Rewrite all elstr- functions to take either elstr or string as arguments.

1 files changed, 102 insertions, 115 deletions

diff --git a/src/ellinika/elmorph.c b/src/ellinika/elmorph.c
index 5785f8d..f55e010 100644
--- a/src/ellinika/elmorph.c
+++ b/src/ellinika/elmorph.c
@@ -104,5 +104,8 @@ _elstr_alloc(const char *instr, int syl)
         elstr->sylmap = NULL;
+        elstr->nsyl = 0;
+        elstr->acc_syl = 0;
+        elstr->acc_pos = 0;
         if (syl)
                 _elstr_syllabize(elstr);
-        
+                
         SCM_RETURN_NEWSMOB(_elstr_tag, elstr);
@@ -142,2 +145,4 @@ _elstr_concat(struct elstr *dest, struct elstr *src, const char *func_name)
 
+        if (src->len == 0)
+                return;
         wp = realloc(dest->str,
@@ -221,2 +226,31 @@ _elstr_init()
 
+static SCM
+force_elstr(struct elstr **ep, SCM scm, int sylopt,
+            int arg, const char *func_name)
+{
+        struct elstr *elstr;
+        
+        if (scm_is_elstr(scm)) {
+                elstr = (struct elstr*) SCM_CDR(scm);
+        } else {
+                SCM newscm;
+                char *str;
+                
+                SCM_ASSERT(scm_is_string(scm), scm, arg, func_name);
+                str = scm_to_locale_string(scm);
+                newscm = _elstr_alloc(str, sylopt);
+                free(str);
+                if (newscm == SCM_EOL)
+                        scm_misc_error(func_name,
+                                       "Invalid input string: ~S",
+                                       scm_list_1(scm));
+                scm = newscm;
+                elstr = (struct elstr*) SCM_CDR(newscm);
+        }
+        if (ep)
+                *ep = elstr;
+        return scm;
+}
+
+
 SCM_DEFINE_PUBLIC(scm_elstr_p, "elstr?", 1, 0, 0,
@@ -239,10 +273,3 @@ SCM_DEFINE_PUBLIC(scm_string__elstr, "string->elstr", 1, 0, 0,
         SCM_ASSERT(scm_is_string(string), string, SCM_ARG1, FUNC_NAME);
-        str = scm_to_locale_string(string);
-        scm = _elstr_alloc(str, 1);
-        free(str);
-        if (scm == SCM_EOL)
-                scm_misc_error(FUNC_NAME,
-                               "Invalid input string: ~S",
-                               scm_list_1(string));
-        return scm;
+        return force_elstr(NULL, string, 1, SCM_ARG1, FUNC_NAME);
 }
@@ -277,5 +304,3 @@ SCM_DEFINE_PUBLIC(scm_elstr_length, "elstr-length", 1, 0, 0,
         struct elstr *elstr;
-
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME);
-        elstr = (struct elstr*) SCM_CDR(el);
+        force_elstr(&elstr, el, 0, SCM_ARG1, FUNC_NAME);
         return scm_from_uint(elstr->len);
@@ -291,5 +316,3 @@ SCM_DEFINE_PUBLIC(scm_elstr_number_of_syllables, "elstr-number-of-syllables",
         struct elstr *elstr;
-
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME);
-        elstr = (struct elstr*) SCM_CDR(el);
+        force_elstr(&elstr, el, 1, SCM_ARG1, FUNC_NAME);
         return scm_from_uint(elstr->nsyl);
@@ -307,4 +330,3 @@ SCM_DEFINE_PUBLIC(scm_elstr_syllable_prop, "elstr-syllable-prop",
         
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME);
-        elstr = (struct elstr*) SCM_CDR(el);
+        force_elstr(&elstr, el, 1, SCM_ARG1, FUNC_NAME);
         SCM_ASSERT(scm_is_integer(n), n, SCM_ARG2, FUNC_NAME);
@@ -332,5 +354,3 @@ SCM_DEFINE_PUBLIC(scm_elstr_accent_position, "elstr-accent-position", 1, 0, 0,
         struct elstr *elstr;
-
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME);
-        elstr = (struct elstr*) SCM_CDR(el);
+        force_elstr(&elstr, el, 0, SCM_ARG1, FUNC_NAME);
         return scm_from_uint(elstr->acc_pos);
@@ -346,5 +366,3 @@ SCM_DEFINE_PUBLIC(scm_elstr_accented_syllable, "elstr-accented-syllable",
         struct elstr *elstr;
-
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME);
-        elstr = (struct elstr*) SCM_CDR(el);
+        force_elstr(&elstr, el, 1, SCM_ARG1, FUNC_NAME);
         return scm_from_uint(elstr->acc_syl);
@@ -364,4 +382,3 @@ SCM_DEFINE_PUBLIC(scm_elstr_syllable, "elstr-syllable",
         
