Move alphabet type into SeqString

[mussa.git] / alg / alphabet.cpp

diff --git a/alg/alphabet.cpp b/alg/alphabet.cpp
index bd29dd2480f47ace4caa84e24decba9549e936cd..56919199b6c7ccf30c05a6471b95187c6075f150 100644 (file)
--- a/alg/alphabet.cpp
+++ b/alg/alphabet.cpp
@@ -1,19 +1,45 @@
+#include <stdexcept>
+
 #include "alg/alphabet.hpp"
 
 // some standard dna alphabets 
 // Include nl (\012), and cr (\015) to make sequence parsing eol 
 // convention independent.
-const Alphabet Alphabet::reduced_dna_alphabet("AaCcGgTtNn\012\015");
-const Alphabet Alphabet::reduced_rna_alphabet("AaCcGgUuNn\012\015");
-const Alphabet Alphabet::reduced_nucleic_alphabet("AaCcGgTtUuNn\012\015");
+const char *Alphabet::reduced_dna_cstr = "AaCcGgTtNn\012\015";
+const char *Alphabet::reduced_rna_cstr = "AaCcGgUuNn\012\015";
+const char *Alphabet::reduced_nucleic_cstr = "AaCcGgTtUuNn\012\015";
 //! this is the general iupac alphabet for nucleotides
-const Alphabet Alphabet::nucleic_alphabet(
-  "AaCcGgTtUuRrYyMmKkSsWwBbDdHhVvNn-~.?\012\015"
-);
+const char *Alphabet::nucleic_cstr =
+  "AaCcGgTtUuRrYyMmKkSsWwBbDdHhVvNn-~.?\012\015";
 //! the protein alphabet
-const Alphabet Alphabet::protein_alphabet(
-  "AaCcDdEeFfGgHhIiKkLlMmNnPpQqRrSsTtVvWwYy\012\015"
-);
+const char *Alphabet::protein_cstr = 
+  "AaCcDdEeFfGgHhIiKkLlMmNnPpQqRrSsTtVvWwYy\012\015";
+const char *Alphabet::empty_cstr = "";
+
+const Alphabet& Alphabet::reduced_dna_alphabet() {
+  static Alphabet *a = new Alphabet(reduced_dna_cstr);
+  return *a;
+}
+const Alphabet& Alphabet::reduced_rna_alphabet() {
+  static Alphabet *a = new Alphabet(reduced_rna_cstr);
+  return *a;
+}
+const Alphabet& Alphabet::reduced_nucleic_alphabet() {
+  static Alphabet *a = new Alphabet(reduced_nucleic_cstr);
+  return *a;
+}
+const Alphabet& Alphabet::nucleic_alphabet() {
+  static Alphabet *a = new Alphabet(nucleic_cstr);
+  return *a;
+}
+const Alphabet& Alphabet::protein_alphabet() {
+  static Alphabet *a = new Alphabet(protein_cstr);
+  return *a;
+}
+const Alphabet& Alphabet::empty_alphabet() {
+  static Alphabet *a = new Alphabet(empty_cstr);
+  return *a;
+}
 
 Alphabet::Alphabet(const char *a) :
   alphabet(a)
@@ -28,12 +54,36 @@ void Alphabet::assign(const Alphabet& a)
   alphabet_set.insert(alphabet.begin(), alphabet.end());
 }
 
-const char *Alphabet::c_str() const
+bool operator==(const Alphabet& x, const Alphabet& y)
+{
+  return x.alphabet == y.alphabet; 
+}
+
+std::ostream& operator<<(std::ostream& out, const Alphabet& a)
 {
-  alphabet.c_str();
+  out << a.alphabet;
 }
 
 bool Alphabet::exists(const char c) const
 {
   return (alphabet_set.find(c) != alphabet_set.end());
+}
+
+const Alphabet& Alphabet::get_alphabet(AlphabetRef alpha)
+{
+  switch (alpha) {
+    case ::reduced_dna_alphabet:
+      return Alphabet::reduced_dna_alphabet();
+    case ::reduced_rna_alphabet:
+      return Alphabet::reduced_rna_alphabet();
+    case ::reduced_nucleic_alphabet:
+      return Alphabet::reduced_nucleic_alphabet();
+    case ::nucleic_alphabet:
+      return Alphabet::nucleic_alphabet();
+    case ::protein_alphabet:
+      return Alphabet::protein_alphabet();    
+    default:
+      throw std::runtime_error("unrecognized alphabet type");
+      break;
+  }
 }
\ No newline at end of file