Fixed charset warning and memory leaks

libisofs/ecma119.c

@@ -429,10 +429,10 @@ write_pri_vol_desc(struct ecma119_write_target *t, uint8_t *buf)
 {
 	struct ecma119_pri_vol_desc *vol = (struct ecma119_pri_vol_desc*)buf;
 	struct iso_volume *volume = t->volset->volume[t->volnum];
-	char *vol_id = wcstoascii(volume->volume_id);
+	char *vol_id = str2ascii(volume->volume_id);
-	char *pub_id = wcstoascii(volume->publisher_id);
+	char *pub_id = str2ascii(volume->publisher_id);
-	char *data_id = wcstoascii(volume->data_preparer_id);
+	char *data_id = str2ascii(volume->data_preparer_id);
-	char *volset_id = wcstoascii(t->volset->volset_id);
+	char *volset_id = str2ascii(t->volset->volset_id);
 
 	vol->vol_desc_type[0] = 1;
 	memcpy(vol->std_identifier, "CD001", 5);

libisofs/joliet.c

@@ -129,6 +129,7 @@ joliet_calc_dir_pos(struct ecma119_write_target *t,
 
 	root->block = t->curblock;
 	t->curblock += div_up(root->len, t->block_size);
+
 	t->dirlist_joliet[t->curfile++] = root;
 	for (i = 0; i < root->nchildren; i++) {
 		ch = root->children[i];
@@ -231,10 +232,10 @@ write_sup_vol_desc(struct ecma119_write_target *t, uint8_t *buf)
 {
 	struct ecma119_sup_vol_desc *vol = (struct ecma119_sup_vol_desc*)buf;
 	struct iso_volume *volume = t->volset->volume[t->volnum];
-	uint16_t *vol_id = wcstoucs(volume->volume_id);
+	uint16_t *vol_id = str2ucs(volume->volume_id);
-	uint16_t *pub_id = wcstoucs(volume->publisher_id);
+	uint16_t *pub_id = str2ucs(volume->publisher_id);
-	uint16_t *data_id = wcstoucs(volume->data_preparer_id);
+	uint16_t *data_id = str2ucs(volume->data_preparer_id);
-	uint16_t *volset_id = wcstoucs(t->volset->volset_id);
+	uint16_t *volset_id = str2ucs(t->volset->volset_id);
 	int vol_id_len = MIN(32, ucslen(vol_id) * 2);
 	int pub_id_len = MIN(128, ucslen(pub_id) * 2);
 	int data_id_len = MIN(128, ucslen(data_id) * 2);

libisofs/tree.c

@@ -86,7 +86,7 @@ iso_tree_add_new_file(struct iso_tree_node *parent, const char *name)
 
 	f->volume = parent ? parent->volume : NULL;
 	f->parent = parent;
-	f->name = parent ? towcs(name) : NULL;
+	f->name = parent ? strdup(name) : NULL;
 	f->attrib = get_attrib(parent);
 	f->attrib.st_mode = 0777 | S_IFREG;
 	f->loc.type = LIBISO_NONE;
@@ -192,7 +192,7 @@ iso_tree_print(const struct iso_tree_node *root, int spaces)
 	memset(sp, ' ', spaces);
 	sp[spaces] = '\0';
 
-	printf("%s%ls\n", sp, root->name);
+	printf("%s%sn", sp, root->name);
 	for (i=0; i < root->nchildren; i++) {
 		iso_tree_print(root->children[i], spaces+2);
 	}
@@ -219,5 +219,5 @@ void
 iso_tree_node_set_name(struct iso_tree_node *file, const char *name)
 {
 	free(file->name);
-	file->name = towcs(name);
+	file->name = strdup(name);
 }

libisofs/tree.h

@@ -51,7 +51,7 @@ struct iso_tree_node
 {
 	struct iso_volume *volume;
 	struct iso_tree_node *parent;
-	wchar_t *name;
+	char *name;
 	struct stat attrib;	/**< The POSIX attributes of this node as
 				  * documented in "man 2 stat". */
 	struct iso_file_location loc;

libisofs/util.c

@@ -13,9 +13,15 @@
 #include <ctype.h>
 #include <assert.h>
 #include <errno.h>
+#include <locale.h>
 
 #include "util.h"
 
+/* avoids warning and names in iso, joliet and rockridge can't be > 255 bytes
+ * anyway. There are at most 31 characters in iso level 1, 255 for rockridge,
+ * 64 characters (* 2 since UCS) for joliet. */
+#define NAME_BUFFER_SIZE 255
+
 int div_up(int n, int div)
 {
 	return (n + div - 1) / div;
@@ -26,85 +32,162 @@ int round_up(int n, int mul)
 	return div_up(n, mul) * mul;
 }
 
