updated pa4

git-svn-id: svn://anubis/gvsu@299 45c1a28c-8058-47b2-ae61-ca45b979098e

cs677/pa4/ElementSet.cc

@@ -10,3 +10,39 @@ ElementSet::ElementSet(char initial)
 {
     mySet.insert(initial);
 }
+
+ElementSet ElementSet::Union(const ElementSet & other)
+{
+    ElementSet es;
+    for (set<char>::iterator it = mySet.begin(); it != mySet.end(); it++)
+        es.mySet.insert(*it);
+    for (set<char>::iterator it = other.mySet.begin(); it != other.mySet.end(); it++)
+        es.mySet.insert(*it);
+    return es;
+}
+
+ElementSet ElementSet::Intersection(const ElementSet & other)
+{
+    ElementSet es;
+    for (set<char>::iterator it = mySet.begin(); it != mySet.end(); it++)
+    {
+        set<char>::iterator find_it = other.mySet.find(*it);
+        if (find_it != other.mySet.end())
+            es.mySet.insert(*it);
+    }
+    return es;
+}
+
+ostream & operator<<(std::ostream & out, const ElementSet & e)
+{
+    int size = e.mySet.size();
+    if (size > 1)
+        out << '{';
+    for (set<char>::iterator it = e.mySet.begin(); it != e.mySet.end(); it++)
+    {
+        out << *it;
+    }
+    if (size > 1)
+        out << '}';
+    return out;
+}

cs677/pa4/ElementSet.h

@@ -3,6 +3,7 @@
 #define ELEMENTSET_H ELEMENTSET_H
 
 #include <set>
+#include <iostream>
 
 class ElementSet
 {
@@ -12,6 +13,11 @@ private:
 public:
     ElementSet();
     ElementSet(char initial);
+    int size() { return mySet.size(); }
+    ElementSet Union(const ElementSet & other);
+    ElementSet Intersection(const ElementSet & other);
+
+    friend std::ostream & operator<<(std::ostream & out, const ElementSet & e);
 };
 
 #endif

cs677/pa4/Sequence.cc

@@ -9,7 +9,13 @@ Sequence::Sequence()
     myMutationCount = 0;
 }
 
-Sequence::Sequence(char * initializer)
+Sequence::Sequence(int size)
+{
+    myMutationCount = 0;
+    myElements = vector<ElementSet>(size);
+}
+
+Sequence::Sequence(const char * initializer)
 {
     myMutationCount = 0;
     int len = strlen(initializer);
@@ -19,3 +25,31 @@ Sequence::Sequence(char * initializer)
         myElements[i] = ElementSet(initializer[i]);
     }
 }
+
+Sequence Sequence::parent(const Sequence & other)
+{
+    int size = myElements.size();
+    Sequence s(size);
+    for (int i = 0; i < size; i++)
+    {
+        ElementSet intersection = myElements[i].Intersection(other.myElements[i]);
+        if (intersection.size() == 0)
+        {
+            s.myElements[i] = myElements[i].Union(other.myElements[i]);
+            s.myMutationCount++;
+        }
+        else
+        {
+            s.myElements[i] = intersection;
+        }
+    }
+    return s;
+}
+
+ostream & operator<<(ostream & out, const Sequence & s)
+{
+    int sz = s.myElements.size();
+    for (int i = 0; i < sz; i++)
+        out << s.myElements[i];
+    return out;
+}

cs677/pa4/Sequence.h

@@ -3,6 +3,7 @@
 #define SEQUENCE_H SEQUENCE_H
 
 #include <vector>
+#include <iostream>
 #include "ElementSet.h"
 
 class Sequence
@@ -13,7 +14,11 @@ private:
 
 public:
     Sequence();
-    Sequence(char * initializer);
+    Sequence(int size);
+    Sequence(const char * initializer);
+    Sequence parent(const Sequence & other);
+
+    friend std::ostream & operator<<(std::ostream & out, const Sequence & s);
 };
 
 #endif

cs677/pa4/maximum-parsimony.cc

@@ -1,18 +1,99 @@
 
 #include <iostream>
+#include <string>
 #include <sys/time.h>
 #include <mpi.h>
+#include <stdlib.h>         /* exit() */
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include "Sequence.h"
 using namespace std;
 
+void usage()
+{
+    cerr << "Usage: maximum-parsimony <sequence-file>" << endl;
+    exit(42);
+}
+
+string getsequence(int fd)
+{
+    string s;
+    char chr;
+    bool saw_name = false;
+    for (;;)
+    {
+        int bytes_read = read(fd, &chr, 1);
+        if (bytes_read < 1)
+        {
+            if (s == "")
+                break;
+            else
+            {
+                cerr << "Unexpected EOF in the middle of a line!" << endl;
+                exit(2);
+            }
+        }
+        else if (chr == '\n')
+        {
+            break;
+        }
+        else if (chr == ' ' || chr == '\t')
+        {
+            saw_name = true;
+        }
+        else
+        {
+            if (saw_name)
+                s += chr;
+        }
+    }
+    return s;
+}
+
 int main(int argc, char * argv[])
 {
     int my_rank;
     int world_size;          /* the number of processes */
+    char * filename = NULL;
 
     MPI_Init(&argc, &argv);
     MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);
     MPI_Comm_size(MPI_COMM_WORLD, &world_size);
 
+    for (int i = 0; i < argc; i++)
+    {
+        filename = argv[i];
+    }
+
+    if (filename == NULL)
+        usage();
+
+    int fd = open(filename, O_RDONLY);
+    if (fd < 0)
+    {
+        cerr << "Couldn't open '" << filename << "'!" << endl;
+        exit(1);
+    }
+
+    vector<Sequence *> sequences;
+    for (;;)
+    {
+        string seqstr = getsequence(fd);
+        if (seqstr == "")
+            break;
+        Sequence * seq = new Sequence(seqstr.c_str());
+        sequences.push_back(seq);
+    }
+
+    close(fd);
+
+    cout << "Found " << sequences.size() << " sequences:" << endl;
+    for (int i = 0, sz = sequences.size(); i < sz; i++)
+    {
+        cout << *sequences[i] << endl;
+    }
+
     MPI_Finalize();
 
     return 0;