New API call iso_write_opts_set_joliet_utf16()

and ability to read Joliet names as UTF-16BE

ChangeLog

@@ -1,3 +1,8 @@
+bzr branch lp:libisofs/for-libisoburn (to become libisofs-1.3.6.tar.gz)
+===============================================================================
+* New API call iso_write_opts_set_joliet_utf16() and ability to read Joliet
+  names as UTF-16BE
+
 libisofs-1.3.4.tar.gz Thu Dec 12 2013
 ===============================================================================
 * Giving sort weight 2 as default to El Torito boot images

libisofs/ecma119.c

@@ -1510,6 +1510,21 @@ ex:;
 }
 
 
+static
+void issue_write_warning_summary(Ecma119Image *target)
+{
+    if (target->joliet_ucs2_failures > ISO_JOLIET_UCS2_WARN_MAX) {
+        iso_msg_submit(-1, ISO_NAME_NOT_UCS2, 0,
+                      "More filenames found which were not suitable for Joliet character set UCS-2");
+    }
+    if (target->joliet_ucs2_failures > 0) {
+        iso_msg_submit(-1, ISO_NAME_NOT_UCS2, 0,
+           "Sum of filenames not suitable for Joliet character set UCS-2: %.f",
+                       (double) target->joliet_ucs2_failures);
+    }
+}
+
+
 static
 void *write_function(void *arg)
 {
@@ -1566,6 +1581,8 @@ void *write_function(void *arg)
     if (res <= 0)
         goto write_error;
 
+    issue_write_warning_summary(target);
+
     target->image->generator_is_running = 0;
 
     /* Give up reference claim made in ecma119_image_new().
@@ -1783,6 +1800,7 @@ int ecma119_image_new(IsoImage *src, IsoWriteOpts *opts, Ecma119Image **img)
     target->relaxed_vol_atts = opts->relaxed_vol_atts;
     target->joliet_longer_paths = opts->joliet_longer_paths;
     target->joliet_long_names = opts->joliet_long_names;
+    target->joliet_utf16 = opts->joliet_utf16;
     target->rrip_version_1_10 = opts->rrip_version_1_10;
     target->rrip_1_10_px_ino = opts->rrip_1_10_px_ino;
     target->aaip_susp_1_10 = opts->aaip_susp_1_10;
@@ -2007,6 +2025,8 @@ int ecma119_image_new(IsoImage *src, IsoWriteOpts *opts, Ecma119Image **img)
     target->filesrc_start = 0;
     target->filesrc_blocks = 0;
 
+    target->joliet_ucs2_failures = 0;
+
     /*
      * 2. Based on those options, create needed writers: iso, joliet...
      * Each writer inits its structures and stores needed info into
@@ -2758,6 +2778,7 @@ int iso_write_opts_new(IsoWriteOpts **opts, int profile)
     wopts->fat = 0;
     wopts->fifo_size = 1024; /* 2 MB buffer */
     wopts->sort_files = 1; /* file sorting is always good */
+    wopts->joliet_utf16 = 0;
     wopts->rr_reloc_dir = NULL;
     wopts->rr_reloc_flags = 0;
     wopts->system_area_data = NULL;
@@ -3035,6 +3056,15 @@ int iso_write_opts_set_joliet_long_names(IsoWriteOpts *opts, int allow)
     return ISO_SUCCESS;
 }
 
+int iso_write_opts_set_joliet_utf16(IsoWriteOpts *opts, int allow)
+{
+    if (opts == NULL) {
+        return ISO_NULL_POINTER;
+    }
+    opts->joliet_utf16 = allow ? 1 : 0;
+    return ISO_SUCCESS;
+}
+
 int iso_write_opts_set_rrip_version_1_10(IsoWriteOpts *opts, int oldvers)
 {
     if (opts == NULL) {

libisofs/ecma119.h

@@ -87,6 +87,12 @@
 #define ISO_GPT_ENTRIES_MAX 248
 
 
+/* How many warnings to issue about writing Joliet names which cannot be
+   properly represented in UCS-2 and thus had to be defaultet to '_'.
+*/
+#define ISO_JOLIET_UCS2_WARN_MAX 3
+
+
 /**
  * Holds the options for the image generation.
  */
@@ -192,6 +198,11 @@ struct iso_write_opts {
      */
     unsigned int joliet_long_names :1;
 