-wchar_t *towcs(const char *str)
+/* this function must always return a name
+ * since the caller never checks if a NULL
+ * is returned. It also avoids some warnings. */
+char *str2ascii(const char *src_arg)
 {
-	size_t len = strlen(str);
+	wchar_t wsrc_[NAME_BUFFER_SIZE];
-	wchar_t *ret = malloc(sizeof(wchar_t) * (len + 1));
+	char *src = (char*)wsrc_;
-	mbstate_t ps;
+	char *ret_;
+	char *ret;
+	mbstate_t state;
+	iconv_t conv;
+	size_t numchars;
+	size_t outbytes;
+	size_t inbytes;
 	size_t n;
 
-	memset(&ps, 0, sizeof(ps));
+	if (!src_arg)
-	n = mbsrtowcs(ret, &str, len, &ps);
-	ret[len] = '\0';
-
-	if (n != len) {
-		free(ret);
-		return NULL;
-	}
-	return ret;
-}
-
-char *wcstoascii(const wchar_t *wsrc)
-{
-	if (!wsrc)
 		return NULL;
 
-	iconv_t conv = iconv_open("ASCII", "WCHAR_T");
+	/* convert the string to a wide character string. Note: outbytes
-	size_t outbytes = wcslen(wsrc);
+	 * is in fact the number of characters in the string and doesn't
-	size_t inbytes = outbytes * sizeof(wchar_t);
+	 * include the last NULL character. */
-	char src_[inbytes + sizeof(wchar_t)];
+	memset(&state, 0, sizeof(state));
-	char *src = src_;
+	numchars = mbsrtowcs(wsrc_, &src_arg, NAME_BUFFER_SIZE-1, &state);
-	char *ret_, *ret;
+	if (numchars < 0)
-	size_t n;
+		return NULL;
 
+	inbytes = numchars * sizeof(wchar_t);
+
+	ret_ = malloc(numchars+1);
+	outbytes = numchars;
+	ret = ret_;
+
+	/* initialize iconv */
+	conv = iconv_open("ASCII", "WCHAR_T");
 	if (conv == (iconv_t)-1)
 		return NULL;
 
-	memcpy(src, wsrc, inbytes + sizeof(wchar_t));
-	ret = malloc(outbytes+1);
-	ret[outbytes] = '\0';
-	ret_ = ret;
-
 	n = iconv(conv, &src, &inbytes, &ret, &outbytes);
-	if (n == -1) {
+	while(n == -1) {
-		free(ret_);
+		/* The destination buffer is too small. Stops here. */
-		return NULL;
+		if(errno == E2BIG)
+			break;
+
+		/* An incomplete multi bytes sequence was found. We 
+		 * can't do anything here. That's quite unlikely. */
+		if(errno == EINVAL)
+			break;
+
+		/* The last possible error is an invalid multi bytes
+		 * sequence. Just replace the character with a "_". 
+		 * Probably the character doesn't exist in ascii like
+		 * "é, è, à, ç, ..." in French. */
+		*ret++ = '_';
+		outbytes--;
+
+		if(!outbytes)
+			break;
+
+		/* There was an error with one character but some other remain
+		 * to be converted. That's probably a multibyte character.
+		 * See above comment. */
+		src += sizeof(wchar_t);
+		inbytes -= sizeof(wchar_t);
+
+		if(!inbytes)
+			break;
+
+		n = iconv(conv, &src, &inbytes, &ret, &outbytes);
 	}
+
+	iconv_close(conv);
+
+	*ret='\0';
 	return ret_;
 }
 
 /* FIXME: C&P */
-uint16_t *wcstoucs(const wchar_t *wsrc)
+uint16_t *str2ucs(const char *src_arg)
 {
-	if (!wsrc)
+	wchar_t wsrc_[NAME_BUFFER_SIZE];
-		return calloc(2, 1); /* empty UCS string */
+	char *src = (char*)wsrc_;
-
+	char *ret_;
-	iconv_t conv = iconv_open("UCS-2BE", "WCHAR_T");
+	char *ret;
-	size_t outbytes = wcslen(wsrc) * 2;
+	mbstate_t state;
-	size_t inbytes = outbytes * sizeof(wchar_t) / 2;
+	iconv_t conv;
-	char src_[inbytes + sizeof(wchar_t)];
+	size_t outbytes;
-	char *src = src_;
+	size_t numchars;
-	char *ret_, *ret;
+	size_t inbytes;
 	size_t n;
 
-	if (conv == (iconv_t)-1)
+	if (!src_arg)
-		return calloc(2, 1);
+		return calloc(2, 1); /* empty UCS string */
 
