X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fdata-out.c;h=df447a6a12cac594c38937725030767c44d7999e;hb=81579d9e9f994fb2908f50af41c3eb033d216e58;hp=bb774374d476cbc7b6c04aa0db60c7ecc452fc04;hpb=4e8257086ffc71bc5a1785fd86610921be677887;p=pspp-builds.git diff --git a/src/data/data-out.c b/src/data/data-out.c index bb774374..df447a6a 100644 --- a/src/data/data-out.c +++ b/src/data/data-out.c @@ -16,7 +16,7 @@ #include -#include "data-out.h" +#include "data/data-out.h" #include #include @@ -24,22 +24,23 @@ #include #include #include - -#include -#include -#include -#include - -#include -#include -#include -#include -#include -#include -#include -#include - -#include "minmax.h" +#include + +#include "data/calendar.h" +#include "data/format.h" +#include "data/settings.h" +#include "data/value.h" +#include "libpspp/assertion.h" +#include "libpspp/cast.h" +#include "libpspp/float-format.h" +#include "libpspp/i18n.h" +#include "libpspp/integer-format.h" +#include "libpspp/message.h" +#include "libpspp/misc.h" +#include "libpspp/pool.h" +#include "libpspp/str.h" + +#include "gl/minmax.h" #include "gettext.h" #define _(msgid) gettext (msgid) @@ -92,57 +93,103 @@ static data_out_converter_func *const converters[FMT_NUMBER_OF_FORMATS] = #include "format.def" }; -/* Similar to data_out. Additionally recodes the output from - native form into the given legacy character ENCODING. - OUTPUT must be provided by the caller and must be at least - FORMAT->w long. No null terminator is appended to OUTPUT. -*/ +/* Converts the INPUT value, encoded in INPUT_ENCODING, according to format + specification FORMAT, appending the output to OUTPUT in OUTPUT_ENCODING. + However, binary formats (FMT_P, FMT_PK, FMT_IB, FMT_PIB, FMT_RB) yield the + binary results, which may not be properly encoded for OUTPUT_ENCODING. + + VALUE must be the correct width for FORMAT, that is, its width must be + fmt_var_width(FORMAT). + + INPUT_ENCODING can normally be obtained by calling dict_get_encoding() on + the dictionary with which INPUT is associated. ENCODING is only important + when FORMAT's type is FMT_A. */ void -data_out_legacy (const union value *input, const char *encoding, - const struct fmt_spec *format, char *output) +data_out_recode (const union value *input, const char *input_encoding, + const struct fmt_spec *format, + struct string *output, const char *output_encoding) { assert (fmt_check_output (format)); + if (format->type == FMT_A) + { + char *in = CHAR_CAST (char *, value_str (input, format->w)); + char *out = recode_string (output_encoding, input_encoding, + in, format->w); + ds_put_cstr (output, out); + free (out); + } + else if (fmt_get_category (format->type) == FMT_CAT_BINARY) + converters[format->type] (input, format, + ds_put_uninit (output, format->w)); + else + { + char *utf8_encoded = data_out (input, input_encoding, format); + char *output_encoded = recode_string (output_encoding, UTF8, + utf8_encoded, -1); + ds_put_cstr (output, output_encoded); + free (output_encoded); + free (utf8_encoded); + } +} - converters[format->type] (input, format, output); - if (0 != strcmp (encoding, C_ENCODING) - && fmt_get_category (format->type) != FMT_CAT_BINARY) +static char * +binary_to_utf8 (const char *in, struct pool *pool) +{ + uint8_t *out = pool_alloc_unaligned (pool, strlen (in) * 2 + 1); + uint8_t *p = out; + + while (*in != '\0') { - char *s = recode_string (encoding, C_ENCODING, output, format->w ); - memcpy (output, s, format->w); - free (s); + uint8_t byte = *in++; + int mblen = u8_uctomb (p, byte, 2); + assert (mblen > 0); + p += mblen; } + *p = '\0'; + + return CHAR_CAST (char *, out); } -/* Converts the INPUT value into a UTF8 encoded string, according - to format specification FORMAT. +/* Converts