add utils tests, small refactor

6 files changed, 351 insertions, 36 deletions

diff --git a/tests/conftest.py b/tests/conftest.py
index b44fbf6..e2d6946 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -1,7 +1,8 @@
 import os
 import sys
 import pytest
-from typing import Callable
+from pathlib import Path
+from typing import Any, Callable
 from pytest import MonkeyPatch
 from argparse import ArgumentParser
 from datetime import datetime, timezone
@@ -70,3 +71,48 @@ def get_fmt_time() -> Callable[..., str]:
     def fmt_time(fmt: str) -> str:
         return datetime.now(tz=timezone.utc).strftime(fmt)
     return fmt_time
+
+
+@pytest.fixture
+def simple_dict() -> dict[str, Any]:
+    return {'define': '$PYSSG_HOME/pyssg/site_example/',
+            'title': 'Example site',
+            'path': {
+                'src': '/tmp/pyssg/pyssg/site_example/src',
+                'dst': '/tmp/pyssg/pyssg/site_example/dst',
+                'plt': '/tmp/pyssg/pyssg/site_example/plt',
+                'db': '/tmp/pyssg/pyssg/site_example/.files'},
+            'url': {
+                'main': 'https://example.com',
+                'static': 'https://static.example.com',
+                'default_image': 'images/default.png'},
+            'fmt': {
+                'date': '%a, %b %d, %Y @ %H:%M %Z',
+                'list_date': '%b %d',
+                'list_sep_date': '%B %Y'},
+            'dirs': {
+                '/': {
+                    'cfg': {
+                        'plt': 'page.html',
+                        'tags': False,
+                        'index': False,
+                        'rss': False,
+                        'sitemap': False,
+                        'exclude_dirs': []}}}}
+
+
+@pytest.fixture(scope='function')
+def tmp_dir_structure(tmp_path: Path) -> Path:
+    root: Path = tmp_path/'dir_str'
+    # order matters
+    dirs: list[Path] = [root,
+                        root/'first',
+                        root/'first/f1',
+                        root/'first/f1/f2',
+                        root/'second',
+                        root/'second/s1']
+    for i, d in enumerate(dirs):
+        d.mkdir()
+        for ext in ['txt', 'md', 'html']:
+            (d/f'f{i}.{ext}').write_text('sample')
+    return root
diff --git a/tests/io_files/multiple.yaml b/tests/io_files/multiple.yaml
new file mode 100644
index 0000000..8d99c40
--- /dev/null
+++ b/tests/io_files/multiple.yaml
@@ -0,0 +1,55 @@
+%YAML 1.2
+---
+define: &root "$PYSSG_HOME/pyssg/site_example/"
+
+title: "Example site"
+path:
+  src: !join [*root, "src"]
+  dst: !join [*root, "dst"]
+  plt: !join [*root, "plt"]
+  db: !join [*root, ".files"]
+url:
+  main: "https://example.com"
+  static: "https://static.example.com"
+  default_image: "images/default.png"
+fmt:
+  date: "%a, %b %d, %Y @ %H:%M %Z"
+  list_date: "%b %d"
+  list_sep_date: "%B %Y"
+dirs:
+  /:
+    cfg:
+      plt: "page.html"
+      tags: False
+      index: False
+      rss: False
+      sitemap: False
+      exclude_dirs: []
+...
+---
+define: &root "$PYSSG_HOME/pyssg/site_example/"
+
+title: "Example site"
+path:
+  src: !join [*root, "src"]
+  dst: !join [*root, "dst"]
+  plt: !join [*root, "plt"]
+  db: !join [*root, ".files"]
+url:
+  main: "https://example.com"
+  static: "https://static.example.com"
+  default_image: "images/default.png"
+fmt:
+  date: "%a, %b %d, %Y @ %H:%M %Z"
+  list_date: "%b %d"
+  list_sep_date: "%B %Y"
+dirs:
+  /:
+    cfg:
+      plt: "page.html"
+      tags: False
+      index: False
+      rss: False
+      sitemap: False
+      exclude_dirs: []
+...
\ No newline at end of file
diff --git a/tests/io_files/multiple_one_doc_error.yaml b/tests/io_files/multiple_one_doc_error.yaml
new file mode 100644
index 0000000..86f6546
--- /dev/null
+++ b/tests/io_files/multiple_one_doc_error.yaml
@@ -0,0 +1,48 @@
+%YAML 1.2
+---
+define: &root "$PYSSG_HOME/pyssg/site_example/"
+
+title: "Example site"
+path:
+  src: !join [*root, "src"]
+  dst: !join [*root, "dst"]
+  plt: !join [*root, "plt"]
+  db: !join [*root, ".files"]
+url:
+  main: "https://example.com"
+  static: "https://static.example.com"
+  default_image: "images/default.png"
+fmt:
+  date: "%a, %b %d, %Y @ %H:%M %Z"
+  list_date: "%b %d"
+  list_sep_date: "%B %Y"
+dirs:
+  /:
+    cfg:
+      plt: "page.html"
+      tags: False
+      index: False
+      rss: False
+      sitemap: False
+      exclude_dirs: []
+...
+---
+define: &root "$PYSSG_HOME/pyssg/site_example/"
+
+title: "Example site"
+path:
+  src: !join [*root, "src"]
+  dst: !join [*root, "dst"]
+  plt: !join [*root, "plt"]
+  db: !join [*root, ".files"]
+url:
+  main: "https://example.com"
+  static: "https://static.example.com"
+  default_image: "images/default.png"
+fmt:
+  date: "%a, %b %d, %Y @ %H:%M %Z"
+  list_date: "%b %d"
+  list_sep_date: "%B %Y"
+dirs:
+  # just removing all paths as it will cause an error
+...
\ No newline at end of file
diff --git a/tests/test_configuration.py b/tests/test_configuration.py
index 0db8094..68f7808 100644
--- a/tests/test_configuration.py
+++ b/tests/test_configuration.py
@@ -1,16 +1,16 @@
 import pytest
 from pytest import LogCaptureFixture
 from typing import Any, Callable
-from logging import DEBUG, INFO, WARNING, ERROR
+from logging import ERROR
 from pyssg.configuration import get_static_config, get_parsed_config
 
