Merge branch 'main' into feature/pr-221-fix

Author: filipchristiansen

src/gitingest/utils/query_parser_utils.py

@@ -60,22 +60,48 @@ def _is_valid_pattern(pattern: str) -> bool:
 
 def _validate_host(host: str) -> None:
     """
-    Validate the given host against the known Git hosts.
+    Validate a hostname.
+
+    The host is accepted if it is either present in the hard-coded `KNOWN_GIT_HOSTS` list or if it satisfies the
+    simple heuristics in `_looks_like_git_host`, which try to recognise common self-hosted Git services (e.g. GitLab
+    instances on sub-domains such as `gitlab.example.com` or `git.example.com`).
 
     Parameters
     ----------
     host : str
-        The host to validate.
+        Hostname (case-insensitive).
 
     Raises
     ------
     ValueError
-        If the host is not a known Git host.
+        If the host cannot be recognised as a probable Git hosting domain.
     """
-    if host not in KNOWN_GIT_HOSTS:
+    host = host.lower()
+    if host not in KNOWN_GIT_HOSTS and not _looks_like_git_host(host):
         raise ValueError(f"Unknown domain '{host}' in URL")
 
 
+def _looks_like_git_host(host: str) -> bool:
+    """
+    Check if the given host looks like a Git host.
+
+    The current heuristic returns `True` when the host starts with `git.` (e.g. `git.example.com`) or starts with
+    `gitlab.` (e.g. `gitlab.company.com`).
+
+    Parameters
+    ----------
+    host : str
+        Hostname (case-insensitive).
+
+    Returns
+    -------
+    bool
+        True if the host looks like a Git host, otherwise False.
+    """
+    host = host.lower()
+    return host.startswith(("git.", "gitlab."))
+
+
 def _validate_url_scheme(scheme: str) -> None:
     """
     Validate the given scheme against the known schemes.
@@ -90,6 +116,7 @@ def _validate_url_scheme(scheme: str) -> None:
     ValueError
         If the scheme is not 'http' or 'https'.
     """
+    scheme = scheme.lower()
     if scheme not in ("https", "http"):
         raise ValueError(f"Invalid URL scheme '{scheme}' in URL")
 

src/server/query_processor.py

@@ -96,6 +96,9 @@ async def process_query(
         if not query.url:
             raise ValueError("The 'url' parameter is required.")
 
+        # Sets the "<user>/<repo>" for the page title
+        context["short_repo_url"] = f"{query.user_name}/{query.repo_name}"
+
         clone_config = query.extract_clone_config()
         await clone_repo(clone_config, token=token)
         summary, tree, content = ingest_query(query)

src/server/templates/base.jinja

@@ -33,7 +33,13 @@
         <meta property="og:url" content="{{ request.url }}">
         <meta property="og:image" content="/static/og-image.png">
         <title>
-            {% block title %}Gitingest{% endblock %}
+            {% block title %}
+                {% if short_repo_url %}
+                    Gitingest - {{ short_repo_url }}
+                {% else %}
+                    Gitingest
+                {% endif %}
+            {% endblock %}
         </title>
         <script src="https://cdn.tailwindcss.com"></script>
         <script src="/static/js/utils.js"></script>

src/static/js/utils.js

@@ -109,6 +109,9 @@ function handleSubmit(event, showLoading = false) {
                     starsElement.textContent = starCount;
                 }
 
+                // Set dynamic title that includes the repo name.
+                document.title = document.body.getElementsByTagName('title')[0].textContent;
+
                 // Scroll to results if they exist
                 const resultsSection = document.querySelector('[data-results]');
                 if (resultsSection) {

tests/query_parser/__init__.py

@@ -10,6 +10,7 @@
 import pytest
 
 from gitingest.query_parsing import parse_query
+from gitingest.utils.query_parser_utils import KNOWN_GIT_HOSTS
 
 # Repository matrix: (host, user, repo)
 _REPOS: List[Tuple[str, str, str]] = [
@@ -18,6 +19,8 @@
     ("bitbucket.org", "na-dna", "llm-knowledge-share"),
     ("gitea.com", "xorm", "xorm"),
     ("codeberg.org", "forgejo", "forgejo"),
+    ("git.rwth-aachen.de", "medialab", "19squared"),
+    ("gitlab.alpinelinux.org", "alpine", "apk-tools"),
 ]
 
 
@@ -43,6 +46,13 @@ async def test_parse_query_without_host(
 
     expected_url = f"https://{host}/{user}/{repo}"
 
+    # For slug form with a custom host (not in KNOWN_GIT_HOSTS) we expect a failure,
+    # because the parser cannot guess which domain to use.
+    if variant == "slug" and host not in KNOWN_GIT_HOSTS:
+        with pytest.raises(ValueError):
+            await parse_query(url, max_file_size=50, from_web=True)
+        return
+
     query = await parse_query(url, max_file_size=50, from_web=True)
 
     # Compare against the canonical dict while ignoring unpredictable fields.

tests/query_parser/test_git_host_agnostic.py

@@ -24,6 +24,9 @@
     "https://gitea.com/user/repo",
     "https://codeberg.org/user/repo",
     "https://gist.github.com/user/repo",
+    "https://git.example.com/user/repo",
+    "https://gitlab.example.com/user/repo",
+    "https://gitlab.example.se/user/repo",
 ]
 
 URLS_HTTP: List[str] = [url.replace("https://", "http://") for url in URLS_HTTPS]