the INPUT value into a UTF-8 encoded string, according to format + specification FORMAT. - VALUE must be the correct width for FORMAT, that is, its - width must be fmt_var_width(FORMAT). + VALUE must be the correct width for FORMAT, that is, its width must be + fmt_var_width(FORMAT). - ENCODING must be the encoding of INPUT. Normally this can - be obtained by calling dict_get_encoding on the dictionary - with which INPUT is associated. + ENCODING must be the encoding of INPUT. Normally this can be obtained by + calling dict_get_encoding() on the dictionary with which INPUT is + associated. ENCODING is only important when FORMAT's type is FMT_A. - The return value is dynamically allocated, and must be freed - by the caller. If POOL is non-null, then the return value is - allocated on that pool. -*/ + The return value is dynamically allocated, and must be freed by the caller. + If POOL is non-null, then the return value is allocated on that pool. */ char * data_out_pool (const union value *input, const char *encoding, const struct fmt_spec *format, struct pool *pool) { - const struct fmt_number_style *style = settings_get_style (format->type); - char *output; - char *t ; assert (fmt_check_output (format)); + if (format->type == FMT_A) + { + char *in = CHAR_CAST (char *, value_str (input, format->w)); + return recode_string_pool (UTF8, encoding, in, format->w, pool); + } + else if (fmt_get_category (format->type) == FMT_CAT_BINARY) + { + char tmp[16]; - output = xmalloc (format->w + style->extra_bytes + 1); - - converters[format->type] (input, format, output); + assert (format->w + 1 <= sizeof tmp); + converters[format->type] (input, format, tmp); + return binary_to_utf8 (tmp, pool); + } + else + { + const struct fmt_number_style *style = settings_get_style (format->type); + size_t size = format->w + style->extra_bytes + 1; + char *output; - t = recode_string_pool (UTF8, encoding, output, format->w, pool); - free (output); - return t; + output = pool_alloc_unaligned (pool, size); + converters[format->type] (input, format, output); + return output; + } } char * @@ -215,11 +262,9 @@ output_Z (const union value *input, const struct fmt_spec *format, char buf[128]; if (input->f == SYSMIS) output_missing (format, output); - else if (fabs (number) >= power10 (format->w) - || sprintf (buf, "%0*.0f", format->w, - fabs (round (number))) != format->w) - output_overflow (format, output); - else + else if (fabs (number) < power10 (format->w) + && sprintf (buf, "%0*.0f", format->w, + fabs (round (number))) == format->w) { if (number < 0 && strspn (buf, "0") < format->w) { @@ -229,6 +274,8 @@ output_Z (const union value *input, const struct fmt_spec *format, memcpy (output, buf, format->w); output[format->w] = '\0'; } + else + output_overflow (format, output); } /* Outputs P format. */ @@ -529,11 +576,10 @@ output_MONTH (const union value *input, const struct fmt_spec *format, /* Outputs A format. */ static void -output_A (const union value *input, const struct fmt_spec *format, - char *output) +output_A (const union value *input UNUSED, + const struct fmt_spec *format UNUSED, char *output UNUSED) { - memcpy (output, value_str (input, format->w), format->w); - output[format->w] = '\0'; + NOT_REACHED (); } /* Outputs AHEX format. */ @@ -682,7 +728,7 @@ output_scientific (double number, const struct fmt_spec *format, int width; int fraction_width; bool add_affixes; - char buf[64], *p; + char *p; /* Allocate minimum required space. */ width = 6 + style->neg_suffix.width; @@ -706,7 +752,7 @@ output_scientific (double number, const struct fmt_spec *format, width += fraction_width; /* Format (except suffix). */ - p = buf; + p = output; if (width < format->w) p = mempset (p, ' ', format->w - width); if (number < 0)