优化huffman.c

.gitignore

@@ -1,5 +1,5 @@
 
 .vs/
 *.exe
-
+*.pkt
 

hello.c

@@ -2,8 +2,9 @@
 #include "stdio.h"
 #include "string.h"
 // #include "coder_lib.h"
-#include "huffman.h"
+// #include "huffman.h"
-
+#include "huffman_.h"
+#include <stdlib.h>
 
 
 // 验证管壳码算法
@@ -30,17 +31,74 @@
 
 // 验证huffman算法
 
+// int main(int argc, char *argv[])
+// {
+//   uint8_t str_in[]="2023 5830628A000005830628A000015830628A000025830628A000035830628A000045830628A000055830628A000065830628A000075830628A000085830628A00009";
+//   array_def *a=arr_creat();
+//   arr_appends(a,str_in,strlen(str_in));
+//   array_def *out=hm_encode(a);
+//   printf("endode:%s\n",arr_string(out));
+//   array_def *de=hm_decode(out);
+//   printf("decde:%s\n",arr_data(de));
+//   return 0;
+// }
+
+
+
+
+long calc_file_size(FILE *stream)
+{
+	long file_size = -1;
+	long cur_offset = ftell(stream);	// 获取当前偏移位置
+	if (cur_offset == -1) {
+		printf("ftell failed :%s\n", strerror(errno));
+		return -1;
+	}
+	if (fseek(stream, 0, SEEK_END) != 0) {	// 移动文件指针到文件末尾
+		printf("fseek failed: %s\n", strerror(errno));
+		return -1;
+	}
+	file_size = ftell(stream);	// 获取此时偏移值，即文件大小
+	if (file_size == -1) {
+		printf("ftell failed :%s\n", strerror(errno));
+	}
+	if (fseek(stream, cur_offset, SEEK_SET) != 0) {	// 将文件指针恢复初始位置
+		printf("fseek failed: %s\n", strerror(errno));
+		return -1;
+	}
+	return file_size;
+}
+
+
+
+int encode_file(const char *filename)
+{
+  FILE *f=fopen(filename, "rb");
+  int file_size=calc_file_size(f);
+  uint8_t *file_data=calloc(file_size+1,sizeof(uint8_t));
+  uint8_t *encode_data=0;
+  int encode_size;
+  fread(file_data,1,file_size,f);
+  hm_encode(file_data,file_size,&encode_data,&encode_size);
+  free(file_data);
+  free(encode_data);
+  fclose(f);
+}
+
+
+
+
+
 int main(int argc, char *argv[])
 {
-  uint8_t str_in[]="2023 5830628A000005830628A000015830628A000025830628A000035830628A000045830628A000055830628A000065830628A000075830628A000085830628A00009";
+  encode_file(argv[1]);
-  array_def *a=arr_creat();
+  // const uint8_t file_data[]="2023 5830628A000005830628A000015830628A000025830628A000035830628A000045830628A000055830628A000065830628A000075830628A000085830628A00009";
-  arr_appends(a,str_in,strlen(str_in));
+  // uint8_t *encode_data=0;
-  array_def *out=hm_encode(a);
+  // int encode_size;
-  printf("endode:%s\n",arr_string(out));
+  // hm_encode(file_data,sizeof(file_data),&encode_data,&encode_size);
-  array_def *de=hm_decode(out);
-  printf("decde:%s\n",arr_data(de));
   return 0;
 }
 
 
 
+

huffman_.c

@@ -3,7 +3,7 @@
 #include "stdio.h"
 #include "string.h"
 #include "huffman_.h"
-
+#include <stdlib.h>
 // huffman编码的实现
 
 #define DBG_WARN printf
@@ -12,7 +12,7 @@
 typedef struct _huff_tree{
   uint8_t data;
   uint8_t pos;// 位置，左为1，右为0
-  uint16_t count;
+  uint32_t count;
   struct _huff_tree *parant;
   struct _huff_tree *left;
   struct _huff_tree *right;
@@ -22,21 +22,41 @@ typedef struct _huff_tree{
 
 typedef struct{
   huff_tree *tree;
-  int index_table_index;
+  uint32_t index_table_index;
   huff_tree *index_table[256];
-  uint16_t count_table[256];
+  uint32_t count_table[256];
   uint8_t *out;
-  int out_len;
+  uint32_t out_len;
-  uint8_t *in;
+  const uint8_t *in;
-  int in_len;
+  uint32_t in_len;
-  int in_bit_count;
+  uint32_t in_bit_count;
-  int arr_bit_index;
+  uint32_t arr_bit_index;
+  /* 以下成员调试时使用 */
+  uint32_t tree_point_num;// 使用的树节点个数
 }huffman_def;
 
 
 static int hm_calc_value_of_tree(huff_tree *t);
 static int hm_calc_deep_of_child(huff_tree* t);
 
