Fix a parse_uri query bug (#1740)

* Fix a parse_uri query bug

Signed-off-by: Haoyang Li <[email protected]>

* Add comments

Signed-off-by: Haoyang Li <[email protected]>

---------

Signed-off-by: Haoyang Li <[email protected]>

Author: thirtiseven

src/main/cpp/src/parse_uri.cu

@@ -500,9 +500,10 @@ __device__ std::pair<string_view, bool> find_query_part(string_view haystack, st
   auto h           = haystack.data();
   auto const end_h = haystack.data() + find_length;
   auto n           = needle.data();
+  bool match       = false;
   while (h < end_h) {
-    bool match = true;
-    for (size_type jdx = 0; match && (jdx < n_bytes); ++jdx) {
+    match = false;  // initialize to false to prevent empty query key
+    for (size_type jdx = 0; (jdx == 0 || match) && (jdx < n_bytes); ++jdx) {
       match = (h[jdx] == n[jdx]);
     }
     if (match) { match = n_bytes < haystack.size_bytes() && h[n_bytes] == '='; }
@@ -519,8 +520,8 @@ __device__ std::pair<string_view, bool> find_query_part(string_view haystack, st
     h++;
   }
 
-  // if h is past the end of the haystack, no match.
-  if (haystack.data() + haystack.size_bytes() <= h || *h != '=') { return {{}, false}; }
+  // if not match or no value, return nothing
+  if (!match || *h != '=') { return {{}, false}; }
 
   // skip over the =
   h++;

src/test/java/com/nvidia/spark/rapids/jni/ParseURITest.java

@@ -214,8 +214,11 @@ void parseURISparkTest() {
       "https://[::1]/?invalid=param&~.=!@&^",
       "[email protected]/path?query=1#Ref",
       "",
-      null};
-  
+      null,
+      "https://www.nvidia.com/?cat=12",
+      "www.nvidia.com/vote.php?pid=50",
+      "https://www.nvidia.com/vote.php?=50",
+    };
 
       String[] queries = {
         "a",
@@ -270,7 +273,11 @@ void parseURISparkTest() {
         "invalid",
         "query",
         "a",
-        "f"};
+        "f",
+        "query",
+        "query",
+        ""
+      };
 
     testProtocol(testData);
     testHost(testData);