sync (#22)

* fix typo

* fix: 32bit be decoding

* chore: add Rust alternative (#18)

* Update C# Link (#20)

* 尝试修复#19

- 移除未使用的参数 int blen
- 完善文档、注释
- 完善文件名错误检测
- 修复 encbuf, decbuf 处理逻辑

* fix: disable mmap in windows

* fix: encode bufsz (#19)

* v2.2.5

* Update Swift Link (#21)

---------

Co-authored-by: 忘忧北萱草 <wybxc@qq.com>
Co-authored-by: LC <64722907+lc6464@users.noreply.github.com>
Co-authored-by: oboard <oboard@outlook.com>

CMakeLists.txt

@@ -2,7 +2,7 @@ cmake_minimum_required(VERSION 2.8.12)
 if (POLICY CMP0048)
   cmake_policy(SET CMP0048 NEW)
 endif (POLICY CMP0048)
-project(base16384 VERSION 2.2.3)
+project(base16384 VERSION 2.2.5)
 
 add_executable(base16384_b base16384.c)
 

README.md

@@ -1,7 +1,7 @@
 # base16384
 > **Note**: This project used the awesome cross-platform binary compiling tool [cosmopolitan](https://github.com/jart/cosmopolitan)
 
-> Alternatives: [Go](https://github.com/fumiama/go-base16384), [Python](https://github.com/synodriver/pybase16384), [Android](https://github.com/fumiama/android-base16384), [TypeScript](https://github.com/shigma/base16384.js), [Lua(binding)](https://github.com/synodriver/lua-base16384), [Lua(pure)](https://github.com/Yiwen-Chan/base16384), [C#](https://github.com/lc6464/Base16384Coder-Console)
+> Alternatives: [Go](https://github.com/fumiama/go-base16384), [Python](https://github.com/synodriver/pybase16384), [Android](https://github.com/fumiama/android-base16384), [TypeScript](https://github.com/shigma/base16384.js), [Lua(binding)](https://github.com/synodriver/lua-base16384), [Lua(pure)](https://github.com/Yiwen-Chan/base16384), [C#](https://github.com/lc6464/Base16384.Net), [Rust](https://github.com/Wybxc/base16384-rs), [Swift](https://github.com/oboard/swift-ui-base16384)
 
 Encode binary file to printable utf16be, and vice versa.
 
@@ -86,7 +86,7 @@ echo -n "1234567" | base16384 -e - - | iconv -f utf-16be -t utf-8
 3. Decode simple text 简单文本解码
 
 ```bash
-echo -n "婌焳廔萷" | iconv -f utf-8 -t utf-16be | ./base16384 -d - -
+echo -n "婌焳廔萷" | iconv -f utf-8 -t utf-16be | base16384 -d - -
 1234567
 ```
 

base1432.c

@@ -1,6 +1,6 @@
 /* base1432.c
  * This file is part of the base16384 distribution (https://github.com/fumiama/base16384).
- * Copyright (c) 2022 Fumiama Minamoto.
+ * Copyright (c) 2022-2023 Fumiama Minamoto.
  * 
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -62,7 +62,7 @@
 
 // #define DEBUG
 
-int base16384_encode(const char* data, int dlen, char* buf, int blen) {
+int base16384_encode(const char* data, int dlen, char* buf) {
 	int outlen = dlen / 7 * 8;
 	int offset = dlen % 7;
 	switch(offset) {	// 算上偏移标志字符占用的2字节
@@ -140,7 +140,7 @@ int base16384_encode(const char* data, int dlen, char* buf, int blen) {
 	return outlen;
 }
 
-int base16384_decode(const char* data, int dlen, char* buf, int blen) {
+int base16384_decode(const char* data, int dlen, char* buf) {
 	int outlen = dlen;
 	int offset = 0;
 	if(data[dlen-2] == '=') {
@@ -194,7 +194,11 @@ int base16384_decode(const char* data, int dlen, char* buf, int blen) {
 				if(offset--) {
 					buf[i] = (sum & 0x0f000000) >> 20;
 					// 这里有读取越界
-					sum = vals[n];
+					#ifdef WORDS_BIGENDIAN
+						sum = __builtin_bswap32(vals[n]);
+					#else
+						sum = vals[n];
+					#endif
 					sum -= 0x0000004e;
 					buf[i++] |= (sum & 0x0000003c) >> 2;
 					if(offset--) {

base1464.c

@@ -1,6 +1,6 @@
 /* base1464.c
  * This file is part of the base16384 distribution (https://github.com/fumiama/base16384).
- * Copyright (c) 2022 Fumiama Minamoto.
+ * Copyright (c) 2022-2023 Fumiama Minamoto.
  * 
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -62,7 +62,7 @@
 
