two urlformat varieties

Author: JasperVanDenBosch

syrinx/config.py

@@ -1,5 +1,5 @@
 from __future__ import annotations
-from typing import Any, Optional, TYPE_CHECKING
+from typing import Optional, TYPE_CHECKING
 from os.path import isfile, join, abspath
 if TYPE_CHECKING:
     from argparse import Namespace
@@ -11,10 +11,11 @@ class SyrinxConfiguration:
     domain: Optional[str]
     environment: str
     verbose: bool
+    urlformat: str
 
     def __str__(self) -> str:
         lines = []
-        for key in ('clean', 'domain', 'environment', 'verbose'):
+        for key in ('clean', 'domain', 'environment', 'verbose', 'urlformat'):
             val = getattr(self, key)
             if isinstance(val, str):
                 val = f'"{val}"'
@@ -30,6 +31,7 @@ def configure(args: Namespace) -> SyrinxConfiguration:
     config.domain = None
     config.verbose = False
     config.environment = 'default'
+    config.urlformat = 'filesystem'
     root_dir = getattr(args, 'root_dir', '.')
     cfg_fpath = join(abspath(root_dir), 'syrinx.cfg')
     if isfile(cfg_fpath):
@@ -49,8 +51,10 @@ def configure(args: Namespace) -> SyrinxConfiguration:
                     config.verbose = val.lower() == 'true'
                 if key == 'environment':
                     config.environment = val
+                if key == 'urlformat':
+                    config.urlformat = val
 
-    for key in ('clean', 'domain', 'verbose', 'environment'):
+    for key in ('clean', 'domain', 'verbose', 'environment', 'urlformat'):
         if hasattr(args, key):
             setattr(config, key, getattr(args, key))
     return config

syrinx/read.py

@@ -21,7 +21,6 @@ class BuildMetaInfo:
 
     def __init__(self, config: SyrinxConfiguration) -> None:
         self.environment = config.environment
-        self.domain = config.domain
         self.timestamp = datetime.now(tz=UTC)
         self.syrinx_version = version('syrinx')
 
@@ -36,12 +35,15 @@ class ContentNode:
     buildPage: bool
     path: str
     meta: BuildMetaInfo
+    config: SyrinxConfiguration
 
-    def __init__(self):
+    def __init__(self, meta: BuildMetaInfo, config: SyrinxConfiguration):
         self.buildPage = False
         self.leaves = []
         self.branches = []
         self.sequenceNumber = SYS_MAX_SIZE
+        self.meta = meta
+        self.config = config
 
     @property
     def title(self) -> str:
@@ -54,8 +56,11 @@ def title(self) -> str:
     def address(self) -> Optional[str]:
         """Full, canonical URL of this node
         """
-        if self.meta.domain is not None:
-            return f'https://{self.meta.domain}{self.path}/'
+        if self.config.domain is not None:
+            if self.config.urlformat == 'clean':
+                return f'https://{self.config.domain}{self.path}'
+            else:
+                return f'https://{self.config.domain}{self.path}/'
 
     @property
     def lastModified(self) -> Optional[str]:
@@ -86,14 +91,13 @@ def read(root_dir: str, config: SyrinxConfiguration) -> ContentNode:
     content_dir = join(root_dir, 'content')
 
     tree: Dict[str, ContentNode] = dict()
-    root = ContentNode()
+    root = ContentNode(meta, config)
     root.name = ''
     root.meta = meta
     for (dirpath, _, fnames) in walk(content_dir):
 
-        indexNode = ContentNode()
+        indexNode = ContentNode(meta, config)
         indexNode.name = basename(dirpath)
-        indexNode.meta = meta
         if dirpath == content_dir:
             indexNode = root
             if 'index.md' not in fnames:
@@ -119,10 +123,9 @@ def read(root_dir: str, config: SyrinxConfiguration) -> ContentNode:
                 node = indexNode
                 node.path = dirpath.replace(content_dir, '')
             else:
-                node = ContentNode()
+                node = ContentNode(meta, config)
                 node.name = name
                 node.path = join(dirpath.replace(content_dir, ''), node.name)
-                node.meta = meta
                 indexNode.leaves.append(node)
 
             node.front = fm_dict

tests/config.py

@@ -25,6 +25,7 @@ def test_defaults(self, _, isfile):
         self.assertFalse(config.verbose)
         self.assertTrue(config.clean)
         self.assertEqual(config.environment, 'default')
+        self.assertEqual(config.urlformat, 'filesystem')
 
     @patch('syrinx.config.isfile')
     @patch('syrinx.config.open', new_callable=mock_open)
@@ -36,13 +37,15 @@ def test_file(self, mocked_open, isfile):
             verbose = true
             environment = "staging"
             clean = false
+            urlformat = "clean"
         """
         args = Namespace()
         config = configure(args)
         self.assertEqual(config.domain, 'some.where.bla')
         self.assertTrue(config.verbose)
         self.assertFalse(config.clean)
         self.assertEqual(config.environment, 'staging')
+        self.assertEqual(config.urlformat, 'clean')
 
