|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
#ifdef HAVE_CONFIG_H |
|
# include <config.h> |
|
#endif |
|
|
|
#include "boolean.h" |
|
#include "chars.h" |
|
#include "gsubr.h" |
|
#include "pairs.h" |
|
#include "srfi-13.h" |
|
#include "strings.h" |
|
#include "symbols.h" |
|
|
|
#include "strorder.h" |
|
|
|
|
|
|
|
|
|
|
|
SCM_C_INLINE_KEYWORD static SCM |
|
srfi13_cmp (SCM s1, SCM s2, SCM (*cmp) (SCM, SCM, SCM, SCM, SCM, SCM)) |
|
{ |
|
if (scm_is_true (cmp (s1, s2, |
|
SCM_UNDEFINED, SCM_UNDEFINED, |
|
SCM_UNDEFINED, SCM_UNDEFINED))) |
|
return SCM_BOOL_T; |
|
else |
|
return SCM_BOOL_F; |
|
} |
|
|
|
static SCM scm_i_string_equal_p (SCM s1, SCM s2, SCM rest); |
|
SCM_DEFINE (scm_i_string_equal_p, "string=?", 0, 2, 1, |
|
(SCM s1, SCM s2, SCM rest), |
|
"Lexicographic equality predicate; return @code{#t} if the two\n" |
|
"strings are the same length and contain the same characters in\n" |
|
"the same positions, otherwise return @code{#f}.\n" |
|
"\n" |
|
"The procedure @code{string-ci=?} treats upper and lower case\n" |
|
"letters as though they were the same character, but\n" |
|
"@code{string=?} treats upper and lower case as distinct\n" |
|
"characters.") |
|
#define FUNC_NAME s_scm_i_string_equal_p |
|
{ |
|
if (SCM_UNBNDP (s1) || SCM_UNBNDP (s2)) |
|
return SCM_BOOL_T; |
|
while (!scm_is_null (rest)) |
|
{ |
|
if (scm_is_false (srfi13_cmp (s1, s2, scm_string_eq))) |
|
return SCM_BOOL_F; |
|
s1 = s2; |
|
s2 = scm_car (rest); |
|
rest = scm_cdr (rest); |
|
} |
|
return srfi13_cmp (s1, s2, scm_string_eq); |
|
} |
|
#undef FUNC_NAME |
|
|
|
SCM scm_string_equal_p (SCM s1, SCM s2) |
|
#define FUNC_NAME s_scm_i_string_equal_p |
|
{ |
|
return srfi13_cmp (s1, s2, scm_string_eq); |
|
} |
|
#undef FUNC_NAME |
|
|
|
static SCM scm_i_string_ci_equal_p (SCM s1, SCM s2, SCM rest); |
|
SCM_DEFINE (scm_i_string_ci_equal_p, "string-ci=?", 0, 2, 1, |
|
(SCM s1, SCM s2, SCM rest), |
|
"Case-insensitive string equality predicate; return @code{#t} if\n" |
|
"the two strings are the same length and their component\n" |
|
"characters match (ignoring case) at each position; otherwise\n" |
|
"return @code{#f}.") |
|
#define FUNC_NAME s_scm_i_string_ci_equal_p |
|
{ |
|
if (SCM_UNBNDP (s1) || SCM_UNBNDP (s2)) |
|
return SCM_BOOL_T; |
|
while (!scm_is_null (rest)) |
|
{ |
|
if (scm_is_false (srfi13_cmp (s1, s2, scm_string_ci_eq))) |
|
return SCM_BOOL_F; |
|
s1 = s2; |
|
s2 = scm_car (rest); |
|
