[BusyBox] PATCH: busybox-0.60.0/sed.c multi-replace problem

[Chang, Shu-Hao]

When I use "GNU sed version 3.02", the following command gets result
"abc xyz" which is what I expacted.

bash$ echo "foo bar" | sed -e 's/foo/abc/' -e 's/bar/xyz/'
abc xyz

But when I use "busybox 0.60.0 sed", I got 2 lines of output.

$ echo "foo bar" | ./busybox sed -e 's/foo/abc/' -e 's/bar/xyz/'
abc bar
foo xyz

This patch fixes this problem.

-- shuhao Aug 21, 2001
-------------- next part --------------
*** busybox-0.60.0/sed.c.old	Tue Aug 21 14:43:05 2001
--- busybox-0.60.0/sed.c	Tue Aug 21 14:59:25 2001
*************** static void load_cmd_file(char *filename
*** 490,497 ****
  	}
  }
  
! static void print_subst_w_backrefs(const char *line, const char *replace, regmatch_t *regmatch)
  {
  	int i;
  
  	/* go through the replacement string */
--- 490,511 ----
  	}
  }
  
! #define PIPE_MAGIC 0x7f
! #define PIPE_GROW 64
! #define pipeputc(c) \
! { if (pipeline[pipeline_idx] == PIPE_MAGIC) { \
! 	pipeline = xrealloc(pipeline, pipeline_len+PIPE_GROW); \
! 	memset(pipeline+pipeline_len, 0, PIPE_GROW); \
! 	pipeline_len += PIPE_GROW; \
! 	pipeline[pipeline_len-1] = PIPE_MAGIC; \
! } \
! pipeline[pipeline_idx++] = (c); }
! 
! static void print_subst_w_backrefs(const char *line, const char *replace, regmatch_t *regmatch, char **pipeline_p, int *pipeline_idx_p, int *pipeline_len_p)
  {
+ 	char *pipeline = *pipeline_p;
+ 	int pipeline_idx = *pipeline_idx_p;
+ 	int pipeline_len = *pipeline_len_p;
  	int i;
  
  	/* go through the replacement string */
*************** static void print_subst_w_backrefs(const
*** 507,519 ****
  			backref = atoi(tmpstr);
  			/* print out the text held in regmatch[backref] */
  			for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
! 				fputc(line[j], stdout);
  		}
  
  		/* if we find a backslash escaped character, print the character */
  		else if (replace[i] == '\\') {
  			++i;
! 			fputc(replace[i], stdout);
  		}
  
  		/* if we find an unescaped '&' print out the whole matched text.
--- 521,533 ----
  			backref = atoi(tmpstr);
  			/* print out the text held in regmatch[backref] */
  			for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
! 				pipeputc(line[j]);
  		}
  
  		/* if we find a backslash escaped character, print the character */
  		else if (replace[i] == '\\') {
  			++i;
! 			pipeputc(replace[i]);
  		}
  
  		/* if we find an unescaped '&' print out the whole matched text.
*************** static void print_subst_w_backrefs(const
*** 523,549 ****
  		else if (replace[i] == '&' && replace[i-1] != '\\') {
  			int j;
  			for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
! 				fputc(line[j], stdout);
  		}
  		/* nothing special, just print this char of the replacement string to stdout */
  		else
! 			fputc(replace[i], stdout);
  	}
  }
  
! static int do_subst_command(const struct sed_cmd *sed_cmd, const char *line)
  {
! 	char *hackline = (char *)line;
  	int altered = 0;
  	regmatch_t *regmatch = NULL;
  
  	/* we only proceed if the substitution 'search' expression matches */
! 	if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == REG_NOMATCH)
  		return 0;
  
  	/* whaddaya know, it matched. get the number of back references */
  	regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
  
