From 70c293a07502f673d1db270e43526999ffe7c11e Mon Sep 17 00:00:00 2001
From: Yurii Chechur <yurii.chechur@student.tuke.sk>
Date: Thu, 26 Dec 2024 15:59:08 +0000
Subject: [PATCH] Update sk1/compressor.c

---
 sk1/compressor.c | 459 +++++++++++++++++++++--------------------------
 1 file changed, 209 insertions(+), 250 deletions(-)

diff --git a/sk1/compressor.c b/sk1/compressor.c
index ab9ed66..01ab31a 100644
--- a/sk1/compressor.c
+++ b/sk1/compressor.c
@@ -1,291 +1,250 @@
-#include "compressor.h"
 #include <stdio.h>
-#include <stdlib.h>
+#include <stdint.h>
 #include <string.h>
+#include <stdlib.h>
 
-// --- Алгоритм Run-Length Encoding (RLE) ---
+#define WINDOW_SIZE 4096
+#define LOOKAHEAD_BUFFER_SIZE 18
 
-int compress_2(const char* input_file_name, const char* output_file_name) {
-    FILE *infile = fopen(input_file_name, "rb");
-    if (!infile) {
+typedef struct {
+    uint16_t offset;
+    uint8_t length;
+    uint8_t next_char;
+} LZ77Triple;
+
+int lz77_compress(const char *input_filename, const char *output_filename) {
+    // Open the input file in binary read mode
+    FILE *input_file = fopen(input_filename, "rb");
+    if (!input_file) {
         perror("Error opening input file");
         return -1;
     }
 
-    FILE *outfile = fopen(output_file_name, "wb");
-    if (!outfile) {
+    // Open the output file in binary write mode
+    FILE *output_file = fopen(output_filename, "wb");
+    if (!output_file) {
         perror("Error opening output file");
-        fclose(infile);
-        return -1;
+        fclose(input_file);
+        return -2;
     }
 
-    unsigned char current_byte, previous_byte;
-    size_t count = 0;
-
-    if (fread(&previous_byte, 1, 1, infile) != 1) {
-        fclose(infile);
-        fclose(outfile);
-        return 0;  // Порожній файл
+    uint8_t *window = (uint8_t *)malloc(WINDOW_SIZE + LOOKAHEAD_BUFFER_SIZE);
+    if (!window) {
+        perror("Memory allocation failed");
+        fclose(input_file);
+        fclose(output_file);
+        return -3;
     }
 
-    count = 1;
+    size_t window_start = 0;
+    size_t lookahead_start = 0;
+    size_t bytes_read;
 