-	memcpy(src, wsrc, inbytes + sizeof(wchar_t));
+	/* convert the string to a wide character string. Note: outbytes
-	ret = malloc(outbytes + sizeof(wchar_t));
+	 * is in fact the number of characters in the string and doesn't
-	ret[outbytes] = 0;
+	 * include the last NULL character. */
-	ret[outbytes+1] = 0;
+	memset(&state, 0, sizeof(state));
-	ret_ = ret;
+	numchars = mbsrtowcs(wsrc_, &src_arg, NAME_BUFFER_SIZE-1, &state);
+	if (numchars < 0)
+		return calloc(2, 1); /* empty UCS string */
+
+	inbytes = numchars * sizeof(wchar_t);
+
+	outbytes = numchars * sizeof(uint16_t);
+	ret_ = malloc ((numchars+1) * sizeof(uint16_t));
+	ret = ret_;
+
+	/* initialize iconv */
+	conv = iconv_open("UCS-2BE", "WCHAR_T");
+	if (conv == (iconv_t)-1)
+		return calloc(2, 1); /* empty UCS string */
 
 	n = iconv(conv, &src, &inbytes, &ret, &outbytes);
-	if (n == -1) {
+	while(n == -1) {
-		perror ("error in iconv conversion");
+		/* The destination buffer is too small. Stops here. */
-		free(ret_);
+		if(errno == E2BIG)
-		return calloc(2, 1);
+			break;
+
+		/* An incomplete multi bytes sequence was found. We 
+		 * can't do anything here. That's quite unlikely. */
+		if(errno == EINVAL)
+			break;
+
+		/* The last possible error is an invalid multi bytes
+		 * sequence. Just replace the character with a "_". 
+		 * Probably the character doesn't exist in ascii like
+		 * "é, è, à, ç, ..." in French. */
+		*((uint16_t*) ret) = '_';
+		ret += sizeof(uint16_t);
+		outbytes -= sizeof(uint16_t);
+
+		if(!outbytes)
+			break;
+
+		/* There was an error with one character but some other remain
+		 * to be converted. That's probably a multibyte character.
+		 * See above comment. */
+		src += sizeof(wchar_t);
+		inbytes -= sizeof(wchar_t);
+
+		if(!inbytes)
+			break;
+
+		n = iconv(conv, &src, &inbytes, &ret, &outbytes);
 	}
+
+	iconv_close(conv);
+
+	/* close the ucs string */
+	*((uint16_t*) ret) = 0;
+
 	return (uint16_t*)ret_;
 }
 
+
 static int valid_d_char(char c)
 {
 	return (c >= '0' && c <= '9') || (c >= 'A' && c <= 'Z') || (c == '_');
@@ -132,9 +215,9 @@ static int valid_p_char(char c)
 				      || (c == '.') || (c == '_') || (c == '-');
 }
 
-static char *iso_dirid(const wchar_t *src, int size)
+static char *iso_dirid(const char *src, int size)
 {
-	char *ret = wcstoascii(src);
+	char *ret = str2ascii(src);
 	size_t len, i;
 
 	if (!ret)
@@ -153,19 +236,19 @@ static char *iso_dirid(const wchar_t *src, int size)
 	return ret;
 }
 
-char *iso_1_dirid(const wchar_t *src)
+char *iso_1_dirid(const char *src)
 {
 	return iso_dirid(src, 8);
 }
 
-char *iso_2_dirid(const wchar_t *src)
+char *iso_2_dirid(const char *src)
 {
 	return iso_dirid(src, 31);
 }
 
-char *iso_1_fileid(const wchar_t *wsrc)
+char *iso_1_fileid(const char *src_arg)
 {
-	char *src = wcstoascii(wsrc);
+	char *src = str2ascii(src_arg);
 	char *dest;
 	char *dot;			/* Position of the last dot in the
 					   filename, will be used to calculate 
@@ -214,9 +297,9 @@ char *iso_1_fileid(const wchar_t *wsrc)
 	return dest;
 }
 
-char *iso_2_fileid(const wchar_t *wsrc)
+char *iso_2_fileid(const char *src_arg)
 {
-	char *src = wcstoascii(wsrc);
+	char *src = str2ascii(src_arg);
 	char *dest;
 	char *dot;
 	int lname, lext, lnname, lnext, pos, i;
@@ -274,9 +357,9 @@ char *iso_2_fileid(const wchar_t *wsrc)
 }
 
 char *
-iso_p_fileid(const wchar_t *src)
+iso_p_fileid(const char *src)
 {
-	char *ret = wcstoascii(src);
+	char *ret = str2ascii(src);
 	size_t i, len;
 
 	if (!ret)
@@ -291,9 +374,9 @@ iso_p_fileid(const wchar_t *src)
 }
 
 uint16_t *
-iso_j_id(const wchar_t *src)
+iso_j_id(const char *src_arg)
 {
-	uint16_t *j_str = wcstoucs(src);
+	uint16_t *j_str = str2ucs(src_arg);
 	size_t len = ucslen(j_str);
 	size_t n;
 

libisofs/util.h

@@ -31,18 +31,18 @@ extern inline int round_up(int n, int mul)
 }
 
 wchar_t *towcs(const char *);
-char *wcstoascii(const wchar_t *);
+char *str2ascii(const char*);
-uint16_t *wcstoucs(const wchar_t*);
+uint16_t *str2ucs(const char*);
 
 /**
  * Create a level 1 directory identifier.
  */
-char *iso_1_dirid(const wchar_t *src);
+char *iso_1_dirid(const char *src);
 
 /**
  * Create a level 2 directory identifier.
  */
-char *iso_2_dirid(const wchar_t *src);
+char *iso_2_dirid(const char *src);
 
 /**
  * Create a level 1 file identifier that consists of a name, extension and
@@ -51,7 +51,7 @@ char *iso_2_dirid(const wchar_t *src);
  * length 3, followed by a separator (;) and a version number (digit 1).
  * @return NULL if the original name and extension both are of length 0.
  */
-char *iso_1_fileid(const wchar_t *src);
+char *iso_1_fileid(const char *src);
 
 /**
  * Create a level 2 file identifier that consists of a name, extension and
@@ -60,7 +60,7 @@ char *iso_1_fileid(const wchar_t *src);
  * followed by a separator (;) and a version number (digit 1).
  * @return NULL if the original name and extension both are of length 0.
  */
-char *iso_2_fileid(const wchar_t *src);
+char *iso_2_fileid(const char *src);
 
 /**
  * Create a Joliet file or directory identifier that consists of a name,
@@ -73,7 +73,7 @@ char *iso_2_fileid(const wchar_t *src);
  * @param size will be set to the size (in bytes) of the identifier.
  * @return NULL if the original name and extension both are of length 0 or the conversion from the current codeset to UCS-2BE is not available.
  */
-uint16_t *iso_j_id(const wchar_t *src);
+uint16_t *iso_j_id(const char *src);
 
 /**
  * FIXME: what are the requirements for these next two? Is this for RR?
@@ -82,14 +82,14 @@ uint16_t *iso_j_id(const wchar_t *src);
  * The resulting file name will not exceed 250 characters.
  * @return NULL if the original name and extension both are of length 0.
  */
-char *iso_p_fileid(const wchar_t *src);
+char *iso_p_fileid(const char *src);
 
 /**
  * Create a POSIX portable directory name.
  * The resulting directory name will not exceed 250 characters.
  * @return NULL if the original name is of length 0.
  */
-char *iso_p_dirid(const wchar_t *src);
+char *iso_p_dirid(const char *src);
 
 void iso_lsb(uint8_t *buf, uint32_t num, int bytes);
 void iso_msb(uint8_t *buf, uint32_t num, int bytes);

libisofs/volume.c

@@ -19,7 +19,7 @@ iso_volset_new(struct iso_volume *vol, const char *id)
 	volset->refcount = 1;
 	volset->volume = malloc(sizeof(void *));
 	volset->volume[0] = vol;
-	volset->volset_id = towcs(id);
+	volset->volset_id = strdup(id);
 
 	vol->refcount++;
 	return volset;
@@ -63,11 +63,11 @@ iso_volume_new_with_root(const char *volume_id,
 	volume->root = root ? root : iso_tree_new_root(volume);
 
 	if (volume_id != NULL)
-		volume->volume_id = towcs(volume_id);
+		volume->volume_id = strdup(volume_id);
 	if (publisher_id != NULL)
-		volume->publisher_id = towcs(publisher_id);
+		volume->publisher_id = strdup(publisher_id);
 	if (data_preparer_id != NULL)
-		volume->data_preparer_id = towcs(data_preparer_id);
+		volume->data_preparer_id = strdup(data_preparer_id);
 	return volume;
 }
 

libisofs/volume.h

@@ -21,9 +21,9 @@ struct iso_volume
 	struct iso_tree_node *root;	/**< Root of the directory tree for the
 					     volume. */
 
-	wchar_t *volume_id;		/**< Volume identifier. */
+	char *volume_id;		/**< Volume identifier. */
-	wchar_t *publisher_id;		/**< Volume publisher. */
+	char *publisher_id;		/**< Volume publisher. */
-	wchar_t *data_preparer_id;	/**< Volume data preparer. */
+	char *data_preparer_id;	/**< Volume data preparer. */
 };
 
 /**
@@ -38,7 +38,7 @@ struct iso_volset
 	int volset_size;		/**< The number of volumes in this
 					     volume set. */
 
-	wchar_t *volset_id;		/**< The id of this volume set, encoded
+	char *volset_id;		/**< The id of this volume set, encoded
 					     in the current locale. */
 };