convert mobi file to epub format successfully, but the epub file format is error, it can't be opened by iBooks and many android epub readers. I check the epub file with calibre-edit, and get the error below:

ERROR: Parsing failed: xmlParseEntityRef: no name, line 1, column 807    [OEBPS/part00000.html]
INFO: File too large    [OEBPS/part00000.html]

123_test.epub.zip

Hello @bfabiszewski I am using your another lib QLMobi combine with libmobi to parse html and images from mobi book. Most book works great, but some books can not get media image. I have try to fix but can not get the point. Hope you can help,this is the last problem for me i think~ Both QLMobi and libmobi are great nearly perfect lib. Thank you very much for your great job~ World of Warcraft - Dawn of the Aspects Part I.mobi.zip

Also i am the developer of Alook Browser - 2x Speed (https://itunes.apple.com/us/app/alook-web-browser-2x-speed/id1261944766?mt=8) if you are using iOS ，and here is a promotional code JWYTH3FE4JJK Forgive my poor english~ Best Regards.

Hey there!

I'd like to report a security issue but cannot find contact instructions on your repository.

If not a hassle, might you kindly add a SECURITY.md file with an email, or another contact method? GitHub recommends this best practice to ensure security issues are responsibly disclosed, and it would serve as a simple instruction for security researchers in the future.

Thank you for your consideration, and I look forward to hearing from you!

(cc @huntr-helper)

Hi first of all thank you for the work you put in this library.

I'm using it in a project where it's statically linked with libmagic and I'm getting this error:

third-party/libscan/third-party/ext_libmobi/src/libmobi/src/.libs//libmobi.a(libmobi_la-buffer.o): In function `buffer_init':
/root/agent/work/f1ce04c709a195f3/third-party/libscan/third-party/ext_libmobi/src/libmobi/src/buffer.c:26: multiple definition of `buffer_init'
/vcpkg/installed/x64-linux/lib/libmagic.a(buffer.o):buffer.c:(.text+0x0): first defined here
collect2: error: ld returned 1 exit status

buffer_init in libmagic: file.h

I managed to work around this by renaming the function to mobi_buffer_init in my fork. I'd appreciate we could merge this upstream (if you really want to keep the name it's fine but it would make my life much easier)

Thanks!

Mobi parse failed but can be open in Kindle app. File is in attachment. Thank you for your great work~ Best Regards. World of Warcraft - Dawn of the Aspects Part I.mobi.zip

Hi Bartek, I'm currently writing a dictionary software and want to add mobi dict support to it. I'm glad to find there is such a great mobi library! This PR is basically what I found i ismail's mobdict project, so credits to him. It adds two useful functions to get the start and length of an entry. And since getting the error string is such a useful function I also exports libmobi_msg. Let me know what do you think of it. :)

The README lists a lot of features, but they're all apparently centered around reading or modifying an existing file.

Can libmobi also create new ebooks from scratch? (For use in an EPUB->MOBI conversion software) If yes, maybe another bullet point in the README clarifying that would be useful :slightly_smiling_face:

Thanks for creating this cool library!

printf("Could not initialize zip archive\n"); Here is the link to the file I tested. https://1drv.ms/u/s!AkaVccfysLmAhI5Odqj2pZ1QCMci6g?e=U9lkC3

This PR adds very basic CMake support -- it just compiles the source code to a static library. It uses zlib and libxml2 libraries on Unix-like OSs, and internal implementations otherwise. I've tested it on Linux and Windows (MinGW). Because I have no experience with autoconf tools, it's difficult for me to came up with a complete configure.ac equivalent CMakeLists.txt right now. My thought is to have basic support now and improve it later.

_buffer_get_varlen I am puzzled by this function, why should I read 7 bit, Stops when byte has bit 7 set, I am also confused about this condition. Should not be a step-by-step read 8 bit

We found with our fuzzer several heap-buffer-overflow errors when compiling libmobi with address sanitizer and run with the command mobitool -i7m $file. Someone else also found a few others here.

We will list them separately in the following issue threads and this is the 1st one.

POC (proof-of-crash) files: https://github.com/ntu-sec/pocs/blob/master/libmobi/hbo_buffer.c%3A212_1.mobi https://github.com/ntu-sec/pocs/blob/master/libmobi/hbo_buffer.c%3A212_2.mobi

gdb output: https://github.com/ntu-sec/pocs/blob/master/libmobi/hbo_buffer.c:212_1.mobi.gdb.txt https://github.com/ntu-sec/pocs/blob/master/libmobi/hbo_buffer.c:212_2.mobi.gdb.txt

When you create a MOBIBuffer object:

    typedef struct {
    size_t offset; /**< Current offset in respect to buffer start */
    size_t maxlen; /**< Length of the buffer data */
    unsigned char *data; /**< Pointer to buffer data */
    MOBI_RET error; /**< MOBI_SUCCESS = 0 if operation on buffer is successful, non-zero value on failure */
} MOBIBuffer;

the initial value of buf->offset is 0:

MOBIBuffer * mobi_buffer_init_null(unsigned char *data, const size_t len) {
    MOBIBuffer *buf = malloc(sizeof(MOBIBuffer));
    if (buf == NULL) {
        debug_print("%s", "Buffer allocation failed\n");
        return NULL;
    }
    buf->data = data;
    buf->offset = 0;
    buf->maxlen = len;
    buf->error = MOBI_SUCCESS;
    return buf;
}

I think there is a problem calling mobi_buffer_get_varlen_internal when direction is -1(read buffer backwards) with a value of buf->offset that is 3. If buf->offset is 3, it should Reads maximum 4 bytes from the buffer. Stops when byte has bit 7 set. so it should read byte number 3, byte number 2, byte number 1, and then byte number 0. but when it comes to read byte number 0, we can see the following check at line 267: if (buf->offset < 1) { it checks if zero is less than 1 and it is, so an error is printed and only the last 3 bytes that have been read return and not the 4. (even though according to pull request it should return 0)

if it needs to read byte number 0 - it should read it and then return without decrementing buf->offset of 0 because if it does it, it will lead to an integer underflow and we will get the max value for size_t in buf->offset, so I suggest checking if it is 0 after reading the byte to the value byte and after updating the value of val, and if buf->offset is 0, we should check byte_count and according to that decide whether to execute

                debug_print("%s", "End of buffer\n");
                buf->error = MOBI_BUFFER_END;
                return 0;

or to set byte to stop_flag so it will stop reading and return val, while keeping buf->offset at 0,