FastSort/src/sort/sam_io.cpp

#include "sam_io.h"

#include <htslib/sam.h>
#include <libdeflate.h>
#include <stdint.h>
#include <stdio.h>
#include <zlib.h>
#include <spdlog/spdlog.h>


namespace nsgv {
extern bool gIsBigEndian;
};

int bgzfUncompress(uint8_t *dst, size_t *dlen, const uint8_t *src, size_t slen, uint32_t expected_crc) {
    struct libdeflate_decompressor *z = libdeflate_alloc_decompressor();
    if (!z) {
        hts_log_error("Call to libdeflate_alloc_decompressor failed");
        return -1;
    }

    int ret = libdeflate_deflate_decompress(z, src, slen, dst, *dlen, dlen);
    libdeflate_free_decompressor(z);

    if (ret != LIBDEFLATE_SUCCESS) {
        hts_log_error("Inflate operation failed: %d", ret);
        return -1;
    }

    uint32_t crc = libdeflate_crc32(0, (unsigned char *)dst, *dlen);
#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
    // Pretend the CRC was OK so the fuzzer doesn't have to get it right
    crc = expected_crc;
#endif
    if (crc != expected_crc) {
        hts_log_error("CRC32 checksum mismatch");
        return -2;
    }

    return 0;
}

void parseSamHeader(FILE *fp, HeaderBuf &hdrBuf) {
    const int kMaxBlockSize = 65535;
    sam_hdr_t *header = NULL;
    uint8_t *fBuf = (uint8_t*)malloc(kMaxBlockSize);
    DataBuffer uData;
    size_t readState = 0;
    int blockLen = 0;
    size_t dlen = 0x10000;  // 65535
    int magicLen;
    int32_t i, nameLen, numNames = 0;
    size_t bufsize;
    ssize_t bytes;

    uData.allocMem(kMaxBlockSize);

    readState = fread(fBuf, 1, BLOCK_HEADER_LENGTH, fp);
    blockLen = unpackInt16(&fBuf[16]) + 1;
    readState = fread(&fBuf[BLOCK_HEADER_LENGTH], 1, blockLen - BLOCK_HEADER_LENGTH, fp);
    header = sam_hdr_init();
    spdlog::info("Header file size: {}", blockLen);

    uint32_t crc = le_to_u32(fBuf + blockLen - 8);
    int ret = bgzfUncompress(uData.data, &dlen, (Bytef *)fBuf + 18, blockLen - 18, crc);
    uData.curLen = dlen;
    // 解析header
    magicLen = 4;
    if (memcmp(uData.data, "BAM\1", magicLen)) {
        spdlog::error("Invalid BAM binary header");
        return;
    }
    uData.readPos += magicLen;
    header->l_text = le_to_u32(uData.data + uData.readPos); uData.readPos += 4;
    header->text = (char *)malloc(header->l_text + 1);
    header->text[header->l_text] = 0;
//    while (uData.readPos + header->l_text > uData.curLen) { // header内容超过了一个block，继续读
//        readState = fread(fBuf, 1, BLOCK_HEADER_LENGTH, fp); // 读压缩块头
//        blockLen = unpackInt16(&fBuf[16]) + 1; // 压缩块大小
//        readState = fread(&fBuf[BLOCK_HEADER_LENGTH], 1, blockLen - BLOCK_HEADER_LENGTH, fp); // 读压缩块剩下的部分
//        size_t newDataSize = uData.maxLen;
//        while (uData.curLen + kMaxBlockSize > uData.maxLen) newDataSize *= 2;
//        uData.reAllocMem(newDataSize);
//        ret = bgzfUncompress(&uData.data[uData.curLen], &dlen, (Bytef *)fBuf + 18, blockLen - 18, crc);
//        uData.curLen += dlen;
//    }
    memcpy(header->text, &uData.data[uData.readPos], header->l_text);
    uData.readPos += header->l_text;

    memcpy(&header->n_targets, &uData.data[uData.readPos], 4); // n_targets
    uData.readPos += 4;
    spdlog::info("num target: {}", header->n_targets);
    if (nsgv::gIsBigEndian) ed_swap_4p(&header->n_targets);
    if (header->n_targets > 0) {
        header->target_name = (char **)calloc(header->n_targets, sizeof(char *));
        header->target_len = (uint32_t *)calloc(header->n_targets, sizeof(uint32_t));
    } else {
        header->target_name = NULL;
        header->target_len = NULL;
    }

    for (int i = 0; i != header->n_targets; ++i) {
        memcpy(&nameLen, &uData.data[uData.readPos], 4);
        uData.readPos += 4;
        if (nsgv::gIsBigEndian) ed_swap_4p(&nameLen);
        header->target_name[i] = (char *)malloc(nameLen);
        ++numNames;
        memcpy(header->target_name[i], &uData.data[uData.readPos], nameLen);
        uData.readPos += nameLen;
        if (header->target_name[i][nameLen - 1] != '\0') {
            /* Fix missing NUL-termination.  Is this being too nice?
               We could alternatively bail out with an error. */
            char *newName = (char*)realloc(header->target_name[i], nameLen + 1);
            header->target_name[i] = newName;
            header->target_name[i][nameLen] = '\0';
        }
        memcpy(&header->target_len[i], &uData.data[uData.readPos], 4);
        uData.readPos += 4;
        if (nsgv::gIsBigEndian) ed_swap_4p(&header->target_len[i]);

        // spdlog::info("nameLen {}, {}, {}", nameLen, header->target_len[i], header->target_name[i]);
    }
    // spdlog::info("test res: {} {} {} {}", header->l_text, dlen, header->n_targets, header->text);


    hdrBuf.header = header;
    free(fBuf);

    // exit(0);
}