Update of /cvsroot/refdb/refdb/src
In directory sc8-pr-cvs1.sourceforge.net:/tmp/cvs-serv28264
Modified Files:
Tag: Release_0_9_5_stable
strfncs.c
Log Message:
count_the_flowers(), replace_char_string(), sgml_entitize(): now handle multibyte letters
Index: strfncs.c
===================================================================
RCS file: /cvsroot/refdb/refdb/src/strfncs.c,v
retrieving revision 1.15.2.4
retrieving revision 1.15.2.5
diff -u -U2 -r1.15.2.4 -r1.15.2.5
--- strfncs.c 11 Oct 2005 20:35:02 -0000 1.15.2.4
+++ strfncs.c 28 Dec 2005 01:31:31 -0000 1.15.2.5
@@ -38,5 +38,5 @@
/* forward declarations of local functions */
-static size_t count_the_flowers(char *buffer, char letter);
+static size_t count_the_flowers(char *buffer, char *letter, size_t len);
static int is_ip(char *address);
@@ -680,5 +680,5 @@
/*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
count_the_flowers() counts the number of occurrences of letter in
- buffer
+ buffer. letter may be a multibyte character
static size_t count_the_flowers returns the number of occurrences of letter
@@ -688,39 +688,55 @@
char* buffer pointer to the string to scan
- char letter the letter to locate
+ char *letter ptr to the letter to locate
+
+ size_t len the number of bytes that represent letter
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++*/
-static size_t count_the_flowers(char *buffer, char letter) {
+static size_t count_the_flowers(char *buffer, char *letter, size_t len) {
size_t numletters = 0;
- int i;
+ size_t i;
+ size_t buflen;
+
+ buflen = strlen(buffer);
- for (i = 0; i < strlen(buffer); i++) {
- if (buffer[i] == letter) {
- numletters++;
+ if (buflen >= len) {
+ buflen = buflen-len+1;
+
+ for (i = 0; i < buflen; i++) {
+ if (!memcmp(buffer+i, letter, len)) {
+ numletters++;
+ }
}
+ return numletters;
+ }
+ else {
+ /* string too short to hold the flower */
+ return 0;
}
- return numletters;
}
/*++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
- replace_char_string() replaces the character that buffer points to
+ replace_char_string() replaces the letter that buffer points to
with an insert of arbitrary size. The calling function
is responsible to allocate enough space for buffer
- so the insertion can be safely executed.
+ so the insertion can be safely executed. The letter may
+ be a multibyte character.
void replace_char_string has no return value
- char* buffer pointer to the string that will be extended
+ char* buffer pointer to the null-terminated string that will be extended
char* insert pointer to the string that will be inserted
+ size_t len length of the letter to replace in bytes
+
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++*/
-void replace_char_string(char *buffer, char *insert) {
- int len_insert;
+void replace_char_string(char *buffer, char *insert, size_t len) {
+ size_t insert_len;
- len_insert = strlen(insert);
+ insert_len = strlen(insert);
- memmove(buffer+len_insert, buffer+1, strlen(buffer));
- memcpy(buffer, insert, len_insert);
+ memmove(buffer+insert_len, buffer+len, strlen(buffer+len)+1);
+ memcpy(buffer, insert, insert_len);
}
@@ -789,9 +805,12 @@
struct charent* the_ents;
- struct charent default_ents[4] = { /* default lookup table for entity replacements */
- {(int)'&', "&"},
- {(int)'<', "<"},
- {(int)'>', ">"},
- {0, ""}
+ struct charent default_ents[7] = { /* default lookup table for entity replacements */
+ {"&", 1, "&"},
+ {"<", 1, "<"},
+ {">", 1, ">"},
+ {{226, 128, 148, 0}, 3, "—"}, /* 0x2014 */
+ {{226, 128, 152, 0}, 3, "‘"}, /* 0x2018 */
+ {{226, 128, 153, 0}, 3, "’"}, /* 0x2019 */
+ {"", 0, ""}
};
@@ -803,8 +822,8 @@
}
- while (the_ents[i].character != 0) {
+ while (the_ents[i].len != 0) {
- numents = count_the_flowers(*buffer, the_ents[i].character);
-/* printf("%d\n", numents); */
+ numents = count_the_flowers(*buffer, the_ents[i].letter, the_ents[i].len);
+/* printf("numents for %s went to %d\n", the_ents[i].letter, numents); */
if (numents > 0) {
@@ -816,5 +835,5 @@
}
- token = strchr(*buffer, the_ents[i].character);
+ token = strstr(*buffer, the_ents[i].letter);
the_end = &((*buffer)[strlen(*buffer)-1]);
@@ -824,5 +843,5 @@
char* next_sp;
-/* printf("token went to:%s<<\n", token); */
+/* printf("token went to:%d<<letter went to %s\n", (int)*token, the_ents[i].letter); */
/* replace ampersand only if it does not start an entity */
/* get pointers to the next ampersand and semicolon, if any,
@@ -832,8 +851,8 @@
next_sp = strchr(token+1, (int)' ');
- if (the_ents[i].character != '&' || compare_ptr(&next_sc, &next_amp) != -1 || compare_ptr(&next_sp, &next_sc) != 1) {
- replace_char_string(token, the_ents[i].entity);
+ if (*(the_ents[i].letter) != '&' || compare_ptr(&next_sc, &next_amp) != -1 || compare_ptr(&next_sp, &next_sc) != 1) {
+ replace_char_string(token, the_ents[i].entity, the_ents[i].len);
}
- token = (token + 1 > the_end + 5) ? NULL:strchr(token+1, the_ents[i].character);
+ token = (token + the_ents[i].len > the_end) ? NULL:strstr(token+the_ents[i].len, the_ents[i].letter);
}
}
@@ -884,5 +903,5 @@
expand with the value of HOME. This way we can call this function
with paths that did not go through shell expansion */
- if ((numtilde = count_the_flowers(the_path, '~')) > 1) {
+ if ((numtilde = count_the_flowers(the_path, "~", 1)) > 1) {
free(homepath);
free(temp);
@@ -892,5 +911,5 @@
myhome = getenv("HOME");
strcpy(homepath, the_path);
- replace_char_string(strchr(homepath, (int)'~'), myhome);
+ replace_char_string(strchr(homepath, (int)'~'), myhome, 1);
}
else {
|