-    while (fread(&current_byte, 1, 1, infile) == 1) {
+    // Initialize the window with data from the input file
+    bytes_read = fread(window + WINDOW_SIZE, 1, LOOKAHEAD_BUFFER_SIZE, input_file);
+
+    while (bytes_read > 0) {
+        size_t best_match_offset = 0;
+        size_t best_match_length = 0;
+
+        // Search for the best match within the sliding window
+        for (size_t i = window_start; i < WINDOW_SIZE + lookahead_start; i++) {
+            size_t match_length = 0;
+
+            while (match_length < bytes_read &&
+                   window[i + match_length] == window[WINDOW_SIZE + match_length]) {
+                match_length++;
+                if (match_length >= LOOKAHEAD_BUFFER_SIZE) {
+                    break;
+                }
+            }
+
+            if (match_length > best_match_length) {
+                best_match_length = match_length;
+                best_match_offset = WINDOW_SIZE + lookahead_start - i;
+            }
+        }
+
+        // Create a triple and write it to the output file
+        LZ77Triple triple;
+        triple.offset = (uint16_t)best_match_offset;
+        triple.length = (uint8_t)best_match_length;
+        triple.next_char = window[WINDOW_SIZE + best_match_length];
+
+        // Write the triple to the output file
+        fwrite(&triple, sizeof(LZ77Triple), 1, output_file);
+
+        // Slide the window
+        window_start = (window_start + best_match_length + 1) % WINDOW_SIZE;
+        lookahead_start = (lookahead_start + best_match_length + 1) % LOOKAHEAD_BUFFER_SIZE;
+
+        // Read new byte into the lookahead buffer
+        bytes_read = fread(window + WINDOW_SIZE, 1, LOOKAHEAD_BUFFER_SIZE - lookahead_start, input_file);
+    }
+
+    // Cleanup and close files
+    fclose(input_file);
+    fclose(output_file);
+    free(window);
+
+    return 0;
+}
+
+int lz77_decompress(const char *input_filename, const char *output_filename) {
+    FILE *input = fopen(input_filename, "rb");
+    FILE *output = fopen(output_filename, "wb");
+    if (!input || !output) {
+        if (input) fclose(input);
+        if (output) fclose(output);
+        return -1; // Помилка відкриття файлу
+    }
+
+    size_t buffer_size = 4096; // Максимальний розмір вікна
+    unsigned char *window = malloc(buffer_size);
+    size_t window_size = 0; // Розмір заповненої частини вікна
+    size_t window_pos = 0;  // Поточна позиція в межах вікна
+
+    if (!window) {
+        fclose(input);
+        fclose(output);
+        return -1; // Помилка пам'яті
+    }
+
+    while (!feof(input)) {
+        unsigned char flag;
+        if (fread(&flag, 1, 1, input) != 1) break;
+
+        for (int i = 0; i < 8 && !feof(input); i++) {
+            if (flag & (1 << i)) { // Літеральний символ
+                unsigned char literal;
+                if (fread(&literal, 1, 1, input) != 1) break;
+
+                fputc(literal, output);
+
+                // Додати символ у вікно
+                window[window_pos] = literal;
+                window_pos = (window_pos + 1) % buffer_size;
+                if (window_size < buffer_size) window_size++;
+            } else { // Посилання
+                unsigned short offset_length;
+                if (fread(&offset_length, 2, 1, input) != 1) break;
+
+                size_t offset = offset_length >> 4;
+                size_t length = (offset_length & 0xF) + 3;
+
+                for (size_t j = 0; j < length; j++) {
+                    unsigned char byte = window[(window_pos - offset + buffer_size) % buffer_size];
+                    fputc(byte, output);
+
+                    // Додати байт у вікно
+                    window[window_pos] = byte;
+                    window_pos = (window_pos + 1) % buffer_size;
+                    if (window_size < buffer_size) window_size++;
+                }
+            }
+        }
+    }
+
+    free(window);
+    fclose(input);
+    fclose(output);
+    return 0; // Успішна декомпресія
+}
+
+
+
+
+void rle_compress(const char *input_filename, const char *output_filename) {
+    FILE *input_file = fopen(input_filename, "rb");
+    if (!input_file) {
+        perror("Error opening input file");
+        return;
+    }
+
+    FILE *output_file = fopen(output_filename, "wb");
+    if (!output_file) {
+        perror("Error opening output file");
+        fclose(input_file);
+        return;
+    }
+
+    uint8_t current_byte, previous_byte;
+    uint8_t count = 1;
+
+    if (fread(&previous_byte, 1, 1, input_file) != 1) {
+        printf("Input file is empty or read error occurred.\n");
+        fclose(input_file);
+        fclose(output_file);
+        return;
+    }
+
+    while (fread(&current_byte, 1, 1, input_file) == 1) {
         if (current_byte == previous_byte && count < 255) {
             count++;
         } else {
-            fwrite(&previous_byte, 1, 1, outfile);
-            fwrite(&count, 1, 1, outfile);
+            fwrite(&previous_byte, 1, 1, output_file);
+            fwrite(&count, 1, 1, output_file);
+            printf("Writing byte: %c with count: %d\n", previous_byte, count);
             previous_byte = current_byte;
             count = 1;
         }
     }
 
-    fwrite(&previous_byte, 1, 1, outfile);
-    fwrite(&count, 1, 1, outfile);
+    fwrite(&previous_byte, 1, 1, output_file);
+    fwrite(&count, 1, 1, output_file);
+    printf("Writing byte: %c with count: %d\n", previous_byte, count);
 
-    fclose(infile);
-    fclose(outfile);
-
-    return 1;
+    fclose(input_file);
+    fclose(output_file);
 }
 