 // #define DEBUG
 
-int base16384_encode(const char* data, int dlen, char* buf, int blen) {
+int base16384_encode(const char* data, int dlen, char* buf) {
 	int outlen = dlen / 7 * 8;
 	int offset = dlen % 7;
 	switch(offset) {	// 算上偏移标志字符占用的2字节
@@ -136,7 +136,7 @@ int base16384_encode(const char* data, int dlen, char* buf, int blen) {
 	return outlen;
 }
 
-int base16384_decode(const char* data, int dlen, char* buf, int blen) {
+int base16384_decode(const char* data, int dlen, char* buf) {
 	int outlen = dlen;
 	int offset = 0;
 	if(data[dlen-2] == '=') {

base16384.1

@@ -1,9 +1,9 @@
-.TH BASE16384 1 "27 April 2022" "GNU" "User Commands"
+.TH BASE16384 1 "26 August 2023" "GNU" "User Commands"
 .SH NAME
 base16384 \- Encode binary files to printable utf16be
 .SH SYNOPSIS
 .B base16384
--[e|d] <\fIinputfile\fR> <\fIoutputfile\fR>
+-[e|d|t] <\fIinputfile\fR> <\fIoutputfile\fR>
 .SH DESCRIPTION
 .LP
 There are
@@ -29,6 +29,12 @@ to
 \fB\-e\fR
 Read data from \fIinputfile\fR and encode them into \fIoutputfile\fR.
 .TP 0.5i
+\fB\-d\fR
+Read data from \fIinputfile\fR and decode them into \fIoutputfile\fR.
+.TP 0.5i
+\fB\-t\fR
+Show spend time.
+.TP 0.5i
 \fBinputfile\fR
 An absolute or relative file path. Specially, pass - to read from stdin.
 .TP 0.5i
@@ -59,6 +65,9 @@ Map input file error.
 .TP 0.5i
 \fB7\fR
 Write file error in mmap.
+.TP 0.5i
+\fB8\fR
+Invalid input/output filename.
 .SH "SEE ALSO"
 https://github.com/fumiama/base16384
 .SH BUGS
@@ -68,7 +77,7 @@ on github.
 .SH AUTHOR
 This manual page contributed by Fumiama Minamoto.
 .SH "COPYRIGHT"
-Copyright \(co 2022, Fumiama Minamoto
+Copyright \(co 2022-2023, Fumiama Minamoto
 This file is part of
 .IR "base16384" .
 .LP

base16384.c

@@ -1,6 +1,6 @@
 /* base16384.c
  * This file is part of the base16384 distribution (https://github.com/fumiama/base16384).
- * Copyright (c) 2022 Fumiama Minamoto.
+ * Copyright (c) 2022-2023 Fumiama Minamoto.
  * 
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -39,7 +39,7 @@ unsigned long get_start_ms() {
 #endif
 
 static void print_usage() {
-	puts("Copyright (c) 2022-2023 Fumiama Minamoto.\nBase16384 2.2.3 (May 18th 2023). Usage:");
+	puts("Copyright (c) 2022-2023 Fumiama Minamoto.\nBase16384 2.2.5 (August 26th 2023). Usage:");
 	puts("base16384 [-edt] [inputfile] [outputfile]");
 	puts("  -e\t\tencode");
 	puts("  -d\t\tdecode");
@@ -97,15 +97,18 @@ int main(int argc, char** argv) {
 			printf("spend time: %lums\n", get_start_ms() - t);
 		#endif
 	}
+	#define print_base16384_err(n) case base16384_err_##n: perror("base16384_err_"#n); break
 	if(exitstat) switch(exitstat) {
-		case base16384_err_get_file_size: perror("base16384_err_get_file_size"); break;
+		print_base16384_err(get_file_size);
-		case base16384_err_fopen_output_file: perror("base16384_err_fopen_output_file"); break;
+		print_base16384_err(fopen_output_file);
-		case base16384_err_fopen_input_file: perror("base16384_err_fopen_input_file"); break;
+		print_base16384_err(fopen_input_file);
-		case base16384_err_write_file: perror("base16384_err_write_file"); break;
+		print_base16384_err(write_file);
-		case base16384_err_open_input_file: perror("base16384_err_open_input_file"); break;
+		print_base16384_err(open_input_file);
-		case base16384_err_map_input_file: perror("base16384_err_map_input_file"); break;
+		print_base16384_err(map_input_file);
-		case base16384_err_read_file: perror("base16384_err_read_file"); break;
+		print_base16384_err(read_file);
+		print_base16384_err(invalid_file_name);
 		default: perror("base16384"); break;
 	}
+	#undef print_base16384_err
     return exitstat;
 }

