Add a support for utf-8 in archive_entry fields

diff --git a/libarchive/archive_entry.c b/libarchive/archive_entry.c
index 293c7016a82cf256f1faf412b336bc6ffb50d7d0..7958a17e31186c2b78b95e0a5fd2490a18c27794 100644 (file)
--- a/libarchive/archive_entry.c
+++ b/libarchive/archive_entry.c
@@ -418,6 +418,18 @@ archive_entry_gname(struct archive_entry *entry)
        return (NULL);
 }
 
+const char *
+archive_entry_gname_utf8(struct archive_entry *entry)
+{
+       const char *p;
+       if (archive_mstring_get_utf8(entry->archive, &entry->ae_gname, &p) == 0)
+               return (p);
+       if (errno == ENOMEM)
+               __archive_errx(1, "No memory");
+       return (NULL);
+}
+
+
 const wchar_t *
 archive_entry_gname_w(struct archive_entry *entry)
 {
@@ -450,6 +462,20 @@ archive_entry_hardlink(struct archive_entry *entry)
        return (NULL);
 }
 
+const char *
+archive_entry_hardlink_utf8(struct archive_entry *entry)
+{
+       const char *p;
+       if ((entry->ae_set & AE_SET_HARDLINK) == 0)
+               return (NULL);
+       if (archive_mstring_get_utf8(
+           entry->archive, &entry->ae_hardlink, &p) == 0)
+               return (p);
+       if (errno == ENOMEM)
+               __archive_errx(1, "No memory");
+       return (NULL);
+}
+
 const wchar_t *
 archive_entry_hardlink_w(struct archive_entry *entry)
 {
@@ -536,6 +562,18 @@ archive_entry_pathname(struct archive_entry *entry)
        return (NULL);
 }
 
+const char *
+archive_entry_pathname_utf8(struct archive_entry *entry)
+{
+       const char *p;
+       if (archive_mstring_get_utf8(
+           entry->archive, &entry->ae_pathname, &p) == 0)
+               return (p);
+       if (errno == ENOMEM)
+               __archive_errx(1, "No memory");
+       return (NULL);
+}
+
 const wchar_t *
 archive_entry_pathname_w(struct archive_entry *entry)
 {
@@ -637,6 +675,20 @@ archive_entry_symlink(struct archive_entry *entry)
        return (NULL);
 }
 
+const char *
+archive_entry_symlink_utf8(struct archive_entry *entry)
+{
+       const char *p;
+       if ((entry->ae_set & AE_SET_SYMLINK) == 0)
+               return (NULL);
+       if (archive_mstring_get_utf8(
+           entry->archive, &entry->ae_symlink, &p) == 0)
+               return (p);
+       if (errno == ENOMEM)
+               __archive_errx(1, "No memory");
+       return (NULL);
+}
+
 const wchar_t *
 archive_entry_symlink_w(struct archive_entry *entry)
 {
@@ -680,6 +732,17 @@ archive_entry_uname(struct archive_entry *entry)
        return (NULL);
 }
 
+const char *
+archive_entry_uname_utf8(struct archive_entry *entry)
+{
+       const char *p;
+       if (archive_mstring_get_utf8(entry->archive, &entry->ae_uname, &p) == 0)
+               return (p);
+       if (errno == ENOMEM)
+               __archive_errx(1, "No memory");
+       return (NULL);
+}
+
 const wchar_t *
 archive_entry_uname_w(struct archive_entry *entry)
 {
@@ -768,6 +831,12 @@ archive_entry_set_gname(struct archive_entry *entry, const char *name)
        archive_mstring_copy_mbs(&entry->ae_gname, name);
 }
 
+void
+archive_entry_set_gname_utf8(struct archive_entry *entry, const char *name)
+{
+       archive_mstring_copy_utf8(&entry->ae_gname, name);
+}
+
 void
 archive_entry_copy_gname(struct archive_entry *entry, const char *name)
 {
@@ -824,6 +893,16 @@ archive_entry_set_hardlink(struct archive_entry *entry, const char *target)
                entry->ae_set &= ~AE_SET_HARDLINK;
 }
 
+void
+archive_entry_set_hardlink_utf8(struct archive_entry *entry, const char *target)
+{
+       archive_mstring_copy_utf8(&entry->ae_hardlink, target);
+       if (target != NULL)
+               entry->ae_set |= AE_SET_HARDLINK;
+       else
+               entry->ae_set &= ~AE_SET_HARDLINK;
+}
+
 void
 archive_entry_copy_hardlink(struct archive_entry *entry, const char *target)
 {
@@ -962,6 +1041,15 @@ archive_entry_set_link(struct archive_entry *entry, const char *target)
                archive_mstring_copy_mbs(&entry->ae_hardlink, target);
 }
 
