From 132a543a674b263b0421d6fe0cc0698417494d3a Mon Sep 17 00:00:00 2001
From: Mark Wardle <mark@wardle.org>
Date: Tue, 30 Mar 2021 22:10:31 +0100
Subject: [PATCH] Search endpoint now supports unlimited results, but constrain
 for web clients.

---
 src/com/eldrix/hermes/impl/search.clj | 28 ++++++++++++++-------------
 src/com/eldrix/hermes/server.clj      | 10 ++++++----
 2 files changed, 21 insertions(+), 17 deletions(-)

diff --git a/src/com/eldrix/hermes/impl/search.clj b/src/com/eldrix/hermes/impl/search.clj
index ad95455a..e5732fca 100644
--- a/src/com/eldrix/hermes/impl/search.clj
+++ b/src/com/eldrix/hermes/impl/search.clj
@@ -278,13 +278,21 @@
   [^IndexSearcher searcher ^ScoreDoc score-doc]
   (doc-id->concept-id searcher (.-doc score-doc)))
 
+(defn do-query-for-results
+  ([^IndexSearcher searcher ^Query q]
+   (->> (search-all searcher q)
+        (map #(.doc searcher %))
+        (map doc->result)))
+  ([^IndexSearcher searcher ^Query q max-hits]
+   (map (partial scoredoc->result searcher) (seq (.-scoreDocs (.search searcher q (int max-hits)))))))
+
 (defn do-search
   "Perform a search against the index.
   Parameters:
   - searcher : the IndexSearcher to use
   - params   : a map of search parameters, which are:
     |- :s                  : search string to use
-    |- :max-hits           : maximum hits (default, 200)
+    |- :max-hits           : maximum hits (if omitted returns unlimited results)
     |- :fuzzy              : fuzziness (0-2, default 0)
     |- :fallback-fuzzy     : if no results, try again with fuzzy search?
     |- :query              : additional ^Query to apply
@@ -302,12 +310,14 @@
   (do-search searcher {:s \"neurologist\"  :properties {snomed/IsA [14679004]}})
 
   A FSN is a fully-specified name and should generally be left out of search."
-  [^IndexSearcher searcher params]
+  [^IndexSearcher searcher {:keys [max-hits] :as params}]
   (let [q1 (make-search-query params)
         q2 (if-let [q (:query params)] (q-and [q1 q]) q1)
-        hits (seq (.-scoreDocs ^TopDocs (.search searcher ^Query q2 (int (or (:max-hits params) 200)))))]
-    (if hits
-      (map (partial scoredoc->result searcher) hits)
+        results (if max-hits
+                  (do-query-for-results searcher q2 (int max-hits))
+                  (do-query-for-results searcher q2))]
+    (if results
+      results
       (let [fuzzy (or (:fuzzy params) 0)
             fallback (or (:fallback-fuzzy params) 0)]
         (when (and (= fuzzy 0) (> fallback 0))
@@ -328,14 +338,6 @@
    (let [topdocs ^TopDocs (.search searcher query ^int max-hits)]
      (topdocs->concept-ids searcher topdocs))))
 
-(defn do-query-for-results
-  ([^IndexSearcher searcher ^Query q]
-   (->> (search-all searcher q)
-        (map #(.doc searcher %))
-        (map doc->result)))
-  ([^IndexSearcher searcher ^Query q max-hits]
-   (map (partial scoredoc->result searcher) (seq (.-scoreDocs (.search searcher q (int max-hits)))))))
-
 (defn q-self
   "Returns a query that will only return documents for the concept specified."
   [concept-id]
diff --git a/src/com/eldrix/hermes/server.clj b/src/com/eldrix/hermes/server.clj
index c95cb44a..293a9433 100644
--- a/src/com/eldrix/hermes/server.clj
+++ b/src/com/eldrix/hermes/server.clj
@@ -179,9 +179,11 @@
   {:name  ::get-search
    :enter (fn [context]
             (let [params (parse-search-params context)
-                  svc (get-in context [:request ::service])]
-              (when (= (:max-hits params) 0) (throw (IllegalArgumentException. "invalid parameter: 0 maxHits")))
-              (assoc context :result (svc/search svc params))))})
+                  svc (get-in context [:request ::service])
+                  max-hits (or (:max-hits params) 200)]
+              (if (< 0 max-hits 10000)
+                (assoc context :result (svc/search svc (assoc params :max-hits max-hits)))
+                (throw (IllegalArgumentException. "invalid parameter: maxHits")))))})
 
 (def common-routes [coerce-body content-neg-intc entity-render])
 (def routes
@@ -189,7 +191,7 @@
     #{["/v1/snomed/concepts/:concept-id" :get (conj common-routes get-concept)]
       ["/v1/snomed/concepts/:concept-id/descriptions" :get (conj common-routes get-concept-descriptions)]
       ["/v1/snomed/concepts/:concept-id/preferred" :get (conj common-routes get-concept-preferred-description)]
-      ["/v1/snomed/concepts/:concept-id/extended" :get (conj common-routes  get-extended-concept)]
+      ["/v1/snomed/concepts/:concept-id/extended" :get (conj common-routes get-extended-concept)]
       ["/v1/snomed/concepts/:concept-id/map/:refset-id" :get (conj common-routes get-map-to)]
       ["/v1/snomed/concepts/:concept-id/subsumed-by/:subsumer-id" :get (conj common-routes subsumed-by?)]
       ["/v1/snomed/crossmap/:refset-id/:code" :get (conj common-routes get-map-from)]