 
 # this test is a bit sketchy, as the way the datetimes are calculated could vary
 #   by milliseconds or even have a difference in seconds
 def test_static_default(rss_date_fmt: str,
-                               sitemap_date_fmt: str,
-                               get_fmt_time: Callable[..., str],
-                               version: str) -> None:
+                        sitemap_date_fmt: str,
+                        get_fmt_time: Callable[..., str],
+                        version: str) -> None:
     rss_run_date: str = get_fmt_time(rss_date_fmt)
     sitemap_run_date: str = get_fmt_time(sitemap_date_fmt)
     sc_dict: dict[str, Any] = {'fmt': {'rss_date': rss_date_fmt,
@@ -22,35 +22,13 @@ def test_static_default(rss_date_fmt: str,
     assert static_config == sc_dict
 
 
-def test_simple(test_dir: str) -> None:
-    yaml_dict: dict[str, Any] = {'define': '$PYSSG_HOME/pyssg/site_example/',
-                                 'title': 'Example site',
-                                 'path': {
-                                     'src': '/tmp/pyssg/pyssg/site_example/src',
-                                     'dst': '/tmp/pyssg/pyssg/site_example/dst',
-                                     'plt': '/tmp/pyssg/pyssg/site_example/plt',
-                                     'db': '/tmp/pyssg/pyssg/site_example/.files'},
-                                 'url': {
-                                     'main': 'https://example.com',
-                                     'static': 'https://static.example.com',
-                                     'default_image': 'images/default.png'},
-                                 'fmt': {
-                                     'date': '%a, %b %d, %Y @ %H:%M %Z',
-                                     'list_date': '%b %d',
-                                     'list_sep_date': '%B %Y'},
-                                 'dirs': {
-                                     '/': {
-                                         'cfg': {
-                                             'plt': 'page.html',
-                                             'tags': False,
-                                             'index': False,
-                                             'rss': False,
-                                             'sitemap': False,
-                                             'exclude_dirs': []}}}}
-    yaml_path: str = f'{test_dir}/io_files/simple.yaml'
+def test_simple(test_dir: str,
+                simple_yaml: str,
+                simple_dict: dict[str, Any]) -> None:
+    yaml_path: str = f'{test_dir}/io_files/{simple_yaml}'
     yaml: list[dict[str, Any]] = get_parsed_config(yaml_path)
     assert len(yaml) == 1
-    assert yaml[0] == yaml_dict
+    assert yaml[0] == simple_dict
 