+void
+archive_entry_set_link_utf8(struct archive_entry *entry, const char *target)
+{
+       if (entry->ae_set & AE_SET_SYMLINK)
+               archive_mstring_copy_utf8(&entry->ae_symlink, target);
+       else
+               archive_mstring_copy_utf8(&entry->ae_hardlink, target);
+}
+
 /* Set symlink if symlink is already set, else set hardlink. */
 void
 archive_entry_copy_link(struct archive_entry *entry, const char *target)
@@ -1051,6 +1139,12 @@ archive_entry_set_pathname(struct archive_entry *entry, const char *name)
        archive_mstring_copy_mbs(&entry->ae_pathname, name);
 }
 
+void
+archive_entry_set_pathname_utf8(struct archive_entry *entry, const char *name)
+{
+       archive_mstring_copy_utf8(&entry->ae_pathname, name);
+}
+
 void
 archive_entry_copy_pathname(struct archive_entry *entry, const char *name)
 {
@@ -1151,6 +1245,16 @@ archive_entry_set_symlink(struct archive_entry *entry, const char *linkname)
                entry->ae_set &= ~AE_SET_SYMLINK;
 }
 
+void
+archive_entry_set_symlink_utf8(struct archive_entry *entry, const char *linkname)
+{
+       archive_mstring_copy_utf8(&entry->ae_symlink, linkname);
+       if (linkname != NULL)
+               entry->ae_set |= AE_SET_SYMLINK;
+       else
+               entry->ae_set &= ~AE_SET_SYMLINK;
+}
+
 void
 archive_entry_copy_symlink(struct archive_entry *entry, const char *linkname)
 {
@@ -1214,6 +1318,12 @@ archive_entry_set_uname(struct archive_entry *entry, const char *name)
        archive_mstring_copy_mbs(&entry->ae_uname, name);
 }
 
+void
+archive_entry_set_uname_utf8(struct archive_entry *entry, const char *name)
+{
+       archive_mstring_copy_utf8(&entry->ae_uname, name);
+}
+
 void
 archive_entry_copy_uname(struct archive_entry *entry, const char *name)
 {