base16384.h

@@ -3,7 +3,7 @@
 
 /* base16384.h
  * This file is part of the base16384 distribution (https://github.com/fumiama/base16384).
- * Copyright (c) 2022 Fumiama Minamoto.
+ * Copyright (c) 2022-2023 Fumiama Minamoto.
  * 
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -24,22 +24,27 @@
 #include <stdio.h>
 #endif
 
+#define define_base16384_err_t(n) base16384_err_##n
+
 // base16384_err_t is the return value of base16384_en/decode_file
 enum base16384_err_t {
-	base16384_err_ok,
+	define_base16384_err_t(ok),
-	base16384_err_get_file_size,
+	define_base16384_err_t(get_file_size),
-	base16384_err_fopen_output_file,
+	define_base16384_err_t(fopen_output_file),
-	base16384_err_fopen_input_file,
+	define_base16384_err_t(fopen_input_file),
-	base16384_err_write_file,
+	define_base16384_err_t(write_file),
-	base16384_err_open_input_file,
+	define_base16384_err_t(open_input_file),
-	base16384_err_map_input_file,
+	define_base16384_err_t(map_input_file),
-	base16384_err_read_file,
+	define_base16384_err_t(read_file),
+	define_base16384_err_t(invalid_file_name),
 };
 // base16384_err_t is the return value of base16384_en/decode_file
 typedef enum base16384_err_t base16384_err_t;
 
-#define BASE16384_ENCBUFSZ (BUFSIZ*1024/7*7)
+#undef define_base16384_err_t
-#define BASE16384_DECBUFSZ (BUFSIZ*1024/8*8+2)
+
+#define BASE16384_ENCBUFSZ (BUFSIZ*1024/7*7+7)
+#define BASE16384_DECBUFSZ (BUFSIZ*1024/8*8+16)
 
 // base16384_encode_len calc min buf size to fill encode result
 static inline int base16384_encode_len(int dlen) {
@@ -75,10 +80,10 @@ static inline int base16384_decode_len(int dlen, int offset) {
 }
 
 // base16384_encode encodes data and write result into buf
-int base16384_encode(const char* data, int dlen, char* buf, int blen);
+int base16384_encode(const char* data, int dlen, char* buf);
 
 // base16384_decode decodes data and write result into buf
-int base16384_decode(const char* data, int dlen, char* buf, int blen);
+int base16384_decode(const char* data, int dlen, char* buf);
 
 // base16384_encode_file encodes input file to output file.
 //    use `-` to specify stdin/stdout

file.c

@@ -1,6 +1,6 @@
 /* file.c
  * This file is part of the base16384 distribution (https://github.com/fumiama/base16384).
- * Copyright (c) 2022 Fumiama Minamoto.
+ * Copyright (c) 2022-2023 Fumiama Minamoto.
  * 
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -20,6 +20,7 @@
 #include <stdio.h>
 #include <stdint.h>
 #include <stdlib.h>
+#include <string.h>
 #include <sys/stat.h>
 #include <fcntl.h>
 #ifdef _WIN32
@@ -48,6 +49,7 @@ base16384_err_t base16384_encode_file(const char* input, const char* output, cha
 	off_t inputsize;
 	FILE* fp = NULL;
 	FILE* fpo;
+	if(!input || !output || strlen(input) <= 0 || strlen(output) <= 0) return base16384_err_invalid_file_name;
 	if(is_standard_io(input)) { // read from stdin
 		inputsize = 0;
 		fp = stdin;
@@ -59,9 +61,9 @@ base16384_err_t base16384_encode_file(const char* input, const char* output, cha
 	if(!fpo) {
 		return base16384_err_fopen_output_file;
 	}
-	if(!inputsize || inputsize > BUFSIZ*1024) { // stdin or big file, use encbuf & fread
+	if(!inputsize || inputsize > BASE16384_ENCBUFSZ) { // stdin or big file, use encbuf & fread
-		inputsize = BUFSIZ*1024/7*7;
+		inputsize = BASE16384_ENCBUFSZ-7;
-		#ifdef _WIN32
+		#if defined _WIN32 || defined __cosmopolitan
 	}
 		#endif
 		if(!fp) fp = fopen(input, "rb");
@@ -69,32 +71,30 @@ base16384_err_t base16384_encode_file(const char* input, const char* output, cha
 			return base16384_err_fopen_input_file;
 		}
 
