Mercurial > hg > xemacs-beta
view src/mule-ccl.c @ 2720:6fa9919a9a0b
[xemacs-hg @ 2005-04-08 23:10:01 by crestani]
ChangeLog addition:
2005-04-01 Marcus Crestani <crestani@xemacs.org>
The new allocator.
New configure flag: `MC_ALLOC':
* configure.ac (XE_COMPLEX_ARG_ENABLE): Add `--enable-mc-alloc' as
a new configure flag.
* configure.in (AC_INIT_PARSE_ARGS): Add `--mc-alloc' as a new
configure flag.
* configure.usage: Add description for `mc-alloc'.
DUMP_IN_EXEC:
* Makefile.in.in: Condition the installation of a separate dump
file on !DUMP_ON_EXEC.
* configure.ac (XE_COMPLEX_ARG_ENABLE): Add
`--enable-dump-in-exec' as a new configure flag.
* configure.ac: DUMP_IN_EXEC is define as default for PDUMP but
not default for MC_ALLOC.
* configure.in (AC_INIT_PARSE_ARGS): Add `--dump-in-exec' as a
new configure flag.
* configure.in: DUMP_IN_EXEC is define as default for PDUMP but
not default for MC_ALLOC.
* configure.usage: Add description for `dump-in-exec'.
lib-src/ChangeLog addition:
2005-04-01 Marcus Crestani <crestani@xemacs.org>
The new allocator.
DUMP_IN_EXEC:
* Makefile.in.in: Only compile insert-data-in-exec if
DUMP_IN_EXEC is defined.
lisp/ChangeLog addition:
2005-04-01 Marcus Crestani <crestani@xemacs.org>
The new allocator.
MEMORY_USAGE_STATS
* diagnose.el: Add new lisp function to pretty print statistics
about the new allocator.
* diagnose.el (show-mc-alloc-memory-usage): New.
modules/ChangeLog addition:
2005-04-01 Marcus Crestani <crestani@xemacs.org>
The new allocator.
Remove Lcrecords:
* postgresql/postgresql.c (allocate_pgconn): Allocate with new
allocator.
* postgresql/postgresql.c (allocate_pgresult): Allocate PGresult
with new allocator.
* postgresql/postgresql.h (struct Lisp_PGconn): Add
lrecord_header.
* postgresql/postgresql.h (struct Lisp_PGresult): Add
lrecord_header.
* ldap/eldap.c (allocate_ldap): Allocate with new allocator.
* ldap/eldap.h (struct Lisp_LDAP): Add lrecord_header.
nt/ChangeLog addition:
2005-04-01 Marcus Crestani <crestani@xemacs.org>
The new allocator.
New configure flag: `MC_ALLOC':
* config.inc.samp: Add new flag `MC_ALLOC'.
* xemacs.mak: Add flag and configuration output for `MC_ALLOC'.
New files:
* xemacs.dsp: Add source files mc-alloc.c and mc-alloc.h.
* xemacs.mak: Add new object file mc-alloc.obj to dependencies.
src/ChangeLog addition:
2005-04-01 Marcus Crestani <crestani@xemacs.org>
The new allocator.
New configure flag: `MC_ALLOC':
* config.h.in: Add new flag `MC_ALLOC'.
New files:
* Makefile.in.in: Add new object file mc-alloc.o.
* depend: Add new files to dependencies.
* mc-alloc.c: New.
* mc-alloc.h: New.
Running the new allocator from XEmacs:
* alloc.c (deadbeef_memory): Moved to mc-alloc.c.
* emacs.c (main_1): Initialize the new allocator and add
syms_of_mc_alloc.
* symsinit.h: Add syms_of_mc_alloc.
New lrecord allocation and free functions:
* alloc.c (alloc_lrecord): New. Allocates an lrecord, includes
type checking and initializing of the lrecord_header.
* alloc.c (noseeum_alloc_lrecord): Same as above, but increments
the NOSEEUM cons counter.
* alloc.c (free_lrecord): New. Calls the finalizer and frees the
lrecord.
* lrecord.h: Add lrecord allocation prototypes and comments.
Remove old lrecord FROB block allocation:
* alloc.c (allocate_lisp_storage): Former function to expand
heap. Not needed anymore, remove.
* alloc.c: Completely remove `Fixed-size type macros'
* alloc.c (release_breathing_space): Remove.
* alloc.c (memory_full): Remove release_breathing_space.
* alloc.c (refill_memory_reserve): Remove.
* alloc.c (TYPE_ALLOC_SIZE): Remove.
* alloc.c (DECLARE_FIXED_TYPE_ALLOC): Remove.
* alloc.c (ALLOCATE_FIXED_TYPE_FROM_BLOCK): Remove.
* alloc.c (ALLOCATE_FIXED_TYPE_1): Remove.
* alloc.c (ALLOCATE_FIXED_TYPE): Remove.
* alloc.c (NOSEEUM_ALLOCATE_FIXED_TYPE): Remove.
* alloc.c (struct Lisp_Free): Remove.
* alloc.c (LRECORD_FREE_P): Remove.
* alloc.c (MARK_LRECORD_AS_FREE): Remove.
* alloc.c (MARK_LRECORD_AS_NOT_FREE): Remove.
* alloc.c (PUT_FIXED_TYPE_ON_FREE_LIST): Remove.
* alloc.c (FREE_FIXED_TYPE): Remove.
* alloc.c (FREE_FIXED_TYPE_WHEN_NOT_IN_GC): Remove.
Allocate old lrecords with new allocator:
* alloc.c: DECLARE_FIXED_TYPE_ALLOC removed for all lrecords
defined in alloc.c.
* alloc.c (Fcons): Allocate with new allocator.
* alloc.c (noseeum_cons): Allocate with new allocator.
* alloc.c (make_float): Allocate with new allocator.
* alloc.c (make_bignum): Allocate with new allocator.
* alloc.c (make_bignum_bg): Allocate with new allocator.
* alloc.c (make_ratio): Allocate with new allocator.
* alloc.c (make_ratio_bg): Allocate with new allocator.
* alloc.c (make_ratio_rt): Allocate with new allocator.
* alloc.c (make_bigfloat): Allocate with new allocator.
* alloc.c (make_bigfloat_bf): Allocate with new allocator.
* alloc.c (make_compiled_function): Allocate with new allocator.
* alloc.c (Fmake_symbol): Allocate with new allocator.
* alloc.c (allocate_extent): Allocate with new allocator.
* alloc.c (allocate_event): Allocate with new allocator.
* alloc.c (make_key_data): Allocate with new allocator.
* alloc.c (make_button_data): Allocate with new allocator.
* alloc.c (make_motion_data): Allocate with new allocator.
* alloc.c (make_process_data): Allocate with new allocator.
* alloc.c (make_timeout_data): Allocate with new allocator.
* alloc.c (make_magic_data): Allocate with new allocator.
* alloc.c (make_magic_eval_data): Allocate with new allocator.
* alloc.c (make_eval_data): Allocate with new allocator.
* alloc.c (make_misc_user_data): Allocate with new allocator.
* alloc.c (Fmake_marker): Allocate with new allocator.
* alloc.c (noseeum_make_marker): Allocate with new allocator.
* alloc.c (make_uninit_string): Allocate with new allocator.
* alloc.c (resize_string): Allocate with new allocator.
* alloc.c (make_string_nocopy): Allocate with new allocator.
Garbage Collection:
* alloc.c (GC_CHECK_NOT_FREE): Remove obsolete assertions.
* alloc.c (SWEEP_FIXED_TYPE_BLOCK): Remove.
* alloc.c (SWEEP_FIXED_TYPE_BLOCK_1): Remove.
* alloc.c (sweep_conses): Remove.
* alloc.c (free_cons): Use new allocator to free.
* alloc.c (sweep_compiled_functions): Remove.
* alloc.c (sweep_floats): Remove.
* alloc.c (sweep_bignums): Remove.
* alloc.c (sweep_ratios): Remove.
* alloc.c (sweep_bigfloats): Remove.
* alloc.c (sweep_symbols): Remove.
* alloc.c (sweep_extents): Remove.
* alloc.c (sweep_events): Remove.
* alloc.c (sweep_key_data): Remove.
* alloc.c (free_key_data): Use new allocator to free.
* alloc.c (sweep_button_data): Remove.
* alloc.c (free_button_data): Use new allocator to free.
* alloc.c (sweep_motion_data): Remove.
* alloc.c (free_motion_data): Use new allocator to free.
* alloc.c (sweep_process_data): Remove.
* alloc.c (free_process_data): Use new allocator to free.
* alloc.c (sweep_timeout_data): Remove.
* alloc.c (free_timeout_data): Use new allocator to free.
* alloc.c (sweep_magic_data): Remove.
* alloc.c (free_magic_data): Use new allocator to free.
* alloc.c (sweep_magic_eval_data): Remove.
* alloc.c (free_magic_eval_data): Use new allocator to free.
* alloc.c (sweep_eval_data): Remove.
* alloc.c (free_eval_data): Use new allocator to free.
* alloc.c (sweep_misc_user_data): Remove.
* alloc.c (free_misc_user_data): Use new allocator to free.
* alloc.c (sweep_markers): Remove.
* alloc.c (free_marker): Use new allocator to free.
* alloc.c (garbage_collect_1): Remove release_breathing_space.
* alloc.c (gc_sweep): Remove all the old lcrecord and lrecord
related stuff. Sweeping now works like this: compact string
chars, finalize, sweep.
* alloc.c (common_init_alloc_early): Remove old lrecord
initializations, remove breathing_space.
* emacs.c (Fdump_emacs): Remove release_breathing_space.
* lisp.h: Remove prototype for release_breathing_space.
* lisp.h: Adjust the special cons mark makros.
Lrecord Finalizer:
* alloc.c: Add finalizer to lrecord definition.
* alloc.c (finalize_string): Add finalizer for string.
* bytecode.c: Add finalizer to lrecord definition.
* bytecode.c (finalize_compiled_function): Add finalizer for
compiled function.
* marker.c: Add finalizer to lrecord definition.
* marker.c (finalize_marker): Add finalizer for marker.
These changes build the interface to mc-alloc:
* lrecord.h (MC_ALLOC_CALL_FINALIZER): Tell mc-alloc how to
finalize lrecords.
* lrecord.h (MC_ALLOC_CALL_FINALIZER_FOR_DISKSAVE): Tell
mc-alloc how to finalize for disksave.
Unify lrecords and lcrecords:
* lisp.h (struct Lisp_String): Adjust string union hack to
new lrecord header.
* lrecord.h: Adjust comments.
* lrecord.h (struct lrecord_header): The new lrecord header
includes type, lisp-readonly, free, and uid.
* lrecord.h (set_lheader_implementation): Adjust to new
lrecord_header.
* lrecord.h (struct lrecord_implementation): The field basic_p
for indication of an old lrecord is not needed anymore, remove.
* lrecord.h (MAKE_LRECORD_IMPLEMENTATION): Remove basic_p.
* lrecord.h (MAKE_EXTERNAL_LRECORD_IMPLEMENTATION): Remove
basic_p.
* lrecord.h (copy_sized_lrecord): Remove distinction between
old lrecords and lcrecords.
* lrecord.h (copy_lrecord): Remove distinction between old
lrecords and lcrecords.
* lrecord.h (zero_sized_lrecord): Remove distinction between
old lrecords and lcrecords.
* lrecord.h (zero_lrecord): Remove distinction between old
lrecords and lcrecords.
Remove lcrecords and lcrecord lists:
* alloc.c (basic_alloc_lcrecord): Not needed anymore, remove.
* alloc.c (very_old_free_lcrecord): Not needed anymore, remove.
* alloc.c (copy_lisp_object): No more distinction between
lrecords and lcrecords.
* alloc.c (all_lcrecords): Not needed anymore, remove.
* alloc.c (make_vector_internal): Allocate as lrecord.
* alloc.c (make_bit_vector_internal): Allocate as lrecord.
* alloc.c: Completely remove `lcrecord lists'.
* alloc.c (free_description): Remove.
* alloc.c (lcrecord_list_description): Remove.
* alloc.c (mark_lcrecord_list): Remove.
* alloc.c (make_lcrecord_list): Remove.
* alloc.c (alloc_managed_lcrecord): Remove.
* alloc.c (free_managed_lcrecord): Remove.
* alloc.c (alloc_automanaged_lcrecord): Remove.
