shortener WIP

Author: ramalho

links/shortener.py

@@ -1,3 +1,16 @@
+"""
+URL shortener for .htaccess redirects
+
+Procedure:
+
+0. create empty dicts named targets and redirects
+1. given a target_url, find it in targets;
+    1.1. if found, use the short_url stored there
+    1.2. if not, generate new short_url and store it in targets and redirects
+
+"""
+
+
 from collections.abc import Iterable, Iterator
 
 
@@ -14,21 +27,23 @@ def parse_htaccess(text: str) -> Iterator[tuple[str, str]]:
 
 
 def choose(a: str, b: str) -> str:
-    def key(k):
+    def key(k: str) -> tuple[int, bool, list[str]]:
         parts = k.split('-')
         if len(parts) > 1:
-            parts = ['z' + p.rjust(8, '0') if p.isnumeric() else p for p in parts]
-        return (len(k), '-' in k, parts)
+            parts = [(f'z{p:>08}' if p.isnumeric() else p) for p in parts]
+        return len(k), '-' in k, parts
     return min(a, b, key=key)
 
 
 def load_redirects(pairs: Iterable[tuple[str, str]]) -> tuple[dict, dict]:
     redirects = {}
-    for key, url in pairs:
-        existing_url = redirects.get(key)
-        if existing_url:
-            if len(existing_url) < len(url):
-                continue
-            elif len(existing_url) == len(url):
-                url = choose(existing_url, url)
-            
+    targets = {}
+    for short_url, url in pairs:
+        url = redirects.setdefault(short_url, url)
+        existing_short_url = targets.get(url)
+        if existing_short_url is None:
+            targets[url] = short_url
+        else:
+            targets[url] = choose(short_url, existing_short_url)
+
+    return redirects, targets

links/test_shortener.py

@@ -1,6 +1,6 @@
 from pytest import mark
 
-from shortener import parse_htaccess, choose
+from shortener import parse_htaccess, choose, load_redirects
 
 
 HTACCESS_1 = """
@@ -13,14 +13,17 @@
 # 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/
 """
 
 PARSED_HTACCESS_1 = [
         ('book', 'https://www.oreilly.com/.../9781492056348/'),
         ('home', 'https://www.fluentpython.com/'),
         ('1-20', 'https://www.fluentpython.com/'),
         ('ora', 'https://www.oreilly.com/.../9781492056348/'),
+        ('2-10', 'http://example.com/'),
+        ('10-2', 'http://example.com/'),
     ]
 
 def test_parse_htaccess():
@@ -40,10 +43,24 @@ def test_choose(a, b, expected):
     assert res == expected
 
 
-# def test_load_redirects():
-#     expected = {
-#         'home': 'https://www.fluentpython.com/',
-#         'ora': 'https://www.oreilly.com/.../9781492056348/'
-#     }
-#     redirects, _ = load_redirects(PARSED_HTACCESS_1)
-#     assert redirects == 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_HTACCESS_1)
+    assert redirects == expected
+
+
+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_HTACCESS_1)
+    assert targets == expected