This repository has been archived on 2026-04-17. You can view files and clone it. You cannot open issues or pull requests or push a commit.
Files
bhlib/doc/HowTo/Utf8Test.md

154 lines
4.2 KiB
Markdown
Raw Normal View History

# HowTo: Transcoding UTF-8 to UTF-8
## Prerequisites
We want to implement a simple command-line utility that can transcode a UTF-8
file into UTF-8 file (or in other words replace any incorrect UTF-8 sequences).
To do this we would run the following command:
```sh
./Utf8Test UTF-8-test.txt UTF-8-out.txt
```
## Includes
To implement this utility, we are going to need to include the following headers:
- `BH/IO.h` to work with files (or input/output devices)
- `BH/String.h` to work with UTF-8 sequences
## Working with Files
Working with files in BHLib is based around the IO device (called `BH_IO`).
Firstly, you need to create an IO device with the `BH_FileNew` function.
Secondly, you need to open the IO device with the `BH_IOOpen` function. While
opening the IO device, you can specify in which mode it will work: reading
(`BH_IO_READ`) or writing (`BH_IO_WRITE`). Additionally, we can specify whether
the IO device (or in our case, the file) should exist before opening
(`BH_IO_EXIST`), be truncated before opening (`BH_IO_TRUNCATE`), should it be
created (`BH_IO_CREATE`), or opened in append mode (`BH_IO_APPEND`).
Here is an example for opening an existing file in read-only mode:
```c
BH_IO *io = BH_FileNew("coolfile.dat");
if (BH_IOOpen(io, BH_IO_READ | BH_IO_EXIST))
{
printf("Can't open file 'coolfile.dat'\n", config.file);
BH_IOFree(io);
return -1;
}
```
## Working with UTF-8
Reading UTF-8/UTF-16/UTF-32 is based around simple loop:
1. Read bytes from input (IO or memory) to some buffer.
2. Call `BH_UnicodeDecodeUtf*`. If return value is 0 - we don't have enough data, so go to step 1. Otherwise remove result bytes from the front of the buffer.
3. If readed codepoint equals -1 - we encountered an error, so replace it with the code 0xFFFD.
Writing UTF-8/UTF-16/UTF-32 is straight forward:
1. Call `BH_UnicodeEncodeUtf*`. If return value is 0 - we can't encode codepoint (either codepoint is surrogate pair or outside valid range).
2. Write data (to IO or memory).
BH_UnicodeDecodeUtf8(inBuffer, inSize, &unit)
```c
while (...)
{
/* Read one byte and try to decode */
if (!inSize || !(outSize = BH_UnicodeDecodeUtf8(inBuffer, inSize, &unit)))
{
BH_IORead(inFile, inBuffer + inSize, 1, &outSize);
inSize += outSize;
continue;
}
/* Remove readed amount */
for (i = 0; i < inSize - outSize; i++)
inBuffer[i] = inBuffer[i + outSize];
inSize -= outSize;
/* Change unit if incorrect and write to output */
if (unit == -1)
unit = 0xFFFD;
outSize = BH_UnicodeEncodeUtf8(unit, outBuffer);
BH_IOWrite(outFile, outBuffer, outSize, NULL);
}
```
## Putting Everything Together
```c
#include <BH/IO.h>
#include <BH/String.h>
#include <stdlib.h>
#include <stdio.h>
void printUsage(void)
{
printf("Utf8Test <input> <output>\n");
exit(1);
}
int main(int argc, char **argv)
{
BH_IO *inFile, *outFile;
char inBuffer[8], outBuffer[8];
uint32_t unit;
size_t i, inSize, outSize;
if (argc < 2)
printUsage();
inFile = BH_FileNew(argv[1]);
outFile = BH_FileNew(argv[2]);
if (!inFile || BH_IOOpen(inFile, BH_IO_READ | BH_IO_EXIST))
return -1;
if (!outFile || BH_IOOpen(outFile, BH_IO_WRITE | BH_IO_TRUNCATE))
return -1;
inSize = 0;
while (!(BH_IOFlags(inFile) & BH_IO_FLAG_EOF))
{
/* Read one byte and try to decode */
if (!inSize || !(outSize = BH_UnicodeDecodeUtf8(inBuffer, inSize, &unit)))
{
BH_IORead(inFile, inBuffer + inSize, 1, &outSize);
inSize += outSize;
continue;
}
/* Remove readed amount */
for (i = 0; i < inSize - outSize; i++)
inBuffer[i] = inBuffer[i + outSize];
inSize -= outSize;
/* Change unit if incorrect and write to output */
if (unit == -1)
unit = 0xFFFD;
outSize = BH_UnicodeEncodeUtf8(unit, outBuffer);
BH_IOWrite(outFile, outBuffer, outSize, NULL);
}
/* Incomplete UTF-8 sequence */
if (inSize)
{
outSize = BH_UnicodeEncodeUtf8(0xFFFD, outBuffer);
BH_IOWrite(outFile, outBuffer, outSize, NULL);
}
BH_IOFree(inFile);
BH_IOFree(outFile);
return 0;
}
```