+
+
+// 生成一个树节点
+static huff_tree *hm_creat_tree_point(huffman_def *h)
+{
+  h->tree_point_num++;
+  return calloc(1,sizeof(huff_tree));
+}
+
+// 删除一个树节点
+static void hm_del_tree_point(huffman_def *h,huff_tree *t)
+{
+  if(h->tree_point_num>0){
+    h->tree_point_num--;
+    free(t);
+  }
+}
+
 // 按出现频次排序
 static void hm_sort_index_table(huff_tree **table,int num)
 {
@@ -98,7 +118,7 @@ static void hm_calc_count(huffman_def *h,const uint8_t *d,const int d_len)
   {
     if(h->count_table[i]>0){
       index=h->index_table_index;
-      h->index_table[index]=calloc(1,sizeof(huff_tree));
+      h->index_table[index]=hm_creat_tree_point(h);
       h->index_table[index]->count=h->count_table[i];
       h->index_table[index]->data=i;
       h->index_table_index++;
@@ -188,7 +208,7 @@ static void hm_creat_tree(huffman_def *h)
     sub1=table[tail-1];
     sub2=table[tail-2];
     // 大在左，小在右
-    temp=calloc(1,sizeof(huff_tree));
+    temp=hm_creat_tree_point(h);
     sub1->parant=temp;
     sub2->parant=temp;
     // 左为1，右为0
@@ -217,13 +237,13 @@ static void hm_creat_tree(huffman_def *h)
 
 
 // 删除树
-static void hm_del_tree(huff_tree *t)
+static void hm_del_tree(huffman_def *h,huff_tree *t)
 {
   if(t->left&&t->right){
-    hm_del_tree(t->left);
+    hm_del_tree(h,t->left);
-    hm_del_tree(t->right);
+    hm_del_tree(h,t->right);
   }
-  free(t);
+  hm_del_tree_point(h,t);
 }
 
 // 数据中添加一个bit
@@ -256,14 +276,10 @@ static int hm_encode_byte(huffman_def *h,uint8_t d)
     DBG_WARN("can not encode.\n");
     exit(-1);
   }
-
   while(t->parant){
     hm_add_bit(h->out,&h->out_len,t->pos,&h->arr_bit_index);
     t=t->parant;
   }
-  // char *str=arr_string(h->out);
-  // DBG_LOG("index:%d,out data:%s\n",h->arr_bit_index,str);
-  // free(str);
   return 0;
 }
 
@@ -274,7 +290,7 @@ static int hm_creat_index_table(huffman_def *h,uint8_t *data,int *data_len)
   int diff;
   int temp_num;
   data[*data_len] = h->index_table_index; (*data_len)++;
-  hm_index_table_print(h);
+  // hm_index_table_print(h);
   for(int i=0;i<h->index_table_index;i++)
   {
     data[*data_len] = h->index_table[i]->data; (*data_len)++;
@@ -313,6 +329,7 @@ int hm_encode(const uint8_t* in, const int in_len, uint8_t** out, int* out_len)
   huffman_def *h=calloc(1,sizeof(huffman_def));
   hm_calc_count(h,in, input_len);
   hm_creat_tree(h);
+  DBG_LOG("huffman tree point num:%d\n",h->tree_point_num);
   output_len = hm_calc_encode_len(h);
   (*out) = calloc(output_len + 1, sizeof(uint8_t));
   hm_creat_index_table(h, *out, &output_index);
@@ -326,7 +343,8 @@ int hm_encode(const uint8_t* in, const int in_len, uint8_t** out, int* out_len)
   (*out)[output_index-1] = h->out_len*8- h->arr_bit_index;
   DBG_LOG("fill with 0 by:%d\n", (*out)[output_index - 1]);
   (*out_len) = output_len;
-  hm_del_tree(h->tree);
+  hm_del_tree(h,h->tree);
+  DBG_LOG("after del tree point num:%d\n",h->tree_point_num);
 
   DBG_LOG("lenth_in:%d,length_encode:%d\n",input_len, output_len);
   free(h);
@@ -342,7 +360,7 @@ static int hm_unpack_count(huffman_def *h,const uint8_t *d,int d_len)
   uint8_t temp;
   for(int i=0;i<num;i++)
   {
-    h->index_table[i]=calloc(1,sizeof(huff_tree));
+    h->index_table[i]=hm_creat_tree_point(h);
     h->index_table[i]->data=d[index];index++;
     do{
       temp= d[index];index++;
@@ -437,6 +455,7 @@ int hm_decode(const uint8_t* in, const int in_len, uint8_t** out, int* out_len)
   hm_creat_tree(h);
   // hm_data_code_print(h);
   // hm_tree_print(h->tree);
+  DBG_LOG("huffman tree point num:%d\n",h->tree_point_num);
   decode_len=hm_calc_decode_len(h);
   decode_index=decode_len;
   decode_data=calloc(decode_len+1,sizeof(uint8_t));
@@ -446,7 +465,8 @@ int hm_decode(const uint8_t* in, const int in_len, uint8_t** out, int* out_len)
     decode_data[decode_index-1]=c;
     decode_index--;
   }
-  hm_del_tree(h->tree);
+  hm_del_tree(h,h->tree);
+  DBG_LOG("after del tree point num:%d\n",h->tree_point_num);
   free(h);
   (*out) = decode_data;
   (*out_len) = decode_len;