toolchain: add option to copy the gconv libraries

The gconv libraries are used to translate between different character sets ('charsets', even 'csets' sometimes). Some packages need them to present text to the user (eg. XBMC Gotham).

In (e)glibc they are implemented by the internal implemenation of iconv, called gconv, and are provided as dlopen-able libraries.

Note that some gconv modules need extra libraries (shared by more than one gconv module), so we must, when adding a subset of modules, scan the installed modules in search of the missing libraries.

Signed-off-by "Yann E. MORIN" <[email protected]>

Cc Maxime Hadjinlian <[email protected]>

Cc Bernd Kuhls <[email protected]>

Signed-off-by Bernd Kuhls <[email protected]>

Cc Eric Limpens <[email protected]>

---
Notes (kept from previous submission):

- this is getting a bit urgent, as without it, XBMC is basically unusable, as it does not render the labels on the menus.

I would really appreciate a review on this, so I have time to respin it before -rc1 if changes are required.

- contrary to what Thomas asked, I did not provide a minimalist list of default gconv modules to install. Thomas was concerned about the size of the installed modules. But those modules are probably only used in GUIs, which would probably already pull one of the graphical frameworks such as Qt or GTK, which are already huge beasts. So, I still think that the default should be to install all modules.

Changes v5 -> v6:
- move the finalise hook to its own .mk file (Thomas) all other toolchain-related finalise-hooks should eventually be moved to that file; can be done in -next
- also install dependable libraries
- generate the expunged gconv-modules file in one shot, by calling the support script only once with all modules (instead of many calls with a single module)
- remove unused variable in the gawk script

Changes v4 -> v5;
- expunge gconv-module from not-installed modules (Thomas)

Change v3 -> v4:
- fix superfluous semi-colon before '||'

Changes v2 -> v3:
- properly fail if a specified lib is missing (e.g. typo)

Changes v1 -> v2:
- add option to install only a select list of gconv libs (Thomas)
- rename the option (Thomas)

---
 Makefile                              |  2 +-
 support/scripts/expunge-gconv-modules | 52 +++++++++++++++++++++++++++++++++++
 toolchain/toolchain-common.in         | 22 +++++++++++++++
 toolchain/toolchain.mk                | 36 ++++++++++++++++++++++++
 4 files changed, 111 insertions(+), 1 deletion(-)
 create mode 100755 support/scripts/expunge-gconv-modules
 create mode 100644 toolchain/toolchain.mk

Makefile +1 -1

index 6bd18e3..d4ef081 100644
@@ -369,7 +369,7 @@ include support/dependencies/dependencies.mk
 # We also need the various per-package makefiles, which also add
 # each selected package to TARGETS if that package was selected
 # in the .config file.
-include toolchain/helpers.mk
+include toolchain/*.mk
 include toolchain/*/*.mk
 
 # Include the package override file if one has been provided in the

support/scripts/expunge-gconv-modules new file +52 -0

new file mode 100755
index 0000000..7810246
@@ -0,0 +1,52 @@
+#!/bin/bash
+
+# The format of gconv-modules is precisely documented in the
+# file itself. It consists of two different directives:
+#   module  FROMSET  TOSET  FILENAME  COST
+#   alias   ALIAS  REALNAME
+# and that's what this script parses and generates.
+#
+# There are two kinds of 'module' directives:
+#   - the first defines conversion of a charset to/from INTERNAL representation
+#   - the second defines conversion of a charset to/from another charset
+# we handle each with slightly different code, since the second never has
+# associated aliases.
+
+gawk -v files="${1}" '
+$1 == "alias" {
+    aliases[$3] = aliases[$3] " " $2;
+}
+$1 == "module" && $2 != "INTERNAL" && $3 == "INTERNAL" {
+    file2internals[$4] = file2internals[$4] " " $2;
+    mod2cost[$2] = $5;
+}
+$1 == "module" && $2 != "INTERNAL" && $3 != "INTERNAL" {
+    file2cset[$4] = file2cset[$4] " " $2 ":" $3;
+    mod2cost[$2] = $5;
+}
+
+END {
+    nb_files = split( files, all_files );
+    for( f=1; f<=nb_files; f++ ) {
+        file = all_files[f];
+        printf( "# Modules and aliases for: %s\n", file );
+        nb_mods = split( file2internals[file], mods );
+        for( i=1; i<=nb_mods; i++ ) {
+            nb_aliases = split( aliases[mods[i]], mod_aliases );
+            for( j=1; j<=nb_aliases; j++ ) {
+                printf( "alias\t%s\t%s\n", mod_aliases[j], mods[i] );
+            }
+            printf( "module\t%s\t%s\t%s\t%d\n", mods[i], "INTERNAL", file, mod2cost[mods[i]] );
+            printf( "module\t%s\t%s\t%s\t%d\n", "INTERNAL", mods[i], file, mod2cost[mods[i]] );
+            printf( "\n" );
+        }
+        printf( "%s", nb_mods!=0?"\n":"" );
+        nb_csets = split( file2cset[file], csets );
+        for( i=1; i<=nb_csets; i++ ) {
+            split( csets[i], cs, ":" );
+            printf( "module\t%s\t%s\t%s\t%d\n", cs[1], cs[2], file, mod2cost[cs[1]] );
+        }
+        printf( "%s", nb_csets!=0?"\n\n":"" );
+    }
+}
+'

toolchain/toolchain-common.in +22 -0

index a91d247..13de9e5 100644
@@ -77,6 +77,28 @@ config BR2_GENERATE_LOCALE
          specified, UTF-8 is assumed. Examples of locales: en_US,
          fr_FR.UTF-8.
 
