mirror of
https://github.com/python/cpython.git
synced 2025-09-26 18:29:57 +00:00
Backport of some of the work in r71665 to trunk. This reworks much of
int, long, and float __format__(), and it keeps their implementation in sync with py3k. Also added PyOS_double_to_string. This is the "fallback" version that's also available in trunk, and should be kept in sync with that code. I'll add an issue to document PyOS_double_to_string in the C API. There are many internal cleanups. Externally visible changes include: - Implement PEP 378, Format Specifier for Thousands Separator, for floats, ints, and longs. - Issue #5515: 'n' formatting for ints, longs, and floats handles leading zero formatting poorly. - Issue #5772: For float.__format__, don't add a trailing ".0" if we're using no type code and we have an exponent.
This commit is contained in:
Eric Smith
parent
cbb5308723
commit
aca19e6a74
9 changed files with 894 additions and 496 deletions
|
|
@ -5,126 +5,208 @@
|
|||
|
||||
#include <locale.h>
|
||||
|
||||
#define MAX(x, y) ((x) < (y) ? (y) : (x))
|
||||
#define MIN(x, y) ((x) < (y) ? (x) : (y))
|
||||
|
||||
typedef struct {
|
||||
const char *grouping;
|
||||
char previous;
|
||||
Py_ssize_t i; /* Where we're currently pointing in grouping. */
|
||||
} GroupGenerator;
|
||||
|
||||
static void
|
||||
_GroupGenerator_init(GroupGenerator *self, const char *grouping)
|
||||
{
|
||||
self->grouping = grouping;
|
||||
self->i = 0;
|
||||
self->previous = 0;
|
||||
}
|
||||
|
||||
/* Returns the next grouping, or 0 to signify end. */
|
||||
static Py_ssize_t
|
||||
_GroupGenerator_next(GroupGenerator *self)
|
||||
{
|
||||
/* Note that we don't really do much error checking here. If a
|
||||
grouping string contains just CHAR_MAX, for example, then just
|
||||
terminate the generator. That shouldn't happen, but at least we
|
||||
fail gracefully. */
|
||||
switch (self->grouping[self->i]) {
|
||||
case 0:
|
||||
return self->previous;
|
||||
case CHAR_MAX:
|
||||
/* Stop the generator. */
|
||||
return 0;
|
||||
default: {
|
||||
char ch = self->grouping[self->i];
|
||||
self->previous = ch;
|
||||
self->i++;
|
||||
return (Py_ssize_t)ch;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/* Fill in some digits, leading zeros, and thousands separator. All
|
||||
are optional, depending on when we're called. */
|
||||
static void
|
||||
fill(STRINGLIB_CHAR **digits_end, STRINGLIB_CHAR **buffer_end,
|
||||
Py_ssize_t n_chars, Py_ssize_t n_zeros, const char* thousands_sep,
|
||||
Py_ssize_t thousands_sep_len)
|
||||
{
|
||||
#if STRINGLIB_IS_UNICODE
|
||||
Py_ssize_t i;
|
||||
#endif
|
||||
|
||||
if (thousands_sep) {
|
||||
*buffer_end -= thousands_sep_len;
|
||||
|
||||
/* Copy the thousands_sep chars into the buffer. */
|
||||
#if STRINGLIB_IS_UNICODE
|
||||
/* Convert from the char's of the thousands_sep from
|
||||
the locale into unicode. */
|
||||
for (i = 0; i < thousands_sep_len; ++i)
|
||||
(*buffer_end)[i] = thousands_sep[i];
|
||||
#else
|
||||
/* No conversion, just memcpy the thousands_sep. */
|
||||
memcpy(*buffer_end, thousands_sep, thousands_sep_len);
|
||||
#endif
|
||||
}
|
||||
|
||||
*buffer_end -= n_chars;
|
||||
*digits_end -= n_chars;
|
||||
memcpy(*buffer_end, *digits_end, n_chars * sizeof(STRINGLIB_CHAR));
|
||||
|
||||
*buffer_end -= n_zeros;
|
||||
STRINGLIB_FILL(*buffer_end, '0', n_zeros);
|
||||
}
|
||||
|
||||
/**
|
||||
* _Py_InsertThousandsGrouping:
|
||||
* @buffer: A pointer to the start of a string.
|
||||
* @n_buffer: The length of the string.
|
||||
* @n_buffer: Number of characters in @buffer.
|
||||
* @digits: A pointer to the digits we're reading from. If count
|
||||
* is non-NULL, this is unused.
|
||||
* @n_digits: The number of digits in the string, in which we want
|
||||
* to put the grouping chars.
|
||||
* @buf_size: The maximum size of the buffer pointed to by buffer.
|
||||
* @count: If non-NULL, points to a variable that will receive the
|
||||
* number of characters we need to insert (and no formatting
|
||||
* will actually occur).
|
||||
* @append_zero_char: If non-zero, put a trailing zero at the end of
|
||||
* of the resulting string, if and only if we modified the
|
||||
* string.
