unescape_xml.c

#include <stdio.h>
#include <string.h>

/*
&apos; is replaced with '
&amp; is replaced with &
&quot; is replaced with "
&lt; is replaced with <
&gt; is replaced with >
*/
int main(int argc, const char* argv[])
{
    int i;
    char inputBuffer[8192];
    char outputBuffer[8192];

    int inputBufferUsed = 0;
    int inputPos = 0;
    int lastInputPos = 0;
    int outputPos = 0;

    for(i = 1; i < argc; i++)
    {
        const char* arg = argv[i];
        inputBufferUsed += snprintf(inputBuffer + inputBufferUsed,
                                     sizeof(inputBuffer) - inputBufferUsed,
                                     "%s\n", arg
        );
    }

    while(inputPos < inputBufferUsed)
    {
        const char* ampBeg = memchr(inputBuffer + inputPos, '&', inputBufferUsed - inputPos);
        if (!ampBeg) break;
        // p is now at the beginning of the next potential escape sequence
        inputPos = (ampBeg - inputBuffer);
        {
            int cpyLen = inputPos - lastInputPos;
            int remLen = (inputBufferUsed - inputPos);
            memcpy(outputBuffer + outputPos, inputBuffer + lastInputPos, cpyLen);
            outputPos += cpyLen;

            if (remLen > 5 && ampBeg[1] == 'a' && ampBeg[2] == 'p' && ampBeg[3] == 'o' && ampBeg[4] == 's' && ampBeg[5] == ';')
            {
                inputPos += 6;
                outputBuffer[outputPos++] = '\'';
            }
            else if (remLen > 4 && ampBeg[1] == 'a' && ampBeg[2] == 'm' && ampBeg[3] == 'p' && ampBeg[4] == ';')
            {
                inputPos += 5;
                outputBuffer[outputPos++] = '&';
            }
            else if (remLen > 5 && ampBeg[1] == 'q' && ampBeg[2] == 'u' && ampBeg[3] == 'o' && ampBeg[4] == 't'&& ampBeg[5] == ';')
            {
                inputPos += 6;
                outputBuffer[outputPos++] = '"';
            }
            else if (remLen > 3 && ampBeg[1] == 'l' && ampBeg[2] == 't'&& ampBeg[3] == ';')
            {
                inputPos += 4;
                outputBuffer[outputPos++] = '<';
            }
            else if (remLen > 3 && ampBeg[1] == 'g' && ampBeg[2] == 't' && ampBeg[3] == ';')
            {
                inputPos += 4;
                outputBuffer[outputPos++] = '>';
            }
            else
            {
                fprintf(stderr, "unknown escape seq: %.*s\n", 4, ampBeg);
            }
            lastInputPos = inputPos;
        }
    }
    // cpy rest
    memcpy(outputBuffer + outputPos, inputBuffer + lastInputPos, (inputBufferUsed - inputPos));
    outputPos += (inputBufferUsed - inputPos);
    outputBuffer[outputPos++] = '\0';
    puts(outputBuffer);
}