rest = scm_cdr (rest); |
|
} |
|
return srfi13_cmp (s1, s2, scm_string_ci_eq); |
|
} |
|
#undef FUNC_NAME |
|
|
|
SCM scm_string_ci_equal_p (SCM s1, SCM s2) |
|
#define FUNC_NAME s_scm_i_string_ci_equal_p |
|
{ |
|
return srfi13_cmp (s1, s2, scm_string_ci_eq); |
|
} |
|
#undef FUNC_NAME |
|
|
|
static SCM scm_i_string_less_p (SCM s1, SCM s2, SCM rest); |
|
SCM_DEFINE (scm_i_string_less_p, "string<?", 0, 2, 1, |
|
(SCM s1, SCM s2, SCM rest), |
|
"Lexicographic ordering predicate; return @code{#t} if @var{s1}\n" |
|
"is lexicographically less than @var{s2}.") |
|
#define FUNC_NAME s_scm_i_string_less_p |
|
{ |
|
if (SCM_UNBNDP (s1) || SCM_UNBNDP (s2)) |
|
return SCM_BOOL_T; |
|
while (!scm_is_null (rest)) |
|
{ |
|
if (scm_is_false (srfi13_cmp (s1, s2, scm_string_lt))) |
|
return SCM_BOOL_F; |
|
s1 = s2; |
|
s2 = scm_car (rest); |
|
rest = scm_cdr (rest); |
|
} |
|
return srfi13_cmp (s1, s2, scm_string_lt); |
|
} |
|
#undef FUNC_NAME |
|
|
|
SCM scm_string_less_p (SCM s1, SCM s2) |
|
#define FUNC_NAME s_scm_i_string_less_p |
|
{ |
|
return srfi13_cmp (s1, s2, scm_string_lt); |
|
} |
|
#undef FUNC_NAME |
|
|
|
static SCM scm_i_string_leq_p (SCM s1, SCM s2, SCM rest); |
|
SCM_DEFINE (scm_i_string_leq_p, "string<=?", 0, 2, 1, |
|
(SCM s1, SCM s2, SCM rest), |
|
"Lexicographic ordering predicate; return @code{#t} if @var{s1}\n" |
|
"is lexicographically less than or equal to @var{s2}.") |
|
#define FUNC_NAME s_scm_i_string_leq_p |
|
{ |
|
if (SCM_UNBNDP (s1) || SCM_UNBNDP (s2)) |
|
return SCM_BOOL_T; |
|
while (!scm_is_null (rest)) |
|
{ |
|
if (scm_is_false (srfi13_cmp (s1, s2, scm_string_le))) |
|
return SCM_BOOL_F; |
|
s1 = s2; |
|
s2 = scm_car (rest); |
|
rest = scm_cdr (rest); |
|
} |
|
return srfi13_cmp (s1, s2, scm_string_le); |
|
} |
|
#undef FUNC_NAME |
|
|
|
SCM scm_string_leq_p (SCM s1, SCM s2) |
|
#define FUNC_NAME s_scm_i_string_leq_p |
|
{ |
|
return srfi13_cmp (s1, s2, scm_string_le); |
|
} |
|
#undef FUNC_NAME |
|
|
|
static SCM scm_i_string_gr_p (SCM s1, SCM s2, SCM rest); |
|
SCM_DEFINE (scm_i_string_gr_p, "string>?", 0, 2, 1, |
|
(SCM s1, SCM s2, SCM rest), |
|
"Lexicographic ordering predicate; return @code{#t} if @var{s1}\n" |
|
"is lexicographically greater than @var{s2}.") |
|
#define FUNC_NAME s_scm_i_string_gr_p |
|
{ |
|
if (SCM_UNBNDP (s1) || SCM_UNBNDP (s2)) |
|
return SCM_BOOL_T; |
|
while (!scm_is_null (rest)) |
|
{ |
|
if (scm_is_false (srfi13_cmp (s1, s2, scm_string_gt))) |