 
 def test_simple_mising_key(test_dir: str,
@@ -67,10 +45,10 @@ def test_simple_mising_key(test_dir: str,
 
 
 def test_simple_mising_dirs(test_dir: str,
-                                caplog: LogCaptureFixture) -> None:
+                            caplog: LogCaptureFixture) -> None:
     err: tuple[str, int, str] = ('pyssg.configuration',
                                  ERROR,
-                                 'config doesn\'t have any dirs configs (dirs.*)')
+                                 'config doesn\'t have any dirs (dirs.*)')
     yaml_path: str = f'{test_dir}/io_files/simple_missing_dirs.yaml'
     with pytest.raises(SystemExit) as system_exit:
         get_parsed_config(yaml_path)
@@ -90,3 +68,29 @@ def test_simple_root_dir(test_dir: str,
     assert system_exit.type == SystemExit
     assert system_exit.value.code == 1
     assert caplog.record_tuples[-1] == err
+
+
+# this really just tests that both documents in the yaml file are read,
+#   multiple.yaml is just simple.yaml with the same document twice,
+#   shouldn't be an issue as the yaml package handles this
+def test_multiple(test_dir: str, simple_dict: dict[str, Any]) -> None:
+    yaml_path: str = f'{test_dir}/io_files/multiple.yaml'
+    yaml: list[dict[str, Any]] = get_parsed_config(yaml_path)
+    assert len(yaml) == 2
+    assert yaml[0] == simple_dict
+    assert yaml[1] == simple_dict
+
+
+# also, this just tests that the checks for a well formed config file are
+#   processed for all documents
+def test_multiple_one_doc_error(test_dir: str,
+                                caplog: LogCaptureFixture) -> None:
+    err: tuple[str, int, str] = ('pyssg.configuration',
+                                 ERROR,
+                                 'config doesn\'t have any dirs (dirs.*)')
+    yaml_path: str = f'{test_dir}/io_files/multiple_one_doc_error.yaml'
+    with pytest.raises(SystemExit) as system_exit:
+        get_parsed_config(yaml_path)
+    assert system_exit.type == SystemExit
+    assert system_exit.value.code == 1
+    assert caplog.record_tuples[-1] == err
diff --git a/tests/test_utils.py b/tests/test_utils.py
new file mode 100644
index 0000000..b7c9754
--- /dev/null
+++ b/tests/test_utils.py
@@ -0,0 +1,162 @@
+import pytest
+from pytest import LogCaptureFixture
+from pathlib import Path
+from logging import INFO
+from pyssg.utils import (get_expanded_path, get_checksum, copy_file, create_dir,
+                         get_dir_structure, get_file_list)
+
+
+# $PYSSG_HOME is the only env var set
+#   in the project settings that resemble a path
+@pytest.mark.parametrize('path, expected_expanded', [
+    ('$PYSSG_HOME', '/tmp/pyssg'),
+    ('$PYSSG_HOME/', '/tmp/pyssg'),
+    ('/test$PYSSG_HOME/', '/test/tmp/pyssg'),
+    ('/test/$PYSSG_HOME/', '/test/tmp/pyssg'),
+    ('/test/$PYSSG_HOME/test', '/test/tmp/pyssg/test')
+])
+def test_path_expansion(path: str, expected_expanded: str) -> None:
+    expanded: str = get_expanded_path(path)
+    assert expanded == expected_expanded
+
+
+@pytest.mark.parametrize('path', [
+    ('$'),
+    ('$NON_EXISTENT_VARIABLE'),
+    ('/path/to/something/$'),
+    ('/path/to/something/$NON_EXISTENT_VARIABLE')
+])