* alloc.c (free_lcrecord): Remove.
* alloc.c (lcrecord_stats): Remove.
* alloc.c (tick_lcrecord_stats): Remove.
* alloc.c (disksave_object_finalization_1): Add call to
mc_finalize_for_disksave. Remove the lcrecord way to visit all
objects.
* alloc.c (kkcc_marking): Remove XD_FLAG_FREE_LISP_OBJECT
* alloc.c (sweep_lcrecords_1): Remove.
* alloc.c (common_init_alloc_early): Remove everything related
to lcrecords, remove old lrecord initializations,
* alloc.c (init_lcrecord_lists): Not needed anymore, remove.
* alloc.c (reinit_alloc_early): Remove everything related to
lcrecords.
* alloc.c (init_alloc_once_early): Remove everything related to
lcrecords.
* buffer.c (allocate_buffer): Allocate as lrecord.
* buffer.c (nuke_all_buffer_slots): Use lrecord functions.
* buffer.c (common_init_complex_vars_of_buffer): Allocate as
lrecord.
* buffer.h (struct buffer): Add lrecord_header.
* casetab.c (allocate_case_table): Allocate as lrecord.
* casetab.h (struct Lisp_Case_Table): Add lrecord_header.
* charset.h (struct Lisp_Charset): Add lrecord_header.
* chartab.c (fill_char_table): Use lrecord functions.
* chartab.c (Fmake_char_table): Allocate as lrecord.
* chartab.c (make_char_table_entry): Allocate as lrecord.
* chartab.c (copy_char_table_entry): Allocate as lrecord.
* chartab.c (Fcopy_char_table): Allocate as lrecord.
* chartab.c (put_char_table): Use lrecord functions.
* chartab.h (struct Lisp_Char_Table_Entry): Add lrecord_header.
* chartab.h (struct Lisp_Char_Table): Add lrecord_header.
* console-impl.h (struct console): Add lrecord_header.
* console-msw-impl.h (struct Lisp_Devmode): Add lrecord_header.
* console-msw-impl.h (struct mswindows_dialog_id): Add
lrecord_header.
* console.c (allocate_console): Allocate as lrecord.
* console.c (nuke_all_console_slots): Use lrecord functions.
* console.c (common_init_complex_vars_of_console): Allocate as
lrecord.
* data.c (make_weak_list): Allocate as lrecord.
* data.c (make_weak_box): Allocate as lrecord.
* data.c (make_ephemeron): Allocate as lrecord.
* database.c (struct Lisp_Database): Add lrecord_header.
* database.c (allocate_database): Allocate as lrecord.
* device-impl.h (struct device): Add lrecord_header.
* device-msw.c (allocate_devmode): Allocate as lrecord.
* device.c (nuke_all_device_slots): Use lrecord functions.
* device.c (allocate_device): Allocate as lrecord.
* dialog-msw.c (handle_question_dialog_box): Allocate as lrecord.
* elhash.c (struct Lisp_Hash_Table): Add lrecord_header.
* elhash.c (make_general_lisp_hash_table): Allocate as lrecord.
* elhash.c (Fcopy_hash_table): Allocate as lrecord.
* event-stream.c: Lcrecord lists Vcommand_builder_free_list and
Vtimeout_free_list are no longer needed. Remove.
* event-stream.c (allocate_command_builder): Allocate as lrecord.
* event-stream.c (free_command_builder): Use lrecord functions.
* event-stream.c (event_stream_generate_wakeup): Allocate as
lrecord.
* event-stream.c (event_stream_resignal_wakeup): Use lrecord
functions.
* event-stream.c (event_stream_disable_wakeup): Use lrecord
functions.
* event-stream.c (reinit_vars_of_event_stream): Lcrecord lists
remove.
* events.h (struct Lisp_Timeout): Add lrecord_header.
* events.h (struct command_builder): Add lrecord_header.
* extents-impl.h (struct extent_auxiliary): Add lrecord_header.
* extents-impl.h (struct extent_info): Add lrecord_header.
* extents.c (allocate_extent_auxiliary): Allocate as lrecord.
* extents.c (allocate_extent_info): Allocate as lrecord.
* extents.c (copy_extent): Allocate as lrecord.
* faces.c (allocate_face): Allocate as lrecord.
* faces.h (struct Lisp_Face): Add lrecord_header.
* file-coding.c (allocate_coding_system): Allocate as lrecord.
* file-coding.c (Fcopy_coding_system): Allocate as lrecord.
* file-coding.h (struct Lisp_Coding_System): Add lrecord_header.
* fns.c (Ffillarray): Allocate as lrecord.
* frame-impl.h (struct frame): Add lrecord_header.
* frame.c (nuke_all_frame_slots): Use lrecord functions.
* frame.c (allocate_frame_core): Allocate as lrecord.
* glyphs.c (allocate_image_instance): Allocate as lrecord.
* glyphs.c (Fcolorize_image_instance): Allocate as lrecord.
* glyphs.c (allocate_glyph): Allocate as lrecord.
* glyphs.h (struct Lisp_Image_Instance): Add lrecord_header.
* glyphs.h (struct Lisp_Glyph): Add lrecord_header.
* gui.c (allocate_gui_item): Allocate as lrecord.
* gui.h (struct Lisp_Gui_Item): Add lrecord_header.
* keymap.c (struct Lisp_Keymap): Add lrecord_header.
* keymap.c (make_keymap): Allocate as lrecord.
* lisp.h (struct Lisp_Vector): Add lrecord_header.
* lisp.h (struct Lisp_Bit_Vector): Add lrecord_header.
* lisp.h (struct weak_box): Add lrecord_header.
* lisp.h (struct ephemeron): Add lrecord_header.
* lisp.h (struct weak_list): Add lrecord_header.
* lrecord.h (struct lcrecord_header): Not used, remove.
* lrecord.h (struct free_lcrecord_header): Not used, remove.
* lrecord.h (struct lcrecord_list): Not needed anymore, remove.
* lrecord.h (lcrecord_list): Not needed anymore, remove.
* lrecord.h: (enum data_description_entry_flags): Remove
XD_FLAG_FREE_LISP_OBJECT.
* lstream.c: Lrecord list Vlstream_free_list remove.
* lstream.c (Lstream_new): Allocate as lrecord.
* lstream.c (Lstream_delete): Use lrecod functions.
* lstream.c (reinit_vars_of_lstream): Vlstream_free_list
initialization remove.
* lstream.h (struct lstream): Add lrecord_header.
* emacs.c (main_1): Remove lstream initialization.
* mule-charset.c (make_charset): Allocate as lrecord.
* objects-impl.h (struct Lisp_Color_Instance): Add
lrecord_header.
* objects-impl.h (struct Lisp_Font_Instance): Add lrecord_header.
* objects.c (Fmake_color_instance): Allocate as lrecord.
* objects.c (Fmake_font_instance): Allocate as lrecord.
* objects.c (reinit_vars_of_objects): Allocate as lrecord.
* opaque.c: Lcreord list Vopaque_ptr_list remove.
* opaque.c (make_opaque): Allocate as lrecord.
* opaque.c (make_opaque_ptr): Allocate as lrecord.
* opaque.c (free_opaque_ptr): Use lrecord functions.
* opaque.c (reinit_opaque_early):
* opaque.c (init_opaque_once_early): Vopaque_ptr_list
initialization remove.
* opaque.h (Lisp_Opaque): Add lrecord_header.
* opaque.h (Lisp_Opaque_Ptr): Add lrecord_header.
* emacs.c (main_1): Remove opaque variable initialization.
* print.c (default_object_printer): Use new lrecord_header.
* print.c (print_internal): Use new lrecord_header.
* print.c (debug_p4): Use new lrecord_header.
* process.c (make_process_internal): Allocate as lrecord.
* procimpl.h (struct Lisp_Process): Add lrecord_header.
* rangetab.c (Fmake_range_table): Allocate as lrecord.
* rangetab.c (Fcopy_range_table): Allocate as lrecord.
* rangetab.h (struct Lisp_Range_Table): Add lrecord_header.
* scrollbar.c (create_scrollbar_instance): Allocate as lrecord.
* scrollbar.h (struct scrollbar_instance): Add lrecord_header.
* specifier.c (make_specifier_internal): Allocate as lrecord.
* specifier.h (struct Lisp_Specifier): Add lrecord_header.
* symbols.c:
* symbols.c (Fmake_variable_buffer_local): Allocate as lrecord.
* symbols.c (Fdontusethis_set_symbol_value_handler): Allocate
as lrecord.
* symbols.c (Fdefvaralias): Allocate as lrecord.
* symeval.h (struct symbol_value_magic): Add lrecord_header.
* toolbar.c (update_toolbar_button): Allocate as lrecord.
* toolbar.h (struct toolbar_button): Add lrecord_header.
* tooltalk.c (struct Lisp_Tooltalk_Message): Add lrecord_header.
* tooltalk.c (make_tooltalk_message): Allocate as lrecord.
* tooltalk.c (struct Lisp_Tooltalk_Pattern): Add lrecord_header.
* tooltalk.c (make_tooltalk_pattern): Allocate as lrecord.
* ui-gtk.c (allocate_ffi_data): Allocate as lrecord.
* ui-gtk.c (allocate_emacs_gtk_object_data): Allocate as lrecord.
* ui-gtk.c (allocate_emacs_gtk_boxed_data): Allocate as lrecord.
* ui-gtk.h (structs): Add lrecord_header.
* window-impl.h (struct window): Add lrecord_header.
* window-impl.h (struct window_mirror): Add lrecord_header.
* window.c (allocate_window): Allocate as lrecord.
* window.c (new_window_mirror): Allocate as lrecord.
* window.c (make_dummy_parent): Allocate as lrecord.
MEMORY_USAGE_STATS
* alloc.c (fixed_type_block_overhead): Not used anymore, remove.
* buffer.c (compute_buffer_usage): Get storage size from new
allocator.
* marker.c (compute_buffer_marker_usage): Get storage size from
new allocator.
* mule-charset.c (compute_charset_usage): Get storage size from
new allocator.
* scrollbar-gtk.c (gtk_compute_scrollbar_instance_usage): Get
storage size from new allocator.
* scrollbar-msw.c (mswindows_compute_scrollbar_instance_usage):
Get storage size from new allocator.
* scrollbar-x.c (x_compute_scrollbar_instance_usage): Get
storage size from new allocator.
* scrollbar.c (compute_scrollbar_instance_usage): Get storage
size from new allocator.
* unicode.c (compute_from_unicode_table_size_1): Get storage
size from new allocator.
* unicode.c (compute_to_unicode_table_size_1): Get storage size
from new allocator.
* window.c (compute_window_mirror_usage): Get storage size from
new allocator.
* window.c (compute_window_usage): Get storage size from new
allocator.
MC_ALLOC_TYPE_STATS:
* alloc.c (alloc_lrecord): Bump lrecord count.
* alloc.c (noseeum_alloc_lrecord): Bump lrecord count.
* alloc.c (struct lrecord_stats): Storage for counts.
* alloc.c (init_lrecord_stats): Zero statistics.
* alloc.c (inc_lrecord_stats): Increase the statistic.
* alloc.c (dec_lrecord_stats): Decrease the statistic.
* alloc.c (gc_plist_hack): Used to print the information.
* alloc.c (Fgarbage_collect): Return the collected information.
* mc-alloc.c (remove_cell): Decrease lrecord count.
* mc-alloc.h: Set flag MC_ALLOC_TYPE_STATS.
* emacs.c (main_1): Init lrecord statistics.
* lrecord.h: Add prototypes for *_lrecord_stats.
Strings:
* alloc.c (Fmake_string): Initialize ascii_begin to zero.
* alloc.c (gc_count_num_short_string_in_use): Remove.
* alloc.c (gc_count_string_total_size): Remove.
* alloc.c (gc_count_short_string_total_size): Remove.
* alloc.c (debug_string_purity): Remove.
* alloc.c (debug_string_purity_print): Remove.
* alloc.c (sweep_strings): Remove.
Remove static C-readonly Lisp objects:
* alloc.c (c_readonly): Not needed anymore, remove.
* alloc.c (GC_CHECK_LHEADER_INVARIANTS): Remove some obsolete
lheader invariants assertions.
* buffer.c (DEFVAR_BUFFER_LOCAL_1): Allocate dynamically.
* console.c (DEFVAR_CONSOLE_LOCAL_1): Allocate dynamically.
