Fix comparison bugs and modernize

pyproject.toml

@@ -9,7 +9,7 @@ authors = [
 maintainers = [
     {name = "Andreas Stefl", email="stefl.andreas@gmail.com"},
 ]
-requires-python = ">=3.7"
+requires-python = ">=3.10"
 readme = "README.md"
 license = {file = "LICENSE.txt"}
 dependencies = [

src/htmlcmp/common.py

@@ -47,9 +47,9 @@ def compare_html(a: Path, b: Path, browser=None, diff_output: Path = None) -> bo
         raise FileNotFoundError("Both arguments must be files")
 
     if browser is None:
-        browser = get_browser()
+        browser = get_browser("firefox")
     diff, (image_a, image_b) = html_render_diff(a, b, browser=browser)
-    result = True if diff.getbbox() is None else False
+    result = diff.getbbox() is None
     if diff_output is not None and not result:
         diff_output.mkdir(parents=True, exist_ok=True)
         image_a.save(diff_output / "a.png")
@@ -64,12 +64,19 @@ def compare_files(a: Path, b: Path, **kwargs) -> bool:
     if not a.is_file() or not b.is_file():
         raise FileNotFoundError("Both arguments must be files")
 
-    if filecmp.cmp(a, b):
+    if filecmp.cmp(a, b, shallow=False):
         return True
     if a.suffix == ".json":
         return compare_json(a, b)
     if a.suffix == ".html":
+        if kwargs.get("browser") is None:
+            # No browser available (e.g. --driver none): the files already
+            # differ at the byte level and we cannot render to check for visual
+            # equality, so report them as different rather than spinning up a
+            # browser here.
+            return False
         return compare_html(a, b, **kwargs)
+    return False
 
 
 def comparable_file(path: Path) -> bool:

src/htmlcmp/compare_output_server.py

@@ -32,6 +32,10 @@ class Config:
     observer = None
     comparator = None
     browser = None
+    # Serializes access to the single shared ``browser`` above: Flask serves
+    # requests from a thread pool and Selenium drivers are not thread-safe, so
+    # concurrent /image_diff requests would otherwise interleave on one driver.
+    browser_lock = threading.Lock()
     thread_local = threading.local()
     log_file: Path = None
 
@@ -722,11 +726,12 @@ def image_diff(path: str):
     if not (Config.path_a / path).is_file() or not (Config.path_b / path).is_file():
         return "Image diff not available: file missing on one side", 404
 
-    diff, _ = html_render_diff(
-        Config.path_a / path,
-        Config.path_b / path,
-        Config.browser,
-    )
+    with Config.browser_lock:
+        diff, _ = html_render_diff(
+            Config.path_a / path,
+            Config.path_b / path,
+            Config.browser,
+        )
     tmp = io.BytesIO()
     diff.save(tmp, "JPEG", quality=70)
     tmp.seek(0)
@@ -786,6 +791,11 @@ def main():
     parser.add_argument("--driver", choices=["chrome", "firefox"])
     parser.add_argument("--max-workers", type=int, default=1)
     parser.add_argument("--compare", action="store_true")
+    parser.add_argument(
+        "--host",
+        default="0.0.0.0",
+        help="Host/interface to bind the server to (default: 0.0.0.0)",
+    )
     parser.add_argument("--port", type=int, default=5000)
     parser.add_argument(
         "-v",
@@ -820,7 +830,7 @@ def main():
         Config.observer = Observer()
         Config.observer.start()
 
-    app.run(host="0.0.0.0", port=args.port)
+    app.run(host=args.host, port=args.port)
 
     if args.compare:
         Config.observer.stop()

src/htmlcmp/html_render_diff.py

@@ -108,8 +108,8 @@ def main():
     parser.add_argument("a", type=Path, help="Path to the first HTML file")
     parser.add_argument("b", type=Path, help="Path to the second HTML file")
     parser.add_argument("--driver", choices=["chrome", "firefox"], default="firefox")
-    parser.add_argument("--max-width", default=1000)
-    parser.add_argument("--max-height", default=10000)
+    parser.add_argument("--max-width", type=int, default=1000)
+    parser.add_argument("--max-height", type=int, default=10000)
     args = parser.parse_args()
 
     browser = get_browser(args.driver, args.max_width, args.max_height)

src/htmlcmp/tidy_output.py

@@ -112,12 +112,12 @@ def tidy_dir(
         "error": [],
     }
 
-    items = [p for p in path.iterdir()]
-    files = sorted([path for path in items if path.is_file() and tidyable_file(path)])
-    dirs = sorted([path for path in items if path.is_dir()])
+    items = list(path.iterdir())
+    files = sorted(p for p in items if p.is_file() and tidyable_file(p))
+    dirs = sorted(p for p in items if p.is_dir())
 
-    for filename in [path.name for path in files]:
-        filepath = path / filename
+    for filepath in files:
+        filename = filepath.name
         tidy = tidy_file(filepath, html_tidy_config=html_tidy_config, verbose=verbose)
         if tidy == 0:
             print(f"{prefix_file}{bcolors.OKGREEN}{filename} ✓{bcolors.ENDC}")
@@ -128,10 +128,10 @@ def tidy_dir(
             print(f"{prefix_file}{bcolors.FAIL}{filename} ✘{bcolors.ENDC}")
             result["error"].append(filepath)
 
-    for dirname in [path.name for path in dirs]:
-        print(prefix + "├── " + dirname)
+    for dirpath in dirs:
+        print(prefix + "├── " + dirpath.name)
         subresult = tidy_dir(
-            path / dirname,
+            dirpath,
             level=level + 1,
             prefix=prefix + "│   ",
             html_tidy_config=html_tidy_config,