/* PSPP - a program for statistical analysis.
- Copyright (C) 2011 Free Software Foundation, Inc.
+ Copyright (C) 2011, 2012, 2013 Free Software Foundation, Inc.
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
of information about encoding detection.
*/
-/* Parses and returns the fallback encoding from ENCODING, which must be in one
- of the forms described at the top of encoding-guesser.h. The returned
- string might be ENCODING itself or a suffix of it, or it might be a
- statically allocated string. */
+/* Returns the encoding specified by ENCODING, which must be in one of the
+ forms described at the top of encoding-guesser.h. The returned string might
+ be ENCODING itself or a suffix of it, or it might be a statically allocated
+ string. */
const char *
encoding_guess_parse_encoding (const char *encoding)
{
get_le32 (const uint8_t *data)
{
return (data[3] << 24) | (data[2] << 16) | (data[1] << 8) | data[0];
-
}
static const char *
return true;
}
+static bool
+is_utf8_bom (const uint8_t *data, size_t n)
+{
+ return n >= 3 && data[0] == 0xef && data[1] == 0xbb && data[2] == 0xbf;
+}
+
+static bool
+is_bom_length (size_t n, size_t w)
+{
+ return n >= ENCODING_GUESS_MIN || (n && n % w == 0);
+}
+
+static bool
+is_utf16le_bom (const uint8_t *data, size_t n)
+{
+ return is_bom_length (n, 2) && get_le16 (data) == 0xfeff;
+}
+
+static bool
+is_utf16be_bom (const uint8_t *data, size_t n)
+{
+ return is_bom_length (n, 2) && get_be16 (data) == 0xfeff;
+}
+
+static bool
+is_utf32le_bom (const uint8_t *data, size_t n)
+{
+ return is_bom_length (n, 4) && get_le32 (data) == 0xfeff;
+}
+
+static bool
+is_utf32be_bom (const uint8_t *data, size_t n)
+{
+ return is_bom_length (n, 4) && get_be32 (data) == 0xfeff;
+}
+
/* Attempts to guess the encoding of a text file based on ENCODING, an encoding
name in one of the forms described at the top of encoding-guesser.h, and
DATA, which contains the first N bytes of the file. Returns the guessed
if (n == 0)
return fallback_encoding;
- if ((n >= ENCODING_GUESS_MIN || n % 4 == 0)
- && (get_be32 (data) == 0xfeff || get_le32 (data) == 0xfeff))
+ if (is_utf32be_bom (data, n) || is_utf32le_bom (data, n))
return "UTF-32";
if (n >= 4)
return "UTF-EBCDIC";
}
- if ((n >= ENCODING_GUESS_MIN || n % 2 == 0)
- && (get_be16 (data) == 0xfeff || get_le16 (data) == 0xfeff))
+ if (is_utf16be_bom (data, n) || is_utf16le_bom (data, n))
return "UTF-16";
- if (n >= 3 && data[0] == 0xef && data[1] == 0xbb && data[2] == 0xbf)
+ if (is_utf8_bom (data, n))
return "UTF-8";
guess = guess_utf16 (data, n);
if (is_utf32 (data, n, get_le32))
return "UTF-32LE";
- if (!is_encoding_ascii_compatible (fallback_encoding)
- || !encoding_guess_tail_is_utf8 (data, n))
+ /* We've tried all the "giveaways" that make the encoding obvious. That
+ rules out, incidentally, all the encodings with multibyte units
+ (e.g. UTF-16, UTF-32). Our remaining goal is to try to distinguish UTF-8
+ from some ASCII-based fallback encoding. */
+
+ /* If the fallback encoding isn't ASCII compatible, give up. */
+ if (!is_encoding_ascii_compatible (fallback_encoding))
return fallback_encoding;
+ /* If the data we have clearly is not UTF-8, give up. */
+ if (!encoding_guess_tail_is_utf8 (data, n))
+ {
+ /* If the fallback encoding is UTF-8, fall back on something else.*/
+ if (is_encoding_utf8 (fallback_encoding))
+ return "windows-1252";
+
+ return fallback_encoding;
+ }
+
return "ASCII";
}
+static bool
+is_encoding_utf16 (const char *encoding)
+{
+ return (!c_strcasecmp (encoding, "utf-16")
+ || !c_strcasecmp (encoding, "utf16"));
+}
+
+static bool
+is_encoding_utf32 (const char *encoding)
+{
+ return (!c_strcasecmp (encoding, "utf-32")
+ || !c_strcasecmp (encoding, "utf32"));
+}
+
+/* If ENCODING is the name of an encoding that could begin with a byte-order
+ mark, and in fact the N bytes in DATA do begin with a byte-order mark,
+ returns the number of bytes in the byte-order mark. Otherwise, returns 0.