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME);
-        elstr = (struct elstr*) SCM_CDR(el);
+        force_elstr(&elstr, el, 1, SCM_ARG1, FUNC_NAME);
         SCM_ASSERT(scm_is_integer(n), n, SCM_ARG2, FUNC_NAME);
@@ -400,4 +417,3 @@ SCM_DEFINE_PUBLIC(scm_elstr_character, "elstr-character",
         
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME);
-        elstr = (struct elstr*) SCM_CDR(el);
+        force_elstr(&elstr, el, 0, SCM_ARG1, FUNC_NAME);
         SCM_ASSERT(scm_is_integer(n), n, SCM_ARG2, FUNC_NAME);
@@ -425,9 +441,12 @@ _elstr_chgcase(SCM el, void (*chgfun)(unsigned *, size_t),
         
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, func_name);
-        elstr = (struct elstr*) SCM_CDR(el);
-        if (destructive)
+        if (destructive) {
+                SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, func_name);
+                elstr = (struct elstr*) SCM_CDR(el);
                 scm = SCM_UNSPECIFIED;
-        else {
-                scm = _elstr_dup(elstr);
-                elstr = (struct elstr*) SCM_CDR(scm);
+        } else {
+                scm = force_elstr(&elstr, el, 1, SCM_ARG1, func_name);
+                if (scm == el) {
+                        scm = _elstr_dup(elstr);
+                        elstr = (struct elstr*) SCM_CDR(scm);
+                }
         }
@@ -484,9 +503,12 @@ _elstr_deaccent(SCM el, int destructive, const char *func_name)
         
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, func_name);
-        elstr = (struct elstr*) SCM_CDR(el);
-        if (destructive)
+        if (destructive) {
+                SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, func_name);
                 scm = SCM_UNSPECIFIED;
-        else {
-                scm = _elstr_dup(elstr);
-                elstr = (struct elstr*) SCM_CDR(scm);
+                elstr = (struct elstr*) SCM_CDR(el);
+        } else {
+                scm = force_elstr(&elstr, el, 1, SCM_ARG1, func_name);
+                if (scm == el) {
+                        scm = _elstr_dup(elstr);
+                        elstr = (struct elstr*) SCM_CDR(scm);
+                }
         }
@@ -528,5 +550,9 @@ _elstr_set_accent(SCM el, SCM n, int destructive, const char *func_name)
         
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, func_name);
+        if (destructive) {
+                SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, func_name);
+                elstr = (struct elstr*) SCM_CDR(el);
+        } else
+                scm = force_elstr(&elstr, el, 1, SCM_ARG1, func_name);
+
         SCM_ASSERT(scm_is_integer(n), n, SCM_ARG2, func_name);
-        elstr = (struct elstr*) SCM_CDR(el);
         num = scm_to_uint(n);
@@ -544,3 +570,3 @@ _elstr_set_accent(SCM el, SCM n, int destructive, const char *func_name)
                 scm = SCM_UNSPECIFIED;
-        else {
+        else if (scm == el) {
                 scm = _elstr_dup(elstr);
@@ -600,5 +626,4 @@ SCM_DEFINE_PUBLIC(scm_elstr_char_prop_bitmask, "elstr-char-prop-bitmask",
         
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, FUNC_NAME);
+        force_elstr(&elstr, el, 0, SCM_ARG1, FUNC_NAME);
         SCM_ASSERT(scm_is_integer(n), n, SCM_ARG2, FUNC_NAME);
-        elstr = (struct elstr*) SCM_CDR(el);
         num = scm_to_int(n);
@@ -686,9 +711,12 @@ _elstr_thema_aoristoy(SCM el, int destructive, const char *func_name)
         
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, func_name);
-        elstr = (struct elstr*) SCM_CDR(el);
-        if (destructive)
+        if (destructive) {
+                SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, func_name);
+                elstr = (struct elstr*) SCM_CDR(el);
                 scm = SCM_UNSPECIFIED;
-        else {
-                scm = _elstr_dup(elstr);
-                elstr = (struct elstr*) SCM_CDR(scm);
+        } else {
+                scm = force_elstr(&elstr, el, 1, SCM_ARG1, func_name);
+                if (scm == el) {
+                        scm = _elstr_dup(elstr);
+                        elstr = (struct elstr*) SCM_CDR(scm);
+                }
         }
@@ -728,6 +756,9 @@ _elstr_slice(SCM el, SCM n, SCM l, int destructive, const char *func_name)
         
