# HG changeset patch # User Kenichi Handa # Date 966389840 0 # Node ID 2db6e42a6ba33d048055bdfa22b7ef503b00b8a6 # Parent b0b1acff966b51afc42c5b2a7b801e617dee6021 (MINIMUM_CONVERSION_BUFFER_SIZE): Macro deleted. (conversion_buffer, conversion_buffer_size): Variables deleted. (get_conversion_buffer): Function deleted. (struct conversion_buffer): New structure. (MAX_ALLOCA): New macro. (allocate_conversion_buffer): New macro. (extend_conversion_buffer, free_conversion_buffer): New functions. (ccl_coding_driver): Set coding->result. (decode_coding): Set coding->result to CODING_FINISH_NORMAL if this is the last block of source. (encode_coding): Likewise. Handle the source block as the last one only when the whole source text is consumed. (decode_coding_string): Handle the case that the output buffer is too small to decode the whole source text. Use allocate_conversion_buffer, extend_conversion_buffer and free_conversion_buffer, not get_conversion_buffer. (encode_coding_string): Likewise. (init_coding): Function deleted. (init_coding_once): Delete code to initialize conversion_buffer_size. diff -r b0b1acff966b -r 2db6e42a6ba3 src/coding.c --- a/src/coding.c Wed Aug 16 01:36:53 2000 +0000 +++ b/src/coding.c Wed Aug 16 01:37:20 2000 +0000 @@ -3870,33 +3870,60 @@ return (src_bytes * magnification + CONVERSION_BUFFER_EXTRA_ROOM); } -#ifndef MINIMUM_CONVERSION_BUFFER_SIZE -#define MINIMUM_CONVERSION_BUFFER_SIZE 1024 -#endif - -char *conversion_buffer; -int conversion_buffer_size; - -/* Return a pointer to a SIZE bytes of buffer to be used for encoding - or decoding. Sufficient memory is allocated automatically. If we - run out of memory, return NULL. */ - -char * -get_conversion_buffer (size) - int size; +/* Working buffer for code conversion. */ +struct conversion_buffer { - if (size > conversion_buffer_size) + int size; /* size of data. */ + int on_stack; /* 1 if allocated by alloca. */ + unsigned char *data; +}; + +/* Don't use alloca for allocating memory space larger than this, lest + we overflow their stack. */ +#define MAX_ALLOCA 16*1024 + +/* Allocate LEN bytes of memory for BUF (struct conversion_buffer). */ +#define allocate_conversion_buffer(buf, len) \ + do { \ + if (len < MAX_ALLOCA) \ + { \ + buf.data = (unsigned char *) alloca (len); \ + buf.on_stack = 1; \ + } \ + else \ + { \ + buf.data = (unsigned char *) xmalloc (len); \ + buf.on_stack = 0; \ + } \ + buf.size = len; \ + } while (0) + +/* Double the allocated memory for *BUF. */ +static void +extend_conversion_buffer (buf) + struct conversion_buffer *buf; +{ + if (buf->on_stack) { - char *buf; - int real_size = conversion_buffer_size * 2; - - while (real_size < size) real_size *= 2; - buf = (char *) xmalloc (real_size); - xfree (conversion_buffer); - conversion_buffer = buf; - conversion_buffer_size = real_size; + unsigned char *save = buf->data; + buf->data = (unsigned char *) xmalloc (buf->size * 2); + bcopy (save, buf->data, buf->size); + buf->on_stack = 0; + } + else + { + buf->data = (unsigned char *) xrealloc (buf->data, buf->size * 2); } - return conversion_buffer; + buf->size *= 2; +} + +/* Free the allocated memory for BUF if it is not on stack. */ +static void +free_conversion_buffer (buf) + struct conversion_buffer *buf; +{ + if (!buf->on_stack) + xfree (buf->data); } int @@ -3929,20 +3956,20 @@ switch (ccl->status) { case CCL_STAT_SUSPEND_BY_SRC: - result = CODING_FINISH_INSUFFICIENT_SRC; + coding->result = CODING_FINISH_INSUFFICIENT_SRC; break; case CCL_STAT_SUSPEND_BY_DST: - result = CODING_FINISH_INSUFFICIENT_DST; + coding->result = CODING_FINISH_INSUFFICIENT_DST; break; case CCL_STAT_QUIT: case CCL_STAT_INVALID_CMD: - result = CODING_FINISH_INTERRUPT; + coding->result = CODING_FINISH_INTERRUPT; break; default: - result = CODING_FINISH_NORMAL; + coding->result = CODING_FINISH_NORMAL; break; } - return result; + return coding->result; } /* Decode EOL format of the text at PTR of BYTES length destructively @@ -4170,6 +4197,7 @@ } coding->consumed = coding->consumed_char = src - source; coding->produced = dst - destination; + coding->result = CODING_FINISH_NORMAL; } if (!coding->dst_multibyte) @@ -4231,7 +4259,8 @@ && coding->consumed == src_bytes) coding->result = CODING_FINISH_NORMAL; - if (coding->mode & CODING_MODE_LAST_BLOCK) + if (coding->mode & CODING_MODE_LAST_BLOCK + && coding->result == CODING_FINISH_INSUFFICIENT_SRC) { unsigned char *src = source + coding->consumed; unsigned char *src_end = src + src_bytes; @@ -4252,6 +4281,7 @@ coding->consumed = src_bytes; } coding->produced = coding->produced_char = dst - destination; + coding->result = CODING_FINISH_NORMAL; } return coding->result; @@ -5197,12 +5227,15 @@ int nocopy; { int len; - char *buf; + struct conversion_buffer buf; int from, to, to_byte; struct gcpro gcpro1; Lisp_Object saved_coding_symbol; int result; int require_decoding; + int shrinked_bytes = 0; + Lisp_Object newstr; + int consumed, produced, produced_char; from = 0; to = XSTRING (str)->size; @@ -5247,12 +5280,11 @@ /* Try to skip the heading and tailing ASCIIs. */ if (require_decoding && coding->type != coding_type_ccl) { - int from_orig = from; - SHRINK_CONVERSION_REGION (&from, &to_byte, coding, XSTRING (str)->data, 0); if (from == to_byte) require_decoding = 0; + shrinked_bytes = from + (STRING_BYTES (XSTRING (str)) - to_byte); } if (!require_decoding) @@ -5271,46 +5303,77 @@ if (coding->composing != COMPOSITION_DISABLED) coding_allocate_composition_data (coding, from); - len = decoding_buffer_size (coding, to_byte - from); - len += from + STRING_BYTES (XSTRING (str)) - to_byte; - GCPRO1 (str); - buf = get_conversion_buffer (len); - UNGCPRO; - - if (from > 0) - bcopy (XSTRING (str)->data, buf, from); - result = decode_coding (coding, XSTRING (str)->data + from, - buf + from, to_byte - from, len); - if (result == CODING_FINISH_INCONSISTENT_EOL) + allocate_conversion_buffer (buf, len); + + consumed = produced = produced_char = 0; + while (1) { - /* We simply try to decode the whole string again but without - eol-conversion this time. */ - coding->eol_type = CODING_EOL_LF; - coding->symbol = saved_coding_symbol; - coding_free_composition_data (coding); - return decode_coding_string (str, coding, nocopy); + result = decode_coding (coding, XSTRING (str)->data + from + consumed, + buf.data + produced, to_byte - from - consumed, + buf.size - produced); + consumed += coding->consumed; + produced += coding->produced; + produced_char += coding->produced_char; + if (result == CODING_FINISH_NORMAL) + break; + if (result == CODING_FINISH_INSUFFICIENT_CMP) + coding_allocate_composition_data (coding, from + produced_char); + else if (result == CODING_FINISH_INSUFFICIENT_DST) + extend_conversion_buffer (&buf); + else if (result == CODING_FINISH_INCONSISTENT_EOL) + { + /* Recover the original EOL format. */ + if (coding->eol_type == CODING_EOL_CR) + { + unsigned char *p; + for (p = buf.data; p < buf.data + produced; p++) + if (*p == '\n') *p = '\r'; + } + else if (coding->eol_type == CODING_EOL_CRLF) + { + int num_eol = 0; + unsigned char *p0, *p1; + for (p0 = buf.data, p1 = p0 + produced; p0 < p1; p0++) + if (*p0 == '\n') num_eol++; + if (produced + num_eol >= buf.size) + extend_conversion_buffer (&buf); + for (p0 = buf.data + produced, p1 = p0 + num_eol; p0 > buf.data;) + { + *--p1 = *--p0; + if (*p0 == '\n') *--p1 = '\r'; + } + produced += num_eol; + produced_char += num_eol; + } + coding->eol_type = CODING_EOL_LF; + coding->symbol = saved_coding_symbol; + } } - bcopy (XSTRING (str)->data + to_byte, buf + from + coding->produced, - STRING_BYTES (XSTRING (str)) - to_byte); - - len = from + STRING_BYTES (XSTRING (str)) - to_byte; if (coding->dst_multibyte) - str = make_multibyte_string (buf, len + coding->produced_char, - len + coding->produced); + newstr = make_uninit_multibyte_string (produced_char + shrinked_bytes, + produced + shrinked_bytes); else - str = make_unibyte_string (buf, len + coding->produced); + newstr = make_uninit_string (produced + shrinked_bytes); + if (from > 0) + bcopy (XSTRING (str)->data, XSTRING (newstr)->data, from); + bcopy (buf.data, XSTRING (newstr)->data + from, produced); + if (shrinked_bytes > from) + bcopy (XSTRING (str)->data + to_byte, + XSTRING (newstr)->data + from + produced, + shrinked_bytes - from); + free_conversion_buffer (&buf); if (coding->cmp_data && coding->cmp_data->used) - coding_restore_composition (coding, str); + coding_restore_composition (coding, newstr); coding_free_composition_data (coding); if (SYMBOLP (coding->post_read_conversion) && !NILP (Ffboundp (coding->post_read_conversion))) - str = run_pre_post_conversion_on_str (str, coding, 0); - - return str; + newstr = run_pre_post_conversion_on_str (newstr, coding, 0); + + return newstr; } Lisp_Object @@ -5320,11 +5383,14 @@ int nocopy; { int len; - char *buf; + struct conversion_buffer buf; int from, to, to_byte; struct gcpro gcpro1; Lisp_Object saved_coding_symbol; int result; + int shrinked_bytes = 0; + Lisp_Object newstr; + int consumed, consumed_char, produced; if (SYMBOLP (coding->pre_write_conversion) && !NILP (Ffboundp (coding->pre_write_conversion))) @@ -5356,32 +5422,44 @@ /* Try to skip the heading and tailing ASCIIs. */ if (coding->type != coding_type_ccl) { - int from_orig = from; - SHRINK_CONVERSION_REGION (&from, &to_byte, coding, XSTRING (str)->data, 1); if (from == to_byte) return (nocopy ? str : Fcopy_sequence (str)); + shrinked_bytes = from + (STRING_BYTES (XSTRING (str)) - to_byte); } len = encoding_buffer_size (coding, to_byte - from); - len += from + STRING_BYTES (XSTRING (str)) - to_byte; - GCPRO1 (str); - buf = get_conversion_buffer (len); - UNGCPRO; - + allocate_conversion_buffer (buf, len); + + consumed = consumed_char = produced = 0; + + while (1) + { + result = encode_coding (coding, XSTRING (str)->data + from + consumed, + buf.data + produced, to_byte - from - consumed, + buf.size - produced); + consumed += coding->consumed; + produced += coding->produced; + if (result == CODING_FINISH_NORMAL) + break; + /* Now result should be CODING_FINISH_INSUFFICIENT_DST. */ + extend_conversion_buffer (&buf); + } + + newstr = make_uninit_string (produced + shrinked_bytes); if (from > 0) - bcopy (XSTRING (str)->data, buf, from); - result = encode_coding (coding, XSTRING (str)->data + from, - buf + from, to_byte - from, len); - bcopy (XSTRING (str)->data + to_byte, buf + from + coding->produced, - STRING_BYTES (XSTRING (str)) - to_byte); - - len = from + STRING_BYTES (XSTRING (str)) - to_byte; - str = make_unibyte_string (buf, len + coding->produced); + bcopy (XSTRING (str)->data, XSTRING (newstr)->data, from); + bcopy (buf.data, XSTRING (newstr)->data + from, produced); + if (shrinked_bytes > from) + bcopy (XSTRING (str)->data + to_byte, + XSTRING (newstr)->data + from + produced, + shrinked_bytes - from); + + free_conversion_buffer (&buf); coding_free_composition_data (coding); - return str; + return newstr; } @@ -6208,12 +6286,6 @@ /*** 9. Post-amble ***/ void -init_coding () -{ - conversion_buffer = (char *) xmalloc (MINIMUM_CONVERSION_BUFFER_SIZE); -} - -void init_coding_once () { int i; @@ -6253,8 +6325,6 @@ iso_code_class[ISO_CODE_SS3] = ISO_single_shift_3; iso_code_class[ISO_CODE_CSI] = ISO_control_sequence_introducer; - conversion_buffer_size = MINIMUM_CONVERSION_BUFFER_SIZE; - setup_coding_system (Qnil, &keyboard_coding); setup_coding_system (Qnil, &terminal_coding); setup_coding_system (Qnil, &safe_terminal_coding);