emo/busybox
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

fix all known regressions with sed and also make it simpler

Browse Source
This commit is contained in:
Denis Vlasenko 2007-01-28 23:26:15 +00:00
parent dccb1a5343
commit 826c85f382
2 changed files with 47 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

70
editors/sed.c
View File

@ -82,7 +82,7 @@ typedef struct sed_cmd_s {
unsigned int in_match:1; /* Next line also included in match? */ unsigned int in_match:1; /* Next line also included in match? */
unsigned int sub_p:1; /* (s) print option */ unsigned int sub_p:1; /* (s) print option */
int last_char; /* Last line written by (sw) had no '\n' */ // int sw_last_char; /* Last line written by (sw) had no '\n' */
/* GENERAL FIELDS */ /* GENERAL FIELDS */
char cmd; /* The command char: abcdDgGhHilnNpPqrstwxy:={} */ char cmd; /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
@ -717,6 +717,11 @@ static void add_input_file(FILE *file)
bbg.input_file_list[bbg.input_file_count++] = file; bbg.input_file_list[bbg.input_file_count++] = file;
} }
enum {
MASK_NO_EOL_CHAR = 0x100,
//MASK_FIRST_LINE = 0x200,
};
/* Get next line of input from bbg.input_file_list, flushing append buffer and /* Get next line of input from bbg.input_file_list, flushing append buffer and
* noting if we ran out of files without a newline on the last line we read. * noting if we ran out of files without a newline on the last line we read.
*/ */
@ -743,13 +748,13 @@ static char *get_next_line(int *last_char)
} }
/* will be returned if last line in the file /* will be returned if last line in the file
* doesn't end with either '\n' or '\0' */ * doesn't end with either '\n' or '\0' */
lc |= 0x100; lc |= MASK_NO_EOL_CHAR;
break; break;
} }
/* Close this file and advance to next one */ /* Close this file and advance to next one */
fclose(bbg.input_file_list[bbg.current_input_file++]); fclose(bbg.input_file_list[bbg.current_input_file++]);
/* "this is the first line from new input file" */ /* "this is the first line from new input file" */
lc |= 0x200; //lc |= MASK_FIRST_LINE;
} }
*last_char = lc; *last_char = lc;
return temp; return temp;
@ -757,7 +762,7 @@ static char *get_next_line(int *last_char)
/* Output line of text. */ /* Output line of text. */
/* Note: /* Note:
* The tricks with 0x200 and last_puts_char are there to emulate gnu sed. * The tricks with MASK_FIRST_LINE and last_puts_char are there to emulate gnu sed.
* Without them, we had this: * Without them, we had this:
* echo -n thingy >z1 * echo -n thingy >z1
* echo -n again >z2 * echo -n again >z2
@ -769,14 +774,14 @@ static char *get_next_line(int *last_char)
* bbox: * bbox:
* 00000000 74 68 7a 6e 67 79 61 67 61 7a 6e |thzngyagazn| * 00000000 74 68 7a 6e 67 79 61 67 61 7a 6e |thzngyagazn|
*/ */
static void puts_maybe_newline(char *s, FILE *file, int last_char)
static int puts_maybe_newline(char *s, FILE *file, int prev_last_char, int last_char)
{ {
static char last_puts_char; static char last_puts_char = '\n';
/* Is this a first line from new file /* Is this a first line from new file
* and old file didn't end with '\n'? */ * and old file didn't end with '\n' or '\0'? */
if ((last_char & 0x200) && last_puts_char != '\n') { // if ((last_char & MASK_FIRST_LINE) && last_puts_char != '\n') {
if (last_puts_char != '\n' && last_puts_char != '\0') {
fputc('\n', file); fputc('\n', file);
last_puts_char = '\n'; last_puts_char = '\n';
} }
@ -784,7 +789,7 @@ static int puts_maybe_newline(char *s, FILE *file, int prev_last_char, int last_
/* why 'x'? - just something which is not '\n' */ /* why 'x'? - just something which is not '\n' */
if (s[0]) if (s[0])
last_puts_char = 'x'; last_puts_char = 'x';
if (!(last_char & 0x100)) { /* had trailing '\n' or '\0'? */ if (!(last_char & MASK_NO_EOL_CHAR)) { /* had trailing '\n' or '\0'? */
last_char &= 0xff; last_char &= 0xff;
fputc(last_char, file); fputc(last_char, file);
last_puts_char = last_char; last_puts_char = last_char;
@ -795,18 +800,18 @@ static int puts_maybe_newline(char *s, FILE *file, int prev_last_char, int last_
bb_error_msg_and_die(bb_msg_write_error); bb_error_msg_and_die(bb_msg_write_error);
} }
return last_char; /* Seems to be unused */
/*return last_char;*/
} }
#define sed_puts(s, n) \ #define sed_puts(s, n) (puts_maybe_newline(s, bbg.nonstdout, n))
(prev_last_char = puts_maybe_newline(s, bbg.nonstdout, prev_last_char, n))
/* Process all the lines in all the files */ /* Process all the lines in all the files */
static void process_files(void) static void process_files(void)
{ {
char *pattern_space, *next_line; char *pattern_space, *next_line;
int linenum = 0, prev_last_char = 0; int linenum = 0;
int last_char, next_last_char = 0; int last_char, next_last_char = 0;
sed_cmd_t *sed_cmd; sed_cmd_t *sed_cmd;
int substituted; int substituted;
@ -873,10 +878,13 @@ restart:
/* Skip blocks of commands we didn't match. */ /* Skip blocks of commands we didn't match. */
if (sed_cmd->cmd == '{') { if (sed_cmd->cmd == '{') {
if (sed_cmd->invert ? matched : !matched) if (sed_cmd->invert ? matched : !matched) {
while (sed_cmd && sed_cmd->cmd != '}') while (sed_cmd->cmd != '}') {
sed_cmd = sed_cmd->next; sed_cmd = sed_cmd->next;
if (!sed_cmd) bb_error_msg_and_die("unterminated {"); if (!sed_cmd)
bb_error_msg_and_die("unterminated {");
}
}
continue; continue;
} }
@ -902,7 +910,8 @@ restart:
if (tmp) { if (tmp) {
*tmp = '\0'; *tmp = '\0';
sed_puts(pattern_space, 1); /* TODO: explain why '\n' below */
sed_puts(pattern_space, '\n');
*tmp = '\n'; *tmp = '\n';
break; break;
} }
@ -911,7 +920,11 @@ restart:
/* Write the current pattern space to output */ /* Write the current pattern space to output */
case 'p': case 'p':
sed_puts(pattern_space, last_char); /* NB: we print this _before_ the last line
* (of current file) is printed. Even if
* that line is nonterminated, we print
* '\n' here (gnu sed does the same) */
sed_puts(pattern_space, (last_char & 0x200) | '\n');
break; break;
/* Delete up through first newline */ /* Delete up through first newline */
case 'D': case 'D':
@ -940,9 +953,9 @@ restart:
sed_puts(pattern_space, last_char); sed_puts(pattern_space, last_char);
/* handle w option */ /* handle w option */
if (sed_cmd->file) if (sed_cmd->file)
sed_cmd->last_char = puts_maybe_newline( /*sed_cmd->sw_last_char =*/ puts_maybe_newline(
pattern_space, sed_cmd->file, pattern_space, sed_cmd->file,
sed_cmd->last_char, last_char); last_char);
break; break;
/* Append line to linked list to be printed later */ /* Append line to linked list to be printed later */
@ -952,14 +965,14 @@ restart:
/* Insert text before this line */ /* Insert text before this line */
case 'i': case 'i':
sed_puts(sed_cmd->string, 1); sed_puts(sed_cmd->string, '\n');
break; break;
/* Cut and paste text (replace) */ /* Cut and paste text (replace) */
case 'c': case 'c':
/* Only triggers on last line of a matching range. */ /* Only triggers on last line of a matching range. */
if (!sed_cmd->in_match) if (!sed_cmd->in_match)
sed_puts(sed_cmd->string, 0); sed_puts(sed_cmd->string, MASK_NO_EOL_CHAR);
goto discard_line; goto discard_line;
/* Read file, append contents to output */ /* Read file, append contents to output */
@ -982,9 +995,9 @@ restart:
/* Write pattern space to file. */ /* Write pattern space to file. */
case 'w': case 'w':
sed_cmd->last_char = puts_maybe_newline( /*sed_cmd->sw_last_char =*/ puts_maybe_newline(
pattern_space, sed_cmd->file, pattern_space, sed_cmd->file,
sed_cmd->last_char, last_char); last_char);
break; break;
/* Read next line from input */ /* Read next line from input */
@ -1123,13 +1136,14 @@ restart:
/* /*
* exit point from sedding... * exit point from sedding...
*/ */
discard_commands: discard_commands:
/* we will print the line unless we were told to be quiet ('-n') /* we will print the line unless we were told to be quiet ('-n')
or if the line was suppressed (ala 'd'elete) */ or if the line was suppressed (ala 'd'elete) */
if (!bbg.be_quiet) sed_puts(pattern_space, last_char); if (!bbg.be_quiet)
sed_puts(pattern_space, last_char);
/* Delete and such jump here. */ /* Delete and such jump here. */
discard_line: discard_line:
flush_append(); flush_append();
free(pattern_space); free(pattern_space);

