Add CSC format code

brucehow · Sep 10, 2019 · 389e269 · 389e269
1 parent 0506ad1
commit 389e269
Show file tree

Hide file tree

Showing 3 changed files with 116 additions and 8 deletions.
diff --git a/format.c b/format.c
@@ -151,4 +151,98 @@ struct CSR csr_format(int rows, int cols, enum mat_type type, char *data) {
         matrix.ia[i+1] = matrix.count;
     }
     return matrix;
+}
+
+struct CSC csc_format(int rows, int cols, enum mat_type type, char *data) {
+    // Structure variable initialisation
+    struct CSC matrix;
+    size_t nnz_size;
+    size_t ja_size = MEMSIZ * sizeof(int);
+    if (type == INT_MAT) {
+        nnz_size = MEMSIZ * sizeof(int);
+        matrix.nnz.i = allocate(nnz_size);
+    }  else {
+        nnz_size = MEMSIZ * sizeof(float);
+        matrix.nnz.f = allocate(nnz_size);
+    }
+    matrix.ja = allocate(ja_size);
+    matrix.ia = allocate(sizeof(int) * (rows+1));
+    matrix.ia[0] = 0; // Conventional
+    matrix.cols = cols;
+    matrix.count = 0;
+
+    // Data reading variables
+    int len, pos = 0;
+    int skip = 0; // Use to read data column down
+    size_t size = MEMSIZ;
+    char *val = allocate(size);
+
+    for (int i = 0; i < cols; i++) {
+        // Figure out where to start col wise
+        skip = 0;
+        pos = 0;
+        while (skip != i) {
+            if (data[pos] == ' ') {
+                    skip++;
+                }
+                pos++;
+        }
+        skip = 0;
+        for (int j = 0; j < rows; j++) {
+            len = 0; // Reset word length to 0
+
+            while (data[pos] != '\0' && data[pos] != ' ') {
+                val[len++] = data[pos++];
+
+                // Dynamic memory reallocation for each digit
+                if ((len * sizeof(char)) == size) {
+                    if (data[pos] != '\0' && data[pos] != ' ') {
+                        size += sizeof(char); // Null byte
+                    } else {
+                        size *= 2;
+                    }
+                    val = reallocate(val, size);
+                }
+            }
+            val[len] = '\0';
+            while (data[pos] != '\0' && skip != cols) { // Iterate column 
+                if (data[pos] == ' ') {
+                    skip++;
+                }
+                pos++;
+            }
+            skip = 0;
+            check_numeric(val);
+
+            // Zero value filter
+            if (type == INT_MAT) {
+                int value = atoi(val);
+                if (value != 0) {
+                    // Dynamically allocate memory for nnz and ja pointers
+                    if ((matrix.count * sizeof(int)) == ja_size) {
+                        nnz_size *= 2;
+                        ja_size *= 2;
+                        matrix.nnz.i = reallocate(matrix.nnz.i, nnz_size);
+                        matrix.ja = reallocate(matrix.ja, ja_size);
+                    }
+                    matrix.nnz.i[matrix.count] = value;
+                    matrix.ja[matrix.count++] = j;
+                }
+            } else {
+                float value = atof(val);
+                if (value != 0.0) {
+                    if ((matrix.count * sizeof(int)) == ja_size) {
+                        nnz_size *= 2;
+                        ja_size *= 2;
+                        matrix.nnz.f = reallocate(matrix.nnz.f, nnz_size);
+                        matrix.ja = reallocate(matrix.ja, ja_size);
+                    }
+                    matrix.nnz.f[matrix.count] = value;
+                    matrix.ja[matrix.count++] = j;
+                }
+            }  
+        }
+        matrix.ia[i+1] = matrix.count;
+    }
+    return matrix;
 }
diff --git a/matrix.c b/matrix.c
@@ -20,7 +20,7 @@ int main(int argc, char *argv[]) {
     data = read_line(fp);
 
     // Matrix processing
-    struct CSR matrix = csr_format(rows, cols, type, data);
+    struct CSC matrix = csc_format(rows, cols, type, data);
 
     /////////////// DEBUG //////////////////
     int limit = matrix.count;
@@ -29,7 +29,7 @@ int main(int argc, char *argv[]) {
         printf("%d,", matrix.nnz.i[i]);
     }
     printf("\nia=");
-    for (int i = 0; i < limit+1; i++) {
+    for (int i = 0; i < cols+1; i++) {
         printf("%d,", matrix.ia[i]);
     }
     printf("\nja=");

diff --git a/matrix.h b/matrix.h
@@ -17,7 +17,6 @@
 
 #define MEMSIZ 8
 
-
 // Matrix type classifier
 enum mat_type {INT_MAT, FLOAT_MAT};
 
@@ -48,6 +47,8 @@ extern struct COO coo_format(int rows, int cols, enum mat_type type, char *data)
 
 extern struct CSR csr_format(int rows, int cols, enum mat_type type, char *data);
 
+extern struct CSC csc_format(int rows, int cols, enum mat_type type, char *data);
+
 /**
  * Allocates memory of a given size using malloc
  * 
@@ -66,7 +67,7 @@ void *allocate(size_t size);
 void *reallocate(void *ptr, size_t size);
 
 
-// COO representation structs
+// COO representation
 struct COO {
     enum mat_type type;
     int count;
@@ -80,15 +81,28 @@ struct COO {
     } *elements;
 };
 
-// CSR representation structs
+// CSR representation
 struct CSR {
     enum mat_type type;
-    int rows;
-    int count; // Number of nrz values
+    int rows; // Used for IA length + 1
+    int count; // Number of nnz values
     union {
         int *i;
         float *f;
     } nnz; // List of non-zero values
     int *ia; // Total number of elements up until specific row
-    int *ja; // List of column index for each nrz value
+    int *ja; // List of column index for each nnz value
+};
+
+// CSC representation
+struct CSC {
+    enum mat_type type;
+    int cols; // Used for IA length + 1
+    int count; // Number of nnz values
+    union {
+        int *i;
+        float *f;
+    } nnz; // List of non-zero values
+    int *ia; // Total number of elements up until specific col
+    int *ja; // List of row index for each nnz value
 };