Fix issue 269, deleting pre-existing files error

.gitignore

@@ -145,3 +145,6 @@ scripts/similarity/config.yml
 # Processed or local files
 /Data/Processed/*
 *.local.pdf
+
+# ASDF version manager
+.tool-versions

resume_matcher/main.py

@@ -18,7 +18,9 @@
 
 
 def get_filenames_from_dir(directory):
-    return [f for f in os.listdir(directory) if os.path.isfile(os.path.join(directory, f))]
+    return [
+        f for f in os.listdir(directory) if os.path.isfile(os.path.join(directory, f))
+    ]
 
 
 def process_files(resume, job_description):
@@ -34,6 +36,3 @@ def process_files(resume, job_description):
         print(r.score)
     print(f"Processing resume: {resume}")
     print(f"Processing job description: {job_description}")
-
-
-

run_first.py

@@ -18,8 +18,13 @@ def read_json(filename):
 
 
 def remove_old_files(files_path):
+    try:
+        filenames = os.listdir(files_path)
+    except FileNotFoundError:
+        logging.info(f"Directory {files_path} does not exist. No files to remove")
+        return
 
-    for filename in os.listdir(files_path):
+    for filename in filenames:
         try:
             file_path = os.path.join(files_path, filename)
 

scripts/JobDescriptionProcessor.py

@@ -4,6 +4,7 @@
 
 from .parsers import ParseJobDesc, ParseResume
 from .ReadPdf import read_single_pdf
+from .utils import make_sure_path_exists
 
 READ_JOB_DESCRIPTION_FROM = "Data/JobDescription/"
 SAVE_DIRECTORY = "Data/Processed/JobDescription"
@@ -41,6 +42,7 @@ def _write_json_file(self, resume_dictionary: dict):
             + ".json"
         )
         save_directory_name = pathlib.Path(SAVE_DIRECTORY) / file_name
+        make_sure_path_exists(SAVE_DIRECTORY)
         json_object = json.dumps(resume_dictionary, sort_keys=True, indent=14)
         with open(save_directory_name, "w+") as outfile:
             outfile.write(json_object)

scripts/ResumeProcessor.py

@@ -4,6 +4,7 @@
 
 from .parsers import ParseJobDesc, ParseResume
 from .ReadPdf import read_single_pdf
+from .utils import make_sure_path_exists
 
 READ_RESUME_FROM = "Data/Resumes/"
 SAVE_DIRECTORY = "Data/Processed/Resumes"
@@ -38,6 +39,7 @@ def _write_json_file(self, resume_dictionary: dict):
             "Resume-" + self.input_file + resume_dictionary["unique_id"] + ".json"
         )
         save_directory_name = pathlib.Path(SAVE_DIRECTORY) / file_name
+        make_sure_path_exists(SAVE_DIRECTORY)
         json_object = json.dumps(resume_dictionary, sort_keys=True, indent=14)
         with open(save_directory_name, "w+") as outfile:
             outfile.write(json_object)

scripts/utils/ReadFiles.py

@@ -8,3 +8,7 @@ def get_filenames_from_dir(directory_path: str) -> list:
         if os.path.isfile(os.path.join(directory_path, f)) and f != ".DS_Store"
     ]
     return filenames
+
+
+def make_sure_path_exists(directory_path: str):
+    os.makedirs(directory_path, exist_ok=True)

scripts/utils/__init__.py

@@ -1,3 +1,3 @@
 from .logger import init_logging_config
-from .ReadFiles import get_filenames_from_dir
+from .ReadFiles import get_filenames_from_dir, make_sure_path_exists
 from .Utils import TextCleaner