Label Loading Fix (#490)

* improve load time

* fix loadTime

* revert memory index parse file

* parse_label_file change for memory

* Revert "parse_label_file change for memory"

This reverts commit 56f65a5.

* fix

* fix bug for loading specific label file

Author: NeelamMahapatro

include/pq_flash_index.h

@@ -116,6 +116,7 @@ template <typename T, typename LabelT = uint32_t> class PQFlashIndex
     DISKANN_DLLEXPORT void generate_random_labels(std::vector<LabelT> &labels, const uint32_t num_labels,
                                                   const uint32_t nthreads);
     void reset_stream_for_reading(std::basic_istream<char> &infile);
+    size_t search_string_range(const std::string &str, char ch, size_t start, size_t end);
     // index info
     // nhood of node `i` is in sector: [i / nnodes_per_sector]
     // offset in sector: [(i % nnodes_per_sector) * max_node_len]

src/pq_flash_index.cpp

@@ -592,7 +592,7 @@ void PQFlashIndex<T, LabelT>::get_label_file_metadata(const std::string &fileCon
         size_t next_lbl_pos = 0;
         while (lbl_pos < next_pos && lbl_pos != std::string::npos)
         {
-            next_lbl_pos = fileContent.find(',', lbl_pos);
+            next_lbl_pos = search_string_range(fileContent, ',', lbl_pos, next_pos);
             if (next_lbl_pos == std::string::npos) // the last label
             {
                 next_lbl_pos = next_pos;
@@ -671,7 +671,7 @@ void PQFlashIndex<T, LabelT>::parse_label_file(std::basic_istream<char> &infile,
         size_t next_lbl_pos = 0;
         while (lbl_pos < next_pos && lbl_pos != std::string::npos)
         {
-            next_lbl_pos = buffer.find(',', lbl_pos);
+            next_lbl_pos = search_string_range(buffer, ',', lbl_pos, next_pos);
             if (next_lbl_pos == std::string::npos) // the last label in the whole file
             {
                 next_lbl_pos = next_pos;
@@ -810,7 +810,7 @@ int PQFlashIndex<T, LabelT>::load_from_separate_paths(uint32_t num_threads, cons
         }
 #endif
         parse_label_file(infile, num_pts_in_label_file);
-        assert(num_pts_in_label_file == this->_num_points);
+        assert(num_pts_in_label_file == this->num_points);
 
 #ifndef EXEC_ENV_OLS
         infile.close();
@@ -1766,6 +1766,20 @@ template <typename T, typename LabelT> char *PQFlashIndex<T, LabelT>::getHeaderB
 }
 #endif
 
+template <typename T, typename LabelT>
+size_t PQFlashIndex<T, LabelT>::search_string_range(const std::string &str, char ch, size_t start, size_t end)
+{
+    for (; start != end; start++)
+    {
+        if (str[start] == ch)
+        {
+            return start;
+        }
+    }
+
+    return std::string::npos;
+}
+
 // instantiations
 template class PQFlashIndex<uint8_t>;
 template class PQFlashIndex<int8_t>;