+    /**
+     * Use UTF-16BE rather than its subset UCS-2
+     */
+    unsigned int joliet_utf16 :1;
+
     /**
      * Write Rock Ridge info as of specification RRIP-1.10 rather than
      * RRIP-1.12: signature "RRIP_1991A" rather than "IEEE_1282",
@@ -540,6 +551,9 @@ struct ecma119_image
     /** Allow Joliet names up to 103 characters rather than 64  */
     unsigned int joliet_long_names :1;
 
+    /** Use UTF-16BE rather than its subset UCS-2 */
+    unsigned int joliet_utf16 :1;
+
     /** Write old fashioned RRIP-1.10 rather than RRIP-1.12 */
     unsigned int rrip_version_1_10 :1;
 
@@ -642,6 +656,7 @@ struct ecma119_image
     uint32_t joliet_path_table_size;
     uint32_t joliet_l_path_table_pos;
     uint32_t joliet_m_path_table_pos;
+    size_t joliet_ucs2_failures;
 
     /*
      * HFS+ related information

libisofs/fs_image.c

@@ -1252,7 +1252,7 @@ char *get_name(_ImageFsData *fsdata, const char *str, size_t len)
             return name;
         } else {
             ret = iso_msg_submit(fsdata->msgid, ISO_FILENAME_WRONG_CHARSET, ret,
-                "Charset conversion error. Cannot convert from %s to %s",
+                "Cannot convert from charset %s to %s",
                 fsdata->input_charset, fsdata->local_charset);
             if (ret < 0) {
                 return NULL; /* aborted */
@@ -1751,7 +1751,7 @@ if (name != NULL && !namecont) {
                 LIBISO_FREE_MEM(msg);
                 LIBISO_ALLOC_MEM(msg, char, 160);
                 sprintf(msg,
-                "Charset conversion error. Cannot convert from %.40s to %.40s",
+                "Cannot convert from charset %.40s to %.40s",
                                  fsdata->input_charset, fsdata->local_charset);
                 ret = iso_rr_msg_submit(fsdata, 17, ISO_FILENAME_WRONG_CHARSET,
                                         ret, msg);
@@ -2906,7 +2906,10 @@ int iso_image_filesystem_new(IsoDataSource *src, struct iso_read_opts *opts,
         if (!opts->nojoliet && opts->preferjoliet && data->joliet) {
             /* if user prefers joliet, that is used */
             iso_msg_debug(data->msgid, "Reading Joliet extensions.");
-            data->input_charset = strdup("UCS-2BE");
+            /* Although Joliet prescribes UCS-2BE, interpret names by its
+               superset UTF-16BE in order to avoid conversion failures.
+            */
+            data->input_charset = strdup("UTF-16BE");
             data->rr = RR_EXT_NO;
             data->iso_root_block = data->svd_root_block;
         } else {
@@ -2919,7 +2922,7 @@ int iso_image_filesystem_new(IsoDataSource *src, struct iso_read_opts *opts,
         if (!opts->nojoliet && data->joliet) {
             /* joliet will be used */
             iso_msg_debug(data->msgid, "Reading Joliet extensions.");
-            data->input_charset = strdup("UCS-2BE");
+            data->input_charset = strdup("UTF-16BE");
             data->iso_root_block = data->svd_root_block;
         } else if (!opts->noiso1999 && data->iso1999) {
             /* we will read ISO 9660:1999 */

libisofs/joliet.c

@@ -31,19 +31,41 @@
 static
 int get_joliet_name(Ecma119Image *t, IsoNode *iso, uint16_t **name)
 {
-    int ret;
+    int ret = ISO_SUCCESS;
-    uint16_t *ucs_name;
+    uint16_t *ucs_name = NULL, *utf16_name = NULL;
     uint16_t *jname = NULL;
 
     if (iso->name == NULL) {
         /* it is not necessarily an error, it can be the root */
+        *name = NULL;
         return ISO_SUCCESS;
     }
 
