shortener WIP

Author: ramalho

links/shortener.py

@@ -22,11 +22,11 @@
 ## Targets in memory
 
 The `targets` dict maps target URLs to short paths.
-It's also loaded from data in the `.htaccess` file,
+It's also computed from data in the `.htaccess` file,
 but the algorithm is more complicated.
 
 The same target URL can be mapped to multiple short paths
-in `.htaccess` when the same target URL was added more
+in `.htaccess` if the same target URL was added more
 than once with different short paths by mistake.
 We cannot fix these mistakes because the redundant
 short paths are printed in Fluent Python Second Edition.
@@ -57,6 +57,13 @@
 
 import itertools
 from collections.abc import Iterable, Iterator
+from typing import NamedTuple
+
+
+class ShortenResult(NamedTuple):
+    url: str
+    path: str
+    new: bool
 
 
 def parse_htaccess(text: str) -> Iterator[tuple[str, str]]:
@@ -95,6 +102,22 @@ def load_redirects(pairs: Iterable[tuple[str, str]]) -> tuple[dict, dict]:
     return redirects, targets
 
 
+NO_PATH = ''
+
+def shorten_one(target: str, path_gen: Iterator[str], redirects: dict, targets: dict) -> ShortenResult:
+    if path := targets.get(target, NO_PATH):
+        return ShortenResult(target, path, False)
+    path = next(path_gen)
+    redirects[path] = target
+    targets[target] = path
+    return ShortenResult(target, path, True)
+
+
+def update_htaccess(new_targets: list[ShortenResult]):
+
+    pass
+
+
 SDIGITS = '23456789abcdefghjkmnpqrstvwxyz'
 
 

links/test_shortener.py

@@ -1,7 +1,33 @@
 from pytest import mark
 
 from shortener import parse_htaccess, choose, load_redirects
-from shortener import gen_short, gen_unused_short
+from shortener import gen_short, gen_unused_short, shorten_one, ShortenResult
+
+
+SAMPLE_HTACCESS = """
+ErrorDocument 404 /404.html
+
+# main resources
+RedirectTemp /book	https://www.oreilly.com/.../9781492056348/
+RedirectTemp /home	https://www.fluentpython.com/  # extra content site
+
+# duplicate targets
+RedirectTemp /1-20	https://www.fluentpython.com/
+RedirectTemp /ora	https://www.oreilly.com/.../9781492056348/
+RedirectTemp /2-10	http://example.com/
+RedirectTemp /10-2	http://example.com/
+
+# shortened
+RedirectTemp /22    http://firstshortened.co
+"""
+
+FROZEN_TIME = '2025-06-07 01:02:03'
+
+UPDATED_SAMPLE_HTACCESS = SAMPLE_HTACCESS + f"""
+# appended: {FROZEN_TIME}
+RedirectTemp /23 https://new.site/
+RedirectTemp /24 https://other.new.site/
+"""
 
 
 PARSED_SAMPLE_HTACCESS = [
@@ -11,13 +37,32 @@
     ('ora', 'https://www.oreilly.com/.../9781492056348/'),
     ('2-10', 'http://example.com/'),
     ('10-2', 'http://example.com/'),
+    ('22', 'http://firstshortened.co')
 ]
 
+# straightforward mapping of .htaccess; some targets may be duplicated.
+SAMPLE_REDIRECTS = {
+    'home': 'https://www.fluentpython.com/',
+    '1-20': 'https://www.fluentpython.com/',
+    '2-10': 'http://example.com/',
+    '10-2': 'http://example.com/',
+    'book': 'https://www.oreilly.com/.../9781492056348/',
+    'ora': 'https://www.oreilly.com/.../9781492056348/',
+    '22': 'http://firstshortened.co',
+}
+
+# the value must be shortest path for that target in the .htaccess
+SAMPLE_TARGETS = {
+    'https://www.fluentpython.com/': 'home',
+    'https://www.oreilly.com/.../9781492056348/': 'ora',
+    'http://example.com/': '2-10',
+    'http://firstshortened.co': '22',
+}
+
 
-def test_parse_htaccess(shared_datadir):
-    with open(shared_datadir / 'sample.htaccess') as fp:
-        text = fp.read()
-    res = list(parse_htaccess(text))
+
+def test_parse_htaccess():
+    res = list(parse_htaccess(SAMPLE_HTACCESS))
     assert res == PARSED_SAMPLE_HTACCESS
 
 
@@ -38,26 +83,43 @@ def test_choose(a, b, expected):
 
 
 def test_load_redirects():
-    expected = {
-        'home': 'https://www.fluentpython.com/',
-        '1-20': 'https://www.fluentpython.com/',
-        '2-10': 'http://example.com/',
-        '10-2': 'http://example.com/',
-        'book': 'https://www.oreilly.com/.../9781492056348/',
-        'ora': 'https://www.oreilly.com/.../9781492056348/',
-    }
-    redirects, _ = load_redirects(PARSED_SAMPLE_HTACCESS)
-    assert redirects == expected
+   redirects, _ = load_redirects(PARSED_SAMPLE_HTACCESS)
+   assert redirects == SAMPLE_REDIRECTS
 
 
 def test_load_redirect_targets():
-    expected = {
-        'https://www.fluentpython.com/': 'home',
-        'https://www.oreilly.com/.../9781492056348/': 'ora',
-        'http://example.com/': '2-10',
-    }
     _, targets = load_redirects(PARSED_SAMPLE_HTACCESS)
-    assert targets == expected
+    assert targets == SAMPLE_TARGETS
+
+
+@mark.parametrize(
+    'target,path,new',
+    [
+        ('https://www.fluentpython.com/', 'home', False),
+        ('https://new.site/', '23', True),
+    ],
+)
+def test_shorten(target, path, new):
+    expected = ShortenResult(target, path, new)
+    redirects = dict(SAMPLE_REDIRECTS)
+    targets = dict(SAMPLE_TARGETS)
+    result = shorten_one(target, gen_unused_short(redirects), redirects, targets)
+    assert result == expected
+    updated = redirects.keys() - SAMPLE_REDIRECTS.keys()
+    if new:
+        assert len(updated) == 1
+        new_path = updated.pop()
+        assert new_path == path
+        assert redirects == SAMPLE_REDIRECTS | {new_path: target}
+        assert targets == SAMPLE_TARGETS | {target: new_path}
+    else:
+        assert len(updated) == 0
+        assert redirects == SAMPLE_REDIRECTS
+        assert targets == SAMPLE_TARGETS
+
+
+def test_update_htaccess():
+    pass
 
 
 def test_gen_short():