refactor: move finding files into separate class

salted/__main__.py

@@ -23,6 +23,7 @@
 from salted import database_io
 from salted import doi_check
 from salted import err
+from salted import file_finder
 from salted import input_handler
 from salted import memory_instance
 from salted import url_check
@@ -154,11 +155,12 @@ def check(self,
             logging.exception(msg)
             raise FileNotFoundError(msg)
 
+        filesearch = file_finder.FileFinder()
         file_io = input_handler.InputHandler(db)
         files_to_check = list()
         if path.is_dir():
             logging.info('Base folder: %s', path)
-            files_to_check = file_io.find_files_by_extensions(path)
+            files_to_check = filesearch.find_files_by_extensions(path)
             if files_to_check:
                 file_io.scan_files(files_to_check)
                 mem_instance.generate_indices()
@@ -168,7 +170,7 @@ def check(self,
                 logging.warning(
                     "No supported files in this folder or its subfolders.")
                 return
-        elif path.is_file() and file_io.is_supported_format(path):
+        elif path.is_file() and filesearch.is_supported_format(path):
             files_to_check.append(path)
         else:
             msg = f"File format of {path} not supported"

salted/file_finder.py

@@ -0,0 +1,72 @@
+#!/usr/bin/python3
+# -*- coding: utf-8 -*-
+
+"""
+Find Files for salted
+~~~~~~~~~~~~~~~~~~~~~
+Source: https://github.com/RuedigerVoigt/salted
+(c) 2020-2021: Rüdiger Voigt
+Released under the Apache License 2.0
+"""
+
+import logging
+import pathlib
+from typing import Final, List, Optional
+
+
+class FileFinder:
+    "Methods to find files in supported formats"
+
+    SUPPORTED_SUFFIX: Final[set] = {".htm", ".html", '.md', '.tex', '.bib'}
+
+    def __init__(self) -> None:
+        return
+
+    def is_supported_format(self,
+                            filepath: pathlib.Path) -> bool:
+        "Checks - using the filename suffix - if the file format is supported."
+        return bool(filepath.suffix in self.SUPPORTED_SUFFIX)
+
+    def find_files_by_extensions(
+            self,
+            path_to_base_folder: pathlib.Path,
+            suffixes: Optional[set] = None) -> List[pathlib.Path]:
+        """Find all files with specific file type suffixes in the base folder
+           and its subfolders. If no file suffix is specified, this will look
+           for all file formats supported by salted."""
+        # self undefined at time of definition. Therefore fallback here:
+        if not suffixes:
+            suffixes = self.SUPPORTED_SUFFIX
+
+        files_to_check = []
+        path_to_check = pathlib.Path(path_to_base_folder)
+        all_files = path_to_check.glob('**/*')
+        for candidate in all_files:
+            if candidate.suffix in suffixes:
+                files_to_check.append(candidate.resolve())
+        logging.debug('Found %s files', len(files_to_check))
+        return files_to_check
+
+    def find_html_files(self,
+                        path_to_base_folder: pathlib.Path
+                        ) -> List[pathlib.Path]:
+        "Find all HTML files in the base folder and its subfolders."
+        return self.find_files_by_extensions(
+            path_to_base_folder,
+            {".htm", ".html"})
+
+    def find_markdown_files(self,
+                            path_to_base_folder: pathlib.Path
+                            ) -> List[pathlib.Path]:
+        "Find all markdown files in the base folder and its subfolders."
+        return self.find_files_by_extensions(
+            path_to_base_folder,
+            {".md"})
+
+    def find_tex_files(self,
+                       path_to_base_folder: pathlib.Path
+                       ) -> List[pathlib.Path]:
+        "Find all .tex files in the base folder and its subfolders."
+        return self.find_files_by_extensions(
+            path_to_base_folder,
+            {".tex"})

salted/input_handler.py

@@ -12,7 +12,7 @@
 from collections import Counter
 import logging
 import pathlib
-from typing import Final, List, Optional
+from typing import List, Optional
 import urllib.parse
 
 import userprovided
@@ -23,65 +23,14 @@
 
 
 class InputHandler:
-    """Methods to find files and the hyperlinks inside them."""
-
-    SUPPORTED_SUFFIX: Final[set] = {".htm", ".html", '.md', '.tex', '.bib'}
+    """read files and extract the hyperlinks inside them."""
 
     def __init__(self,
                  db: database_io.DatabaseIO):
         self.db = db
         self.cnt: Counter = Counter()
         self.parser = parser.Parser()
 
-    def is_supported_format(self,
-                            filepath: pathlib.Path) -> bool:
-        "Checks - using the filename suffix - if the file format is supported."
-        return bool(filepath.suffix in self.SUPPORTED_SUFFIX)
-
-    def find_files_by_extensions(
-            self,
-            path_to_base_folder: pathlib.Path,
-            suffixes: Optional[set] = None) -> List[pathlib.Path]:
-        """Find all files with specific file type suffixes in the base folder
-           and its subfolders. If no file suffix is specified, this will look
-           for all file formats supported by salted."""
-        # self undefined at time of definition. Therefore fallback here:
-        if not suffixes:
-            suffixes = self.SUPPORTED_SUFFIX
-
-        files_to_check = []
-        path_to_check = pathlib.Path(path_to_base_folder)
-        all_files = path_to_check.glob('**/*')
-        for candidate in all_files:
-            if candidate.suffix in suffixes:
-                files_to_check.append(candidate.resolve())
-        logging.debug('Found %s files', len(files_to_check))
-        return files_to_check
-
-    def find_html_files(self,
-                        path_to_base_folder: pathlib.Path
-                        ) -> List[pathlib.Path]:
-        "Find all HTML files in the base folder and its subfolders."
-        return self.find_files_by_extensions(
-            path_to_base_folder,
-            {".htm", ".html"})
-
-    def find_markdown_files(self,
-                            path_to_base_folder: pathlib.Path
-                            ) -> List[pathlib.Path]:
-        "Find all markdown files in the base folder and its subfolders."
-        return self.find_files_by_extensions(
-            path_to_base_folder,
-            {".md"})
-
-    def find_tex_files(self,
-                       path_to_base_folder: pathlib.Path
-                       ) -> List[pathlib.Path]:
-        "Find all .tex files in the base folder and its subfolders."
-        return self.find_files_by_extensions(
-            path_to_base_folder,
-            {".tex"})
-
     def read_file_content(self,
                           path_to_file: pathlib.Path) -> Optional[str]:
         "Return the file content or log an error if file cannot be accessed."

tests.py

@@ -214,14 +214,14 @@ def test_file_discovery(fs):
     fs.create_file('/fake/fake/noextension')
     fs.create_file('/fake/fake/foo.htmlandmore')
     fs.create_file('/fake/fake/fake/foo.bib')
-    test_io = salted.input_handler.InputHandler(None)
-    supported_files = test_io.find_files_by_extensions('/fake')
+    filesearch = salted.file_finder.FileFinder()
+    supported_files = filesearch.find_files_by_extensions('/fake')
     assert len(supported_files) == 7
-    html_files = test_io.find_html_files('/fake')
+    html_files = filesearch.find_html_files('/fake')
     assert len(html_files) == 2
-    md_files = test_io.find_markdown_files('/fake')
+    md_files = filesearch.find_markdown_files('/fake')
     assert len(md_files) == 2
-    tex_files = test_io.find_tex_files('/fake')
+    tex_files = filesearch.find_tex_files('/fake')
     assert len(tex_files) == 2