|
||||
* @min_width: The minimum width of the digits in the output string.
|
||||
* Output will be zero-padded on the left to fill.
|
||||
* @grouping: see definition in localeconv().
|
||||
* @thousands_sep: see definition in localeconv().
|
||||
*
|
||||
* Inserts thousand grouping characters (as defined in the current
|
||||
* locale) into the string between buffer and buffer+n_digits. If
|
||||
* count is non-NULL, don't do any formatting, just count the number
|
||||
* of characters to insert. This is used by the caller to
|
||||
* appropriately resize the buffer, if needed. If count is non-NULL,
|
||||
* buffer can be NULL (it is not dereferenced at all in that case).
|
||||
* There are 2 modes: counting and filling. If @buffer is NULL,
|
||||
* we are in counting mode, else filling mode.
|
||||
* If counting, the required buffer size is returned.
|
||||
* If filling, we know the buffer will be large enough, so we don't
|
||||
* need to pass in the buffer size.
|
||||
* Inserts thousand grouping characters (as defined by grouping and
|
||||
* thousands_sep) into the string between buffer and buffer+n_digits.
|
||||
*
|
||||
* Return value: 0 on error, else 1. Note that no error can occur if
|
||||
* count is non-NULL.
|
||||
*
|
||||
* This name won't be used, the includer of this file should define
|
||||
* it to be the actual function name, based on unicode or string.
|
||||
*
|
||||
* As closely as possible, this code mimics the logic in decimal.py's
|
||||
_insert_thousands_sep().
|
||||
**/
|
||||
int
|
||||
Py_ssize_t
|
||||
_Py_InsertThousandsGrouping(STRINGLIB_CHAR *buffer,
|
||||
Py_ssize_t n_buffer,
|
||||
Py_ssize_t n_digits,
|
||||
Py_ssize_t buf_size,
|
||||
Py_ssize_t *count,
|
||||
int append_zero_char)
|
||||
Py_ssize_t n_buffer,
|
||||
STRINGLIB_CHAR *digits,
|
||||
Py_ssize_t n_digits,
|
||||
Py_ssize_t min_width,
|
||||
const char *grouping,
|
||||
const char *thousands_sep)
|
||||
{
|
||||
struct lconv *locale_data = localeconv();
|
||||
const char *grouping = locale_data->grouping;
|
||||
const char *thousands_sep = locale_data->thousands_sep;
|
||||
Py_ssize_t thousands_sep_len = strlen(thousands_sep);
|
||||
STRINGLIB_CHAR *pend = NULL; /* current end of buffer */
|
||||
STRINGLIB_CHAR *pmax = NULL; /* max of buffer */
|
||||
char current_grouping;
|
||||
Py_ssize_t remaining = n_digits; /* Number of chars remaining to
|
||||
be looked at */
|
||||
Py_ssize_t count = 0;
|
||||
Py_ssize_t n_zeros;
|
||||
int loop_broken = 0;
|
||||
int use_separator = 0; /* First time through, don't append the
|
||||
separator. They only go between
|
||||
groups. */
|
||||
STRINGLIB_CHAR *buffer_end = NULL;
|
||||
STRINGLIB_CHAR *digits_end = NULL;
|
||||
Py_ssize_t l;
|
||||
Py_ssize_t n_chars;
|
||||
Py_ssize_t thousands_sep_len = strlen(thousands_sep);
|
||||
Py_ssize_t remaining = n_digits; /* Number of chars remaining to
|
||||
be looked at */
|
||||
/* A generator that returns all of the grouping widths, until it
|
||||
returns 0. */
|
||||
GroupGenerator groupgen;
|
||||
_GroupGenerator_init(&groupgen, grouping);
|
||||
|
||||
/* Initialize the character count, if we're just counting. */
|
||||
if (count)
|
||||
*count = 0;
|
||||
else {
|
||||
/* We're not just counting, we're modifying buffer */
|
||||
pend = buffer + n_buffer;
|
||||
pmax = buffer + buf_size;
|
||||
}
|
||||
if (buffer) {
|
||||
buffer_end = buffer + n_buffer;
|
||||
digits_end = digits + n_digits;
|
||||
}
|
||||
|
||||
/* Starting at the end and working right-to-left, keep track of
|
||||
what grouping needs to be added and insert that. */
|
||||
current_grouping = *grouping++;
|
||||
while ((l = _GroupGenerator_next(&groupgen)) > 0) {
|
||||
l = MIN(l, MAX(MAX(remaining, min_width), 1));
|
||||
n_zeros = MAX(0, l - remaining);
|
||||
n_chars = MAX(0, MIN(remaining, l));
|
||||
|
||||
/* If the first character is 0, perform no grouping at all. */
|
||||
if (current_grouping == 0)
|
||||
return 1;
|
||||
