Upload app.py

app.py

@@ -57,10 +57,14 @@ API_CACHE_TTL_SECONDS = 120
 SEARCH_CACHE_TTL_SECONDS = 300
 API_CACHE_MAX_ENTRIES = 500
 SEARCH_CACHE_MAX_ENTRIES = 500
+PREVIEW_CACHE_MAX_ENTRIES = 64
+PREVIEW_CACHE_MAX_ENTRY_BYTES = 2 * 1024 * 1024
 api_response_cache: dict[tuple, tuple[float, object]] = {}
 api_response_cache_lock = threading.Lock()
 search_response_cache: dict[tuple, tuple[float, object]] = {}
 search_response_cache_lock = threading.Lock()
+preview_text_cache: dict[str, str] = {}
+preview_text_cache_lock = threading.Lock()
 ZIP_TOKEN_TTL_SECONDS = 600
 ZIP_TOKEN_MAX_ENTRIES = 32
 zip_download_tokens: dict[str, tuple[float, list[str]]] = {}
@@ -1119,6 +1123,23 @@ def fulltext_search(q="", sources_filter=None, folders=None, min_size=None, max_
 def get_doc_storage_path(doc: dict) -> Path:
     return BASE_DIR / doc["storage_root"] / doc["storage_rel_path"]
 
+def read_preview_text(doc_id: str, file_path: Path) -> str:
+    with preview_text_cache_lock:
+        cached = preview_text_cache.get(doc_id)
+        if cached is not None:
+            return cached
+    try:
+        size = file_path.stat().st_size
+    except OSError:
+        size = PREVIEW_CACHE_MAX_ENTRY_BYTES + 1
+    text = file_path.read_text(encoding="utf-8", errors="ignore")
+    if size <= PREVIEW_CACHE_MAX_ENTRY_BYTES:
+        with preview_text_cache_lock:
+            preview_text_cache[doc_id] = text
+            while len(preview_text_cache) > PREVIEW_CACHE_MAX_ENTRIES:
+                preview_text_cache.pop(next(iter(preview_text_cache)), None)
+    return text
+
 def get_folder_contents(source_slug: str, path: str) -> dict:
     source_browser = folder_browser_data.get(source_slug, {})
     if path in source_browser:
@@ -1412,7 +1433,7 @@ def api_preview(doc_id: str):
     file_path = get_doc_storage_path(rec)
     if not file_path.exists():
         return JSONResponse({"error": f"missing file: {file_path.name}"}, status_code=404)
-    text = file_path.read_text(encoding="utf-8", errors="ignore")
+    text = read_preview_text(doc_id, file_path)
     return JSONResponse({"doc_id": doc_id, "title": rec["display_name"], "path": rec["display_rel_path"], "source": rec["source_name"], "text": text})
 @app.get("/api/snippet/{doc_id}")
 
@@ -1445,6 +1466,7 @@ def api_download(doc_id: str):
         media_type="text/plain; charset=utf-8",
         filename=f"{rec['display_name']}.txt",
         content_disposition_type="attachment",
+        headers={"Content-Encoding": "identity"},
     )
 @app.post("/api/zip")
 
@@ -1505,14 +1527,22 @@ app.mount("/static", StaticFiles(directory=str(BASE_DIR / "static"), html=True),
 app.mount("/icons", StaticFiles(directory=str(BASE_DIR / "static/icons")), name="icons")
 app.mount("/data/initial", StaticFiles(directory=str(BASE_DIR / "data/initial")), name="initial-data")
 app.mount("/data/sidebar", StaticFiles(directory=str(BASE_DIR / "data/sidebar")), name="sidebar-data")
+@lru_cache(maxsize=1)
+def _manifest_payload() -> dict:
+    return json.loads((BASE_DIR / "static/manifest.json").read_text(encoding="utf-8"))
+
 @app.get("/manifest.json")
 
 def serve_manifest():
-    return JSONResponse(json.loads((BASE_DIR / "static/manifest.json").read_text(encoding="utf-8")))
+    return JSONResponse(_manifest_payload())
+@lru_cache(maxsize=1)
+def _sw_source() -> str:
+    return (BASE_DIR / "static/sw.js").read_text(encoding="utf-8")
+
 @app.get("/sw.js")
 
 def serve_sw():
-    return PlainTextResponse((BASE_DIR / "static/sw.js").read_text(encoding="utf-8"), media_type="application/javascript")
+    return PlainTextResponse(_sw_source(), media_type="application/javascript")
 
 def refresh_fulltext_loop() -> None:
     while not fulltext_refresh_stop.wait(FULLTEXT_REFRESH_INTERVAL_SECONDS):
@@ -1544,6 +1574,7 @@ def sidebar_payload_for_path(rest_of_path: str) -> str | None:
         return None
     return payload_path.read_text(encoding="utf-8")
 
+@lru_cache(maxsize=64)
 def render_index_html(rest_of_path: str = "") -> str:
     html = (BASE_DIR / "static/index.html").read_text(encoding="utf-8")
     initial_payload = initial_payload_for_path(rest_of_path)