|
return SCM_BOOL_F; |
|
s1 = s2; |
|
s2 = scm_car (rest); |
|
rest = scm_cdr (rest); |
|
} |
|
return srfi13_cmp (s1, s2, scm_string_gt); |
|
} |
|
#undef FUNC_NAME |
|
|
|
SCM scm_string_gr_p (SCM s1, SCM s2) |
|
#define FUNC_NAME s_scm_i_string_gr_p |
|
{ |
|
return srfi13_cmp (s1, s2, scm_string_gt); |
|
} |
|
#undef FUNC_NAME |
|
|
|
static SCM scm_i_string_geq_p (SCM s1, SCM s2, SCM rest); |
|
SCM_DEFINE (scm_i_string_geq_p, "string>=?", 0, 2, 1, |
|
(SCM s1, SCM s2, SCM rest), |
|
"Lexicographic ordering predicate; return @code{#t} if @var{s1}\n" |
|
"is lexicographically greater than or equal to @var{s2}.") |
|
#define FUNC_NAME s_scm_i_string_geq_p |
|
{ |
|
if (SCM_UNBNDP (s1) || SCM_UNBNDP (s2)) |
|
return SCM_BOOL_T; |
|
while (!scm_is_null (rest)) |
|
{ |
|
if (scm_is_false (srfi13_cmp (s1, s2, scm_string_ge))) |
|
return SCM_BOOL_F; |
|
s1 = s2; |
|
s2 = scm_car (rest); |
|
rest = scm_cdr (rest); |
|
} |
|
return srfi13_cmp (s1, s2, scm_string_ge); |
|
} |
|
#undef FUNC_NAME |
|
|
|
SCM scm_string_geq_p (SCM s1, SCM s2) |
|
#define FUNC_NAME s_scm_i_string_geq_p |
|
{ |
|
return srfi13_cmp (s1, s2, scm_string_ge); |
|
} |
|
#undef FUNC_NAME |
|
|
|
static SCM scm_i_string_ci_less_p (SCM s1, SCM s2, SCM rest); |
|
SCM_DEFINE (scm_i_string_ci_less_p, "string-ci<?", 0, 2, 1, |
|
(SCM s1, SCM s2, SCM rest), |
|
"Case insensitive lexicographic ordering predicate; return\n" |
|
"@code{#t} if @var{s1} is lexicographically less than @var{s2}\n" |
|
"regardless of case.") |
|
#define FUNC_NAME s_scm_i_string_ci_less_p |
|
{ |
|
if (SCM_UNBNDP (s1) || SCM_UNBNDP (s2)) |
|
return SCM_BOOL_T; |
|
while (!scm_is_null (rest)) |
|
{ |
|
if (scm_is_false (srfi13_cmp (s1, s2, scm_string_ci_lt))) |
|
return SCM_BOOL_F; |
|
s1 = s2; |
|
s2 = scm_car (rest); |
|
rest = scm_cdr (rest); |
|
} |
|
return srfi13_cmp (s1, s2, scm_string_ci_lt); |
|
} |
|
#undef FUNC_NAME |
|
|
|
SCM scm_string_ci_less_p (SCM s1, SCM s2) |
|
#define FUNC_NAME s_scm_i_string_ci_less_p |
|
{ |
|
return srfi13_cmp (s1, s2, scm_string_ci_lt); |
|
} |
|
#undef FUNC_NAME |
|
|
|
static SCM scm_i_string_ci_leq_p (SCM s1, SCM s2, SCM rest); |
|
SCM_DEFINE (scm_i_string_ci_leq_p, "string-ci<=?", 0, 2, 1, |
|
(SCM s1, SCM s2, SCM rest), |
|
"Case insensitive lexicographic ordering predicate; return\n" |
|
"@code{#t} if @var{s1} is lexicographically less than or equal\n" |
|
"to @var{s2} regardless of case.") |
|
#define FUNC_NAME s_scm_i_string_ci_leq_p |