* gpmevent.c: Indirection via MC_ALLOC_Freceive_gpm_event.
* gpmevent.c (Fgpm_enable): Allocate dynamically.
* gpmevent.c (syms_of_gpmevent): Allocate dynamically.
* lisp.h (C_READONLY): Not needed anymore, remove.
* lisp.h (DEFUN): Allocate dynamically.
* lrecord.h (C_READONLY_RECORD_HEADER_P): Not needed anymore,
remove.
* lrecord.h (SET_C_READONLY_RECORD_HEADER): Not needed anymore,
remove.
* symbols.c (guts_of_unbound_marker):
* symeval.h (defsubr): Allocate dynamically.
* symeval.h (DEFSUBR_MACRO): Allocate dynamically.
* symeval.h (DEFVAR_ SYMVAL_FWD): Allocate dynamically.
* tests.c (TESTS_DEFSUBR): Allocate dynamically.
Definition of mcpro:
* lisp.h: Add mcpro prototypes.
* alloc.c (common_init_alloc_early): Add initialization for
mcpros.
* alloc.c (mcpro_description_1): New.
* alloc.c (mcpro_description): New.
* alloc.c (mcpros_description_1): New.
* alloc.c (mcpros_description): New.
* alloc.c (mcpro_one_name_description_1): New.
* alloc.c (mcpro_one_name_description): New.
* alloc.c (mcpro_names_description_1): New.
* alloc.c (mcpro_names_description): New.
* alloc.c (mcpros): New.
* alloc.c (mcpro_names): New.
* alloc.c (mcpro_1): New.
* alloc.c (mc_pro): New.
* alloc.c (garbage_collect_1): Add mcpros to root set.
Usage of mcpro:
* alloc.c (make_string_nocopy): Add string to root set.
* symbols.c (init_symbols_once_early): Add Qunbound to root set.
Changes to the Portable Dumper:
* alloc.c (FREE_OR_REALLOC_BEGIN): Since dumped objects can be
freed with the new allocator, remove assertion for !DUMPEDP.
* dumper.c: Adjust comments, increase PDUMP_HASHSIZE.
* dumper.c (pdump_make_hash): Shift address only 2 bytes, to
avoid collisions.
* dumper.c (pdump_objects_unmark): No more mark bits within
the object, remove.
* dumper.c (mc_addr_elt): New. Element data structure for mc
hash table.
* dumper.c (pdump_mc_hash): New hash table: `lookup table'.
* dumper.c (pdump_get_mc_addr): New. Lookup for hash table.
* dumper.c (pdump_get_indirect_mc_addr): New. Lookup for
convertibles.
* dumper.c (pdump_put_mc_addr): New. Putter for hash table.
* dumper.c (pdump_dump_mc_data): New. Writes the table for
relocation at load time to the dump file.
* dumper.c (pdump_scan_lisp_objects_by_alignment): New.
Visits all dumped Lisp objects.
* dumper.c (pdump_scan_non_lisp_objects_by_alignment): New.
Visits all other dumped objects.
* dumper.c (pdump_reloc_one_mc): New. Updates all pointers
of an object by using the hash table pdump_mc_hash.
* dumper.c (pdump_reloc_one): Replaced by pdump_reloc_one_mc.
* dumper.c (pdump): Change the structure of the dump file, add
the mc post dump relocation table to dump file.
* dumper.c (pdump_load_finish): Hand all dumped objects to the
new allocator and use the mc post dump relocation table for
relocating the dumped objects at dump file load time, free not
longer used data structures.
* dumper.c (pdump_load): Free the dump file.
* dumper.h: Remove pdump_objects_unmark.
* lrecord.h (DUMPEDP): Dumped objects can be freed, remove.
DUMP_IN_EXEC:
* Makefile.in.in: Linking for and with dump in executable only if
DUMP_IN_EXEC is defined.
* config.h.in: Add new flag `DUMP_IN_EXEC'
* emacs.c: Condition dump-data.h on DUMP_IN_EXEC.
* emacs.c (main_1): Flag `-si' only works if dump image is
written into executable.
Miscellanious
* lrecord.h (enum lrecord_type): Added numbers to all types,
very handy for debugging.
* xemacs.def.in.in: Add mc-alloc functions to make them visible
to the modules.
| author | crestani |
|---|---|
| date | Fri, 08 Apr 2005 23:11:35 +0000 |
| parents | cf4a5f5e87ea |
| children | 714b354cef67 |
line wrap: on
line source
/* CCL (Code Conversion Language) interpreter. Copyright (C) 1995, 1997 Electrotechnical Laboratory, JAPAN. Copyright (C) 2002 Ben Wing. Licensed to the Free Software Foundation. This file is part of XEmacs. XEmacs is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. XEmacs is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with XEmacs; see the file COPYING. If not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ /* Synched up with : FSF Emacs 21.0.90 except TranslateCharacter */ #include <config.h> #include "lisp.h" #include "buffer.h" #include "charset.h" #include "mule-ccl.h" #include "file-coding.h" Lisp_Object Qccl_error; /* This contains all code conversion map available to CCL. */ Lisp_Object Vcode_conversion_map_vector; /* Alist of fontname patterns vs corresponding CCL program. */ Lisp_Object Vfont_ccl_encoder_alist; /* This symbol is a property which associates with ccl program vector. Ex: (get 'ccl-big5-encoder 'ccl-program) returns ccl program vector. */ Lisp_Object Qccl_program; /* These symbols are properties which associate with code conversion map and their ID respectively. */ Lisp_Object Qcode_conversion_map; Lisp_Object Qcode_conversion_map_id; /* Symbols of ccl program have this property, a value of the property is an index for Vccl_program_table. */ Lisp_Object Qccl_program_idx; /* Table of registered CCL programs. Each element is a vector of NAME, CCL_PROG, and RESOLVEDP where NAME (symbol) is the name of the program, CCL_PROG (vector) is the compiled code of the program, RESOLVEDP (t or nil) is the flag to tell if symbols in CCL_PROG is already resolved to index numbers or not. */ Lisp_Object Vccl_program_table; /* CCL (Code Conversion Language) is a simple language which has operations on one input buffer, one output buffer, and 7 registers. The syntax of CCL is described in `ccl.el'. Emacs Lisp function `ccl-compile' compiles a CCL program and produces a CCL code which is a vector of integers. The structure of this vector is as follows: The 1st element: buffer-magnification, a factor for the size of output buffer compared with the size of input buffer. The 2nd element: address of CCL code to be executed when encountered with end of input stream. The 3rd and the remaining elements: CCL codes. */ /* Header of CCL compiled code */ #define CCL_HEADER_BUF_MAG 0 #define CCL_HEADER_EOF 1 #define CCL_HEADER_MAIN 2 /* CCL code is a sequence of 28-bit non-negative integers (i.e. the MSB is always 0), each contains CCL command and/or arguments in the following format: |----------------- integer (28-bit) ------------------| |------- 17-bit ------|- 3-bit --|- 3-bit --|- 5-bit -| |--constant argument--|-register-|-register-|-command-| ccccccccccccccccc RRR rrr XXXXX or |------- relative address -------|-register-|-command-| cccccccccccccccccccc rrr XXXXX or |------------- constant or other args ----------------| cccccccccccccccccccccccccccc where, `cc...c' is a non-negative integer indicating constant value (the left most `c' is always 0) or an absolute jump address, `RRR' and `rrr' are CCL register number, `XXXXX' is one of the following CCL commands. */ /* CCL commands Each comment fields shows one or more lines for command syntax and the following lines for semantics of the command. In semantics, IC stands for Instruction Counter. */ #define CCL_SetRegister 0x00 /* Set register a register value: 1:00000000000000000RRRrrrXXXXX ------------------------------ reg[rrr] = reg[RRR]; */ #define CCL_SetShortConst 0x01 /* Set register a short constant value: 1:CCCCCCCCCCCCCCCCCCCCrrrXXXXX ------------------------------ reg[rrr] = CCCCCCCCCCCCCCCCCCC; */ #define CCL_SetConst 0x02 /* Set register a constant value: 1:00000000000000000000rrrXXXXX 2:CONSTANT ------------------------------ reg[rrr] = CONSTANT; IC++; */ #define CCL_SetArray 0x03 /* Set register an element of array: 1:CCCCCCCCCCCCCCCCCRRRrrrXXXXX 2:ELEMENT[0] 3:ELEMENT[1] ... ------------------------------ if (0 <= reg[RRR] < CC..C) reg[rrr] = ELEMENT[reg[RRR]]; IC += CC..C; */ #define CCL_Jump 0x04 /* Jump: 1:A--D--D--R--E--S--S-000XXXXX ------------------------------ IC += ADDRESS; */ /* Note: If CC..C is greater than 0, the second code is omitted. */ #define CCL_JumpCond 0x05 /* Jump conditional: 1:A--D--D--R--E--S--S-rrrXXXXX ------------------------------ if (!reg[rrr]) IC += ADDRESS; */ #define CCL_WriteRegisterJump 0x06 /* Write register and jump: 1:A--D--D--R--E--S--S-rrrXXXXX ------------------------------ write (reg[rrr]); IC += ADDRESS; */ #define CCL_WriteRegisterReadJump 0x07 /* Write register, read, and jump: 1:A--D--D--R--E--S--S-rrrXXXXX 2:A--D--D--R--E--S--S-rrrYYYYY ----------------------------- write (reg[rrr]); IC++; read (reg[rrr]); IC += ADDRESS; */ /* Note: If read is suspended, the resumed execution starts from the second code (YYYYY == CCL_ReadJump). */ #define CCL_WriteConstJump 0x08 /* Write constant and jump: 1:A--D--D--R--E--S--S-000XXXXX 2:CONST ------------------------------ write (CONST); IC += ADDRESS; */ #define CCL_WriteConstReadJump 0x09 /* Write constant, read, and jump: 1:A--D--D--R--E--S--S-rrrXXXXX 2:CONST 3:A--D--D--R--E--S--S-rrrYYYYY ----------------------------- write (CONST); IC += 2; read (reg[rrr]); IC += ADDRESS; */ /* Note: If read is suspended, the resumed execution starts from the second code (YYYYY == CCL_ReadJump). */ #define CCL_WriteStringJump 0x0A /* Write string and jump: 1:A--D--D--R--E--S--S-000XXXXX 2:LENGTH 3:0000STRIN[0]STRIN[1]STRIN[2] ... ------------------------------ write_string (STRING, LENGTH); IC += ADDRESS; */ #define CCL_WriteArrayReadJump 0x0B /* Write an array element, read, and jump: 1:A--D--D--R--E--S--S-rrrXXXXX 2:LENGTH 3:ELEMENET[0] 4:ELEMENET[1] ... N:A--D--D--R--E--S--S-rrrYYYYY ------------------------------ if (0 <= reg[rrr] < LENGTH) write (ELEMENT[reg[rrr]]); IC += LENGTH + 2; (... pointing at N+1) read (reg[rrr]); IC += ADDRESS; */ /* Note: If read is suspended, the resumed execution starts from the Nth code (YYYYY == CCL_ReadJump). */ #define CCL_ReadJump 0x0C /* Read and jump: 1:A--D--D--R--E--S--S-rrrYYYYY ----------------------------- read (reg[rrr]); IC += ADDRESS; */ #define CCL_Branch 0x0D /* Jump by branch table: 1:CCCCCCCCCCCCCCCCCCCCrrrXXXXX 2:A--D--D--R--E-S-S[0]000XXXXX 3:A--D--D--R--E-S-S[1]000XXXXX ... ------------------------------ if (0 <= reg[rrr] < CC..C) IC += ADDRESS[reg[rrr]]; else IC += ADDRESS[CC..C]; */ #define CCL_ReadRegister 0x0E /* Read bytes into registers: 1:CCCCCCCCCCCCCCCCCCCCrrrXXXXX 2:CCCCCCCCCCCCCCCCCCCCrrrXXXXX ... ------------------------------ while (CCC--) read (reg[rrr]); */ #define CCL_WriteExprConst 0x0F /* write result of expression: 1:00000OPERATION000RRR000XXXXX 2:CONSTANT ------------------------------ write (reg[RRR] OPERATION CONSTANT); IC++; */ /* Note: If the Nth read is suspended, the resumed execution starts from the Nth code. */ #define CCL_ReadBranch 0x10 /* Read one byte into a register, and jump by branch table: 1:CCCCCCCCCCCCCCCCCCCCrrrXXXXX 2:A--D--D--R--E-S-S[0]000XXXXX 3:A--D--D--R--E-S-S[1]000XXXXX ... ------------------------------ read (read[rrr]); if (0 <= reg[rrr] < CC..C) IC += ADDRESS[reg[rrr]]; else IC += ADDRESS[CC..C]; */ #define CCL_WriteRegister 0x11 /* Write registers: 1:CCCCCCCCCCCCCCCCCCCrrrXXXXX 2:CCCCCCCCCCCCCCCCCCCrrrXXXXX ... ------------------------------ while (CCC--) write (reg[rrr]); ... */ /* Note: If the Nth write is suspended, the resumed execution starts from the Nth code. */ #define CCL_WriteExprRegister 0x12 /* Write result of expression 1:00000OPERATIONRrrRRR000XXXXX ------------------------------ write (reg[RRR] OPERATION reg[Rrr]); */ #define CCL_Call 0x13 /* Call the CCL program whose ID is CC..C or cc..c. 1:CCCCCCCCCCCCCCCCCCCCFFFXXXXX [2:00000000cccccccccccccccccccc] ------------------------------ if (FFF) call (cc..c) IC++; else call (CC..C) */ #define CCL_WriteConstString 0x14 /* Write a constant or a string: 1:CCCCCCCCCCCCCCCCCCCCrrrXXXXX [2:0000STRIN[0]STRIN[1]STRIN[2]] [...] ----------------------------- if (!rrr) write (CC..C) else write_string (STRING, CC..C); IC += (CC..C + 2) / 3; */ #define CCL_WriteArray 0x15 /* Write an element of array: 1:CCCCCCCCCCCCCCCCCCCCrrrXXXXX 2:ELEMENT[0] 3:ELEMENT[1] ... ------------------------------ if (0 <= reg[rrr] < CC..C) write (ELEMENT[reg[rrr]]); IC += CC..C; */ #define CCL_End 0x16 /* Terminate: 1:00000000000000000000000XXXXX ------------------------------ terminate (); */ /* The following two codes execute an assignment arithmetic/logical operation. The form of the operation is like REG OP= OPERAND. */ #define CCL_ExprSelfConst 0x17 /* REG OP= constant: 1:00000OPERATION000000rrrXXXXX 2:CONSTANT ------------------------------ reg[rrr] OPERATION= CONSTANT; */ #define CCL_ExprSelfReg 0x18 /* REG1 OP= REG2: 1:00000OPERATION000RRRrrrXXXXX ------------------------------ reg[rrr] OPERATION= reg[RRR]; */ /* The following codes execute an arithmetic/logical operation. The form of the operation is like REG_X = REG_Y OP OPERAND2. */ #define CCL_SetExprConst 0x19 /* REG_X = REG_Y OP constant: 1:00000OPERATION000RRRrrrXXXXX 2:CONSTANT ------------------------------ reg[rrr] = reg[RRR] OPERATION CONSTANT; IC++; */ #define CCL_SetExprReg 0x1A /* REG1 = REG2 OP REG3: 1:00000OPERATIONRrrRRRrrrXXXXX ------------------------------ reg[rrr] = reg[RRR] OPERATION reg[Rrr]; */ #define CCL_JumpCondExprConst 0x1B /* Jump conditional according to an operation on constant: 1:A--D--D--R--E--S--S-rrrXXXXX 2:OPERATION 3:CONSTANT ----------------------------- reg[7] = reg[rrr] OPERATION CONSTANT; if (!(reg[7])) IC += ADDRESS; else IC += 2 */ #define CCL_JumpCondExprReg 0x1C /* Jump conditional according to an operation on register: 1:A--D--D--R--E--S--S-rrrXXXXX 2:OPERATION 3:RRR ----------------------------- reg[7] = reg[rrr] OPERATION reg[RRR]; if (!reg[7]) IC += ADDRESS; else IC += 2; */ #define CCL_ReadJumpCondExprConst 0x1D /* Read and jump conditional according to an operation on constant: 1:A--D--D--R--E--S--S-rrrXXXXX 2:OPERATION 3:CONSTANT ----------------------------- read (reg[rrr]); reg[7] = reg[rrr] OPERATION CONSTANT; if (!reg[7]) IC += ADDRESS; else IC += 2; */ #define CCL_ReadJumpCondExprReg 0x1E /* Read and jump conditional according to an operation on register: 1:A--D--D--R--E--S--S-rrrXXXXX 2:OPERATION 3:RRR ----------------------------- read (reg[rrr]); reg[7] = reg[rrr] OPERATION reg[RRR]; if (!reg[7]) IC += ADDRESS; else IC += 2; */ #define CCL_Extension 0x1F /* Extended CCL code 1:ExtendedCOMMNDRrrRRRrrrXXXXX 2:ARGUMENT 3:... ------------------------------ extended_command (rrr,RRR,Rrr,ARGS) */ /* Here after, Extended CCL Instructions. Bit length of extended command is 14. Therefore, the instruction code range is 0..16384(0x3fff). */ /* Read a multibyte characeter. A code point is stored into reg[rrr]. A charset ID is stored into reg[RRR]. */ #define CCL_ReadMultibyteChar2 0x00 /* Read Multibyte Character 1:ExtendedCOMMNDRrrRRRrrrXXXXX */ /* Write a multibyte character. Write a character whose code point is reg[rrr] and the charset ID is reg[RRR]. */ #define CCL_WriteMultibyteChar2 0x01 /* Write Multibyte Character 1:ExtendedCOMMNDRrrRRRrrrXXXXX */ /* Translate a character whose code point is reg[rrr] and the charset ID is reg[RRR] by a translation table whose ID is reg[Rrr]. A translated character is set in reg[rrr] (code point) and reg[RRR] (charset ID). */ #define CCL_TranslateCharacter 0x02 /* Translate a multibyte character 1:ExtendedCOMMNDRrrRRRrrrXXXXX */ /* Translate a character whose code point is reg[rrr] and the charset ID is reg[RRR] by a translation table whose ID is ARGUMENT. A translated character is set in reg[rrr] (code point) and reg[RRR] (charset ID). */ #define CCL_TranslateCharacterConstTbl 0x03 /* Translate a multibyte character 1:ExtendedCOMMNDRrrRRRrrrXXXXX 2:ARGUMENT(Translation Table ID) */ /* Iterate looking up MAPs for reg[rrr] starting from the Nth (N = reg[RRR]) MAP until some value is found. Each MAP is a Lisp vector whose element is number, nil, t, or lambda. If the element is nil, ignore the map and proceed to the next map. If the element is t or lambda, finish without changing reg[rrr]. If the element is a number, set reg[rrr] to the number and finish. Detail of the map structure is described in the comment for CCL_MapMultiple below. */ #define CCL_IterateMultipleMap 0x10 /* Iterate multiple maps 1:ExtendedCOMMNDXXXRRRrrrXXXXX 2:NUMBER of MAPs 3:MAP-ID1 4:MAP-ID2 ... */ /* Map the code in reg[rrr] by MAPs starting from the Nth (N = reg[RRR]) map. MAPs are supplied in the succeeding CCL codes as follows: When CCL program gives this nested structure of map to this command: ((MAP-ID11 MAP-ID12 (MAP-ID121 MAP-ID122 MAP-ID123) MAP-ID13) (MAP-ID21 (MAP-ID211 (MAP-ID2111) MAP-ID212) MAP-ID22)), the compiled CCL code has this sequence: CCL_MapMultiple (CCL code of this command) 16 (total number of MAPs and SEPARATORs) -7 (1st SEPARATOR) MAP-ID11 MAP-ID12 -3 (2nd SEPARATOR) MAP-ID121 MAP-ID122 MAP-ID123 MAP-ID13 -7 (3rd SEPARATOR) MAP-ID21 -4 (4th SEPARATOR) MAP-ID211 -1 (5th SEPARATOR) MAP_ID2111 MAP-ID212 MAP-ID22 A value of each SEPARATOR follows this rule: MAP-SET := SEPARATOR [(MAP-ID | MAP-SET)]+ SEPARATOR := -(number of MAP-IDs and SEPARATORs in the MAP-SET) (*)....Nest level of MAP-SET must not be over than MAX_MAP_SET_LEVEL. When some map fails to map (i.e. it doesn't have a value for reg[rrr]), the mapping is treated as identity. The mapping is iterated for all maps in each map set (set of maps separated by SEPARATOR) except in the case that lambda is encountered. More precisely, the mapping proceeds as below: At first, VAL0 is set to reg[rrr], and it is translated by the first map to VAL1. Then, VAL1 is translated by the next map to VAL2. This mapping is iterated until the last map is used. The result of the mapping is the last value of VAL?. When the mapping process reached to the end of the map set, it moves to the next map set. If the next does not exit, the mapping process terminates, and regard the last value as a result. But, when VALm is mapped to VALn and VALn is not a number, the mapping proceeds as follows: If VALn is nil, the lastest map is ignored and the mapping of VALm proceeds to the next map. In VALn is t, VALm is reverted to reg[rrr] and the mapping of VALm proceeds to the next map. If VALn is lambda, move to the next map set like reaching to the end of the current map set. If VALn is a symbol, call the CCL program refered by it. Then, use reg[rrr] as a mapped value except for -1, -2 and -3. Such special values are regarded as nil, t, and lambda respectively. Each map is a Lisp vector of the following format (a) or (b): (a)......[STARTPOINT VAL1 VAL2 ...] (b)......