mini-player读取和展示音频元数据作的一些尝试-CSDN博客

本文链接：https://blog.csdn.net/jacky_zz/article/details/148522803

0x00 起源

音频播放器播放列表展示音频的元数据如title、artist、album等信息是一个常见的功能，mini-player使用FFmpeg作为解码器，也自带读取音频元数据的功能，样例代码如下：

// 定回调函数
typedef void (*metadata_callback)(void *user, const char *key, const char *value);

int32_t metadata(struct decoder *pdecoder, const char *filename, metadata_callback cb, void *user)
{
    const AVDictionary *m = NULL;
    ...
    m = is->ic->metadata;
    if (m && !(av_dict_count(m) == 1 && av_dict_get(m, "language", NULL, 0)))
    {
        while ((tag = av_dict_iterate(m, tag)))
        {
            if (strcmp("language", tag->key))
            {
                if (cb)
                    cb(user, tag->key, tag->value);
            }
        }
    }
    ....
}

本想着FFmpeg的这个功能能快速的实现音频元数据的读取和展示，如果音频文件不涉及中文（GBK或GB18030）倒是没什么问题，但实际是不太可能的。为了正确展示音频元数据，我先后尝试了taglib（https://taglib.org/）和自己写代码解析，效果都不是很好。自己写代码解析了解一下细节可以，但重复造轮子不是我想的，还是立足FFmpeg现有的基础吧。

0x01 捋清思路

我的解决思路是先用utf8cpp（https://github.com/nemtrif/utfcpp）判断字符串编码是否为utf-8，如果编码不是utf-8，假设字符串编码为GBK或GB18030，将其转换为utf-8；如果编码是utf-8，这里需要解决两个问题：一个问题是GBK或GB18030编码的字符串被ISO-8859-1再次编码（典型特征是字符串中含有成对的0xC2、0xC3，如：[0xC2, 0xA3, 0xC2, 0xA8, 0xC3, 0x81, 0xC3, 0xB5, 0xC2, 0xBB, 0xC2, 0xB6, 0xC2, 0xA3]），将其转换为被ISO-8859-1编码前的字符串，另一个问题是GBK或GB18030编码的字符（如：路远，字节码为[0xC2, 0xB7, 0xD4, 0xB6]）会被误判为utf-8。解决了以上两个问题，基本可以解决中文显示乱码的问题。

0x02 编码转换函数

编码转换基于iconv，代码如下：

int convert(const char *inFmt, const char *inStr, size_t inSize, const char *outFmt, char *outStr, size_t outSize)
{
    iconv_t cd = iconv_open(outFmt, inFmt);
    if (cd == (iconv_t)-1)
    {
        perror("iconv_open");
        return -1;
    }

    char *inptr = (char *)inStr;
    size_t inbytesleft = inSize;
    char *outptr = outStr;
    size_t outbytesleft = outSize;
    size_t ret = 0;

    if ((ret = iconv(cd, &inptr, &inbytesleft, &outptr, &outbytesleft)) == (size_t)-1)
    {
        perror("iconv");
        iconv_close(cd);
        return -1;
    }

    iconv_close(cd);
    return (outSize - outbytesleft);
}

0x03 解决0xC2、0xC3问题

直接上代码：

uint8_t is_encoded_by_iso8859_1(const uint8_t *data, size_t size)
{
    assert(data);
    assert(size > 0);

    size_t i = 0;
    const uint8_t *p = data;
    uint8_t ret = 0;

    while (i < size)
    {
        if (*p < 0x80)
        {
            ++p;
            ++i;
            continue;
        }

        if ((*p == 0xC2) || (*p == 0xC3))
        {
            if ((*(p + 2) == 0xC2) || (*(p + 2) == 0xC3))
            {
                ret = 1;
                break;
            }
            else
            {
                ++p;
                ++i;
            }
        }
        else
        {
            ++p;
            ++i;
        }
    }

    return ret;
}

#define DEFAULT_BUF_SIZE 2048
void metadata_callback2(void *user, const char *key, const char *value)
{
    size_t value_size = strlen(value);
    std::string str = value;
    auto end_it = utf8::find_invalid(str.begin(), str.end());
    if (end_it == str.end())
    {
        if (is_encoded_by_iso8859_1((uint8_t *)value, value_size))
        {
            char buf[DEFAULT_BUF_SIZE] = {0};
            size_t size = convert("UTF-8", value, value_size, "ISO-8859-1", buf, sizeof(buf));
            ...
        }
    }
}

0x04 解决GBK或GB18030误判为UTF-8问题

因为GBK或GB18030编码规则与UTF-8编码规则部分重叠，会导致GBK或GB18030编码的字符串被误认为UTF-8，这里使用查表法判断，代码如下：

#include "gb18030_unicode.h"

typedef struct check_result
{
    uint32_t ascii_count;
    uint32_t utf8_count;
    uint32_t gb18030_count;
} check_result;

void recheck_utf8(const uint8_t *data, size_t in_size, check_result *result)
{
    assert(data);
    assert(result);

    size_t table_size = sizeof(gb18030_unicode_table) / sizeof(gb18030_unicode);
    const uint8_t *p = data;
    uint16_t val = 0;

    memset(result, 0, sizeof(check_result));

    for (size_t i = 0; i < in_size; ++i)
    {
        if (p[i] <= 0x7F)
        {
            ++result->ascii_count;
            continue;
        }

        if ((p[i] >= 0xC0 && p[i] <= 0xDF) && ((i + 2) <= in_size))
        {
            val = p[i] << 8 | p[i + 1];
            for (size_t j = 0; j < table_size; ++j)
            {
                if (gb18030_unicode_table[j].utf8 == val)
                {
                    ++result->utf8_count;
                }

                if (gb18030_unicode_table[j].gb18030 == val)
                {
                    ++result->gb18030_count;
                }
            }
        }

        if ((p[i] >= 0xE0 && p[i] <= 0xEF) && ((i + 3) <= in_size))
        {
            ++result->utf8_count;
        }

        if ((p[i] >= 0xF0 && p[i] <= 0xF7) && ((i + 4) <= in_size))
        {
            ++result->utf8_count;
        }

        if ((p[i] >= 0xF8 && p[i] <= 0xFB) && ((i + 5) <= in_size))
        {
            ++result->utf8_count;
        }

        if ((p[i] >= 0xFC && p[i] <= 0xFD) && ((i + 6) <= in_size))
        {
            ++result->utf8_count;
        }
    }
}

#define DEFAULT_BUF_SIZE 2048
void metadata_callback2(void *user, const char *key, const char *value)
{
    size_t value_size = strlen(value);
    std::string str = value;
    auto end_it = utf8::find_invalid(str.begin(), str.end());
    if (end_it == str.end())
    {
        if (is_encoded_by_iso8859_1((uint8_t *)value, value_size))
        {
            ...
        }
        else
        {
            check_result result;
            recheck_utf8((const uint8_t *)value, value_size, &result);

            if (result.gb18030_count > result.utf8_count)
            {
                char buf[DEFAULT_BUF_SIZE] = {0};
                size_t size = convert("GB18030", value, value_size, "UTF-8", buf, sizeof(buf));
                ....
            }
        }
    }
}

gb18030_unicode.h下载地址：GitHub - jackyxinli/mini-player