Move page number method to path utils

src/corppa/ocr/collate_txt.py

@@ -12,21 +12,7 @@
 
 from tqdm import tqdm
 
-from corppa.utils.path_utils import find_relative_paths
-
-
-def page_number(filename: pathlib.Path) -> str:
-    """Extract and return the page number from a :class:`pathlib.Path`
-    file for content from a single page (e.g., image or text). Returns the
-    page number as a string with leading zeros. (Note: logic is curently
-    specific to Gale/ECCO file naming conventions.)"""
-    # NOTE: this logic is currently specific to Gale/ECCO files,
-    # which look like CW0112029406_00180.txt
-
-    # split the file base/stem name by _ and take the last part
-    pagenum = filename.stem.split("_")[-1]
-    # return the number as a string; strip extra trailing zero
-    return pagenum[:-1]  # strip trailing zero
+from corppa.utils.path_utils import find_relative_paths, page_number
 
 
 def collate_txt(

src/corppa/utils/path_utils.py

@@ -118,6 +118,20 @@ def get_image_relpath(work_id, page_num):
         raise ValueError(f"Unsupported source '{source}'")
 
 
+def page_number(filename: pathlib.Path) -> str:
+    """Extract and return the page number from a :class:`pathlib.Path`
+    file for content from a single page (e.g., image or text). Returns the
+    page number as a string with leading zeros. (Note: logic is curently
+    specific to Gale/ECCO file naming conventions.)"""
+    # NOTE: this logic is currently specific to Gale/ECCO files,
+    # which look like CW0112029406_00180.txt
+
+    # split the file base/stem name by _ and take the last part
+    pagenum = filename.stem.split("_")[-1]
+    # return the number as a string; strip extra trailing zero
+    return pagenum[:-1]  # strip trailing zero
+
+
 def find_relative_paths(
     base_dir, exts, follow_symlinks=True, group_by_dir=False
 ) -> Iterator[pathlib.Path] | Iterator[tuple[pathlib.Path, list]]:

test/test_ocr/test_collate_text.py

@@ -4,11 +4,7 @@
 
 import pytest
 
-from corppa.ocr.collate_txt import collate_txt, main, page_number
-
-
-def test_page_number():
-    assert page_number(pathlib.Path("CW0112029406_00180.txt")) == "0018"
+from corppa.ocr.collate_txt import collate_txt, main
 
 
 def test_collate_txt(tmp_path, capsys):

test/test_utils/test_path_utils.py

@@ -13,6 +13,7 @@
     get_stub_dir,
     get_vol_dir,
     get_volume_id,
+    page_number,
 )
 
 
@@ -105,6 +106,10 @@ def test_get_volume_id():
     assert get_volume_id("coo1.ark:/13960/t4bp0n867-p3") == "coo1.ark:/13960/t4bp0n867"
 
 
+def test_page_number():
+    assert page_number(pathlib.Path("CW0112029406_00180.txt")) == "0018"
+
+
 @patch("corppa.utils.path_utils.get_volume_id", return_value="vol_id")
 @patch("corppa.utils.path_utils.get_vol_dir", return_value=pathlib.Path("vol_dir"))
 @patch("corppa.utils.path_utils.get_ppa_source")