X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flibpspp%2Fi18n.c;h=4934617121c786f5b19c6db88cc20e6f2033f116;hb=d01cc07b11b5919369bf4e8989360c2b4fe0380c;hp=06c67d947928122ec8e75f665a4e6f9b62d9df45;hpb=fb1c27ed0cbc16d151d182dd3f2be5eb3189a045;p=pspp diff --git a/src/libpspp/i18n.c b/src/libpspp/i18n.c index 06c67d9479..4934617121 100644 --- a/src/libpspp/i18n.c +++ b/src/libpspp/i18n.c @@ -15,31 +15,30 @@ along with this program. If not, see . */ #include -#include + +#include "libpspp/i18n.h" + #include +#include +#include +#include +#include #include -#include #include +#include #include -#include -#include -#include -#include -#include "assertion.h" -#include "hmapx.h" -#include "hash-functions.h" -#include "pool.h" - -#include "i18n.h" -#include "version.h" +#include "libpspp/assertion.h" +#include "libpspp/hmapx.h" +#include "libpspp/hash-functions.h" +#include "libpspp/pool.h" +#include "libpspp/str.h" +#include "libpspp/version.h" -#include -#include "xstrndup.h" - -#if HAVE_NL_LANGINFO -#include -#endif +#include "gl/localcharset.h" +#include "gl/xalloc.h" +#include "gl/relocatable.h" +#include "gl/xstrndup.h" struct converter { @@ -98,6 +97,61 @@ recode_string (const char *to, const char *from, } +/* Uses CONV to convert the INBYTES starting at IP into the OUTBYTES starting + at OP, and appends a null terminator to the output. + + Returns the output length if successful, -1 if the output buffer is too + small. */ +static ssize_t +try_recode (iconv_t conv, + const char *ip, size_t inbytes, + char *op_, size_t outbytes) +{ + /* FIXME: Need to ensure that this char is valid in the target encoding */ + const char fallbackchar = '?'; + char *op = op_; + + /* Put the converter into the initial shift state, in case there was any + state information left over from its last usage. */ + iconv (conv, NULL, 0, NULL, 0); + + while (iconv (conv, (ICONV_CONST char **) &ip, &inbytes, + &op, &outbytes) == -1) + switch (errno) + { + case EINVAL: + if (outbytes < 2) + return -1; + *op++ = fallbackchar; + *op = '\0'; + return op - op_; + + case EILSEQ: + if (outbytes == 0) + return -1; + *op++ = fallbackchar; + outbytes--; + ip++; + inbytes--; + break; + + case E2BIG: + return -1; + + default: + /* should never happen */ + fprintf (stderr, "Character conversion error: %s\n", strerror (errno)); + NOT_REACHED (); + break; + } + + if (outbytes == 0) + return -1; + + *op = '\0'; + return op - op_; +} + /* Converts the string TEXT, which should be encoded in FROM-encoding, to a dynamically allocated string in TO-encoding. Any characters which cannot be converted will be represented by '?'. @@ -114,22 +168,34 @@ char * recode_string_pool (const char *to, const char *from, const char *text, int length, struct pool *pool) { - char *outbuf = 0; - size_t outbufferlength; - size_t result; - char *op ; - size_t inbytes = 0; - size_t outbytes ; - iconv_t conv ; - - /* FIXME: Need to ensure that this char is valid in the target encoding */ - const char fallbackchar = '?'; + struct substring out; if ( text == NULL ) return NULL; if ( length == -1 ) - length = strlen(text); + length = strlen (text); + + out = recode_substring_pool (to, from, ss_buffer (text, length), pool); + return out.string; +} + +/* Converts the string TEXT, which should be encoded in FROM-encoding, to a + dynamically allocated string in TO-encoding. Any characters which cannot be + converted will be represented by '?'. + + The returned string will be null-terminated and allocated on POOL. + + This function's behaviour differs from that of g_convert_with_fallback + provided by GLib. The GLib function will fail (returns NULL) if any part of + the input string is not valid in the declared input encoding. This function + however perseveres even in the presence of badly encoded input. */ +struct substring +recode_substring_pool (const char *to, const char *from, + struct substring text, struct pool *pool) +{ + size_t outbufferlength; + iconv_t conv ; if (to == NULL) to = default_encoding; @@ -140,79 +206,26 @@ recode_string_pool (const char *to, const char *from, conv = create_iconv (to, from); if ( (iconv_t) -1 == conv ) - return xstrdup (text); - - /* Put the converter into the initial shift state, in case there was any - state information left over from its last usage. */ - iconv (conv, NULL, 0, NULL, 0); + { + struct substring out; + ss_alloc_substring (&out, text); + return out; + } for ( outbufferlength = 1 ; outbufferlength != 0; outbufferlength <<= 1 ) - if ( outbufferlength > length) - break; - - outbuf = pool_malloc (pool, outbufferlength); - op = outbuf; - - outbytes = outbufferlength; - inbytes = length; - - - do { - const char *ip = text; - result = iconv (conv, (ICONV_CONST char **) &text, &inbytes, - &op, &outbytes); - - if ( -1 == result ) + if ( outbufferlength > text.length) { - int the_error = errno; - - switch (the_error) - { - case EILSEQ: - case EINVAL: - if ( outbytes > 0 ) - { - *op++ = fallbackchar; - outbytes--; - text++; - inbytes--; - break; - } - /* Fall through */ - case E2BIG: - iconv (conv, NULL, 0, NULL, 0); - pool_free (pool, outbuf); - outbufferlength <<= 1; - outbuf = pool_malloc (pool, outbufferlength); - op = outbuf; - outbytes = outbufferlength; - inbytes = length; - text = ip; - break; - default: - /* should never happen */ - fprintf (stderr, "Character conversion error: %s\n", - strerror (the_error)); - NOT_REACHED (); - break; - } + char *output = pool_malloc (pool, outbufferlength); + ssize_t output_len = try_recode (conv, text.string, text.length, + output, outbufferlength); + if (output_len >= 0) + return ss_buffer (output, output_len); + pool_free (pool, output); } - } while ( -1 == result ); - - if (outbytes == 0 ) - { - char *const oldaddr = outbuf; - outbuf = pool_realloc (pool, outbuf, outbufferlength + 1); - - op += (outbuf - oldaddr) ; - } - - *op = '\0'; - return outbuf; + NOT_REACHED (); } - void i18n_init (void) { @@ -316,7 +329,7 @@ i18n_done (void) bool valid_encoding (const char *enc) { - iconv_t conv = iconv_open ("UTF8", enc); + iconv_t conv = iconv_open (UTF8, enc); if ( conv == (iconv_t) -1) return false;