-		int outputsize = base16384_encode_len(inputsize)+16;
 		size_t cnt = 0;
 		fputc(0xFE, fpo);
 		fputc(0xFF, fpo);
 		while((cnt = fread(encbuf, sizeof(char), inputsize, fp)) > 0) {
-			int n = base16384_encode(encbuf, cnt, decbuf, outputsize);
+			int n = base16384_encode(encbuf, cnt, decbuf);
 			if(fwrite(decbuf, n, 1, fpo) <= 0) {
 				return base16384_err_write_file;
 			}
 		}
 		if(!is_standard_io(output)) fclose(fpo);
 		if(!is_standard_io(input)) fclose(fp);
-	#ifndef _WIN32
+	#if !defined _WIN32 && !defined __cosmopolitan
 	} else { // small file, use mmap & fwrite
 		int fd = open(input, O_RDONLY);
 		if(fd < 0) {
 			return base16384_err_open_input_file;
 		}
-		char *input_file = mmap(NULL, (size_t)inputsize, PROT_READ, MAP_PRIVATE, fd, 0);
+		char *input_file = mmap(NULL, (size_t)inputsize+16, PROT_READ, MAP_PRIVATE, fd, 0);
 		if(input_file == MAP_FAILED) {
 			return base16384_err_map_input_file;
 		}
-		int outputsize = base16384_encode_len(inputsize)+16;
 		fputc(0xFE, fpo);
 		fputc(0xFF, fpo);
-		int n = base16384_encode(input_file, (int)inputsize, decbuf, outputsize);
+		int n = base16384_encode(input_file, (int)inputsize, decbuf);
 		if(fwrite(decbuf, n, 1, fpo) <= 0) {
 			return base16384_err_write_file;
 		}
@@ -113,13 +113,12 @@ base16384_err_t base16384_encode_fp(FILE* input, FILE* output, char* encbuf, cha
 	if(!output) {
 		return base16384_err_fopen_output_file;
 	}
-	off_t inputsize = BUFSIZ*1024/7*7;
+	off_t inputsize = BASE16384_ENCBUFSZ-7;
-	int outputsize = base16384_encode_len(inputsize)+16;
 	size_t cnt = 0;
 	fputc(0xFE, output);
 	fputc(0xFF, output);
 	while((cnt = fread(encbuf, sizeof(char), inputsize, input)) > 0) {
-		int n = base16384_encode(encbuf, cnt, decbuf, outputsize);
+		int n = base16384_encode(encbuf, cnt, decbuf);
 		if(fwrite(decbuf, n, 1, output) <= 0) {
 			return base16384_err_write_file;
 		}
@@ -134,12 +133,11 @@ base16384_err_t base16384_encode_fd(int input, int output, char* encbuf, char* d
 	if(output < 0) {
 		return base16384_err_fopen_output_file;
 	}
-	off_t inputsize = BUFSIZ*1024/7*7;
+	off_t inputsize = BASE16384_ENCBUFSZ-7;
-	int outputsize = base16384_encode_len(inputsize)+16;
 	size_t cnt = 0;
 	write(output, "\xfe\xff", 2);
 	while((cnt = read(input, encbuf, inputsize)) > 0) {
-		int n = base16384_encode(encbuf, cnt, decbuf, outputsize);
+		int n = base16384_encode(encbuf, cnt, decbuf);
 		if(write(output, decbuf, n) < n) {
 			return base16384_err_write_file;
 		}
@@ -167,6 +165,7 @@ base16384_err_t base16384_decode_file(const char* input, const char* output, cha
 	off_t inputsize;
 	FILE* fp = NULL;
 	FILE* fpo;
+	if(!input || !output || strlen(input) <= 0 || strlen(output) <= 0) return base16384_err_invalid_file_name;
 	if(is_standard_io(input)) { // read from stdin
 		inputsize = 0;
 		fp = stdin;
@@ -178,16 +177,15 @@ base16384_err_t base16384_decode_file(const char* input, const char* output, cha
 	if(!fpo) {
 		return base16384_err_fopen_output_file;
 	}
-	if(!inputsize || inputsize > BUFSIZ*1024) { // stdin or big file, use decbuf & fread