|
{ |
|
if (SCM_UNBNDP (s1) || SCM_UNBNDP (s2)) |
|
return SCM_BOOL_T; |
|
while (!scm_is_null (rest)) |
|
{ |
|
if (scm_is_false (srfi13_cmp (s1, s2, scm_string_ci_le))) |
|
return SCM_BOOL_F; |
|
s1 = s2; |
|
s2 = scm_car (rest); |
|
rest = scm_cdr (rest); |
|
} |
|
return srfi13_cmp (s1, s2, scm_string_ci_le); |
|
} |
|
#undef FUNC_NAME |
|
|
|
SCM scm_string_ci_leq_p (SCM s1, SCM s2) |
|
#define FUNC_NAME s_scm_i_string_ci_leq_p |
|
{ |
|
return srfi13_cmp (s1, s2, scm_string_ci_le); |
|
} |
|
#undef FUNC_NAME |
|
|
|
static SCM scm_i_string_ci_gr_p (SCM s1, SCM s2, SCM rest); |
|
SCM_DEFINE (scm_i_string_ci_gr_p, "string-ci>?", 0, 2, 1, |
|
(SCM s1, SCM s2, SCM rest), |
|
"Case insensitive lexicographic ordering predicate; return\n" |
|
"@code{#t} if @var{s1} is lexicographically greater than\n" |
|
"@var{s2} regardless of case.") |
|
#define FUNC_NAME s_scm_i_string_ci_gr_p |
|
{ |
|
if (SCM_UNBNDP (s1) || SCM_UNBNDP (s2)) |
|
return SCM_BOOL_T; |
|
while (!scm_is_null (rest)) |
|
{ |
|
if (scm_is_false (srfi13_cmp (s1, s2, scm_string_ci_gt))) |
|
return SCM_BOOL_F; |
|
s1 = s2; |
|
s2 = scm_car (rest); |
|
rest = scm_cdr (rest); |
|
} |
|
return srfi13_cmp (s1, s2, scm_string_ci_gt); |
|
} |
|
#undef FUNC_NAME |
|
|
|
SCM scm_string_ci_gr_p (SCM s1, SCM s2) |
|
#define FUNC_NAME s_scm_i_string_ci_gr_p |
|
{ |
|
return srfi13_cmp (s1, s2, scm_string_ci_gt); |
|
} |
|
#undef FUNC_NAME |
|
|
|
static SCM scm_i_string_ci_geq_p (SCM s1, SCM s2, SCM rest); |
|
SCM_DEFINE (scm_i_string_ci_geq_p, "string-ci>=?", 0, 2, 1, |
|
(SCM s1, SCM s2, SCM rest), |
|
"Case insensitive lexicographic ordering predicate; return\n" |
|
"@code{#t} if @var{s1} is lexicographically greater than or\n" |
|
"equal to @var{s2} regardless of case.") |
|
#define FUNC_NAME s_scm_i_string_ci_geq_p |
|
{ |
|
if (SCM_UNBNDP (s1) || SCM_UNBNDP (s2)) |
|
return SCM_BOOL_T; |
|
while (!scm_is_null (rest)) |
|
{ |
|
if (scm_is_false (srfi13_cmp (s1, s2, scm_string_ci_ge))) |
|
return SCM_BOOL_F; |
|
s1 = s2; |
|
s2 = scm_car (rest); |
|
rest = scm_cdr (rest); |
|
} |
|
return srfi13_cmp (s1, s2, scm_string_ci_ge); |
|
} |
|
#undef FUNC_NAME |
|
|
|
SCM scm_string_ci_geq_p (SCM s1, SCM s2) |
|
#define FUNC_NAME s_scm_i_string_ci_geq_p |
|
{ |
|
return srfi13_cmp (s1, s2, scm_string_ci_ge); |
|
} |
|
#undef FUNC_NAME |
|
|
|
|
|
|
|
void |
|
scm_init_strorder () |
|
{ |
|
#include "strorder.x" |
|
} |
|
|
|
|