_name)
     else:
         save_dir_name = "tracker-" + re.sub(r"\-| |\:", "", str(datetime.datetime.now())[2:16])
-        os.mkdir(save_dir_name)
+    os.mkdir(save_dir_name)
 
 
     email_messages = athr_srch_emails_and_reports.create_project_reports_and_emails(authors_by_project_dict, publication_dict, config_dict, save_dir_name)
@@ -222,9 +217,3 @@ def save_and_send_reports_and_emails(authors_by_project_dict, publication_dict,
     return save_dir_name
 
 
-
-
-
-
-
-

src/academic_tracker/ref_srch_modularized.py

@@ -40,10 +40,6 @@ def input_reading_and_checking(config_json_filepath, ref_path_or_URL, MEDLINE_re
     ## read in config file
     config_dict = fileio.load_json(config_json_filepath)
 
-<<<<<<< Updated upstream:academic_tracker/ref_srch_modularized.py
-    ## Get inputs from config file and check them for errors.
-    user_input_checking.ref_config_file_check(config_dict, no_Crossref)
-=======
     if not "Crossref_search" in config_dict:
         no_Crossref = True
 
@@ -53,7 +49,6 @@ def input_reading_and_checking(config_json_filepath, ref_path_or_URL, MEDLINE_re
     ## Get inputs from config file and check them for errors.
     user_input_checking.ref_config_file_check(config_dict, no_Crossref, no_PubMed)
     user_input_checking.config_report_check(config_dict)
->>>>>>> Stashed changes:src/academic_tracker/ref_srch_modularized.py
 
     if not prev_pub_filepath or prev_pub_filepath.lower() == "ignore":
         prev_pubs = {}
@@ -150,10 +145,9 @@ def save_and_send_reports_and_emails(config_dict, tokenized_citations, publicati
     ## Build the save directory name.
     if test:
         save_dir_name = "tracker-test-" + re.sub(r"\-| |\:", "", str(datetime.datetime.now())[2:16])
-        os.mkdir(save_dir_name)
     else:
         save_dir_name = "tracker-" + re.sub(r"\-| |\:", "", str(datetime.datetime.now())[2:16])
-        os.mkdir(save_dir_name)
+    os.mkdir(save_dir_name)
 
 
     if "summary_report" in config_dict:
@@ -195,5 +189,3 @@ def save_and_send_reports_and_emails(config_dict, tokenized_citations, publicati
 
 
 
-
-