5
testsuite/sed.tests
View File

@ -146,6 +146,11 @@ testing "sed match EOF" "sed -e '"'$p'"'" "hello\nthere\nthere" "" \
"hello\nthere" "hello\nthere"
testing "sed match EOF two files" "sed -e '"'$p'"' input -" \ testing "sed match EOF two files" "sed -e '"'$p'"' input -" \
"one\ntwo\nthree\nfour\nfour" "one\ntwo" "three\nfour" "one\ntwo\nthree\nfour\nfour" "one\ntwo" "three\nfour"
# sed match EOF inline: gnu sed 4.1.5 outputs this:
#00000000 6f 6e 65 0a 6f 6f 6b 0a 6f 6f 6b 0a 74 77 6f 0a |one.ook.ook.two.|
#00000010 0a 74 68 72 65 65 0a 6f 6f 6b 0a 6f 6f 6b 0a 66 |.three.ook.ook.f|
#00000020 6f 75 72 |our|
# which looks buggy to me.
echo -ne "three\nfour" > input2 echo -ne "three\nfour" > input2
testing "sed match EOF inline" \ testing "sed match EOF inline" \
"sed -e '"'$i ook'"' -i input input2 && cat input input2" \ "sed -e '"'$i ook'"' -i input input2 && cat input input2" \