[t VAL STARTPOINT ENDPOINT], where STARTPOINT is an offset to be used for indexing a map, ENDPOINT is a maximum index number of a map, VAL and VALn is a number, nil, t, or lambda. Valid index range of a map of type (a) is: STARTPOINT <= index < STARTPOINT + map_size - 1 Valid index range of a map of type (b) is: STARTPOINT <= index < ENDPOINT */ #define CCL_MapMultiple 0x11 /* Mapping by multiple code conversion maps 1:ExtendedCOMMNDXXXRRRrrrXXXXX 2:N-2 3:SEPARATOR_1 (< 0) 4:MAP-ID_1 5:MAP-ID_2 ... M:SEPARATOR_x (< 0) M+1:MAP-ID_y ... N:SEPARATOR_z (< 0) */ #define MAX_MAP_SET_LEVEL 30 typedef struct { int rest_length; int orig_val; } tr_stack; static tr_stack mapping_stack[MAX_MAP_SET_LEVEL]; static tr_stack *mapping_stack_pointer; /* If this variable is non-zero, it indicates the stack_idx of immediately called by CCL_MapMultiple. */ static int stack_idx_of_map_multiple; #define PUSH_MAPPING_STACK(restlen, orig) \ do { \ mapping_stack_pointer->rest_length = (restlen); \ mapping_stack_pointer->orig_val = (orig); \ mapping_stack_pointer++; \ } while (0) #define POP_MAPPING_STACK(restlen, orig) \ do { \ mapping_stack_pointer--; \ (restlen) = mapping_stack_pointer->rest_length; \ (orig) = mapping_stack_pointer->orig_val; \ } while (0) #define CCL_CALL_FOR_MAP_INSTRUCTION(symbol, ret_ic) \ do { \ struct ccl_program called_ccl; \ if (stack_idx >= 256 \ || (setup_ccl_program (&called_ccl, (symbol)) != 0)) \ { \ if (stack_idx > 0) \ { \ ccl_prog = ccl_prog_stack_struct[0].ccl_prog; \ ic = ccl_prog_stack_struct[0].ic; \ } \ CCL_INVALID_CMD; \ } \ ccl_prog_stack_struct[stack_idx].ccl_prog = ccl_prog; \ ccl_prog_stack_struct[stack_idx].ic = (ret_ic); \ stack_idx++; \ ccl_prog = called_ccl.prog; \ ic = CCL_HEADER_MAIN; \ /* The "if (1)" prevents warning \ "end-of loop code not reached" */ \ if (1) goto ccl_repeat; \ } while (0) #define CCL_MapSingle 0x12 /* Map by single code conversion map 1:ExtendedCOMMNDXXXRRRrrrXXXXX 2:MAP-ID ------------------------------ Map reg[rrr] by MAP-ID. If some valid mapping is found, set reg[rrr] to the result, else set reg[RRR] to -1. */ /* CCL arithmetic/logical operators. */ #define CCL_PLUS 0x00 /* X = Y + Z */ #define CCL_MINUS 0x01 /* X = Y - Z */ #define CCL_MUL 0x02 /* X = Y * Z */ #define CCL_DIV 0x03 /* X = Y / Z */ #define CCL_MOD 0x04 /* X = Y % Z */ #define CCL_AND 0x05 /* X = Y & Z */ #define CCL_OR 0x06 /* X = Y | Z */ #define CCL_XOR 0x07 /* X = Y ^ Z */ #define CCL_LSH 0x08 /* X = Y << Z */ #define CCL_RSH 0x09 /* X = Y >> Z */ #define CCL_LSH8 0x0A /* X = (Y << 8) | Z */ #define CCL_RSH8 0x0B /* X = Y >> 8, r[7] = Y & 0xFF */ #define CCL_DIVMOD 0x0C /* X = Y / Z, r[7] = Y % Z */ #define CCL_LS 0x10 /* X = (X < Y) */ #define CCL_GT 0x11 /* X = (X > Y) */ #define CCL_EQ 0x12 /* X = (X == Y) */ #define CCL_LE 0x13 /* X = (X <= Y) */ #define CCL_GE 0x14 /* X = (X >= Y) */ #define CCL_NE 0x15 /* X = (X != Y) */ #define CCL_DECODE_SJIS 0x16 /* X = HIGHER_BYTE (DE-SJIS (Y, Z)) r[7] = LOWER_BYTE (DE-SJIS (Y, Z)) */ #define CCL_ENCODE_SJIS 0x17 /* X = HIGHER_BYTE (SJIS (Y, Z)) r[7] = LOWER_BYTE (SJIS (Y, Z) */ /* Terminate CCL program successfully. */ #define CCL_SUCCESS \ do { \ ccl->status = CCL_STAT_SUCCESS; \ /* The "if (1)" inhibits the warning \ "end-of loop code not reached" */ \ if (1) goto ccl_finish; \ } while (0) /* Suspend CCL program because of reading from empty input buffer or writing to full output buffer. When this program is resumed, the same I/O command is executed. */ #define CCL_SUSPEND(stat) \ do { \ ic--; \ ccl->status = (stat); \ /* The "if (1)" inhibits the warning \ "end-of loop code not reached" */ \ if (1) goto ccl_finish; \ } while (0) /* Terminate CCL program because of invalid command. Should not occur in the normal case. */ #define CCL_INVALID_CMD \ do { \ ccl->status = CCL_STAT_INVALID_CMD; \ /* enable this to debug invalid cmd errors */ \ /* debug_break (); */ \ /* The "if (1)" inhibits the warning \ "end-of loop code not reached" */ \ if (1) goto ccl_error_handler; \ } while (0) /* Encode one character CH to multibyte form and write to the current output buffer. At encoding time, if CH is less than 256, CH is written as is. At decoding time, if CH cannot be regarded as an ASCII character, write it in multibyte form. */ #define CCL_WRITE_CHAR(ch) \ do { \ if (!destination) \ CCL_INVALID_CMD; \ if (conversion_mode == CCL_MODE_ENCODING) \ { \ if ((ch) == '\n') \ { \ if (ccl->eol_type == CCL_CODING_EOL_CRLF) \ { \ Dynarr_add (destination, '\r'); \ Dynarr_add (destination, '\n'); \ } \ else if (ccl->eol_type == CCL_CODING_EOL_CR) \ Dynarr_add (destination, '\r'); \ else \ Dynarr_add (destination, '\n'); \ } \ else if ((ch) < 0x100) \ { \ Dynarr_add (destination, ch); \ } \ else \ { \ Ibyte work[MAX_ICHAR_LEN]; \ int len; \ len = non_ascii_set_itext_ichar (work, ch); \ Dynarr_add_many (destination, work, len); \ } \ } \ else \ { \ if (!ichar_multibyte_p(ch)) \ { \ Dynarr_add (destination, ch); \ } \ else \ { \ Ibyte work[MAX_ICHAR_LEN]; \ int len; \ len = non_ascii_set_itext_ichar (work, ch); \ Dynarr_add_many (destination, work, len); \ } \ } \ } while (0) /* Write a string at ccl_prog[IC] of length LEN to the current output buffer. But this macro treat this string as a binary. Therefore, cannot handle a multibyte string except for Control-1 characters. */ #define CCL_WRITE_STRING(len) \ do { \ Ibyte work[MAX_ICHAR_LEN]; \ int ch; \ if (!destination) \ CCL_INVALID_CMD; \ else if (conversion_mode == CCL_MODE_ENCODING) \ { \ for (i = 0; i < (len); i++) \ { \ ch = ((XINT (ccl_prog[ic + (i / 3)])) \ >> ((2 - (i % 3)) * 8)) & 0xFF; \ if (ch == '\n') \ { \ if (ccl->eol_type == CCL_CODING_EOL_CRLF) \ { \ Dynarr_add (destination, '\r'); \ Dynarr_add (destination, '\n'); \ } \ else if (ccl->eol_type == CCL_CODING_EOL_CR) \ Dynarr_add (destination, '\r'); \ else \ Dynarr_add (destination, '\n'); \ } \ if (ch < 0x100) \ { \ Dynarr_add (destination, ch); \ } \ else \ { \ non_ascii_set_itext_ichar (work, ch); \ Dynarr_add_many (destination, work, len); \ } \ } \ } \ else \ { \ for (i = 0; i < (len); i++) \ { \ ch = ((XINT (ccl_prog[ic + (i / 3)])) \ >> ((2 - (i % 3)) * 8)) & 0xFF; \ if (!ichar_multibyte_p(ch)) \ { \ Dynarr_add (destination, ch); \ } \ else \ { \ non_ascii_set_itext_ichar (work, ch); \ Dynarr_add_many (destination, work, len); \ } \ } \ } \ } while (0) /* Read one byte from the current input buffer into Rth register. */ #define CCL_READ_CHAR(r) \ do { \ if (!src) \ CCL_INVALID_CMD; \ if (src < src_end) \ (r) = *src++; \ else \ { \ if (ccl->last_block) \ { \ ic = ccl->eof_ic; \ goto ccl_repeat; \ } \ else \ CCL_SUSPEND (CCL_STAT_SUSPEND_BY_SRC); \ } \ } while (0) /* Set C to the character code made from CHARSET and CODE. This is like make_ichar but check the validity of CHARSET and CODE. If they are not valid, set C to (CODE & 0xFF) because that is usually the case that CCL_ReadMultibyteChar2 read an invalid code and it set CODE to that invalid byte. */ /* On XEmacs, TranslateCharacter is not supported. Thus, this macro is not used. */ #if 0 #define CCL_MAKE_CHAR(charset, code, c) \ do { \ if ((charset) == CHARSET_ASCII) \ (c) = (code) & 0xFF; \ else if (CHARSET_DEFINED_P (charset) \ && ((code) & 0x7F) >= 32 \ && ((code) < 256 || ((code >> 7) & 0x7F) >= 32)) \ { \ int c1 = (code) & 0x7F, c2 = 0; \ \ if ((code) >= 256) \ c2 = c1, c1 = ((code) >> 7) & 0x7F; \ (c) = make_ichar (charset, c1, c2); \ } \ else \ (c) = (code) & 0xFF; \ } while (0) #endif /* Execute CCL code on SRC_BYTES length text at SOURCE. The resulting text goes to a place pointed by DESTINATION, the length of which should not exceed DST_BYTES. The bytes actually processed is returned as *CONSUMED. The return value is the length of the resulting text. As a side effect, the contents of CCL registers are updated. If SOURCE or DESTINATION is NULL, only operations on registers are permitted. */ #ifdef CCL_DEBUG #define CCL_DEBUG_BACKTRACE_LEN 256 int ccl_backtrace_table[CCL_BACKTRACE_TABLE]; int ccl_backtrace_idx; #endif struct ccl_prog_stack { Lisp_Object *ccl_prog; /* Pointer to an array of CCL code. */ int ic; /* Instruction Counter. */ }; /* For the moment, we only support depth 256 of stack. */ static struct ccl_prog_stack ccl_prog_stack_struct[256]; int ccl_driver (struct ccl_program *ccl, const unsigned char *source, unsigned_char_dynarr *destination, int src_bytes, int *consumed, int conversion_mode) { register int *reg = ccl->reg; register int ic = ccl->ic; register int code = -1; register int field1, field2; register Lisp_Object *ccl_prog = ccl->prog; const unsigned char *src = source, *src_end = src + src_bytes; int jump_address; int i, j, op; int stack_idx = ccl->stack_idx; /* Instruction counter of the current CCL code. */ int this_ic = 0; if (ic >= ccl->eof_ic) ic = CCL_HEADER_MAIN; if (ccl->buf_magnification ==0) /* We can't produce any bytes. */ destination = NULL; /* Set mapping stack pointer. */ mapping_stack_pointer = mapping_stack; #ifdef CCL_DEBUG ccl_backtrace_idx = 0; #endif for (;;) { ccl_repeat: #ifdef CCL_DEBUG ccl_backtrace_table[ccl_backtrace_idx++] = ic; if (ccl_backtrace_idx >= CCL_DEBUG_BACKTRACE_LEN) ccl_backtrace_idx = 0; ccl_backtrace_table[ccl_backtrace_idx] = 0; #endif if (!NILP (Vquit_flag) && NILP (Vinhibit_quit)) { /* We can't just signal Qquit, instead break the loop as if the whole data is processed. Don't reset Vquit_flag, it must be handled later at a safer place. */ if (consumed) src = source + src_bytes; ccl->status = CCL_STAT_QUIT; break; } this_ic = ic; code = XINT (ccl_prog[ic]); ic++; field1 = code >> 8; field2 = (code & 0xFF) >> 5; #define rrr field2 #define RRR (field1 & 7) #define Rrr ((field1 >> 3) & 7) #define ADDR field1 #define EXCMD (field1 >> 6) switch (code & 0x1F) { case CCL_SetRegister: /* 00000000000000000RRRrrrXXXXX */ reg[rrr] = reg[RRR]; break; case CCL_SetShortConst: /* CCCCCCCCCCCCCCCCCCCCrrrXXXXX */ reg[rrr] = field1; break; case CCL_SetConst: /* 00000000000000000000rrrXXXXX */ reg[rrr] = XINT (ccl_prog[ic]); ic++; break; case CCL_SetArray: /* CCCCCCCCCCCCCCCCCCCCRRRrrrXXXXX */ i = reg[RRR]; j = field1 >> 3; /* #### it's non-obvious to me that we need these casts, but the left one was already there so clearly the intention was an unsigned comparison. --ben */ if ((unsigned int) i < (unsigned int) j) reg[rrr] = XINT (ccl_prog[ic + i]); ic += j; break; case CCL_Jump: /* A--D--D--R--E--S--S-000XXXXX */ ic += ADDR; break; case CCL_JumpCond: /* A--D--D--R--E--S--S-rrrXXXXX */ if (!reg[rrr]) ic += ADDR; break; case CCL_WriteRegisterJump: /* A--D--D--R--E--S--S-rrrXXXXX */ i = reg[rrr]; CCL_WRITE_CHAR (i); ic += ADDR; break; case CCL_WriteRegisterReadJump: /* A--D--D--R--E--S--S-rrrXXXXX */ i = reg[rrr]; CCL_WRITE_CHAR (i); ic++; CCL_READ_CHAR (reg[rrr]); ic += ADDR - 1; break; case CCL_WriteConstJump: /* A--D--D--R--E--S--S-000XXXXX */ i = XINT (ccl_prog[ic]); CCL_WRITE_CHAR (i); ic += ADDR; break; case CCL_WriteConstReadJump: /* A--D--D--R--E--S--S-rrrXXXXX */ i = XINT (ccl_prog[ic]); CCL_WRITE_CHAR (i); ic++; CCL_READ_CHAR (reg[rrr]); ic += ADDR - 1; break; case CCL_WriteStringJump: /* A--D--D--R--E--S--S-000XXXXX */ j = XINT (ccl_prog[ic]); ic++; CCL_WRITE_STRING (j); ic += ADDR - 1; break; case CCL_WriteArrayReadJump: /* A--D--D--R--E--S--S-rrrXXXXX */ i = reg[rrr]; j = XINT (ccl_prog[ic]); /* #### see comment at CCL_SetArray */ if ((unsigned int) i < (unsigned int) j) { i = XINT (ccl_prog[ic + 1 + i]); CCL_WRITE_CHAR (i); } ic += j + 2; CCL_READ_CHAR (reg[rrr]); ic += ADDR - (j + 2); break; case CCL_ReadJump: /* A--D--D--R--E--S--S-rrrYYYYY */ CCL_READ_CHAR (reg[rrr]); ic += ADDR; break; case CCL_ReadBranch: /* CCCCCCCCCCCCCCCCCCCCrrrXXXXX */ CCL_READ_CHAR (reg[rrr]); /* fall through ... */ case CCL_Branch: /* CCCCCCCCCCCCCCCCCCCCrrrXXXXX */ /* #### see comment at CCL_SetArray */ if ((unsigned int) reg[rrr] < (unsigned int) field1) ic += XINT (ccl_prog[ic + reg[rrr]]); else ic += XINT (ccl_prog[ic + field1]); break; case CCL_ReadRegister: /* CCCCCCCCCCCCCCCCCCCCrrXXXXX */ while (1) { CCL_READ_CHAR (reg[rrr]); if (!field1) break; code = XINT (ccl_prog[ic]); ic++; field1 = code >> 8; field2 = (code & 0xFF) >> 5; } break; case CCL_WriteExprConst: /* 1:00000OPERATION000RRR000XXXXX */ rrr = 7; i = reg[RRR]; j = XINT (ccl_prog[ic]); op = field1 >> 6; jump_address = ic + 1; goto ccl_set_expr; case CCL_WriteRegister: /* CCCCCCCCCCCCCCCCCCCrrrXXXXX */ while (1) { i = reg[rrr]; CCL_WRITE_CHAR (i); if (!field1) break; code = XINT (ccl_prog[ic]); ic++; field1 = code >> 8; field2 = (code & 0xFF) >> 5; } break; case CCL_WriteExprRegister: /* 1:00000OPERATIONRrrRRR000XXXXX */ rrr = 7; i = reg[RRR]; j = reg[Rrr]; op = field1 >> 6; jump_address = ic; goto ccl_set_expr; case CCL_Call: /* 1:CCCCCCCCCCCCCCCCCCCCFFFXXXXX */ { Lisp_Object slot; int prog_id; /* If FFF is nonzero, the CCL program ID is in the following code. */ if (rrr) { prog_id = XINT (ccl_prog[ic]); ic++; } else prog_id = field1; if (stack_idx >= 256 || prog_id < 0 || prog_id >= XVECTOR (Vccl_program_table)->size || (slot = XVECTOR (Vccl_program_table)->contents[prog_id], !VECTORP (slot)) || !VECTORP (XVECTOR (slot)->contents[1])) { if (stack_idx > 0) { ccl_prog = ccl_prog_stack_struct[0].ccl_prog; ic = ccl_prog_stack_struct[0].ic; } CCL_INVALID_CMD; } ccl_prog_stack_struct[stack_idx].ccl_prog = ccl_prog; ccl_prog_stack_struct[stack_idx].ic = ic; stack_idx++; ccl_prog = XVECTOR (XVECTOR (slot)->contents[1])->contents; ic = CCL_HEADER_MAIN; } break; case CCL_WriteConstString: /* CCCCCCCCCCCCCCCCCCCCrrrXXXXX */ if (!rrr) CCL_WRITE_CHAR (field1); else { CCL_WRITE_STRING (field1); ic += (field1 + 2) / 3; } break; case CCL_WriteArray: /* CCCCCCCCCCCCCCCCCCCCrrrXXXXX */ i = reg[rrr]; /* #### see comment at CCL_SetArray */ if ((unsigned int) i < (unsigned int) field1) { j = XINT (ccl_prog[ic + i]); CCL_WRITE_CHAR (j); } ic += field1; break; case CCL_End: /* 0000000000000000000000XXXXX */ if (stack_idx > 0) { stack_idx--; ccl_prog = ccl_prog_stack_struct[stack_idx].ccl_prog; ic = ccl_prog_stack_struct[stack_idx].ic; break; } if (src) src = src_end; /* ccl->ic should points to this command code again to suppress further processing. */ ic--; CCL_SUCCESS; case CCL_ExprSelfConst: /* 00000OPERATION000000rrrXXXXX */ i = XINT (ccl_prog[ic]); ic++; op = field1 >> 6; goto ccl_expr_self; case CCL_ExprSelfReg: /* 00000OPERATION000RRRrrrXXXXX */ i = reg[RRR]; op = field1 >> 6; ccl_expr_self: switch (op) { case CCL_PLUS: reg[rrr] += i; break; case CCL_MINUS: reg[rrr] -= i; break; case CCL_MUL: reg[rrr] *= i; break; case CCL_DIV: reg[rrr] /= i; break; case CCL_MOD: reg[rrr] %= i; break; case CCL_AND: reg[rrr] &= i; break; case CCL_OR: reg[rrr] |= i; break; case CCL_XOR: reg[rrr] ^= i; break; case CCL_LSH: reg[rrr] <<= i; break; case CCL_RSH: reg[rrr] >>= i; break; case CCL_LSH8: reg[rrr] <<= 8; reg[rrr] |= i; break; case CCL_RSH8: reg[7] = reg[rrr] & 0xFF; reg[rrr] >>= 8; break; case CCL_DIVMOD: reg[7] = reg[rrr] % i; reg[rrr] /= i; break; case CCL_LS: reg[rrr] = reg[rrr] < i; break; case CCL_GT: reg[rrr] = reg[rrr] > i; break; case CCL_EQ: reg[rrr] = reg[rrr] == i; break; case CCL_LE: reg[rrr] = reg[rrr] <= i; break; case CCL_GE: reg[rrr] = reg[rrr] >= i; break; case CCL_NE: reg[rrr] = reg[rrr] != i; break; default: CCL_INVALID_CMD; } break; case CCL_SetExprConst: /* 00000OPERATION000RRRrrrXXXXX */ i = reg[RRR]; j = XINT (ccl_prog[ic]); op = field1 >> 6; jump_address = ++ic; goto ccl_set_expr; case CCL_SetExprReg: /* 00000OPERATIONRrrRRRrrrXXXXX */ i = reg[RRR]; j = reg[Rrr]; op = field1 >> 6; jump_address = ic; goto ccl_set_expr; case CCL_ReadJumpCondExprConst: /* A--D--D--R--E--S--S-rrrXXXXX */ CCL_READ_CHAR (reg[rrr]); case CCL_JumpCondExprConst: /* A--D--D--R--E--S--S-rrrXXXXX */ i = reg[rrr]; op = XINT (ccl_prog[ic]); jump_address = ic++ + ADDR; j = XINT (ccl_prog[ic]); ic++; rrr = 7; goto ccl_set_expr; case CCL_ReadJumpCondExprReg: /* A--D--D--R--E--S--S-rrrXXXXX */ CCL_READ_CHAR (reg[rrr]); case CCL_JumpCondExprReg: i = reg[rrr]; op = XINT (ccl_prog[ic]); jump_address = ic++ + ADDR; j = reg[XINT (ccl_prog[ic])]; ic++; rrr = 7; ccl_set_expr: switch (op) { case CCL_PLUS: reg[rrr] = i + j; break; case CCL_MINUS: reg[rrr] = i - j; break; case CCL_MUL: reg[rrr] = i * j; break; case CCL_DIV: reg[rrr] = i / j; break; case CCL_MOD: reg[rrr] = i % j; break; case CCL_AND: reg[rrr] = i & j; break; case CCL_OR: reg[rrr] = i | j; break; case CCL_XOR: reg[rrr] = i ^ j;; break; case CCL_LSH: reg[rrr] = i << j; break; case CCL_RSH: reg[rrr] = i >> j; break; case CCL_LSH8: reg[rrr] = (i << 8) | j; break; case CCL_RSH8: reg[rrr] = i >> 8; reg[7] = i & 0xFF; break; case CCL_DIVMOD: reg[rrr] = i / j; reg[7] = i % j; break; case CCL_LS: reg[rrr] = i < j; break; case CCL_GT: reg[rrr] = i > j; break; case CCL_EQ: reg[rrr] = i == j; break; case CCL_LE: reg[rrr] = i <= j; break; case CCL_GE: reg[rrr] = i >= j; break; case CCL_NE: reg[rrr] = i != j; break; case CCL_DECODE_SJIS: /* DECODE_SHIFT_JIS set MSB for internal format as opposed to Emacs. */ DECODE_SHIFT_JIS (i, j, reg[rrr], reg[7]); reg[rrr] &= 0x7F; reg[7] &= 0x7F; break; case CCL_ENCODE_SJIS: /* ENCODE_SHIFT_JIS assumes MSB of SHIFT-JIS-char is set as opposed to Emacs. */ ENCODE_SHIFT_JIS (i | 0x80, j | 0x80, reg[rrr], reg[7]); break; default: CCL_INVALID_CMD; } code &= 0x1F; if (code == CCL_WriteExprConst || code == CCL_WriteExprRegister) { i = reg[rrr]; CCL_WRITE_CHAR (i); ic = jump_address; } else if (!reg[rrr]) ic = jump_address; break; case CCL_Extension: switch (EXCMD) { case CCL_ReadMultibyteChar2: if (!src) CCL_INVALID_CMD; if (src >= src_end) { src++; goto ccl_read_multibyte_character_suspend; } i = *src++; if (i < 0x80) { /* ASCII */ reg[rrr] = i; reg[RRR] = LEADING_BYTE_ASCII; } else if (i <= MAX_LEADING_BYTE_OFFICIAL_1) { if (src >= src_end) goto ccl_read_multibyte_character_suspend; reg[RRR] = i; reg[rrr] = (*src++ & 0x7F); } else if (LEADING_BYTE_CONTROL_1 == i) { if (src >= src_end) goto ccl_read_multibyte_character_suspend; reg[RRR] = i; reg[rrr] = (*src++ - 0xA0); } else if (i <= MAX_LEADING_BYTE_OFFICIAL_2) { if ((src + 1) >= src_end) goto ccl_read_multibyte_character_suspend; reg[RRR] = i; i = (*src++ & 0x7F); reg[rrr] = ((i << 7) | (*src & 0x7F)); src++; } else if (i == PRE_LEADING_BYTE_PRIVATE_1) { if ((src + 1) >= src_end) goto ccl_read_multibyte_character_suspend; reg[RRR] = *src++; reg[rrr] = (*src++ & 0x7F); } else if (i == PRE_LEADING_BYTE_PRIVATE_2) { if ((src + 2) >= src_end) goto ccl_read_multibyte_character_suspend; reg[RRR] = *src++; i = (*src++ & 0x7F); reg[rrr] = ((i << 7) | (*src & 0x7F)); src++; } else { /* INVALID CODE. Return a single byte character. */ reg[RRR] = LEADING_BYTE_ASCII; reg[rrr] = i; } break; ccl_read_multibyte_character_suspend: src--; if (ccl->last_block) { ic = ccl->eof_ic; goto ccl_repeat; } else CCL_SUSPEND (CCL_STAT_SUSPEND_BY_SRC); break; case CCL_WriteMultibyteChar2: i = reg[RRR]; /* charset */ if (i == LEADING_BYTE_ASCII || i == LEADING_BYTE_CONTROL_1) i = reg[rrr] & 0xFF; else if (XCHARSET_DIMENSION (charset_by_leading_byte (i)) == 1) i = (((i - FIELD2_TO_OFFICIAL_LEADING_BYTE) << 7) | (reg[rrr] & 0x7F)); else if (i < MAX_LEADING_BYTE_OFFICIAL_2) i = ((i - FIELD1_TO_OFFICIAL_LEADING_BYTE) << 14) | reg[rrr]; else i = ((i - FIELD1_TO_PRIVATE_LEADING_BYTE) << 14) | reg[rrr]; CCL_WRITE_CHAR (i); break; case CCL_TranslateCharacter: #if 0 /* XEmacs does not have translate_char, and its equivalent nor. We do nothing on this operation. */ CCL_MAKE_CHAR (reg[RRR], reg[rrr], i); op = translate_char (GET_TRANSLATION_TABLE (reg[Rrr]), i, -1, 0, 0); SPLIT_CHAR (op, reg[RRR], i, j); if (j != -1) i = (i << 7) | j; reg[rrr] = i; #endif break; case CCL_TranslateCharacterConstTbl: #if 0 /* XEmacs does not have translate_char or an equivalent. We do nothing on this operation. */ op = XINT (ccl_prog[ic]); /* table */ ic++; CCL_MAKE_CHAR (reg[RRR], reg[rrr], i); op = translate_char (GET_TRANSLATION_TABLE (op), i, -1, 0, 0); SPLIT_CHAR (op, reg[RRR], i, j); if (j != -1) i = (i << 7) | j; reg[rrr] = i; #endif break; case CCL_IterateMultipleMap: { Lisp_Object map, content, attrib, value; int point, size, fin_ic; j = XINT (ccl_prog[ic++]); /* number of maps. */ fin_ic = ic + j; op = reg[rrr]; if ((j > reg[RRR]) && (j >= 0)) { ic += reg[RRR]; i = reg[RRR]; } else { reg[RRR] = -1; ic = fin_ic; break; } for (;i < j;i++) { size = XVECTOR (Vcode_conversion_map_vector)->size; point = XINT (ccl_prog[ic++]); if (point >= size) continue; map = XVECTOR (Vcode_conversion_map_vector)->contents[point]; /* Check map validity. */ if (!CONSP (map)) continue; map = XCDR (map); if (!VECTORP (map)) continue; size = XVECTOR (map)->size; if (size <= 1) continue; content = XVECTOR (map)->contents[0]; /* check map type, [STARTPOINT VAL1 VAL2 ...] or [t ELEMENT STARTPOINT ENDPOINT] */ if (INTP (content)) { point = XUINT (content); point = op - point + 1; if (!