+def test_path_expansion_failure(path: str) -> None:
+    with pytest.raises(SystemExit) as system_exit:
+        get_expanded_path(path)
+    assert system_exit.type == SystemExit
+    assert system_exit.value.code == 1
+
+
+def test_checksum(test_dir: str, simple_yaml: str) -> None:
+    path: str = f'{test_dir}/io_files/{simple_yaml}'
+    simple_yaml_checksum: str = 'd4f0a3ed56fd530d3ea485dced25534c'
+    checksum: str = get_checksum(path)
+    assert checksum == simple_yaml_checksum
+
+
+def test_copy_file(tmp_path: Path, caplog: LogCaptureFixture) -> None:
+    src: Path = tmp_path/'src'
+    dst: Path = tmp_path/'dst'
+    src.mkdir()
+    dst.mkdir()
+    src_file: Path = src/'tmp_file.txt'
+    dst_file: Path = dst/'tmp_file.txt'
+    src_file.write_text('something')
+    inf: tuple[str, int, str] = ('pyssg.utils',
+                                 INFO,
+                                 f'copied file "{src_file}" to "{dst_file}"')
+    copy_file(str(src_file), str(dst_file))
+    assert caplog.record_tuples[-1] == inf
+
+
+def test_copy_file_failure(tmp_path: Path, caplog: LogCaptureFixture) -> None:
+    src: Path = tmp_path/'src'
+    dst: Path = tmp_path/'dst'
+    src.mkdir()
+    dst.mkdir()
+    src_file: Path = src/'tmp_file.txt'
+    dst_file: Path = dst/'tmp_file.txt'
+    src_file.write_text('something')
+    dst_file.write_text('something')
+    inf: tuple[str, int, str] = ('pyssg.utils',
+                                 INFO,
+                                 f'file "{dst_file}" already exists, ignoring')
+    copy_file(str(src_file), str(dst_file))
+    assert caplog.record_tuples[-1] == inf
+
+
+def test_create_dir(tmp_path: Path, caplog: LogCaptureFixture) -> None:
+    path: Path = tmp_path/'new_dir'
+    inf: tuple[str, int, str] = ('pyssg.utils',
+                                 INFO,
+                                 f'created directory "{path}"')
+    assert path.exists() is False
+    create_dir(str(path), False, False)
+    assert path.exists() is True
+    assert caplog.record_tuples[-1] == inf
+
+
+def test_create_dir_failure(tmp_path: Path, caplog: LogCaptureFixture) -> None:
+    path: Path = tmp_path/'new_dir'
+    inf: tuple[str, int, str] = ('pyssg.utils',
+                                 INFO,
+                                 f'directory "{path}" exists, ignoring')
+    path.mkdir()
+    create_dir(str(path), False, False)
+    assert caplog.record_tuples[-1] == inf
+
+
+def test_create_dirs(tmp_path: Path, caplog: LogCaptureFixture) -> None:
+    path: Path = tmp_path/'new_dir'
+    sub_path: Path = path/'sub_dir'
+    inf: tuple[str, int, str] = ('pyssg.utils',
+                                 INFO,
+                                 f'created directory "{sub_path}"')
+    assert path.exists() is False
+    assert sub_path.exists() is False
+    create_dir(str(sub_path), True, False)
+    assert path.exists() is True
+    assert sub_path.exists() is True
+    assert caplog.record_tuples[-1] == inf
+
+
+def test_create_dirs_failure(tmp_path: Path, caplog: LogCaptureFixture) -> None:
+    path: Path = tmp_path/'new_dir'
+    sub_path: Path = path/'sub_dir'
+    inf: tuple[str, int, str] = ('pyssg.utils',