-        SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, func_name);
+        if (destructive) {
+                SCM_ASSERT(scm_is_elstr(el), el, SCM_ARG1, func_name);
+                elstr = (struct elstr*) SCM_CDR(el);
+        } else
+                scm = force_elstr(&elstr, el, 1, SCM_ARG1, func_name);
         SCM_ASSERT(scm_is_integer(n), n, SCM_ARG2, func_name);
         SCM_ASSERT(scm_is_integer(l), l, SCM_ARG3, func_name);
-        elstr = (struct elstr*) SCM_CDR(el);
         num = scm_to_int(n);
@@ -744,3 +775,3 @@ _elstr_slice(SCM el, SCM n, SCM l, int destructive, const char *func_name)
                 scm = SCM_UNSPECIFIED;
-        else {
+        else if (scm == el) {
                 scm = _elstr_dup(elstr);
@@ -783,29 +814,9 @@ SCM_DEFINE_PUBLIC(scm_elstr_index, "elstr-index",
 {
-        struct elstr *elstr;
-        unsigned *wc, *wtmp = NULL, *p;
-        unsigned wlen;
-        
-        SCM_ASSERT(scm_is_elstr(word), word, SCM_ARG1, FUNC_NAME);
-        elstr = (struct elstr*) SCM_CDR(word);
-        if (scm_is_elstr(needle)) {
-                struct elstr *ep = (struct elstr*) SCM_CDR(needle);
-                wc = ep->str;
-                wlen = ep->len;
-        } else {
-                SCM scm;
-                char *str;
-                
-                SCM_ASSERT(scm_is_string(needle), needle, SCM_ARG2, FUNC_NAME);
-                str = scm_to_locale_string(needle);
-                if (utf8_mbstr_to_wc(str, &wtmp, &wlen)) {
-                        free(str);
-                        scm_misc_error(FUNC_NAME,
-                                       "Invalid needle string: ~S",
-                                       scm_list_1(needle));
-                }
-                free(str);
-                wc = wtmp;
-        }
-        p = (unsigned*)utf8_wc_strnstr(elstr->str, elstr->len, wc, wlen);
-        free(wtmp);
+        struct elstr *elstr, *ep;
+        unsigned *p;
+
+        force_elstr(&elstr, word, 0, SCM_ARG1, FUNC_NAME);
+        force_elstr(&ep, needle, 0, SCM_ARG2, FUNC_NAME);
+        p = (unsigned*)utf8_wc_strnstr(elstr->str, elstr->len,
+                                       ep->str, ep->len);
         if (p)
@@ -817,18 +828,4 @@ SCM_DEFINE_PUBLIC(scm_elstr_index, "elstr-index",
 static int
-_suffix_matches(struct elstr *elstr, SCM suffix, int arg, const char *func_name)
+_suffix_matches(struct elstr *elstr, struct elstr *ep)
 {
-        struct elstr *ep;
-        
-        if (scm_is_elstr(suffix)) {
-                ep = (struct elstr*) SCM_CDR(suffix);
-        } else {
-                SCM scm;
-                char *str;
-                
-                SCM_ASSERT(scm_is_string(suffix), suffix, arg, func_name);
-                str = scm_to_locale_string(suffix);
-                scm = _elstr_alloc(str, 0);
-                free(str);
-                ep = (struct elstr*) SCM_CDR(scm);
-        }
         return (ep->len < elstr->len &&
@@ -845,7 +842,7 @@ SCM_DEFINE_PUBLIC(scm_elstr_suffix_p, "elstr-suffix?",
 {
-        struct elstr *elstr;
+        struct elstr *elstr, *ep;
 
-        SCM_ASSERT(scm_is_elstr(word), word, SCM_ARG1, FUNC_NAME);
-        elstr = (struct elstr*) SCM_CDR(word);
-        if (_suffix_matches(elstr, suffix, SCM_ARG2, FUNC_NAME))
+        force_elstr(&elstr, word, 0, SCM_ARG1, FUNC_NAME);
+        force_elstr(&ep, suffix, 0, SCM_ARG2, FUNC_NAME);
+        if (_suffix_matches(elstr, ep))
                 return suffix;
@@ -854,3 +851,4 @@ SCM_DEFINE_PUBLIC(scm_elstr_suffix_p, "elstr-suffix?",
                 SCM val = SCM_CAR(rest);
-                if (_suffix_matches(elstr, val, SCM_ARGn, FUNC_NAME))
+                force_elstr(&ep, val, 0, SCM_ARGn, FUNC_NAME);
+                if (_suffix_matches(elstr, ep))
                         return val;
@@ -871,18 +869,7 @@ SCM_DEFINE_PUBLIC(scm_elstr_append, "elstr-append",
         for (; !scm_is_null(rest); rest = SCM_CDR(rest)) {
+                struct elstr *elt;
                 SCM val = SCM_CAR(rest);
-                if (scm_is_elstr(val)) {
-                        struct elstr *elt = (struct elstr*) SCM_CDR(val);
-                        _elstr_concat(elstr, elt, FUNC_NAME);
-                } else if (scm_is_string(val)) {
-                        char *s = scm_to_locale_string(val);
-                        if (s[0]) {
-                                SCM tmp = _elstr_alloc(s, 0);
-                                free(s);
-                                _elstr_concat(elstr,
-                                              (struct elstr*) SCM_CDR(tmp),
-                                              FUNC_NAME);
-                        } else
-                                free(s);
-                } else
-                        scm_wrong_type_arg(FUNC_NAME, SCM_ARGn, rest);
+
+                force_elstr(&elt, val, 0, SCM_ARGn, FUNC_NAME);
+                _elstr_concat(elstr, elt, FUNC_NAME);
         }