  	/* and now, as long as we've got a line to try matching and if we can match
  	 * the search string, we make substitutions */
  	while (*hackline && (regexec(sed_cmd->sub_match, hackline,
--- 537,577 ----
  		else if (replace[i] == '&' && replace[i-1] != '\\') {
  			int j;
  			for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
! 				pipeputc(line[j]);
  		}
  		/* nothing special, just print this char of the replacement string to stdout */
  		else
! 			pipeputc(replace[i]);
  	}
+ 	*pipeline_p = pipeline;
+ 	*pipeline_idx_p = pipeline_idx;
+ 	*pipeline_len_p = pipeline_len;
  }
  
! static int do_subst_command(const struct sed_cmd *sed_cmd, char **line)
  {
! 	char *hackline = *line;
! 	char *pipeline = 0;
! 	int pipeline_idx = 0;
! 	int pipeline_len = 0;
  	int altered = 0;
  	regmatch_t *regmatch = NULL;
  
  	/* we only proceed if the substitution 'search' expression matches */
! 	if (regexec(sed_cmd->sub_match, hackline, 0, NULL, 0) == REG_NOMATCH)
  		return 0;
  
  	/* whaddaya know, it matched. get the number of back references */
  	regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
  
+ 	/* allocate more PIPE_GROW bytes
+ 	   if replaced string is larger than original */
+ 	pipeline_len = strlen(hackline)+PIPE_GROW;
+ 	pipeline = xmalloc(pipeline_len);
+ 	memset(pipeline, 0, pipeline_len);
+ 	/* buffer magic */
+ 	pipeline[pipeline_len-1] = PIPE_MAGIC;
+ 
  	/* and now, as long as we've got a line to try matching and if we can match
  	 * the search string, we make substitutions */
  	while (*hackline && (regexec(sed_cmd->sub_match, hackline,
*************** static int do_subst_command(const struct
*** 552,561 ****
  
  		/* print everything before the match */
  		for (i = 0; i < regmatch[0].rm_so; i++)
! 			fputc(hackline[i], stdout);
  
  		/* then print the substitution string */
! 		print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch);
  
  		/* advance past the match */
  		hackline += regmatch[0].rm_eo;
--- 580,589 ----
  
  		/* print everything before the match */
  		for (i = 0; i < regmatch[0].rm_so; i++)
! 			pipeputc(hackline[i]);
  
  		/* then print the substitution string */
! 		print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch, &pipeline, &pipeline_idx, &pipeline_len);
  
  		/* advance past the match */
  		hackline += regmatch[0].rm_eo;
*************** static int do_subst_command(const struct
*** 567,577 ****
  			break;
  	}
  
! 	puts(hackline);
  
  	/* cleanup */
  	free(regmatch);
  
  	return altered;
  }
  
--- 595,608 ----
  			break;
  	}
  
! 	for (; *hackline; hackline++) pipeputc(*hackline);
! 	if (pipeline[pipeline_idx] == PIPE_MAGIC) pipeline[pipeline_idx] = 0;
  
  	/* cleanup */
  	free(regmatch);
  
+ 	free(*line);
+ 	*line = pipeline;
  	return altered;
  }
  
*************** static void process_file(FILE *file)
*** 650,661 ****
  
  						/* we print the line once, unless we were told to be quiet */
  						if (!be_quiet)
! 							altered |= do_subst_command(&sed_cmds[i], line);
  
  						/* we also print the line if we were given the 'p' flag
  						 * (this is quite possibly the second printing) */
  						if (sed_cmds[i].sub_p)
! 							altered |= do_subst_command(&sed_cmds[i], line);
  
  						break;
  
--- 681,694 ----
  
  						/* we print the line once, unless we were told to be quiet */
  						if (!be_quiet)
! 							altered |= do_subst_command(&sed_cmds[i], &line);
  
  						/* we also print the line if we were given the 'p' flag
  						 * (this is quite possibly the second printing) */
  						if (sed_cmds[i].sub_p)
! 							altered |= do_subst_command(&sed_cmds[i], &line);
! 						if (altered && (i+1 >= ncmds || sed_cmds[i+1].cmd != 's'))
! 							puts(line);
  
  						break;