-    ret = str2ucs(t->input_charset, iso->name, &ucs_name);
+    if (t->joliet_utf16) {
-    if (ret < 0) {
+        ret = str2utf16be(t->input_charset, iso->name, &ucs_name);
-        iso_msg_debug(t->image->id, "Can't convert %s", iso->name);
+        if (ret < 0) {
-        return ret;
+            iso_msg_debug(t->image->id, "Cannot convert to UTF-16 : \"%s\"",
+                          iso->name);
+            goto ex;
+        }
+    } else {
+        ret = str2ucs(t->input_charset, iso->name, &ucs_name);
+        if (ret < 0) {
+            iso_msg_debug(t->image->id, "Cannot convert to UCS-2 : \"%s\"",
+                          iso->name);
+            goto ex;
+        }
+        ret = str2utf16be(t->input_charset, iso->name, &utf16_name);
+        if (ret == ISO_SUCCESS) {
+            if (ucscmp(ucs_name, utf16_name) != 0) {
+                t->joliet_ucs2_failures++;
+                if (t->joliet_ucs2_failures <= ISO_JOLIET_UCS2_WARN_MAX) {
+                    iso_msg_submit(t->image->id, ISO_NAME_NOT_UCS2, 0,
+               "Filename not suitable for Joliet character set UCS-2 : \"%s\"",
+                                   iso->name);
+                }
+            }
+        }
     }
     if (iso->type == LIBISO_DIR) {
         jname = iso_j_dir_id(ucs_name, t->joliet_long_names << 1);
@@ -51,8 +73,17 @@ int get_joliet_name(Ecma119Image *t, IsoNode *iso, uint16_t **name)
         jname = iso_j_file_id(ucs_name,
                        (t->joliet_long_names << 1) | !!(t->no_force_dots & 2));
     }
-    free(ucs_name);
+    ret = ISO_SUCCESS;
-    if (jname != NULL) {
+ex:;
+    if (ucs_name != NULL)
+        free(ucs_name);
+    if (utf16_name != NULL)
+        free(utf16_name);
+    if (ret != ISO_SUCCESS) {
+        if (jname != NULL)
+            free(jname);
+        return ret;
+    } else if (jname != NULL) {
         *name = jname;
         return ISO_SUCCESS;
     } else {
@@ -828,18 +859,22 @@ void ucsncpy_pad(uint16_t *dest, const uint16_t *src, size_t max)
     csrc = (char*)src;
 
     if (src != NULL) {
-        len = MIN(ucslen(src) * 2, max);
+        len = MIN(ucslen(src) * 2, max - (max % 2));
     } else {
         len = 0;
     }
 
     for (i = 0; i < len; ++i)
         cdest[i] = csrc[i];
+    if (len >= 2)
+        iso_handle_split_utf16(dest + (len / 2 - 1));
 
-    for (i = len; i < max; i += 2) {
+    for (i = len; i + 1 < max; i += 2) {
         cdest[i] = '\0';
         cdest[i + 1] = ' ';
     }
+    if (max % 2)
+        cdest[max - 1] = 0;
 }
 
 int joliet_writer_write_vol_desc(IsoImageWriter *writer)

libisofs/libisofs.h

@@ -1771,6 +1771,16 @@ int iso_write_opts_set_joliet_longer_paths(IsoWriteOpts *opts, int allow);
  */
 int iso_write_opts_set_joliet_long_names(IsoWriteOpts *opts, int allow);
 
+/**
+ * Use character set UTF-16BE with Joliet, which is a superset of the
+ * actually prescribed character set UCS-2.
+ * This breaks Joliet specification with exotic characters which would
+ * elsewise be mapped to underscore '_'. Use with caution.
+ *
+ * @since 1.3.6
+ */
+int iso_write_opts_set_joliet_utf16(IsoWriteOpts *opts, int allow);
+
 /**
  * Write Rock Ridge info as of specification RRIP-1.10 rather than RRIP-1.12:
  * signature "RRIP_1991A" rather than "IEEE_1282", field PX without file
@@ -7544,6 +7554,9 @@ int iso_image_hfsplus_get_blessed(IsoImage *img, IsoNode ***blessed_nodes,
 /** Unrecognized file type in ISO image            (FAILURE, HIGH, -396) */
 #define ISO_BAD_ISO_FILETYPE        0xE830FE74
 
+/** Filename not suitable for character set UCS-2  (WARNING, HIGH, -397) */
+#define ISO_NAME_NOT_UCS2           0xD030FE73
+
 
