+ errorf(&lexer_token.source_position, "%s", msg);
+}
+
+/**
+ * Prints an internal error message at the current token.
+ *
+ * @param msg the error message
+ */
+static NORETURN internal_error(const char *msg)
+{
+ internal_errorf(&lexer_token.source_position, "%s", msg);
+}
+
+static size_t read_block(unsigned char *const read_buf, size_t const n)
+{
+ size_t const s = fread(read_buf, 1, n, input);
+ if (s == 0) {
+ /* on OS/X ferror appears to return true on eof as well when running
+ * the application in gdb... */
+ if (!feof(input) && ferror(input))
+ parse_error("read from input failed");
+ buf[MAX_PUTBACK] = EOF;
+ bufpos = buf + MAX_PUTBACK;
+ bufend = buf + MAX_PUTBACK + 1;
+ }
+ return s;
+}
+
+static void decode_iso_8859_1(void)
+{
+ unsigned char read_buf[BUF_SIZE];
+ size_t const s = read_block(read_buf, sizeof(read_buf));
+ if (s == 0)
+ return;
+
+ unsigned char const *src = read_buf;
+ unsigned char const *end = read_buf + s;
+ utf32 *dst = buf + MAX_PUTBACK;
+ while (src != end)
+ *dst++ = *src++;
+
+ bufpos = buf + MAX_PUTBACK;
+ bufend = dst;
+}
+
+static void decode_iso_8859_15(void)
+{
+ unsigned char read_buf[BUF_SIZE];
+ size_t const s = read_block(read_buf, sizeof(read_buf));
+ if (s == 0)
+ return;
+
+ unsigned char const *src = read_buf;
+ unsigned char const *end = read_buf + s;
+ utf32 *dst = buf + MAX_PUTBACK;
+ while (src != end) {
+ utf32 tc = *src++;
+ switch (tc) {
+ case 0xA4: tc = 0x20AC; break; // €
+ case 0xA6: tc = 0x0160; break; // Š
+ case 0xA8: tc = 0x0161; break; // š
+ case 0xB4: tc = 0x017D; break; // Ž
+ case 0xB8: tc = 0x017E; break; // ž
+ case 0xBC: tc = 0x0152; break; // Œ
+ case 0xBD: tc = 0x0153; break; // œ
+ case 0xBE: tc = 0x0178; break; // Ÿ
+ }
+ *dst++ = tc;
+ }
+
+ bufpos = buf + MAX_PUTBACK;
+ bufend = dst;
+}
+
+static void decode_utf8(void)
+{
+ static utf32 part_decoded_min_code;
+ static utf32 part_decoded_char;
+ static size_t part_decoded_rest_len;
+
+ do {
+ unsigned char read_buf[BUF_SIZE];
+ size_t const s = read_block(read_buf, sizeof(read_buf));
+ if (s == 0) {
+ if (part_decoded_rest_len > 0)
+ parse_error("incomplete input char at end of input");
+ return;
+ }
+
+ unsigned char const *src = read_buf;
+ unsigned char const *end = read_buf + s;
+ utf32 *dst = buf + MAX_PUTBACK;
+ utf32 decoded;
+ utf32 min_code;
+
+ if (part_decoded_rest_len != 0) {
+ min_code = part_decoded_min_code;
+ decoded = part_decoded_char;
+ size_t const rest_len = part_decoded_rest_len;
+ part_decoded_rest_len = 0;
+ switch (rest_len) {
+ case 4: goto realign;
+ case 3: goto three_more;
+ case 2: goto two_more;
+ default: goto one_more;
+ }
+ }
+
+ while (src != end) {
+ if ((*src & 0x80) == 0) {
+ decoded = *src++;
+ } else if ((*src & 0xE0) == 0xC0) {
+ min_code = 0x80;
+ decoded = *src++ & 0x1F;
+one_more:
+ if (src == end) {
+ part_decoded_min_code = min_code;
+ part_decoded_char = decoded;
+ part_decoded_rest_len = 1;
+ break;
+ }
+ if ((*src & 0xC0) == 0x80) {
+ decoded = (decoded << 6) | (*src++ & 0x3F);
+ } else {
+ goto invalid_char;
+ }
+ if (decoded < min_code ||
+ decoded > 0x10FFFF ||
+ (0xD800 <= decoded && decoded < 0xE000) || // high/low surrogates
+ (0xFDD0 <= decoded && decoded < 0xFDF0) || // noncharacters
+ (decoded & 0xFFFE) == 0xFFFE) { // noncharacters
+ parse_error("invalid byte sequence in input");
+ }
+ } else if ((*src & 0xF0) == 0xE0) {
+ min_code = 0x800;
+ decoded = *src++ & 0x0F;
+two_more:
+ if (src == end) {
+ part_decoded_min_code = min_code;
+ part_decoded_char = decoded;
+ part_decoded_rest_len = 2;
+ break;
+ }
+ if ((*src & 0xC0) == 0x80) {
+ decoded = (decoded << 6) | (*src++ & 0x3F);
+ } else {
+ goto invalid_char;
+ }
+ goto one_more;
+ } else if ((*src & 0xF8) == 0xF0) {
+ min_code = 0x10000;
+ decoded = *src++ & 0x07;