+config BR2_TOOLCHAIN_GLIBC_GCONV_LIBS_COPY
+       bool "Copy gconv libraries"
+       depends on BR2_TOOLCHAIN_USES_GLIBC
+       help
+         The gconv libraries are used to convert between different
+         character sets (charsets).
+
+         Say 'y' if you need to store and/or display different charsets.
+
+config BR2_TOOLCHAIN_GLIBC_GCONV_LIBS_LIST
+       string "Gconv libraries to copy"
+       depends on BR2_TOOLCHAIN_GLIBC_GCONV_LIBS_COPY
+       help
+         Set to the list of gconv libraries to copy.
+         Leave empty to copy all gconv libraries.
+
+         Specify only the basename of the libraries, leave
+         out the .so extension. Eg.:
+           IBM850 ISO8859-15 UNICODE
+
+         Note: the full set of gconv libs are ~8MiB (on ARM).
+
 # glibc and eglibc directly include gettext, so a separatly compiled
 # gettext isn't needed and shouldn't be built to avoid conflicts. Some
 # packages always need gettext, other packages only need gettext when

toolchain/toolchain.mk new file +36 -1

new file mode 100644
index 0000000..8fe06ff
@@ -0,0 +1,36 @@
+# This file contains toolchain-related customisation of the content
+# of the target/ directory. Those customisations are added to the
+# TARGET_FINALIZE_HOOKS, to be applied just after all packages
+# have been built.
+
+# Install the gconv modules
+ifeq ($(BR2_TOOLCHAIN_GLIBC_GCONV_LIBS_COPY),y)
+GCONV_LIBS = $(call qstrip,$(BR2_TOOLCHAIN_GLIBC_GCONV_LIBS_LIST))
+define COPY_GCONV_LIBS
+       $(Q)if [ -z "$(GCONV_LIBS)" ]; then \
+               $(INSTALL) -m 0644 -D $(STAGING_DIR)/usr/lib/gconv/gconv-modules \
+                                     $(TARGET_DIR)/usr/lib/gconv/gconv-modules; \
+               $(INSTALL) -m 0644 $(STAGING_DIR)/usr/lib/gconv/*.so \
+                                  $(TARGET_DIR)/usr/lib/gconv \
+               || exit 1; \
+       else \
+               for l in $(GCONV_LIBS); do \
+                       $(INSTALL) -m 0644 -D $(STAGING_DIR)/usr/lib/gconv/$${l}.so \
+                                             $(TARGET_DIR)/usr/lib/gconv/$${l}.so \
+                       || exit 1; \
+                       $(TARGET_READELF) -d $(STAGING_DIR)/usr/lib/gconv/$${l}.so |\
+                       sort -u |\
+                       sed -e '/.*(NEEDED).*\[\(.*\.so\)\]$$/!d; s//\1/;' |\
+                       while read lib; do \
+                                $(INSTALL) -m 0644 -D $(STAGING_DIR)/usr/lib/gconv/$${lib} \
+                                                      $(TARGET_DIR)/usr/lib/gconv/$${lib} \
+                                || exit 1; \
+                       done; \
+               done; \
+               ./support/scripts/expunge-gconv-modules "$(GCONV_LIBS)" \
+                       <$(STAGING_DIR)/usr/lib/gconv/gconv-modules \
+                       >$(TARGET_DIR)/usr/lib/gconv/gconv-modules; \
+       fi
+endef
+TARGET_FINALIZE_HOOKS += COPY_GCONV_LIBS
+endif
--

1.9.1

The gconv libraries are used to translate between different character sets ('charsets', even 'csets' sometimes). Some packages need them to present text to the user (eg. XBMC Gotham).

In (e)glibc they are implemented by the internal implemenation of iconv, called gconv, and are provided as dlopen-able libraries.

Note that some gconv modules need extra libraries (shared by more than one gconv module), so we must, when adding a subset of modules, scan the installed modules in search of the missing libraries.

Signed-off-by "Yann E. MORIN" <[email protected]>

Cc Maxime Hadjinlian <[email protected]>

Cc Bernd Kuhls <[email protected]>

Signed-off-by Bernd Kuhls <[email protected]>

Cc Eric Limpens <[email protected]>