-int decompress_2(const char* input_file_name, const char* output_file_name) {
-    FILE *infile = fopen(input_file_name, "rb");
-    if (!infile) {
+
+
+int rle_decompress(const char *input_filename, const char *output_filename) {
+    // Open the input file in binary read mode
+    FILE *input_file = fopen(input_filename, "rb");
+    if (!input_file) {
         perror("Error opening input file");
         return -1;
     }
 
-    FILE *outfile = fopen(output_file_name, "wb");
-    if (!outfile) {
+    // Open the output file in binary write mode
+    FILE *output_file = fopen(output_filename, "wb");
+    if (!output_file) {
         perror("Error opening output file");
-        fclose(infile);
-        return -1;
+        fclose(input_file);
+        return -2;
     }
 
-    unsigned char current_byte;
-    unsigned char count;
+    uint8_t byte;
+    uint8_t count;
+    size_t decompressed_size = 0;
 
-    while (fread(&current_byte, 1, 1, infile) == 1) {
-        if (fread(&count, 1, 1, infile) != 1) {
-            perror("Malformed input file");
-            fclose(infile);
-            fclose(outfile);
-            return -1;
+    // Read [byte, count] pairs from the input file
+    while (fread(&byte, 1, 1, input_file) == 1) {
+        if (fread(&count, 1, 1, input_file) != 1) {
+            // Handle malformed input file
+            fprintf(stderr, "Error: Malformed input file\n");
+            fclose(input_file);
+            fclose(output_file);
+            return -3;
         }
 
-        for (size_t i = 0; i < count; i++) {
-            fwrite(&current_byte, 1, 1, outfile);
-        }
-    }
-
-    fclose(infile);
-    fclose(outfile);
-
-    return 1;
-}
-
-// --- Алгоритм Хаффмана (Huffman Coding) ---
-
-typedef struct Node {
-    unsigned char symbol;
-    size_t frequency;
-    struct Node *left, *right;
-} Node;
-
-typedef struct {
-    unsigned char symbol;
-    char *code;
-} HuffmanCode;
-
-int compare_nodes(const void *a, const void *b) {
-    return (*(Node**)a)->frequency - (*(Node**)b)->frequency;
-}
-
-Node* create_huffman_tree(unsigned char *data, size_t size) {
-    size_t freq[256] = {0};
-    for (size_t i = 0; i < size; i++) {
-        freq[data[i]]++;
-    }
-
-    Node *nodes[256];
-    size_t node_count = 0;
-
-    for (int i = 0; i < 256; i++) {
-        if (freq[i] > 0) {
-            nodes[node_count] = malloc(sizeof(Node));
-            if (!nodes[node_count]) {
-                perror("Memory allocation failed");
-                return NULL;
+        // Write 'count' occurrences of 'byte' to the output file
+        for (uint8_t i = 0; i < count; i++) {
+            if (fwrite(&byte, 1, 1, output_file) != 1) {
+                perror("Error writing to output file");
+                fclose(input_file);
+                fclose(output_file);
+                return -4;
             }
-            nodes[node_count]->symbol = (unsigned char)i;
-            nodes[node_count]->frequency = freq[i];
-            nodes[node_count]->left = nodes[node_count]->right = NULL;
-            node_count++;
+            decompressed_size++;
         }
     }
 
-    while (node_count > 1) {
-        qsort(nodes, node_count, sizeof(Node*), compare_nodes);
+    // Clean up and close files
+    fclose(input_file);
+    fclose(output_file);
 
