(SUBSTITUTE): New macro.
(seen_list): New variable.
(substitute_object_in_subtree): New function.
(substitute_object_recurse): New function.
(substitute_in_interval): New function.
(prev_saved_doc_string*): New variables.
(Fload): Initalize prev_saved_doc_string.
(read1): Copy saved_doc_string to prev_saved_doc_string
before storing a new string in saved_doc_string.
(read_list): Look in prev_saved_doc_string as well as
in saved_doc_string.
(read1): Swap saved_doc_string_length and
prev_saved_doc_string_length.
(read_list): Negate docstring position if negative before checking
against saved_doc_string_length et al.
(read_vector): Add bytecodeflag parameter, which is
nonzero when reading a bytecode object. If
`load-force-doc-strings' is t when reading a lazily-loaded
bytecode vector, the loaded docstring must be treated as unibyte
and passed to Fread to obtain the actual bytecode string and
constants vector.
(read1): Add extra parameter to read_vector calls.
(read1): Enable saving of doc strings on WINDOWSNT.
(read_list): Call get_doc_string instead of read_doc_string, when
forced to load doc strings.
to read a character byte by byte, if its byte sequence
does not really match the character code.
(unreadchar): Handle readchar_backlog.
(readevalloop, Fread): Initialize readchar_backlog to -1.
platforms. Avoid calling fdopen only for WINDOWSNT, use fdopen
for the the MSDOS version.
(read1): Avoid recording doc strings positions only for WINDOWSNT,
not for MSDOS.
(read_from_string_index): Now counts characters.
(readchar, unreadchar, Fread_from_string): Changed accordingly.
(readchar): Read a multibyte char all at once
from a buffer, marker or string.
(unreadchar): Unread a multibyte char all at once.
(read1): Properly handle non-escaped multibyte chars.
They force a string to be multibyte.
When reading direct from a file, any multibyte sequence means
a multibyte string.
Insist on MAX_LENGTH_OF_MULTI_BYTE_FORM bytes when checking
for read_buffer full; this way need not check specially for multibyte.
Use it to bind load_convert_to_unibyte.
(Feval_buffer): New arg UNIBYTE. Pass it to readevalloop.
(readevalloop_1): New unwind-protect subroutine.
(load_convert_to_unibyte): New variable.
(syms_of_lread): Set up Lisp var.
(read1): If load_convert_to_unibyte, convert strings to unibyte.
Otherwise buffer_defaults->enable_multibyte_characters decides.
Only set force_singlebyte when a \-sequence is in the relevant
range.
(read_escape): New arg STRINGP. `\ ' is ignored only when STRINGP
(normally, when reading a string). Calls changed.
(oblookup): Take args SIZE and SIZE_BYTE. Callers changed.
(init_obarray, read1, make_symbol):
Pass new arg to make_pure_string.
(Fintern, oblookup, Fintern_soft, intern): Handle size_byte.
(dir_warning): Pass new arg to message_dolog.
(read1): PCH is now int *.
Declare ch as int.
(read0): Declare c as int.
(read_list): Declare ch as int.
(read0): Use Fmake_string not make_string.
(read1): When reading a string, maybe use make_unibyte_string.
(Fread_from_string): Convert string indices to/from bytes.
(readchar): When fetching from buffer or marker,
use readchar_backlog to fetch bytes from a character.
(unreadchar): Increment readchar_backlog.
(readevalloop, Fread): Init readchar_backlog.