((point >= 1) && (point < size))) continue; content = XVECTOR (map)->contents[point]; } else if (EQ (content, Qt)) { if (size != 4) continue; /* #### see comment at CCL_SetArray; in this case the casts are added but the XUINT was already present */ if (((unsigned int) op >= XUINT (XVECTOR (map)->contents[2])) && ((unsigned int) op < XUINT (XVECTOR (map)->contents[3]))) content = XVECTOR (map)->contents[1]; else continue; } else continue; if (NILP (content)) continue; else if (INTP (content)) { reg[RRR] = i; reg[rrr] = XINT(content); break; } else if (EQ (content, Qt) || EQ (content, Qlambda)) { reg[RRR] = i; break; } else if (CONSP (content)) { attrib = XCAR (content); value = XCDR (content); if (!INTP (attrib) || !INTP (value)) continue; reg[RRR] = i; reg[rrr] = XUINT (value); break; } else if (SYMBOLP (content)) CCL_CALL_FOR_MAP_INSTRUCTION (content, fin_ic); else CCL_INVALID_CMD; } if (i == j) reg[RRR] = -1; ic = fin_ic; } break; case CCL_MapMultiple: { Lisp_Object map, content, attrib, value; int point, size, map_vector_size; int map_set_rest_length, fin_ic; int current_ic = this_ic; /* inhibit recursive call on MapMultiple. */ if (stack_idx_of_map_multiple > 0) { if (stack_idx_of_map_multiple <= stack_idx) { stack_idx_of_map_multiple = 0; mapping_stack_pointer = mapping_stack; CCL_INVALID_CMD; } } else mapping_stack_pointer = mapping_stack; stack_idx_of_map_multiple = 0; map_set_rest_length = XINT (ccl_prog[ic++]); /* number of maps and separators. */ fin_ic = ic + map_set_rest_length; op = reg[rrr]; if ((map_set_rest_length > reg[RRR]) && (reg[RRR] >= 0)) { ic += reg[RRR]; i = reg[RRR]; map_set_rest_length -= i; } else { ic = fin_ic; reg[RRR] = -1; mapping_stack_pointer = mapping_stack; break; } if (mapping_stack_pointer <= (mapping_stack + 1)) { /* Set up initial state. */ mapping_stack_pointer = mapping_stack; PUSH_MAPPING_STACK (0, op); reg[RRR] = -1; } else { /* Recover after calling other ccl program. */ int orig_op; POP_MAPPING_STACK (map_set_rest_length, orig_op); POP_MAPPING_STACK (map_set_rest_length, reg[rrr]); switch (op) { case -1: /* Regard it as Qnil. */ op = orig_op; i++; ic++; map_set_rest_length--; break; case -2: /* Regard it as Qt. */ op = reg[rrr]; i++; ic++; map_set_rest_length--; break; case -3: /* Regard it as Qlambda. */ op = orig_op; i += map_set_rest_length; ic += map_set_rest_length; map_set_rest_length = 0; break; default: /* Regard it as normal mapping. */ i += map_set_rest_length; ic += map_set_rest_length; POP_MAPPING_STACK (map_set_rest_length, reg[rrr]); break; } } map_vector_size = XVECTOR (Vcode_conversion_map_vector)->size; do { for (;map_set_rest_length > 0;i++, ic++, map_set_rest_length--) { point = XINT(ccl_prog[ic]); if (point < 0) { /* +1 is for including separator. */ point = -point + 1; if (mapping_stack_pointer >= mapping_stack + countof (mapping_stack)) CCL_INVALID_CMD; PUSH_MAPPING_STACK (map_set_rest_length - point, reg[rrr]); map_set_rest_length = point; reg[rrr] = op; continue; } if (point >= map_vector_size) continue; map = (XVECTOR (Vcode_conversion_map_vector) ->contents[point]); /* Check map validity. */ if (!CONSP (map)) continue; map = XCDR (map); if (!VECTORP (map)) continue; size = XVECTOR (map)->size; if (size <= 1) continue; content = XVECTOR (map)->contents[0]; /* check map type, [STARTPOINT VAL1 VAL2 ...] or [t ELEMENT STARTPOINT ENDPOINT] */ if (INTP (content)) { point = XUINT (content); point = op - point + 1; if (!((point >= 1) && (point < size))) continue; content = XVECTOR (map)->contents[point]; } else if (EQ (content, Qt)) { if (size != 4) continue; /* #### see comment at CCL_SetArray; in this case the casts are added but the XUINT was already present */ if (((unsigned int) op >= XUINT (XVECTOR (map)->contents[2])) && ((unsigned int) op < XUINT (XVECTOR (map)->contents[3]))) content = XVECTOR (map)->contents[1]; else continue; } else continue; if (NILP (content)) continue; reg[RRR] = i; if (INTP (content)) { op = XINT (content); i += map_set_rest_length - 1; ic += map_set_rest_length - 1; POP_MAPPING_STACK (map_set_rest_length, reg[rrr]); map_set_rest_length++; } else if (CONSP (content)) { attrib = XCAR (content); value = XCDR (content); if (!INTP (attrib) || !INTP (value)) continue; op = XUINT (value); i += map_set_rest_length - 1; ic += map_set_rest_length - 1; POP_MAPPING_STACK (map_set_rest_length, reg[rrr]); map_set_rest_length++; } else if (EQ (content, Qt)) { op = reg[rrr]; } else if (EQ (content, Qlambda)) { i += map_set_rest_length; ic += map_set_rest_length; break; } else if (SYMBOLP (content)) { if (mapping_stack_pointer >= mapping_stack + countof (mapping_stack)) CCL_INVALID_CMD; PUSH_MAPPING_STACK (map_set_rest_length, reg[rrr]); PUSH_MAPPING_STACK (map_set_rest_length, op); stack_idx_of_map_multiple = stack_idx + 1; CCL_CALL_FOR_MAP_INSTRUCTION (content, current_ic); } else CCL_INVALID_CMD; } if (mapping_stack_pointer <= (mapping_stack + 1)) break; POP_MAPPING_STACK (map_set_rest_length, reg[rrr]); i += map_set_rest_length; ic += map_set_rest_length; POP_MAPPING_STACK (map_set_rest_length, reg[rrr]); } while (1); ic = fin_ic; } reg[rrr] = op; break; case CCL_MapSingle: { Lisp_Object map, attrib, value, content; int size, point; j = XINT (ccl_prog[ic++]); /* map_id */ op = reg[rrr]; if (j >= XVECTOR (Vcode_conversion_map_vector)->size) { reg[RRR] = -1; break; } map = XVECTOR (Vcode_conversion_map_vector)->contents[j]; if (!CONSP (map)) { reg[RRR] = -1; break; } map = XCDR (map); if (!VECTORP (map)) { reg[RRR] = -1; break; } size = XVECTOR (map)->size; point = XUINT (XVECTOR (map)->contents[0]); point = op - point + 1; reg[RRR] = 0; if ((size <= 1) || (!((point >= 1) && (point < size)))) reg[RRR] = -1; else { reg[RRR] = 0; content = XVECTOR (map)->contents[point]; if (NILP (content)) reg[RRR] = -1; else if (INTP (content)) reg[rrr] = XINT (content); else if (EQ (content, Qt)); else if (CONSP (content)) { attrib = XCAR (content); value = XCDR (content); if (!INTP (attrib) || !INTP (value)) continue; reg[rrr] = XUINT(value); break; } else if (SYMBOLP (content)) CCL_CALL_FOR_MAP_INSTRUCTION (content, ic); else reg[RRR] = -1; } } break; default: CCL_INVALID_CMD; } break; default: CCL_INVALID_CMD; } } ccl_error_handler: if (destination) { /* We can insert an error message only if DESTINATION is specified and we still have a room to store the message there. */ char msg[256]; switch (ccl->status) { case CCL_STAT_INVALID_CMD: sprintf(msg, "\nCCL: Invalid command %x (ccl_code = %x) at %d.", code & 0x1F, code, this_ic); #ifdef CCL_DEBUG { int i = ccl_backtrace_idx - 1; int j; Dynarr_add_many (destination, (unsigned char *) msg, strlen (msg)); for (j = 0; j < CCL_DEBUG_BACKTRACE_LEN; j++, i--) { if (i < 0) i = CCL_DEBUG_BACKTRACE_LEN - 1; if (ccl_backtrace_table[i] == 0) break; sprintf(msg, " %d", ccl_backtrace_table[i]); Dynarr_add_many (destination, (unsigned char *) msg, strlen (msg)); } goto ccl_finish; } #endif break; case CCL_STAT_QUIT: sprintf(msg, "\nCCL: Exited."); break; default: sprintf(msg, "\nCCL: Unknown error type (%d).", ccl->status); } Dynarr_add_many (destination, (unsigned char *) msg, strlen (msg)); } ccl_finish: ccl->ic = ic; ccl->stack_idx = stack_idx; ccl->prog = ccl_prog; if (consumed) *consumed = src - source; if (!destination) return 0; return Dynarr_length (destination); } /* Resolve symbols in the specified CCL code (Lisp vector). This function converts symbols of code conversion maps and character translation tables embedded in the CCL code into their ID numbers. The return value is a vector (CCL itself or a new vector in which all symbols are resolved), Qt if resolving of some symbol failed, or nil if CCL contains invalid data. */ static Lisp_Object resolve_symbol_ccl_program (Lisp_Object ccl) { int i, veclen, unresolved = 0; Lisp_Object result, contents, val; result = ccl; veclen = XVECTOR (result)->size; for (i = 0; i < veclen; i++) { contents = XVECTOR (result)->contents[i]; if (INTP (contents)) continue; else if (CONSP (contents) && SYMBOLP (XCAR (contents)) && SYMBOLP (XCDR (contents))) { /* This is the new style for embedding symbols. The form is (SYMBOL . PROPERTY). (get SYMBOL PROPERTY) should give an index number. */ if (EQ (result, ccl)) result = Fcopy_sequence (ccl); val = Fget (XCAR (contents), XCDR (contents), Qnil); if (NATNUMP (val)) XVECTOR (result)->contents[i] = val; else unresolved = 1; continue; } else if (SYMBOLP (contents)) { /* This is the old style for embedding symbols. This style may lead to a bug if, for instance, a translation table and a code conversion map have the same name. */ if (EQ (result, ccl)) result = Fcopy_sequence (ccl); val = Fget (contents, Qcode_conversion_map_id, Qnil); if (NATNUMP (val)) XVECTOR (result)->contents[i] = val; else { val = Fget (contents, Qccl_program_idx, Qnil); if (NATNUMP (val)) XVECTOR (result)->contents[i] = val; else unresolved = 1; } continue; } return Qnil; } return (unresolved ? Qt : result); } /* Return the compiled code (vector) of CCL program CCL_PROG. CCL_PROG is a name (symbol) of the program or already compiled code. If necessary, resolve symbols in the compiled code to index numbers. If we failed to get the compiled code or to resolve symbols, return Qnil. */ static Lisp_Object ccl_get_compiled_code (Lisp_Object ccl_prog) { Lisp_Object val, slot; if (VECTORP (ccl_prog)) { val = resolve_symbol_ccl_program (ccl_prog); return (VECTORP (val) ? val : Qnil); } if (!SYMBOLP (ccl_prog)) return Qnil; val = Fget (ccl_prog, Qccl_program_idx, Qnil); if (! NATNUMP (val) || XINT (val) >= XVECTOR_LENGTH (Vccl_program_table)) return Qnil; slot = XVECTOR_DATA (Vccl_program_table)[XINT (val)]; if (! VECTORP (slot) || XVECTOR (slot)->size != 3 || ! VECTORP (XVECTOR_DATA (slot)[1])) return Qnil; if (NILP (XVECTOR_DATA (slot)[2])) { val = resolve_symbol_ccl_program (XVECTOR_DATA (slot)[1]); if (! VECTORP (val)) return Qnil; XVECTOR_DATA (slot)[1] = val; XVECTOR_DATA (slot)[2] = Qt; } return XVECTOR_DATA (slot)[1]; } /* Setup fields of the structure pointed by CCL appropriately for the execution of CCL program CCL_PROG. CCL_PROG is the name (symbol) of the CCL program or the already compiled code (vector). Return 0 if we succeed this setup, else return -1. If CCL_PROG is nil, we just reset the structure pointed by CCL. */ int setup_ccl_program (struct ccl_program *ccl, Lisp_Object ccl_prog) { xzero (*ccl); /* XEmacs change */ if (! NILP (ccl_prog)) { ccl_prog = ccl_get_compiled_code (ccl_prog); if (! VECTORP (ccl_prog)) return -1; ccl->size = XVECTOR_LENGTH (ccl_prog); ccl->prog = XVECTOR_DATA (ccl_prog); ccl->eof_ic = XINT (XVECTOR_DATA (ccl_prog)[CCL_HEADER_EOF]); ccl->buf_magnification = XINT (XVECTOR_DATA (ccl_prog)[CCL_HEADER_BUF_MAG]); } ccl->ic = CCL_HEADER_MAIN; ccl->eol_type = CCL_CODING_EOL_LF; return 0; } #ifdef emacs DEFUN ("ccl-program-p", Fccl_program_p, 1, 1, 0, /* Return t if OBJECT is a CCL program name or a compiled CCL program code. See the documentation of `define-ccl-program' for the detail of CCL program. */ (object)) { Lisp_Object val; if (VECTORP (object)) { val = resolve_symbol_ccl_program (object); return (VECTORP (val) ? Qt : Qnil); } if (!SYMBOLP (object)) return Qnil; val = Fget (object, Qccl_program_idx, Qnil); return ((! NATNUMP (val) || XINT (val) >= XVECTOR_LENGTH (Vccl_program_table)) ? Qnil : Qt); } DEFUN ("ccl-execute", Fccl_execute, 2, 2, 0, /* Execute CCL-PROGRAM with registers initialized by REGISTERS. CCL-PROGRAM is a CCL program name (symbol) or a compiled code generated by `ccl-compile' (for backward compatibility, in this case, the overhead of the execution is bigger than the former case). No I/O commands should appear in CCL-PROGRAM. REGISTERS is a vector of [R0 R1 ... R7] where RN is an initial value of Nth register. As side effect, each element of REGISTERS holds the value of corresponding register after the execution. See the documentation of `define-ccl-program' for the detail of CCL program. */ (ccl_prog, reg)) { struct ccl_program ccl; int i; if (setup_ccl_program (&ccl, ccl_prog) < 0) syntax_error ("Invalid CCL program", Qunbound); CHECK_VECTOR (reg); if (XVECTOR_LENGTH (reg) != 8) syntax_error ("Length of vector REGISTERS is not 8", Qunbound); for (i = 0; i < 8; i++) ccl.reg[i] = (INTP (XVECTOR_DATA (reg)[i]) ? XINT (XVECTOR_DATA (reg)[i]) : 0); ccl_driver (&ccl, (const unsigned char *)0, (unsigned_char_dynarr *)0, 0, (int *)0, CCL_MODE_ENCODING); QUIT; if (ccl.status != CCL_STAT_SUCCESS) signal_error (Qccl_error, "Error in CCL program at code numbered ...", make_int (ccl.ic)); for (i = 0; i < 8; i++) XVECTOR (reg)->contents[i] = make_int (ccl.reg[i]); return Qnil; } DEFUN ("ccl-execute-on-string", Fccl_execute_on_string, 3, 4, 0, /* Execute CCL-PROGRAM with initial STATUS on STRING. CCL-PROGRAM is a symbol registered by register-ccl-program, or a compiled code generated by `ccl-compile' (for backward compatibility, in this case, the execution is slower). Read buffer is set to STRING, and write buffer is allocated automatically. STATUS is a vector of [R0 R1 ... R7 IC], where R0..R7 are initial values of corresponding registers, IC is the instruction counter specifying from where to start the program. If R0..R7 are nil, they are initialized to 0. If IC is nil, it is initialized to head of the CCL program. If optional 4th arg CONTINUE is non-nil, keep IC on read operation when read buffer is exhausted, else, IC is always set to the end of CCL-PROGRAM on exit. It returns the contents of write buffer as a string, and as side effect, STATUS is updated. See the documentation of `define-ccl-program' for the detail of CCL program. */ (ccl_prog, status, string, continue_)) { Lisp_Object val; struct ccl_program ccl; int i, produced; unsigned_char_dynarr *outbuf; struct gcpro gcpro1, gcpro2; if (setup_ccl_program (&ccl, ccl_prog) < 0) syntax_error ("Invalid CCL program", Qunbound); CHECK_VECTOR (status); if (XVECTOR (status)->size != 9) syntax_error ("Length of vector STATUS is not 9", Qunbound); CHECK_STRING (string); GCPRO2 (status, string); for (i = 0; i < 8; i++) { if (NILP (XVECTOR_DATA (status)[i])) XVECTOR_DATA (status)[i] = make_int (0); if (INTP (XVECTOR_DATA (status)[i])) ccl.reg[i] = XINT (XVECTOR_DATA (status)[i]); } if (INTP (XVECTOR (status)->contents[i])) { i = XINT (XVECTOR_DATA (status)[8]); if (ccl.ic < i && i < ccl.size) ccl.ic = i; } outbuf = Dynarr_new (unsigned_char); ccl.last_block = NILP (continue_); produced = ccl_driver (&ccl, XSTRING_DATA (string), outbuf, XSTRING_LENGTH (string), (int *) 0, CCL_MODE_DECODING); for (i = 0; i < 8; i++) XVECTOR_DATA (status)[i] = make_int (ccl.reg[i]); XVECTOR_DATA (status)[8] = make_int (ccl.ic); UNGCPRO; val = make_string (Dynarr_atp (outbuf, 0), produced); Dynarr_free (outbuf); QUIT; if (ccl.status == CCL_STAT_SUSPEND_BY_DST) signal_error (Qccl_error, "Output buffer for the CCL programs overflow", Qunbound); if (ccl.status != CCL_STAT_SUCCESS && ccl.status != CCL_STAT_SUSPEND_BY_SRC) signal_error (Qccl_error, "Error in CCL program at code numbered...", make_int (ccl.ic)); return val; } DEFUN ("register-ccl-program", Fregister_ccl_program, 2, 2, 0, /* Register CCL program CCL-PROG as NAME in `ccl-program-table'. CCL-PROG should be a compiled CCL program (vector), or nil. If it is nil, just reserve NAME as a CCL program name. Return index number of the registered CCL program. */ (name, ccl_prog)) { int len = XVECTOR_LENGTH (Vccl_program_table); int idx; Lisp_Object resolved; CHECK_SYMBOL (name); resolved = Qnil; if (!NILP (ccl_prog)) { CHECK_VECTOR (ccl_prog); resolved = resolve_symbol_ccl_program (ccl_prog); if (! NILP (resolved)) { ccl_prog = resolved; resolved = Qt; } } for (idx = 0; idx < len; idx++) { Lisp_Object slot; slot = XVECTOR_DATA (Vccl_program_table)[idx]; if (!VECTORP (slot)) /* This is the first unused slot. Register NAME here. */ break; if (EQ (name, XVECTOR_DATA (slot)[0])) { /* Update this slot. */ XVECTOR_DATA (slot)[1] = ccl_prog; XVECTOR_DATA (slot)[2] = resolved; return make_int (idx); } } if (idx == len) { /* Extend the table. */ Lisp_Object new_table; int j; new_table = Fmake_vector (make_int (len * 2), Qnil); for (j = 0; j < len; j++) XVECTOR_DATA (new_table)[j] = XVECTOR_DATA (Vccl_program_table)[j]; Vccl_program_table = new_table; } { Lisp_Object elt; elt = Fmake_vector (make_int (3), Qnil); XVECTOR_DATA (elt)[0] = name; XVECTOR_DATA (elt)[1] = ccl_prog; XVECTOR_DATA (elt)[2] = resolved; XVECTOR_DATA (Vccl_program_table)[idx] = elt; } Fput (name, Qccl_program_idx, make_int (idx)); return make_int (idx); } /* Register code conversion map. A code conversion map consists of numbers, Qt, Qnil, and Qlambda. The first element is start code point. The rest elements are mapped numbers. Symbol t means to map to an original number before mapping. Symbol nil means that the corresponding element is empty. Symbol lambda means to terminate mapping here. */ DEFUN ("register-code-conversion-map", Fregister_code_conversion_map, 2, 2, 0, /* Register SYMBOL as code conversion map MAP. Return index number of the registered map. */ (symbol, map)) { int len = XVECTOR_LENGTH (Vcode_conversion_map_vector); int i; Lisp_Object idx; CHECK_SYMBOL (symbol); CHECK_VECTOR (map); for (i = 0; i < len; i++) { Lisp_Object slot = XVECTOR_DATA (Vcode_conversion_map_vector)[i]; if (!CONSP (slot)) break; if (EQ (symbol, XCAR (slot))) { idx = make_int (i); XCDR (slot) = map; Fput (symbol, Qcode_conversion_map, map); Fput (symbol, Qcode_conversion_map_id, idx); return idx; } } if (i == len) { Lisp_Object new_vector = Fmake_vector (make_int (len * 2), Qnil); int j; for (j = 0; j < len; j++) XVECTOR_DATA (new_vector)[j] = XVECTOR_DATA (Vcode_conversion_map_vector)[j]; Vcode_conversion_map_vector = new_vector; } idx = make_int (i); Fput (symbol, Qcode_conversion_map, map); Fput (symbol, Qcode_conversion_map_id, idx); XVECTOR_DATA (Vcode_conversion_map_vector)[i] = Fcons (symbol, map); return idx; } void syms_of_mule_ccl (void) { DEFERROR_STANDARD (Qccl_error, Qconversion_error); DEFSUBR (Fccl_program_p); DEFSUBR (Fccl_execute); DEFSUBR (Fccl_execute_on_string); DEFSUBR (Fregister_ccl_program); DEFSUBR (Fregister_code_conversion_map); } void vars_of_mule_ccl (void) { staticpro (&Vccl_program_table); Vccl_program_table = Fmake_vector (make_int (32), Qnil); DEFSYMBOL (Qccl_program); DEFSYMBOL (Qccl_program_idx); DEFSYMBOL (Qcode_conversion_map); DEFSYMBOL (Qcode_conversion_map_id); DEFVAR_LISP ("code-conversion-map-vector", &Vcode_conversion_map_vector /* Vector of code conversion maps. */ ); Vcode_conversion_map_vector = Fmake_vector (make_int (16), Qnil); DEFVAR_LISP ("font-ccl-encoder-alist", &Vfont_ccl_encoder_alist /* Alist of fontname patterns vs corresponding CCL program. Each element looks like (REGEXP . CCL-CODE), where CCL-CODE is a compiled CCL program. When a font whose name matches REGEXP is used for displaying a character, CCL-CODE is executed to calculate the code point in the font from the charset number and position code(s) of the character which are set in CCL registers R0, R1, and R2 before the execution. The code point in the font is set in CCL registers R1 and R2 when the execution terminated. If the font is single-byte font, the register R2 is not used. */ ); Vfont_ccl_encoder_alist = Qnil; } #endif /* emacs */
