feat: mdgrep 0.1.0 – boolean full-text search for Markdown

- query language: substrings, phrases, /regex/, AND/OR/NOT (+ UND/ODER/NICHT),
  parentheses; case-insensitive; ReDoS protection via regex timeouts;
  syntax errors with stable code and position
- SearchEngine over in-memory blocks: filters, predicate, grouping, sorting,
  paging, time budget, automatic reload, highlighted snippets
- Markdown loader with split modes file/heading/list/paragraph
- grep-like CLI (mdgrep), JSON output, grep exit codes
- Flask blueprint/app with JSON API and widget assets, dev server
- framework-free JS widget: live search, URL state, filters, paging,
  syntax help, en/de labels, CSS custom properties for theming

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

tests/test_web.py

@@ -0,0 +1,80 @@
+import pytest
+
+from mdgrep import Block, SearchEngine
+from mdgrep.web import create_app
+
+BLOCKS = [
+    Block("1", "rust news", title="Rust", meta={"lang": "en", "url": "u1", "markdown": "**rust**"}),
+    Block("2", "rust nachrichten", title="Rust", meta={"lang": "de", "url": "u1", "markdown": "x"}),
+    Block("3", "go news", title="Go", meta={"lang": "en", "url": "u2", "markdown": "y"}),
+    Block("4", "rust again", title="Rust", meta={"lang": "en", "url": "u1", "markdown": "z"}),
+]
+
+
+@pytest.fixture
+def client():
+    app = create_app(SearchEngine(BLOCKS), filters=["lang"],
+                     group_by=lambda b: (b.meta["lang"], b.meta["url"]))
+    return app.test_client()
+
+
+def test_search_ok(client):
+    data = client.get("/search?q=rust").get_json()
+    assert data["total_blocks"] == 3 and data["total_groups"] == 2
+    first = data["results"][0]
+    assert first["block"]["id"] == "1" and [b["id"] for b in first["also"]] == ["4"]
+    assert "markdown" not in first["block"]["meta"]
+    assert "markdown" not in first["also"][0]["meta"]
+
+
+def test_filters(client):
+    assert client.get("/search?q=rust&lang=de").get_json()["total_blocks"] == 1
+    assert client.get("/search?q=rust&lang=all").get_json()["total_blocks"] == 3
+    assert client.get("/search?q=rust&lang=de,en").get_json()["total_blocks"] == 3
+
+
+def test_paging_and_limit_clamping(client):
+    data = client.get("/search?q=news OR rust&limit=1").get_json()
+    assert len(data["results"]) == 1 and data["next_offset"] == 1
+    data = client.get("/search?q=rust&limit=9999&offset=-5").get_json()
+    assert data["limit"] == 100 and data["offset"] == 0
+
+
+def test_empty_query(client):
+    data = client.get("/search?q=").get_json()
+    assert data["results"] == [] and data["total_groups"] == 0
+
+
+def test_syntax_error_400(client):
+    resp = client.get("/search?q=(rust")
+    assert resp.status_code == 400
+    assert resp.get_json()["error"] == {"code": "unclosed_paren",
+                                        "message": "Missing closing parenthesis", "position": 0}
+
+
+def test_too_long_query(client):
+    resp = client.get("/search?q=" + "a" * 400)
+    assert resp.status_code == 400 and resp.get_json()["error"]["code"] == "too_long"
+
+
+def test_unicode_not_escaped(client):
+    app = create_app(SearchEngine([Block("x", "Größe")]))
+    body = app.test_client().get("/search?q=größe").get_data(as_text=True)
+    assert "Größe" in body
+
+
+def test_widget_assets_served(client):
+    assert client.get("/mdgrep-static/mdgrep-widget.js").status_code == 200
+    assert client.get("/mdgrep-static/mdgrep-widget.css").status_code == 200
+
+
+def test_static_dir_and_prefix(tmp_path):
+    (tmp_path / "index.html").write_text("<h1>home</h1>")
+    (tmp_path / "sub").mkdir()
+    (tmp_path / "sub" / "index.html").write_text("sub page")
+    app = create_app(SearchEngine(BLOCKS), url_prefix="/api", static_dir=tmp_path)
+    c = app.test_client()
+    assert c.get("/").get_data(as_text=True) == "<h1>home</h1>"
+    assert c.get("/sub/").get_data(as_text=True) == "sub page"
+    assert c.get("/api/search?q=go").get_json()["total_blocks"] == 1
+    assert c.get("/../etc/passwd").status_code == 404