diff --git a/libarchive/archive_entry.h b/libarchive/archive_entry.h
index ff5350a8edff94d6dc0e942166b3ad79cf6582ad..9f7fafecd4cb90adb952751070ec663622cacfca 100644 (file)
--- a/libarchive/archive_entry.h
+++ b/libarchive/archive_entry.h
@@ -208,8 +208,10 @@ __LA_DECL void              archive_entry_fflags(struct archive_entry *,
 __LA_DECL const char   *archive_entry_fflags_text(struct archive_entry *);
 __LA_DECL __LA_INT64_T  archive_entry_gid(struct archive_entry *);
 __LA_DECL const char   *archive_entry_gname(struct archive_entry *);
+__LA_DECL const char   *archive_entry_gname_utf8(struct archive_entry *);
 __LA_DECL const wchar_t        *archive_entry_gname_w(struct archive_entry *);
 __LA_DECL const char   *archive_entry_hardlink(struct archive_entry *);
+__LA_DECL const char   *archive_entry_hardlink_utf8(struct archive_entry *);
 __LA_DECL const wchar_t        *archive_entry_hardlink_w(struct archive_entry *);
 __LA_DECL __LA_INT64_T  archive_entry_ino(struct archive_entry *);
 __LA_DECL __LA_INT64_T  archive_entry_ino64(struct archive_entry *);
@@ -220,6 +222,7 @@ __LA_DECL long               archive_entry_mtime_nsec(struct archive_entry *);
 __LA_DECL int           archive_entry_mtime_is_set(struct archive_entry *);
 __LA_DECL unsigned int  archive_entry_nlink(struct archive_entry *);
 __LA_DECL const char   *archive_entry_pathname(struct archive_entry *);
+__LA_DECL const char   *archive_entry_pathname_utf8(struct archive_entry *);
 __LA_DECL const wchar_t        *archive_entry_pathname_w(struct archive_entry *);
 __LA_DECL __LA_MODE_T   archive_entry_perm(struct archive_entry *);
 __LA_DECL dev_t                 archive_entry_rdev(struct archive_entry *);
@@ -231,9 +234,11 @@ __LA_DECL __LA_INT64_T      archive_entry_size(struct archive_entry *);
 __LA_DECL int           archive_entry_size_is_set(struct archive_entry *);
 __LA_DECL const char   *archive_entry_strmode(struct archive_entry *);
 __LA_DECL const char   *archive_entry_symlink(struct archive_entry *);
+__LA_DECL const char   *archive_entry_symlink_utf8(struct archive_entry *);
 __LA_DECL const wchar_t        *archive_entry_symlink_w(struct archive_entry *);
 __LA_DECL __LA_INT64_T  archive_entry_uid(struct archive_entry *);
 __LA_DECL const char   *archive_entry_uname(struct archive_entry *);
+__LA_DECL const char   *archive_entry_uname_utf8(struct archive_entry *);
 __LA_DECL const wchar_t        *archive_entry_uname_w(struct archive_entry *);
 __LA_DECL int archive_entry_is_data_encrypted(struct archive_entry *);
 __LA_DECL int archive_entry_is_metadata_encrypted(struct archive_entry *);
@@ -271,16 +276,19 @@ __LA_DECL const wchar_t *archive_entry_copy_fflags_text_w(struct archive_entry *
            const wchar_t *);
 __LA_DECL void archive_entry_set_gid(struct archive_entry *, __LA_INT64_T);
 __LA_DECL void archive_entry_set_gname(struct archive_entry *, const char *);
+__LA_DECL void archive_entry_set_gname_utf8(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_gname(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_gname_w(struct archive_entry *, const wchar_t *);
 __LA_DECL int  archive_entry_update_gname_utf8(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_set_hardlink(struct archive_entry *, const char *);
+__LA_DECL void archive_entry_set_hardlink_utf8(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_hardlink(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_hardlink_w(struct archive_entry *, const wchar_t *);
 __LA_DECL int  archive_entry_update_hardlink_utf8(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_set_ino(struct archive_entry *, __LA_INT64_T);
 __LA_DECL void archive_entry_set_ino64(struct archive_entry *, __LA_INT64_T);
 __LA_DECL void archive_entry_set_link(struct archive_entry *, const char *);
+__LA_DECL void archive_entry_set_link_utf8(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_link(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_link_w(struct archive_entry *, const wchar_t *);
 __LA_DECL int  archive_entry_update_link_utf8(struct archive_entry *, const char *);
@@ -289,6 +297,7 @@ __LA_DECL void      archive_entry_set_mtime(struct archive_entry *, time_t, long);
 __LA_DECL void  archive_entry_unset_mtime(struct archive_entry *);
 __LA_DECL void archive_entry_set_nlink(struct archive_entry *, unsigned int);
 __LA_DECL void archive_entry_set_pathname(struct archive_entry *, const char *);
+__LA_DECL void archive_entry_set_pathname_utf8(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_pathname(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_pathname_w(struct archive_entry *, const wchar_t *);
 __LA_DECL int  archive_entry_update_pathname_utf8(struct archive_entry *, const char *);
@@ -301,11 +310,13 @@ __LA_DECL void    archive_entry_unset_size(struct archive_entry *);
 __LA_DECL void archive_entry_copy_sourcepath(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_sourcepath_w(struct archive_entry *, const wchar_t *);
 __LA_DECL void archive_entry_set_symlink(struct archive_entry *, const char *);
+__LA_DECL void archive_entry_set_symlink_utf8(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_symlink(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_symlink_w(struct archive_entry *, const wchar_t *);
 __LA_DECL int  archive_entry_update_symlink_utf8(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_set_uid(struct archive_entry *, __LA_INT64_T);
 __LA_DECL void archive_entry_set_uname(struct archive_entry *, const char *);
+__LA_DECL void archive_entry_set_uname_utf8(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_uname(struct archive_entry *, const char *);
 __LA_DECL void archive_entry_copy_uname_w(struct archive_entry *, const wchar_t *);
 __LA_DECL int  archive_entry_update_uname_utf8(struct archive_entry *, const char *);

diff --git a/libarchive/archive_string.c b/libarchive/archive_string.c
index 87f9288f19661a0483e100b8d1068f448e8a619f..eebd7642bec14163d6a058bb5b0c6df31b2688a6 100644 (file)
--- a/libarchive/archive_string.c
+++ b/libarchive/archive_string.c
@@ -4061,6 +4061,19 @@ archive_mstring_copy_wcs(struct archive_mstring *aes, const wchar_t *wcs)
                                wcs == NULL ? 0 : wcslen(wcs));
 }
 
+int
+archive_mstring_copy_utf8(struct archive_mstring *aes, const char *utf8)
+{
+  if (utf8 == NULL) {
+    aes->aes_set = 0;
+  }
+  aes->aes_set = AES_SET_UTF8;
+  archive_string_empty(&(aes->aes_mbs));
+  archive_string_empty(&(aes->aes_wcs));
+  archive_strncpy(&(aes->aes_utf8), utf8, strlen(utf8));
+  return strlen(utf8);
+}
+
 int
 archive_mstring_copy_wcs_len(struct archive_mstring *aes, const wchar_t *wcs,
     size_t len)