+three_more:
+ if (src == end) {
+ part_decoded_min_code = min_code;
+ part_decoded_char = decoded;
+ part_decoded_rest_len = 3;
+ break;
+ }
+ if ((*src & 0xC0) == 0x80) {
+ decoded = (decoded << 6) | (*src++ & 0x3F);
+ } else {
+ goto invalid_char;
+ }
+ goto two_more;
+ } else {
+invalid_char:
+ parse_error("invalid byte sequence in input");
+realign:
+ do {
+ ++src;
+ if (src == end) {
+ part_decoded_rest_len = 4;
+ break;
+ }
+ } while ((*src & 0xC0) == 0x80 || (*src & 0xF8) == 0xF8);
+ continue;
+ }
+ *dst++ = decoded;
+ }
+
+ bufpos = buf + MAX_PUTBACK;
+ bufend = dst;
+ } while (bufpos == bufend);
+}
+
+static void decode_windows_1252(void)
+{
+ unsigned char read_buf[BUF_SIZE];
+ size_t const s = read_block(read_buf, sizeof(read_buf));
+ if (s == 0)
+ return;
+
+ unsigned char const *src = read_buf;
+ unsigned char const *end = read_buf + s;
+ utf32 *dst = buf + MAX_PUTBACK;
+ while (src != end) {
+ utf32 tc = *src++;
+ switch (tc) {
+ case 0x80: tc = 0x20AC; break; // €
+ case 0x82: tc = 0x201A; break; // ‚
+ case 0x83: tc = 0x0192; break; // ƒ
+ case 0x84: tc = 0x201E; break; // „
+ case 0x85: tc = 0x2026; break; // …
+ case 0x86: tc = 0x2020; break; // †
+ case 0x87: tc = 0x2021; break; // ‡
+ case 0x88: tc = 0x02C6; break; // ˆ
+ case 0x89: tc = 0x2030; break; // ‰
+ case 0x8A: tc = 0x0160; break; // Š
+ case 0x8B: tc = 0x2039; break; // ‹
+ case 0x8C: tc = 0x0152; break; // Œ
+ case 0x8E: tc = 0x017D; break; // Ž
+ case 0x91: tc = 0x2018; break; // ‘
+ case 0x92: tc = 0x2019; break; // ’
+ case 0x93: tc = 0x201C; break; // “
+ case 0x94: tc = 0x201D; break; // ”
+ case 0x95: tc = 0x2022; break; // •
+ case 0x96: tc = 0x2013; break; // –
+ case 0x97: tc = 0x2014; break; // —
+ case 0x98: tc = 0x02DC; break; // ˜
+ case 0x99: tc = 0x2122; break; // ™
+ case 0x9A: tc = 0x0161; break; // š
+ case 0x9B: tc = 0x203A; break; // ›
+ case 0x9C: tc = 0x0153; break; // œ
+ case 0x9E: tc = 0x017E; break; // ž
+ case 0x9F: tc = 0x0178; break; // Ÿ
+ }
+ *dst++ = tc;
+ }
+
+ bufpos = buf + MAX_PUTBACK;
+ bufend = dst;
+}
+
+typedef void (*decoder_t)(void);
+
+static decoder_t decoder = decode_utf8;
+
+typedef struct named_decoder_t {
+ char const *name;
+ decoder_t decoder;
+} named_decoder_t;
+
+static named_decoder_t const decoders[] = {
+ { "CP819", decode_iso_8859_1 }, // offical alias
+ { "IBM819", decode_iso_8859_1 }, // offical alias
+ { "ISO-8859-1", decode_iso_8859_1 }, // offical alias
+ { "ISO-8859-15", decode_iso_8859_15 }, // offical name
+ { "ISO8859-1", decode_iso_8859_1 },
+ { "ISO8859-15", decode_iso_8859_15 },
+ { "ISO_8859-1", decode_iso_8859_1 }, // offical alias
+ { "ISO_8859-15", decode_iso_8859_15 }, // offical alias
+ { "ISO_8859-1:1987", decode_iso_8859_1 }, // offical name
+ { "Latin-9", decode_iso_8859_15 }, // offical alias
+ { "UTF-8", decode_utf8 }, // offical name
+ { "csISOLatin1", decode_iso_8859_1 }, // offical alias
+ { "cp1252", decode_windows_1252 },
+ { "iso-ir-100", decode_iso_8859_1 }, // offical alias
+ { "l1", decode_iso_8859_1 }, // offical alias
+ { "latin1", decode_iso_8859_1 }, // offical alias
+ { "windows-1252", decode_windows_1252 }, // official name
+
+ { NULL, NULL }
+};
+
+/** strcasecmp is not part of C99 so we need our own implementation here */
+static int my_strcasecmp(const char *s1, const char *s2)
+{
+ for ( ; *s1 != 0; ++s1, ++s2) {
+ if (tolower(*s1) != tolower(*s2))
+ break;
+ }
+ return (unsigned char)*s1 - (unsigned char)*s2;
+}
+
+void select_input_encoding(char const* const encoding)
+{
+ for (named_decoder_t const *i = decoders; i->name != NULL; ++i) {
+ if (my_strcasecmp(encoding, i->name) != 0)
+ continue;
+ decoder = i->decoder;
+ return;
+ }
+ fprintf(stderr, "error: input encoding \"%s\" not supported\n", encoding);