Merge pull request #2629 from akohlmey/citeme-use-hash

Use a 64-bit hash in the CiteMe class to identify unique citations

src/citeme.cpp

@@ -15,6 +15,8 @@
 #include "comm.h"
 #include "universe.h"
 
+#include <functional>
+
 using namespace LAMMPS_NS;
 
 static const char cite_separator[] =
@@ -26,6 +28,9 @@ static const char cite_nagline[] =
 static const char cite_file[] = "The {} {} lists these citations in "
                                 "BibTeX format.\n\n";
 
+// define hash function
+static std::hash<std::string> get_hash;
+
 /* ---------------------------------------------------------------------- */
 
 CiteMe::CiteMe(LAMMPS *lmp, int _screen, int _logfile, const char *_file)
@@ -68,14 +73,16 @@ CiteMe::~CiteMe()
    process an added citation so it will be shown only once and as requested
 ------------------------------------------------------------------------- */
 
-void CiteMe::add(const char *ref)
+void CiteMe::add(const std::string &reference)
 {
   if (comm->me != 0) return;
-  if (cs->find(ref) != cs->end()) return;
+
-  cs->insert(ref);
+  std::size_t crc = get_hash(reference);
+  if (cs->find(crc) != cs->end()) return;
+  cs->insert(crc);
 
   if (fp) {
-    fputs(ref,fp);
+    fputs(reference.c_str(),fp);
     fflush(fp);
   }
 
@@ -93,7 +100,6 @@ void CiteMe::add(const char *ref)
     if (logfile_flag == VERBOSE) logbuffer += "\n";
   }
 
-  std::string reference = ref;
   std::size_t found = reference.find_first_of("\n");
   std::string header = reference.substr(0,found+1);
   if (screen_flag == VERBOSE) scrbuffer += "- " + reference;

src/citeme.h

@@ -23,7 +23,7 @@ class CiteMe : protected Pointers {
  public:
   CiteMe(class LAMMPS *, int, int, const char *);
   virtual ~CiteMe();
-  void add(const char *);       // register publication for output
+  void add(const std::string &); // register publication for output
   void flush();                  // flush buffers to screen and logfile
   enum {VERBOSE, TERSE};
 
@@ -34,7 +34,7 @@ class CiteMe : protected Pointers {
   int logfile_flag;            // determine whether verbose or terse output
   std::string scrbuffer;       // output buffer for screen
   std::string logbuffer;       // output buffer for logfile
-  typedef std::set<const char *> citeset;
+  typedef std::set<std::size_t> citeset;
   citeset *cs;                 // registered set of publications
 };
 }

unittest/commands/test_simple_commands.cpp

@@ -13,6 +13,7 @@
 
 #include "lammps.h"
 
+#include "citeme.h"
 #include "force.h"
 #include "info.h"
 #include "input.h"
@@ -368,6 +369,51 @@ TEST_F(SimpleCommandsTest, Shell)
     ASSERT_THAT(other_var, StrEq("2"));
 }
 
+TEST_F(SimpleCommandsTest, CiteMe)
+{
+    ASSERT_EQ(lmp->citeme, nullptr);
+
+    lmp->citeme = new LAMMPS_NS::CiteMe(lmp, CiteMe::TERSE, CiteMe::TERSE, nullptr);
+
+    ::testing::internal::CaptureStdout();
+    lmp->citeme->add("test citation one:\n 1\n");
+    lmp->citeme->add("test citation two:\n 2\n");
+    lmp->citeme->add("test citation one:\n 1\n");
+    lmp->citeme->flush();
+    std::string text = ::testing::internal::GetCapturedStdout();
+    if (verbose) std::cout << text;
+
+    // find the two unique citations, but not the third
+    ASSERT_THAT(text, MatchesRegex(".*one.*two.*"));
+    ASSERT_THAT(text, Not(MatchesRegex(".*one.*two.*one.*")));
+
+    ::testing::internal::CaptureStdout();
+    lmp->citeme->add("test citation one:\n 0\n");
+    lmp->citeme->add("test citation two:\n 2\n");
+    lmp->citeme->add("test citation three:\n 3\n");
+    lmp->citeme->flush();
+
+    text = ::testing::internal::GetCapturedStdout();
+    if (verbose) std::cout << text;
+
+    // find the forth (only differs in long citation) and sixth added citation
+    ASSERT_THAT(text, MatchesRegex(".*one.*three.*"));
+    ASSERT_THAT(text, Not(MatchesRegex(".*two.*")));
+
+    ::testing::internal::CaptureStdout();
+    lmp->citeme->add("test citation one:\n 1\n");
+    lmp->citeme->add("test citation two:\n 2\n");
+    lmp->citeme->add("test citation one:\n 0\n");
+    lmp->citeme->add("test citation two:\n 2\n");
+    lmp->citeme->add("test citation three:\n 3\n");
+    lmp->citeme->flush();
+
+    text = ::testing::internal::GetCapturedStdout();
+    if (verbose) std::cout << text;
+
+    // no new citation. no CITE-CITE-CITE- lines
+    ASSERT_THAT(text, Not(MatchesRegex(".*CITE-CITE-CITE-CITE.*")));
+}
 } // namespace LAMMPS_NS
 
 int main(int argc, char **argv)