     @patch('syrinx.config.isfile')
     @patch('syrinx.config.open', new_callable=mock_open)
@@ -54,17 +57,20 @@ def test_cli(self, mocked_open, isfile):
             domain = "some.where.bla"
             verbose = true
             environment = "staging"
+            urlformat = "clean"
         """
         args = Namespace()
         args.domain = 'not.there.bla'
         args.verbose = False
         args.environment = 'production'
         args.clean = True
+        args.urlformat = 'clean'
         config = configure(args)
         self.assertEqual(config.domain, 'not.there.bla')
         self.assertFalse(config.verbose)
         self.assertEqual(config.environment, 'production')
         self.assertTrue(config.clean)
+        self.assertEqual(config.urlformat, 'clean')
 
     def test_configuration_stringifiable(self):
         """SyrinxConfiguration objects should convert to readable 
@@ -76,5 +82,11 @@ def test_configuration_stringifiable(self):
         config.environment = 'default'
         config.verbose = False
         config.clean = True
+        config.urlformat = 'filesystem'
         self.assertEqual(str(config), 
-            '\tclean = true\n\tdomain = "some.where.bla"\n\tenvironment = "default"\n\tverbose = false')
+            '\tclean = true\n'
+            '\tdomain = "some.where.bla"\n'
+            '\tenvironment = "default"\n'
+            '\tverbose = false\n'
+            '\turlformat = "filesystem"'
+        )

tests/content_node.py

@@ -5,7 +5,7 @@ class ContentNodeTests(TestCase):
 
     def test_follows_buildPage(self):
         from syrinx.read import ContentNode
-        node = ContentNode()
+        node = ContentNode(None, None)
         node.name = 'foo_bar'
         node.front = dict()
         self.assertEqual(node.title, 'Foo Bar')

tests/read.py

@@ -70,8 +70,11 @@ def test_read_adds_build_info(self, datetime, read_file, walk):
 
     @patch('syrinx.read.walk')
     @patch('syrinx.read.read_file')
-    def test_read_address(self, read_file, walk):
-        """
+    def test_read_address_filesystem(self, read_file, walk):
+        """With the "filesystem" style, terminal branches have trailing slashes,
+        and leaves have file extensions.
+
+        This style typically works out-of-the-box with web server software.
         """
         read_file.return_value = dict(), ''
         walk.return_value = [
@@ -82,8 +85,43 @@ def test_read_address(self, read_file, walk):
         from syrinx.read import read
         config = Mock()
         config.domain = 'loop.xyz'
+        config.urlformat = 'filesystem'
         root = read('/pth', config)
-        self.assertEqual(root.address, 'https://loop.xyz/')
+        self.assertEqual(root.address, 'https://loop.xyz')
         self.assertEqual(root.branches[0].address, 'https://loop.xyz/foo/')
+        self.assertEqual(root.branches[0].leaves[0].address,
+                         'https://loop.xyz/foo/boz.html')
         self.assertEqual(root.branches[0].branches[0].address,
             'https://loop.xyz/foo/bar/')
+        
+    @patch('syrinx.read.walk')
+    @patch('syrinx.read.read_file')
+    def test_read_address_clean(self, read_file, walk):
+        """With the "clean" style, terminal branches and leaves
+        do not have trailing slashes or file extensions
+        """
+        read_file.return_value = dict(), ''
+        walk.return_value = [
+            ('/pth/content', None, ['index.md']),
+            ('/pth/content/bar', None, ['index.md', 'boz.md']),
+            ('/pth/content/foo', None, ['index.md']),
+            ('/pth/content/foo/bar', None, ['index.md']),
+        ]
+        from syrinx.read import read
+        config = Mock()
+        config.domain = 'loop.xyz'
+        config.urlformat = 'clean'
+        root = read('/pth', config)
+        self.assertEqual(root.address, 'https://loop.xyz')
+        # this page has sub-branches, i.e. is a directory, 
+        # so comes with trailing slash:
+        self.assertEqual(root.branches[0].address, 'https://loop.xyz/bar/')
+        # this page has leaves, i.e. is a directory, 
+        # so comes with trailing slash:
+        self.assertEqual(root.branches[1].address, 'https://loop.xyz/foo/')
+        # leaves dont get a trailing slash:
+        self.assertEqual(root.branches[0].leaves[0].address,
+                         'https://loop.xyz/foo/boz')
+        # branches without sub-branches don't get a slash
+        self.assertEqual(root.branches[0].branches[0].address,
+                         'https://loop.xyz/foo/bar')