+#ifdef _WIN32
+#include "charconv.h"
+#endif
+
+/*
+ * Set the encoding.
+ */
+#ifdef _WIN32
+/*
+ * True if we should use UTF-8.
+ */
+static int use_utf_8;
+
+void
+pcapint_fmt_set_encoding(unsigned int opts)
+{
+ if (opts == PCAP_CHAR_ENC_UTF_8)
+ use_utf_8 = 1;
+}
+#else
+void
+pcapint_fmt_set_encoding(unsigned int opts _U_)
+{
+ /*
+ * Nothing to do here.
+ */
+}
+#endif
+
+#ifdef _WIN32
+/*
+ * Convert a null-terminated UTF-16LE string to UTF-8, putting it into
+ * a buffer starting at the specified location and stopping if we go
+ * past the specified size. This will only put out complete UTF-8
+ * sequences.
+ *
+ * We do this ourselves because Microsoft doesn't offer a "convert and
+ * stop at a UTF-8 character boundary if we run out of space" routine.
+ */
+#define IS_LEADING_SURROGATE(c) \
+ ((c) >= 0xd800 && (c) < 0xdc00)
+#define IS_TRAILING_SURROGATE(c) \
+ ((c) >= 0xdc00 && (c) < 0xe000)
+#define SURROGATE_VALUE(leading, trailing) \
+ (((((leading) - 0xd800) << 10) | ((trailing) - 0xdc00)) + 0x10000)
+#define REPLACEMENT_CHARACTER 0x0FFFD
+
+static char *
+utf_16le_to_utf_8_truncated(const wchar_t *utf_16, char *utf_8,
+ size_t utf_8_len)
+{
+ wchar_t c, c2;
+ uint32_t uc;
+
+ if (utf_8_len == 0) {
+ /*
+ * Not even enough room for a trailing '\0'.
+ * Don't put anything into the buffer.
+ */
+ return (utf_8);
+ }
+
+ while ((c = *utf_16++) != '\0') {
+ if (IS_LEADING_SURROGATE(c)) {
+ /*
+ * Leading surrogate. Must be followed by
+ * a trailing surrogate.
+ */
+ c2 = *utf_16;
+ if (c2 == '\0') {
+ /*
+ * Oops, string ends with a lead
+ * surrogate. Try to drop in
+ * a REPLACEMENT CHARACTER, and
+ * don't move the string pointer,
+ * so on the next trip through
+ * the loop we grab the terminating
+ * '\0' and quit.
+ */
+ uc = REPLACEMENT_CHARACTER;
+ } else {
+ /*
+ * OK, we can consume this 2-octet
+ * value.
+ */
+ utf_16++;
+ if (IS_TRAILING_SURROGATE(c2)) {
+ /*
+ * Trailing surrogate.
+ * This calculation will,
+ * for c being a leading
+ * surrogate and c2 being
+ * a trailing surrogate,
+ * produce a value between
+ * 0x100000 and 0x10ffff,
+ * so it's always going to be
+ * a valid Unicode code point.
+ */
+ uc = SURROGATE_VALUE(c, c2);
+ } else {
+ /*
+ * Not a trailing surrogate;
+ * try to drop in a
+ * REPLACEMENT CHARACTER.
+ */
+ uc = REPLACEMENT_CHARACTER;
+ }
+ }
+ } else {
+ /*
+ * Not a leading surrogate.
+ */
+ if (IS_TRAILING_SURROGATE(c)) {
+ /*
+ * Trailing surrogate without
+ * a preceding leading surrogate.
+ * Try to drop in a REPLACEMENT
+ * CHARACTER.
+ */
+ uc = REPLACEMENT_CHARACTER;
+ } else {
+ /*
+ * This is a valid BMP character;
+ * drop it in.
+ */
+ uc = c;
+ }
+ }
+
+ /*
+ * OK, uc is a valid Unicode character; how
+ * many bytes worth of UTF-8 does it require?
+ */
+ if (uc < 0x0080) {
+ /* 1 byte. */
+ if (utf_8_len < 2) {
+ /*
+ * Not enough room for that byte
+ * plus a trailing '\0'.
+ */
+ break;
+ }
+ *utf_8++ = (char)uc;
+ utf_8_len--;
+ } else if (uc < 0x0800) {
+ /* 2 bytes. */
+ if (utf_8_len < 3) {
+ /*
+ * Not enough room for those bytes
+ * plus a trailing '\0'.
+ */
+ break;
+ }
+ *utf_8++ = ((uc >> 6) & 0x3F) | 0xC0;
+ *utf_8++ = ((uc >> 0) & 0x3F) | 0x80;
+ utf_8_len -= 2;
+ } else if (uc < 0x010000) {
+ /* 3 bytes. */
+ if (utf_8_len < 4) {
+ /*
+ * Not enough room for those bytes
+ * plus a trailing '\0'.
+ */
+ break;
+ }
+ *utf_8++ = ((uc >> 12) & 0x0F) | 0xE0;
+ *utf_8++ = ((uc >> 6) & 0x3F) | 0x80;
+ *utf_8++ = ((uc >> 0) & 0x3F) | 0x80;
+ utf_8_len -= 3;
+ } else {
+ /* 4 bytes. */
+ if (utf_8_len < 5) {
+ /*
+ * Not enough room for those bytes
+ * plus a trailing '\0'.
+ */
+ break;
+ }
+ *utf_8++ = ((uc >> 18) & 0x03) | 0xF0;
+ *utf_8++ = ((uc >> 12) & 0x3F) | 0x80;
+ *utf_8++ = ((uc >> 6) & 0x3F) | 0x80;
+ *utf_8++ = ((uc >> 0) & 0x3F) | 0x80;
+ utf_8_len -= 3;
+ }
+ }
+
+ /*
+ * OK, we have enough room for (at least) a trailing '\0'.
+ * (We started out with enough room, thanks to the test
+ * for a zero-length buffer at the beginning, and if
+ * there wasn't enough room for any character we wanted
+ * to put into the buffer *plus* a trailing '\0',
+ * we'd have quit before putting it into the buffer,
+ * and thus would have left enough room for the trailing
+ * '\0'.)
+ *
+ * Drop it in.
+ */
+ *utf_8 = '\0';
+
+ /*
+ * Return a pointer to the terminating '\0', in case we
+ * want to drop something in after that.
+ */
+ return (utf_8);
+}
+#endif /* _WIN32 */
+
projects / libpcap / blobdiff