Add mailto option

app.py

@@ -128,17 +128,17 @@ def get_index(dir: Path, search_time_s: float) -> Dataset:
     return index
 
 
-def execute_request(ids: list[str]) -> list[Work]:
+def execute_request(ids: list[str], mailto: str | None) -> list[Work]:
     if len(ids) > 100:
         raise ValueError("querying /works endpoint with more than 100 works")
 
     # query with the /works endpoint with a specific list of IDs and fields
     search_filter = f"openalex_id:{"|".join(ids)}"
     search_select = ",".join(["id"] + Work.get_raw_fields())
-    response = requests.get(
-        "https://api.openalex.org/works",
-        {"filter": search_filter, "select": search_select, "per-page": 100}
-    )
+    params = {"filter": search_filter, "select": search_select, "per-page": 100}
+    if mailto is not None:
+        params["mailto"] = mailto
+    response = requests.get("https://api.openalex.org/works", params)
     response.raise_for_status()
 
     # the response is not necessarily ordered, so order them
@@ -210,7 +210,8 @@ def main():
     model_name = get_env_var("MODEL_NAME", default="all-MiniLM-L6-v2")
     dir = get_env_var("DIR", Path, default=Path("index"))
     search_time_s = get_env_var("SEARCH_TIME_S", int, default=1)
-    k = get_env_var("K", int, default=20)
+    k = get_env_var("K", int, default=20)  # TODO: can't go higher than 20 yet
+    mailto = get_env_var("MAILTO", str, None)
 
     model = get_model(model_name, "cpu")
     index = get_index(dir, search_time_s)
@@ -222,7 +223,7 @@ def main():
 
         faiss_ids_flat = list(chain(*faiss_ids))
         openalex_ids_flat = index[faiss_ids_flat]["idxs"]
-        works_flat = execute_request(openalex_ids_flat)
+        works_flat = execute_request(openalex_ids_flat, mailto)
         works = [list(batch) for batch in batched(works_flat, k)]
 
         result_strings = [format_response(w, d) for w, d in zip(works, distances)]