 /* Internal developer note: 
    Place new error codes directly above this comment. 

libisofs/libisofs.ver

@@ -303,6 +303,7 @@ iso_write_opts_set_iso_level;
 iso_write_opts_set_joliet;
 iso_write_opts_set_joliet_long_names;
 iso_write_opts_set_joliet_longer_paths;
+iso_write_opts_set_joliet_utf16;
 iso_write_opts_set_max_37_char_filenames;
 iso_write_opts_set_ms_block;
 iso_write_opts_set_no_force_dots;

libisofs/messages.c

@@ -503,6 +503,8 @@ const char *iso_error_to_msg(int errcode)
         return "Too many chained symbolic links";
     case ISO_BAD_ISO_FILETYPE:
         return "Unrecognized file type in ISO image";
+    case ISO_NAME_NOT_UCS2:
+         return "Filename not suitable for character set UCS-2";
     default:
         return "Unknown error";
     }

libisofs/util.c

@@ -1197,7 +1197,7 @@ uint16_t *iso_j_file_id(const uint16_t *src, int flag)
 {
     uint16_t *dot, *retval = NULL;
     size_t lname, lext, lnname, lnext, pos, i, maxchar = 64;
-    uint16_t *dest = NULL;
+    uint16_t *dest = NULL, c;
 
     LIBISO_ALLOC_MEM_VOID(dest, uint16_t, LIBISO_JOLIET_NAME_MAX);
                                /* was: 66 = 64 (name + ext) + 1 (.) + 1 (\0) */
@@ -1237,7 +1237,7 @@ uint16_t *iso_j_file_id(const uint16_t *src, int flag)
 
     /* Convert up to lnname characters of the filename. */
     for (i = 0; i < lnname; i++) {
-        uint16_t c = src[i];
+        c = src[i];
         if (valid_j_char(c)) {
             dest[pos++] = c;
         } else {
@@ -1245,6 +1245,7 @@ uint16_t *iso_j_file_id(const uint16_t *src, int flag)
             pos++;
         }
     }
+    iso_handle_split_utf16(dest + (pos - 1));
 
     if ((flag & 1) && lnext <= 0)
         goto is_done;
@@ -1262,6 +1263,7 @@ uint16_t *iso_j_file_id(const uint16_t *src, int flag)
             pos++;
         }
     }
+    iso_handle_split_utf16(dest + (pos - 1));
 
 is_done:;
     set_ucsbe(dest + pos, '\0');
@@ -1298,6 +1300,7 @@ uint16_t *iso_j_dir_id(const uint16_t *src, int flag)
             set_ucsbe(dest + i, '_');
         }
     }
+    iso_handle_split_utf16(dest + (len - 1));
     set_ucsbe(dest + len, '\0');
     retval = ucsdup(dest);
 ex:
@@ -1379,6 +1382,8 @@ uint16_t *ucsncpy(uint16_t *dest, const uint16_t *src, size_t n)
 {
     n = MIN(n, ucslen(src) + 1);
     memcpy(dest, src, n*2);
+    if (n >= 2)
+        iso_handle_split_utf16(dest + (n - 2));
     return dest;
 }
 
@@ -2209,3 +2214,16 @@ uint16_t iso_htons(uint16_t v)
     return ret;
 }   
 
+
+/* If an UTF-16 surrogate pair was split : Change to UTF-16 '_'.
+   (UCS-2 is promised to reserve 0xd800 to 0xdbff for UTF-16).
+*/
+void iso_handle_split_utf16(uint16_t *utf_word)
+{
+    unsigned char *hb;
+
+    hb = (unsigned char *) utf_word;
+    if ((hb[0] & 0xfc) == 0xd8)
+        set_ucsbe(utf_word, '_');
+}
+

libisofs/util.h

@@ -236,6 +236,12 @@ uint16_t *ucscpy(uint16_t *dest, const uint16_t *src);
  */
 uint16_t *ucsncpy(uint16_t *dest, const uint16_t *src, size_t n);
 
+/**
+ * Check whether utf_word is the first surrogate word of a pair.
+ * If so, change it to UTF-16 character '_'.
+ */
+void iso_handle_split_utf16(uint16_t *utf_word);
+
 /**
  * Convert a given input string to d-chars.
  * @return