-        Node* left = nodes[0];
-        Node* right = nodes[1];
-
-        Node* parent = malloc(sizeof(Node));
-        if (!parent) {
-            perror("Memory allocation failed");
-            return NULL;
-        }
-
-        parent->symbol = 0;
-        parent->frequency = left->frequency + right->frequency;
-        parent->left = left;
-        parent->right = right;
-
-        memmove(nodes, nodes + 2, (node_count - 2) * sizeof(Node*));
-        nodes[node_count - 2] = parent;
-        node_count--;
-    }
-
-    return nodes[0];
+    return (int)decompressed_size;
 }
-
-void generate_huffman_codes(Node* root, HuffmanCode* codes, char* current_code, int depth) {
-    if (!root) return;
-
-    if (root->left == NULL && root->right == NULL) {
-        current_code[depth] = '\0';
-        codes[root->symbol].symbol = root->symbol;
-        codes[root->symbol].code = strdup(current_code);
-        return;
-    }
-
-    current_code[depth] = '0';
-    generate_huffman_codes(root->left, codes, current_code, depth + 1);
-
-    current_code[depth] = '1';
-    generate_huffman_codes(root->right, codes, current_code, depth + 1);
-}
-
-void serialize_huffman_tree(Node* root, FILE* outfile) {
-    if (!root) return;
-
-    if (root->left == NULL && root->right == NULL) {
-        fputc('L', outfile);
-        fputc(root->symbol, outfile);
-    } else {
-        fputc('I', outfile);
-        serialize_huffman_tree(root->left, outfile);
-        serialize_huffman_tree(root->right, outfile);
-    }
-}
-
-Node* rebuild_huffman_tree(unsigned char* tree_data, size_t size) {
-    (void)size;  // Позначаємо параметр як тимчасово невикористаний
-    size_t index = 0;
-
-    Node* build_tree_recursively(unsigned char* data, size_t* index) {
-        if (data[*index] == 'L') { // Лист (Leaf)
-            (*index)++;
-            Node* leaf = malloc(sizeof(Node));
-            if (!leaf) {
-                perror("Memory allocation failed");
-                return NULL;
-            }
-            leaf->symbol = data[*index];
-            leaf->frequency = 0; // частота не потрібна для декомпресії
-            leaf->left = leaf->right = NULL;
-            (*index)++;
-            return leaf;
-        } else if (data[*index] == 'I') { // Вузол (Internal)
-            (*index)++;
-            Node* internal = malloc(sizeof(Node));
-            if (!internal) {
-                perror("Memory allocation failed");
-                return NULL;
-            }
-            internal->symbol = 0; // внутрішні вузли не мають символів
-            internal->frequency = 0;
-            internal->left = build_tree_recursively(data, index);
-            internal->right = build_tree_recursively(data, index);
-            return internal;
-        }
-        return NULL;
-    }
-
-    return build_tree_recursively(tree_data, &index);
-}
-
-void free_huffman_tree(Node* root) {
-    if (!root) return;
-    free_huffman_tree(root->left);
-    free_huffman_tree(root->right);
-    free(root);
-}
-
-int compress_1(const char* input_file_name, const char* output_file_name) {
-    FILE *infile = fopen(input_file_name, "rb");
-    if (!infile) {
-        perror("Error opening input file");
-        return -1;
-    }
-
-    fseek(infile, 0, SEEK_END);
-    size_t file_size = ftell(infile);
-    fseek(infile, 0, SEEK_SET);
-
-    unsigned char* data = malloc(file_size);
-    if (!data) {
-        fclose(infile);
-        perror("Memory allocation failed");
-        return -1;
-    }
-
-    fread(data, 1, file_size, infile);
-    fclose(infile);
-
-    Node* root = create_huffman_tree(data, file_size);
-    if (!root) {
-        free(data);
-        return -1;
-    }
-
-    HuffmanCode codes[256] = {0};
-    char current_code[256];
-    generate_huffman_codes(root, codes, current_code, 0);
-
-    FILE *outfile = fopen(output_file_name, "wb");
-    if (!outfile) {
-        free(data);
-        free_huffman_tree(root);
-        return -1;
-    }
-
-    serialize_huffman_tree(root, outfile);
-
-    for (size_t i = 0; i < file_size; i++) {
-        const char* code = codes[data[i]].code;
-        for (size_t j = 0; code[j] != '\0'; j++) {
-            fputc(code[j], outfile);
-        }
-    }
-
-    fclose(outfile);
-    free(data);
-    free_huffman_tree(root);
-
-    for (int i = 0; i < 256; i++) {
-        free(codes[i].code);
-    }
-
-    return 1;
-}
-
-// Декомпресія (приклад потребує уточнень залежно від специфіки формату)
-int decompress_1(const char* input_file_name, const char* output_file_name) {
-    (void)input_file_name;
-    (void)output_file_name;
-    return -1;  // Поки що не реалізовано.
-}
-