+	if(!inputsize || inputsize > BASE16384_DECBUFSZ) { // stdin or big file, use decbuf & fread
-		inputsize = BUFSIZ*1024/8*8;
+		inputsize = BASE16384_DECBUFSZ/8*8;
-		#ifdef _WIN32
+		#if defined _WIN32 || defined __cosmopolitan
 	}
 		#endif
 		if(!fp) fp = fopen(input, "rb");
 		if(!fp) {
 			return base16384_err_fopen_input_file;
 		}
-		int outputsize = base16384_decode_len(inputsize, 0)+16;
 		int cnt = 0;
 		int end = 0;
 		rm_head(fp);
@@ -196,25 +194,24 @@ base16384_err_t base16384_decode_file(const char* input, const char* output, cha
 				decbuf[cnt++] = '=';
 				decbuf[cnt++] = end;
 			}
-			if(fwrite(encbuf, base16384_decode(decbuf, cnt, encbuf, outputsize), 1, fpo) <= 0) {
+			if(fwrite(encbuf, base16384_decode(decbuf, cnt, encbuf), 1, fpo) <= 0) {
 				return base16384_err_write_file;
 			}
 		}
 		if(!is_standard_io(output)) fclose(fpo);
 		if(!is_standard_io(input)) fclose(fp);
-	#ifndef _WIN32
+	#if !defined _WIN32 && !defined __cosmopolitan
 	} else { // small file, use mmap & fwrite
 		int fd = open(input, O_RDONLY);
 		if(fd < 0) {
 			return base16384_err_open_input_file;
 		}
-		char *input_file = mmap(NULL, (size_t)inputsize, PROT_READ, MAP_PRIVATE, fd, 0);
+		char *input_file = mmap(NULL, (size_t)inputsize+16, PROT_READ, MAP_PRIVATE, fd, 0);
 		if(input_file == MAP_FAILED) {
 			return base16384_err_map_input_file;
 		}
-		int outputsize = base16384_decode_len(inputsize, 0)+16;
 		int off = skip_offset(input_file);
-		if(fwrite(encbuf, base16384_decode(input_file+off, inputsize-off, encbuf, outputsize), 1, fpo) <= 0) {
+		if(fwrite(encbuf, base16384_decode(input_file+off, inputsize-off, encbuf), 1, fpo) <= 0) {
 			return base16384_err_write_file;
 		}
 		munmap(input_file, (size_t)inputsize);
@@ -232,8 +229,7 @@ base16384_err_t base16384_decode_fp(FILE* input, FILE* output, char* encbuf, cha
 	if(!output) {
 		return base16384_err_fopen_output_file;
 	}
-	off_t inputsize = BUFSIZ*1024/8*8;
+	off_t inputsize = BASE16384_DECBUFSZ/8*8;
-	int outputsize = base16384_decode_len(inputsize, 0)+16;
 	int cnt = 0;
 	int end = 0;
 	rm_head(input);
@@ -242,7 +238,7 @@ base16384_err_t base16384_decode_fp(FILE* input, FILE* output, char* encbuf, cha
 			decbuf[cnt++] = '=';
 			decbuf[cnt++] = end;
 		}
-		if(fwrite(encbuf, base16384_decode(decbuf, cnt, encbuf, outputsize), 1, output) <= 0) {
+		if(fwrite(encbuf, base16384_decode(decbuf, cnt, encbuf), 1, output) <= 0) {
 			return base16384_err_write_file;
 		}
 	}
@@ -265,8 +261,7 @@ base16384_err_t base16384_decode_fd(int input, int output, char* encbuf, char* d
 	if(output < 0) {
 		return base16384_err_fopen_output_file;
 	}
-	off_t inputsize = BUFSIZ*1024/8*8;
+	off_t inputsize = BASE16384_DECBUFSZ/8*8;
-	int outputsize = base16384_decode_len(inputsize, 0)+16;
 	int cnt = 0;
 	int end = 0;
 	decbuf[0] = 0;
@@ -283,7 +278,7 @@ base16384_err_t base16384_decode_fd(int input, int output, char* encbuf, char* d
 				end = 0;
 			} else end = 1;
 		} else end = 0;
-		cnt = base16384_decode(decbuf, cnt, encbuf, outputsize);
+		cnt = base16384_decode(decbuf, cnt, encbuf);
 		if(write(output, encbuf, cnt) < cnt) {
 			return base16384_err_write_file;
 		}