Wrap cityhash in namespace and add accessor methods for metagraph compatibility

Changes for metagraph integration:

1. Wrap cityhash in namespace to avoid typedef pollution
   - cityhash.hpp and cityhash.cpp: Wrapped in 'cityhash' namespace
   - Prevents conflicts with rollinghashcpp uint64 typedef
   - hash_util.hpp: Use cityhash::CityHash128WithSeed instead of static CityMurmur

2. Add public accessor methods to dictionary
   - Added kmer_type typedef for easier template parameter extraction
   - Added strings() accessor returning m_spss.strings bit vector
   - Added strings_offsets() accessor returning m_spss.strings_offsets
   - Moved strings_offsets() next to string_offsets() for better organization

3. Build system changes
   - CMakeLists.txt: Added cityhash.cpp to SSHASH_SOURCES for linking

4. Update pthash submodule
   - Points to fix for compute_empirical_entropy multiple definition error

Author: adamant-pwn

CMakeLists.txt

@@ -60,6 +60,10 @@ set(Z_LIB_SOURCES
   external/gz/zip_stream.cpp
 )
 
+set(CITYHASH_SOURCES
+  external/cityhash/cityhash.cpp
+)
+
 set(SSHASH_SOURCES
   src/build.cpp
   src/dictionary.cpp
@@ -81,6 +85,7 @@ set(SSHASH_INCLUDE_DIRS
 # Create a static lib
 add_library(sshash_static STATIC
   ${Z_LIB_SOURCES}
+  ${CITYHASH_SOURCES}
   ${SSHASH_SOURCES}
 )
 

external/cityhash/cityhash.cpp

@@ -34,6 +34,8 @@
 #include <algorithm>
 #include <string.h>  // for memcpy and memset
 
+namespace cityhash {
+
 using namespace std;
 
 static uint64 UNALIGNED_LOAD64(const char* p) {
@@ -442,4 +444,6 @@ uint128 CityHashCrc128(const char* s, size_t len) {
     }
 }
 
-#endif
+#endif
+
+}  // namespace cityhash

external/cityhash/cityhash.hpp

@@ -48,6 +48,8 @@
 #include <stdlib.h>  // for size_t.
 #include <utility>
 
+namespace cityhash {
+
 // Microsoft Visual Studio may not have stdint.h.
 #if defined(_MSC_VER) && (_MSC_VER < 1600)
 typedef unsigned char uint8_t;
@@ -112,4 +114,6 @@ void CityHashCrc256(const char* s, size_t len, uint64* result);
 
 #endif  // __SSE4_2__
 
+}  // namespace cityhash
+
 #endif  // CITY_HASH_H_

external/pthash

@@ -68,6 +68,9 @@ struct dictionary  //
     /* Return the string of the kmer whose id is kmer_id. */
     void access(uint64_t kmer_id, char* string_kmer) const;
 
+    /* Accessor for internal bit vector */
+    bits::bit_vector const& strings() const { return m_spss.strings; }
+
     /* Membership queries. */
     bool is_member(char const* string_kmer, bool check_reverse_complement = true) const;
     bool is_member(Kmer uint_kmer, bool check_reverse_complement = true) const;
@@ -104,6 +107,9 @@ struct dictionary  //
         return m_spss.string_offsets(string_id);
     }
 
+    /* Accessor for internal offsets structure */
+    Offsets const& strings_offsets() const { return m_spss.strings_offsets; }
+
     iterator at_string_id(const uint64_t string_id) const {
         assert(string_id < num_strings());
         auto [begin, end] = string_offsets(string_id);

include/dictionary.hpp

@@ -1,7 +1,7 @@
 #pragma once
 
 #include "external/pthash/include/pthash.hpp"
-#include "external/cityhash/cityhash.cpp"
+#include "external/cityhash/cityhash.hpp"
 #include "constants.hpp"
 
 namespace sshash {
@@ -10,7 +10,7 @@ struct minimizers_city_hasher_128 {
     typedef pthash::hash128 hash_type;
 
     static inline pthash::hash128 hash(uint64_t const minimizer, uint64_t seed) {
-        auto ret = CityMurmur(reinterpret_cast<char const*>(&minimizer),  //
+        auto ret = cityhash::CityHash128WithSeed(reinterpret_cast<char const*>(&minimizer),  //
                               sizeof(minimizer), {seed, ~seed});
         return {ret.first, ret.second};
     }
@@ -60,7 +60,7 @@ struct kmers_city_hasher_128 {
     typedef pthash::hash128 hash_type;
 
     static inline pthash::hash128 hash(Kmer const x, uint64_t seed) {
-        auto ret = CityMurmur(reinterpret_cast<char const*>(&(x.bits)),  //
+        auto ret = cityhash::CityHash128WithSeed(reinterpret_cast<char const*>(&(x.bits)),  //
                               sizeof(x.bits), {seed, ~seed});
         return {ret.first, ret.second};
     }