/* Use n_zero zero's and n_chars chars */
|
||||
|
||||
while (remaining > current_grouping) {
|
||||
/* Always leave buffer and pend valid at the end of this
|
||||
loop, since we might leave with a return statement. */
|
||||
/* Count only, don't do anything. */
|
||||
count += (use_separator ? thousands_sep_len : 0) + n_zeros + n_chars;
|
||||
|
||||
remaining -= current_grouping;
|
||||
if (count) {
|
||||
/* We're only counting, not touching the memory. */
|
||||
*count += thousands_sep_len;
|
||||
}
|
||||
else {
|
||||
/* Do the formatting. */
|
||||
if (buffer) {
|
||||
/* Copy into the output buffer. */
|
||||
fill(&digits_end, &buffer_end, n_chars, n_zeros,
|
||||
use_separator ? thousands_sep : NULL, thousands_sep_len);
|
||||
}
|
||||
|
||||
STRINGLIB_CHAR *plast = buffer + remaining;
|
||||
/* Use a separator next time. */
|
||||
use_separator = 1;
|
||||
|
||||
/* Is there room to insert thousands_sep_len chars? */
|
||||
if (pmax - pend < thousands_sep_len)
|
||||
/* No room. */
|
||||
return 0;
|
||||
remaining -= n_chars;
|
||||
min_width -= l;
|
||||
|
||||
/* Move the rest of the string down. */
|
||||
memmove(plast + thousands_sep_len,
|
||||
plast,
|
||||
(pend - plast) * sizeof(STRINGLIB_CHAR));
|
||||
/* Copy the thousands_sep chars into the buffer. */
|
||||
#if STRINGLIB_IS_UNICODE
|
||||
/* Convert from the char's of the thousands_sep from
|
||||
the locale into unicode. */
|
||||
{
|
||||
Py_ssize_t i;
|
||||
for (i = 0; i < thousands_sep_len; ++i)
|
||||
plast[i] = thousands_sep[i];
|
||||
}
|
||||
#else
|
||||
/* No conversion, just memcpy the thousands_sep. */
|
||||
memcpy(plast, thousands_sep, thousands_sep_len);
|
||||
#endif
|
||||
}
|
||||
if (remaining <= 0 && min_width <= 0) {
|
||||
loop_broken = 1;
|
||||
break;
|
||||
}
|
||||
min_width -= thousands_sep_len;
|
||||
}
|
||||
if (!loop_broken) {
|
||||
/* We left the loop without using a break statement. */
|
||||
|
||||
/* Adjust end pointer. */
|
||||
pend += thousands_sep_len;
|
||||
l = MAX(MAX(remaining, min_width), 1);
|
||||
n_zeros = MAX(0, l - remaining);
|
||||
n_chars = MAX(0, MIN(remaining, l));
|
||||
|
||||
/* Move to the next grouping character, unless we're
|
||||
repeating (which is designated by a grouping of 0). */
|
||||
if (*grouping != 0) {
|
||||
current_grouping = *grouping++;
|
||||
if (current_grouping == CHAR_MAX)
|
||||
/* We're done. */
|
||||
break;
|
||||
}
|
||||
}
|
||||
if (append_zero_char) {
|
||||
/* Append a zero character to mark the end of the string,
|
||||
if there's room. */
|
||||
if (pend - (buffer + remaining) < 1)
|
||||
/* No room, error. */
|
||||
return 0;
|
||||
*pend = 0;
|
||||
}
|
||||
return 1;
|
||||
/* Use n_zero zero's and n_chars chars */
|
||||
count += (use_separator ? thousands_sep_len : 0) + n_zeros + n_chars;
|
||||
if (buffer) {
|
||||
/* Copy into the output buffer. */
|
||||
fill(&digits_end, &buffer_end, n_chars, n_zeros,
|
||||
use_separator ? thousands_sep : NULL, thousands_sep_len);
|
||||
}
|
||||
}
|
||||
return count;
|
||||
}
|
||||
|
||||
/**
|
||||
* _Py_InsertThousandsGroupingLocale:
|
||||
* @buffer: A pointer to the start of a string.
|
||||
* @n_digits: The number of digits in the string, in which we want
|
||||
* to put the grouping chars.
|
||||
*
|
||||
* Reads thee current locale and calls _Py_InsertThousandsGrouping().
|
||||
**/
|
||||
Py_ssize_t
|
||||
_Py_InsertThousandsGroupingLocale(STRINGLIB_CHAR *buffer,
|
||||
Py_ssize_t n_buffer,
|
||||
STRINGLIB_CHAR *digits,
|
||||
Py_ssize_t n_digits,
|
||||
Py_ssize_t min_width)
|
||||
{
|
||||
struct lconv *locale_data = localeconv();
|
||||
const char *grouping = locale_data->grouping;
|
||||
const char *thousands_sep = locale_data->thousands_sep;
|
||||
|
||||
return _Py_InsertThousandsGrouping(buffer, n_buffer, digits, n_digits,
|
||||
min_width, grouping, thousands_sep);
|
||||
}
|
||||
#endif /* STRINGLIB_LOCALEUTIL_H */
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue