New API call iso_conv_name_chars(), provisory implementation

libisofs/ecma119.c

@@ -3492,3 +3492,102 @@ int iso_write_opts_set_hfsp_block_size(IsoWriteOpts *opts,
 }
 
 
+/*
+ * @param flag
+ *      Bitfield for control purposes.
+ *        bit0-bit7= Name space
+ *                   0= generic (to_charset is valid, no reserved characters,
+ *                               no length limits)
+ *                   1= Rock Ridge (to_charset is valid)
+ *                   2= Joliet (to_charset gets overriden by UCS-2 or UTF-16)
+ *                   3= ECMA-119 (dull ISO 9660 character set)
+ *                   4= HFS+ (to_charset gets overridden by UTF-16BE) 
+ *        bit15= Reverse operation (best to be done only with results of
+ *               previous conversions)
+ */
+int iso_conv_name_chars(IsoWriteOpts *opts, char *in_name, size_t name_len,
+                        char **result, size_t *result_len, int flag)
+{
+    int name_space, ret, reverse;
+    size_t i;
+    char *from_charset, *to_charset, *conved = NULL, *smashed = NULL, *name;
+    char *tr;
+    size_t conved_len;
+    uint16_t *ucs;
+
+    name = in_name;
+    from_charset = iso_get_local_charset(0);
+    name_space = flag & 0xff;
+    reverse = !!(flag & (1 << 15));
+    if (name_space == 0) { /* generic */
+        to_charset = opts->output_charset;
+
+    } else if (name_space == 1) { /* Rock Ridge */
+        to_charset = opts->output_charset;
+        if (!reverse) {
+            LIBISO_ALLOC_MEM(smashed, char, name_len + 1);
+            memcpy(smashed, name, name_len);
+            smashed[name_len] = 0;
+            for (i = 0; i < name_len; i++)
+                 if (smashed[i] == '/')
+                     smashed[i] = '_';
+            name = smashed;
+
+            /* >>> ??? truncate to 255 chars */
+         }
+
+    } else if (name_space == 2) { /* Joliet */
+        if (opts->joliet_utf16)
+            to_charset = "UTF-16BE";
+        else
+            to_charset = "UCS-2BE";
+        /* ( Smashing happens after conversion ) */
+
+    } else if (name_space == 3) { /* ECMA-119 */
+        to_charset = "ASCII";
+
+        /* >>> ??? Use an  IsoWriteOpts+char of get_iso_name() */
+
+    } else if (name_space == 4) { /* HFS+ */
+        to_charset= "UTF-16BE";
+
+        /* >>> ??? any other conversions for HFS+ ? */;
+
+    } else {
+        ret = ISO_WRONG_ARG_VALUE;
+        goto ex;
+    }
+    if (reverse) {
+        tr = from_charset;
+        from_charset = to_charset;
+        to_charset = tr;
+    }
+
+    ret = strnconvl(name, from_charset, to_charset, name_len,
+                    &conved, &conved_len);
+    if (ret != ISO_SUCCESS)
+        goto ex;
+
+    if (name_space == 2 && !reverse) { /* Joliet */
+
+        /* >>> ??? Rather use iso_j_dir_id()/iso_j_file_id()
+               ??? Or even an IsoWriteOpts version of get_joliet_name() ?
+        */
+
+        ucs = (uint16_t *) conved;
+        iso_smash_chars_for_joliet(ucs);
+    } else if (name_space == 3 && !reverse) { /* ECMA-119 */
+
+        /* >>> smash all forbidden characters ? (or use get_iso_name() ) */;
+
+    }
+
+    *result = conved;
+    *result_len = conved_len;
+    return ISO_SUCCESS;
+ex:
+    LIBISO_FREE_MEM(smashed);
+    return ret;
+}
+
+

libisofs/fs_image.c

@@ -1258,6 +1258,9 @@ char *get_name(_ImageFsData *fsdata, const char *str, size_t len)
                 return NULL; /* aborted */
             }
             /* fallback */
+
+            /* >>> create a hopefully unique name */;
+
         }
     }
 

libisofs/libisofs.h

@@ -7109,6 +7109,51 @@ int iso_image_hfsplus_get_blessed(IsoImage *img, IsoNode ***blessed_nodes,
                                   int *bless_max, int flag);
 
