Update sk1/compressor.c

sk1/compressor.c

@@ -1,21 +1,140 @@
 #include <stdio.h>
 
-int copy_file(const char *a, const char *b)
+#define WSIZE 250
+#define BSIZE 50
+
+typedef struct {
+    unsigned char pos;
+    unsigned char n;
+    char symb;
+} LZ77Token;
+
+int compress_1(const char* input_file, const char* output_file)
 {
-    FILE *x = fopen(a, "rb");  if (!x) return -1;
-    FILE *y = fopen(b, "wb");  if (!y) { fclose(x); return -1; }
-    int c, n = 0;
-    while ((c = fgetc(x)) != EOF)
+    size_t compr_size = 0, sz=0;
+    FILE* input = fopen(input_file, "rb");
+    FILE* output = fopen(output_file, "wb");
+
+    if (!input || !output)
     {      
-        fputc(c, y);
-        n++;
+        exit(-1);
     }
-    fclose(x);
-    fclose(y);
-    return n;
+
+    char window[WSIZE] = { 0 };
+    char buffer[BSIZE] = { 0 };
+
+    int w_start = 0;
+    int w_end = 0;
+
+    while (!feof(input))
+    {
+        size_t blen = fread(buffer, 1, BSIZE, input);
+
+        for (int bpos = 0; bpos < blen; ++bpos)
+        {
+            int best_pos = 0;
+            int best_n = 0;
+
+            for (int spos = w_start; spos < w_end; ++spos)
+            {
+                int mlength = 0;
+
+                while (mlength < blen - bpos &&
+                    mlength < WSIZE &&
+                    (spos + mlength) < w_end &&
+                    window[(spos + mlength) % WSIZE] == buffer[bpos + mlength])
+                {
+                    ++mlength;
+                }
+
+                if (mlength > best_n)
+                {
+                    best_pos = w_end - spos;
+                    best_n = mlength;
+                }
+            }
+
+            LZ77Token token;
+            token.pos = best_pos;
+            token.n = best_n;
+
+            if (bpos + best_n < blen)
+            {
+                token.symb = buffer[bpos + best_n];
+            }
+            else
+            {
+                token.symb = '\0';
+            }
+
+            sz=fwrite(&token, sizeof(LZ77Token), 1, output);
+            compr_size += sizeof(LZ77Token) * sz;
+
+            for (int j = 0; j <= best_n && bpos < blen; ++j)
+            {
+                if (bpos + j < blen)
+                {
+                    window[w_end % WSIZE] = buffer[bpos + j];
+                    w_end++;
+
+                    if (w_end - w_start > WSIZE)
+                    {
+                        w_start++;
+                    }
+                }
+            }
+            bpos += best_n;
+        }
+    }
+
+    fclose(input);
+    fclose(output);
+   
+    return compr_size;
+}
+
+int decompress_1(const char* input_file, const char* output_file)
+{
+    size_t compr_size = 0, sz = 0;
+    FILE* input = fopen(input_file, "rb");
+    FILE* output = fopen(output_file, "wb");
+
+    if (!input || !output)
+    {
+        exit(-1);
+    }
+
+    char window[WSIZE] = { 0 };
+    int window_end = 0;
+
+    LZ77Token token;
+    while (fread(&token, sizeof(LZ77Token), 1, input))
+    {
+        if (token.pos == 0 && token.n == 0 && token.symb == '\0')
+        {
+            continue;
+        }
+
+        int start_index = (window_end - token.pos + WSIZE) % WSIZE;
+
+        for (int i = 0; i < token.n; ++i)
+        {
+            char ch = window[(start_index + i) % WSIZE];            
+            compr_size += fwrite(&ch, 1, 1, output);
+            window[window_end % WSIZE] = ch;
+            window_end = (window_end + 1) % WSIZE;
+        }
+
+        if (token.symb != '\0')
+        {
+            compr_size += fwrite(&token.symb, 1, 1, output);
+            window[window_end % WSIZE] = token.symb;
+            window_end = (window_end + 1) % WSIZE;
+        }
+    }
+    fclose(input);
+    fclose(output);
+
+    return compr_size;
 }
 
-int compress_1(const char *a, const char *b){return copy_file(a,b);}
-int compress_2(const char *a, const char *b){return copy_file(a,b);}
-int decompress_1(const char *a, const char *b){return copy_file(a,b);}
-int decompress_2(const char *a, const char *b){return copy_file(a,b);}