resume_gen/unicode_to_pdf.c

#include <stdlib.h>
#include <string.h>

#include "unicode_to_pdf.h"
#include "utf8_m.h"

#define ASCII_BYTE(x) (x > 9 ? x - 10 + 'A' : x + '0')

unsigned short
to16(const char *a)
{
	unsigned short val;
	switch (utf8_octet_count(*a)) {
	case 4:
		val = ((a[0] & 0x7) << 18) + ((a[1]) << 12) +
			((a[2] & 0x3f) << 6) + ((a[3] & 0x3f));
		break;
	case 3:
		val = ((a[0] & 0xf) << 12) + ((a[1] & 0x3f) << 6) +
		    ((a[2] & 0x3f));
		break;
	case 2:
		val = ((a[0] & 0x1f) << 6) + ((a[1] & 0x3f));
		break;
	case 1:
		val = *a;
		break;
	default:
		val = 0u;
	}
	return val;
}

unsigned short
asciibyte(char c)
{
	char hihalf = (c & 0xf0) >> 4;
	char lohalf = c & 0xf;
	hihalf = ASCII_BYTE(hihalf);
	lohalf = ASCII_BYTE(lohalf);
	return hihalf | (lohalf << 8);
}

enum conversion_status
convert_to_pdf16(const char **curr, int *in_bytes_count, int *bytes_used,
		write_callback write_fn, void *userptr)
{
	/* Convert up to 512 of octets in one turn. */
	char reprbuf[512 * 4];
	union {
		unsigned short utf16code;
		struct _tag_u {
			char u_lo;
			char u_hi;
		} s_u_code;
	} u_code;
	int s_count;
	char *it = reprbuf;
	const char *r_it = reprbuf + sizeof(reprbuf) - 1;
	int inbuf_count = 0;
	while (*in_bytes_count > 0) {
		s_count = utf8_octet_count(**curr);
		u_code.utf16code = to16(*curr);
		if (u_code.utf16code == 0u)
			return CONV_STATUS_ERROR;
		((unsigned short *)it)[1] = asciibyte(u_code.s_u_code.u_lo);
		((unsigned short *)it)[0] = asciibyte(u_code.s_u_code.u_hi);
		it += sizeof(unsigned short) * 2;
		if (it > r_it) {
			write_fn(reprbuf, sizeof(reprbuf), userptr);
			return CONV_STATUS_MORE_DATA;
		}
		inbuf_count += sizeof(unsigned short) * 2;
		*bytes_used += sizeof(unsigned short) * 2;
		*in_bytes_count -= s_count;
		*curr += s_count;
	}
	write_fn(reprbuf, inbuf_count, userptr);
	return CONV_STATUS_DONE;
}

int
convert_to16(const char *str, unsigned short *charpool,
		unsigned poolcount, unsigned short **opt_out)
{
	const char *cptr;
	unsigned u8count = utf8_symcount(str);
	unsigned short *target = charpool;
	if (poolcount < (u8count + 1)) {
		*opt_out = calloc(u8count, sizeof(unsigned short) + 1);
		target = *opt_out;
	} else {
		*opt_out = NULL;
	}
	target[u8count] = 0u;
	for (cptr = str; *cptr; cptr += utf8_octet_count(*cptr), ++ target) {
		if ((*target = to16(cptr)) == 0u) {
			if (*opt_out)
				free(*opt_out);
			return -1;
		}
	}
	return 0;
}