+
+ N must be at least ENCODING_GUESS_MIN, unless the file is shorter than
+ that. */
+size_t
+encoding_guess_bom_length (const char *encoding,
+ const void *data_, size_t n)
+{
+ const uint8_t *data = data_;
+
+ return (is_utf8_bom (data, n) && is_encoding_utf8 (encoding) ? 3
+ : is_utf16le_bom (data, n) && is_encoding_utf16 (encoding) ? 2
+ : is_utf16be_bom (data, n) && is_encoding_utf16 (encoding) ? 2
+ : is_utf32le_bom (data, n) && is_encoding_utf32 (encoding) ? 4
+ : is_utf32be_bom (data, n) && is_encoding_utf32 (encoding) ? 4
+ : 0);
+}
+
/* Returns an encoding guess based on ENCODING and the N bytes of text starting
at DATA. DATA should start with the first non-ASCII text character (as
determined by encoding_guess_is_ascii_text()) found in the input.
encoding_guess_tail_encoding (const char *encoding,
const void *data, size_t n)
{
- return (encoding_guess_tail_is_utf8 (data, n)
- ? "UTF-8"
- : encoding_guess_parse_encoding (encoding));
+
+ if (encoding_guess_tail_is_utf8 (data, n) != 0)
+ return "UTF-8";
+ else
+ {
+ /* The data is not UTF-8. */
+ const char *fallback_encoding = encoding_guess_parse_encoding (encoding);
+
+ /* If the fallback encoding is UTF-8, fall back on something else.*/
+ if (is_encoding_utf8 (fallback_encoding))
+ return "windows-1252";
+
+ return fallback_encoding;
+ }
+
}
-/* Same as encoding_guess_tail_encoding() but returns true for UTF-8 or false
- for the fallback encoding. */
-bool
+/* Returns an encoding guess based on ENCODING and the N bytes of text starting
+ at DATA. DATA should start with the first non-ASCII text character (as
+ determined by encoding_guess_is_ascii_text()) found in the input.
+
+ The return value is:
+
+ 0, if the encoding is definitely not UTF-8 (because the input contains
+ byte sequences that are not valid in UTF-8).
+
+ 1, if the encoding appears to be UTF-8 (because the input contains valid
+ UTF-8 multibyte sequences).
+
+ -1, if the input contains only ASCII characters. (This means that the
+ input may be treated as UTF-8, since ASCII is a subset of UTF-8.)
+
+ See encoding-guesser.h for intended use of this function.
+
+ N must be at least ENCODING_GUESS_MIN, unless the file has fewer bytes than
+ that starting with the first non-ASCII text character. */
+int
encoding_guess_tail_is_utf8 (const void *data, size_t n)
{
+ /* If all the bytes are in the ASCII range, it's just ASCII. */
+ if (encoding_guess_count_ascii (data, n) == n)
+ return -1;
+
return (n < ENCODING_GUESS_MIN
? u8_check (data, n) == NULL
: is_all_utf8_text (data, n));
guess = encoding_guess_head_encoding (encoding, text, size);
if (!strcmp (guess, "ASCII") && encoding_guess_encoding_is_auto (encoding))
- {
- size_t ofs = encoding_guess_count_ascii (text, size);
- if (ofs < size)
- return encoding_guess_tail_encoding (encoding,
- (const char *) text + ofs,
- size - ofs);
- else
- return encoding_guess_parse_encoding (encoding);
- }
+ return encoding_guess_tail_encoding (encoding, text, size);
else
return guess;
}