+                                 INFO,
+                                 f'directory "{sub_path}" exists, ignoring')
+    path.mkdir()
+    sub_path.mkdir()
+    create_dir(str(sub_path), True, False)
+    assert caplog.record_tuples[-1] == inf
+
+
+@pytest.mark.parametrize('exclude, exp_dir_str', [
+    ([], ['second/s1', 'first/f1/f2']),
+    (['f2'], ['second/s1', 'first/f1']),
+    (['f1'], ['second/s1', 'first']),
+    (['second'], ['first/f1/f2']),
+    (['s1', 'f2'], ['second', 'first/f1']),
+    (['s1', 'f1'], ['second', 'first']),
+    (['s1', 'first'], ['second'])
+])
+def test_dir_structure(tmp_dir_structure: Path,
+                       exclude: list[str],
+                       exp_dir_str: list[str]) -> None:
+    dir_str: list[str] = get_dir_structure(str(tmp_dir_structure), exclude)
+    # order doesn't matter
+    assert sorted(dir_str) == sorted(exp_dir_str)
+
+
+@pytest.mark.parametrize('exts, exclude_dirs, exp_flist', [
+    (('txt',), [], ['f0.txt', 'second/f4.txt',
+                    'second/s1/f5.txt', 'first/f1.txt',
+                    'first/f1/f2.txt', 'first/f1/f2/f3.txt']),
+    (('txt', 'html'), [], ['f0.html', 'f0.txt',
+                           'second/f4.txt', 'second/f4.html',
+                           'second/s1/f5.html', 'second/s1/f5.txt',
+                           'first/f1.html', 'first/f1.txt',
+                           'first/f1/f2.txt', 'first/f1/f2.html',
+                           'first/f1/f2/f3.txt', 'first/f1/f2/f3.html']),
+    (('md',), [], ['f0.md', 'second/f4.md',
+                   'second/s1/f5.md', 'first/f1.md',
+                   'first/f1/f2.md', 'first/f1/f2/f3.md']),
+    (('md',), ['first'], ['f0.md', 'second/f4.md', 'second/s1/f5.md']),
+    (('md',), ['first', 's1'], ['f0.md', 'second/f4.md']),
+    (('md',), ['f2', 's1'], ['f0.md', 'second/f4.md',
+                             'first/f1.md', 'first/f1/f2.md',]),
+])
+def test_file_list(tmp_dir_structure: Path,
+                   exts: tuple[str],
+                   exclude_dirs: list[str],
+                   exp_flist: list[str]) -> None:
+    flist: list[str] = get_file_list(str(tmp_dir_structure), exts, exclude_dirs)
+    # order doesn't matter
+    assert sorted(flist) == sorted(exp_flist)
diff --git a/tests/test_yaml_parser.py b/tests/test_yaml_parser.py
index dd4c6d4..906c7e6 100644
--- a/tests/test_yaml_parser.py
+++ b/tests/test_yaml_parser.py
@@ -5,7 +5,7 @@ from pyssg.yaml_parser import get_parsed_yaml
 #   and test the join functionality
 
 
-def test_yaml_resource_read(simple_yaml:str, test_resource: str) -> None:
+def test_yaml_resource_read(simple_yaml: str, test_resource: str) -> None:
     yaml: list[dict[str, Any]] = get_parsed_yaml(simple_yaml, test_resource)
     assert len(yaml) == 1
 
@@ -16,7 +16,7 @@ def test_yaml_path_read(test_dir: str) -> None:
     assert len(yaml) == 1
 
 
-def test_yaml_join(simple_yaml:str, test_resource: str) -> None:
+def test_yaml_join(simple_yaml: str, test_resource: str) -> None:
     yaml: dict[str, Any] = get_parsed_yaml(simple_yaml, test_resource)[0]
     define_str: str = '$PYSSG_HOME/pyssg/site_example/'
     assert yaml['define'] == define_str