 
+/* ----------------------------- Character sets ---------------------------- */
+
+/**
+ * >>>              *** NOT COMPLETELY IMPLEMENTED YET ***
+ * >>> Convert the characters in name from local charset to another charset or
+ * >>> to the representation of a particular ISO image name space.
+ * >>> In the latter case it is assumed that the conversion result does not
+ * >>> collide with any other converted name in the same directory.
+ * >>> I.e. this function does not take into respect possible name changes
+ * >>> due to collision handling.
+ *
+ * @param opts
+ *      Defines output charset, UCS-2 versus UTF-16 for Joliet,
+ *      and naming restrictions.
+ * @param name
+ *      The input text which shall be converted.
+ * @param result
+ *      Will return the conversion result. Terminated by a trailing zero byte.
+ *      Use free() to dispose it when no longer needed.
+ * @param result_len
+ *      Will return the number of bytes in result (excluding trailing zero)
+ * @param flag
+ *      Bitfield for control purposes.
+ *        bit0-bit7= Name space
+ *                   0= generic (to_charset is valid,
+ *                               no reserved characters, no length limits)
+ *                   1= Rock Ridge (to_charset is valid)
+ *                   2= Joliet (to_charset gets overridden by UCS-2 or UTF-16)
+ *                   3= ECMA-119 (to_charset gets overridden by 
+ *                                dull ISO 9660 subset of ASCII)
+ *                   4= HFS+ (to_charset gets overridden by UTF-16BE)
+ *        bit8=  Input text is a directory name
+ *               (matters for Joliet and ECMA-119)
+ *        bit15= Reverse operation (best to be done only with results of
+ *               previous conversions)
+ * @return
+ *      1 means success, <0 means error
+ *
+ * @since 1.3.6
+ */
+int iso_conv_name_chars(IsoWriteOpts *opts, char *name, size_t name_len,
+                        char **result, size_t *result_len, int flag);
+
+
+
 /************ Error codes and return values for libisofs ********************/
 
 /** successfully execution */

libisofs/libisofs.ver

@@ -19,6 +19,7 @@ el_torito_set_load_seg;
 el_torito_set_load_size;
 el_torito_set_no_bootable;
 el_torito_set_selection_crit;
+iso_conv_name_chars;
 iso_data_source_new_from_file;
 iso_data_source_ref;
 iso_data_source_unref;

libisofs/util.c

@@ -241,8 +241,8 @@ ex:;
     return retval;
 }
 
-int strnconv(const char *str, const char *icharset, const char *ocharset,
+int strnconvl(const char *str, const char *icharset, const char *ocharset,
-             size_t len, char **output)
+              size_t len, char **output, size_t *out_len)
 {
     size_t inbytes;
     size_t outbytes;
@@ -291,6 +291,15 @@ ex:;
     return retval;
 }
 
+int strnconv(const char *str, const char *icharset, const char *ocharset,
+             size_t len, char **output)
+{
+    size_t l;
+
+    return strnconvl(str, icharset, ocharset, len, output, &l);
+}
+
+
 /**
  * Convert a str in a specified codeset to WCHAR_T. 
  * The result must be free() when no more needed
@@ -2227,3 +2236,13 @@ void iso_handle_split_utf16(uint16_t *utf_word)
         set_ucsbe(utf_word, '_');
 }
 
+
+void iso_smash_chars_for_joliet(uint16_t *name)
+{
+    size_t i;
+
+    for (i = 0; name[i] != 0; i++)
+        if (! valid_j_char(name[i]))
+            set_ucsbe(name + i, '_');
+}
+

libisofs/util.h

@@ -56,9 +56,16 @@ int iso_init_locale(int flag);
 int strconv(const char *input, const char *icharset, const char *ocharset,
             char **output);
 
+/* Like strconv but processing len input bytes rather than strlen(input)
+ */
 int strnconv(const char *str, const char *icharset, const char *ocharset,
              size_t len, char **output);
 
+/* Like strnconv but also returning the number of bytes in *output.
+ */
+int strnconvl(const char *str, const char *icharset, const char *ocharset,
+              size_t len, char **output, size_t *out_len);
+
 /**
  * Convert a given string from any input charset to ASCII
  * 
@@ -204,6 +211,11 @@ uint16_t *iso_j_file_id(const uint16_t *src, int flag);
  */
 uint16_t *iso_j_dir_id(const uint16_t *src, int flag);
 
+/**
+ * Maps forbidden Joliet characters to UCS-2 '_'
+ */
+void iso_smash_chars_for_joliet(uint16_t *name);
+
 /**
  * Like strlen, but for Joliet strings.
  */