aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorMark Whitley <markw@lineo.com>2000-11-03 19:47:00 +0000
committerMark Whitley <markw@lineo.com>2000-11-03 19:47:00 +0000
commit2dc192fd991ff24a2e1e469b988391d7830ad4b6 (patch)
treef093280113a03e33ad105928373a4bb2546f0d2c
parenta75466e9819a7172ba48b06958a868022f4510a9 (diff)
downloadbusybox-2dc192fd991ff24a2e1e469b988391d7830ad4b6.tar.gz
Re-worked the support for s///p, fixed a backref limit buglet, and cleaned up
a few other ugly places (do_subst_command got a much-needed overhaul). Also took out BB_FEATURE_SED_PATTERN_SPACE from Config.h[.Hurd] as the 'p' is now a standard feature (adds almost no bloat).
-rw-r--r--Config.h3
-rw-r--r--Config.h.Hurd3
-rw-r--r--editors/sed.c157
-rw-r--r--sed.c157
4 files changed, 172 insertions, 148 deletions
diff --git a/Config.h b/Config.h
index a111feeda..b2baac2a2 100644
--- a/Config.h
+++ b/Config.h
@@ -223,9 +223,6 @@
// Enable support for "--exclude" for excluding files
#define BB_FEATURE_TAR_EXCLUDE
//
-// Enable support for s///p pattern matching
-#define BB_FEATURE_SED_PATTERN_SPACE
-//
//// Enable reverse sort
#define BB_FEATURE_SORT_REVERSE
//
diff --git a/Config.h.Hurd b/Config.h.Hurd
index 79fdff6ad..9238761a2 100644
--- a/Config.h.Hurd
+++ b/Config.h.Hurd
@@ -218,9 +218,6 @@
// Enable support for "--exclude" for excluding files
#define BB_FEATURE_TAR_EXCLUDE
//
-// Enable support for s///p pattern matching
-#define BB_FEATURE_SED_PATTERN_SPACE
-//
//// Enable reverse sort
#define BB_FEATURE_SORT_REVERSE
//
diff --git a/editors/sed.c b/editors/sed.c
index 75435f67a..eaca9ada5 100644
--- a/editors/sed.c
+++ b/editors/sed.c
@@ -62,13 +62,12 @@ extern char *optarg; /* ditto */
/* options */
static int be_quiet = 0;
-static const int SUB_G = 1 << 0;
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
-static const int SUB_P = 1 << 1;
-#endif
struct sed_cmd {
+
+ /* GENERAL FIELDS */
+
/* address storage */
int beg_line; /* 'sed 1p' 0 == no begining line, apply commands to all lines */
int end_line; /* 'sed 1,3p' 0 == no end line, use only beginning. -1 == $ */
@@ -78,19 +77,21 @@ struct sed_cmd {
/* the command */
char cmd; /* p,d,s (add more at your leisure :-) */
- /* substitution command specific fields */
- regex_t *sub_match; /* sed -e 's/sub_match/replace/' */
- char *replace; /* sed -e 's/sub_match/replace/' XXX: who will hold the \1 \2 \3s? */
+
+ /* SUBSTITUTION COMMAND SPECIFIC FIELDS */
+
+ /* sed -e 's/sub_match/replace/' */
+ regex_t *sub_match;
+ char *replace;
unsigned int num_backrefs:4; /* how many back references (\1..\9) */
/* Note: GNU/POSIX sed does not save more than nine backrefs, so
* we only use 4 bits to hold the number */
-#ifndef BB_FEATURE_SED_PATTERN_SPACE
- unsigned int sub_flags:1; /* sed -e 's/foo/bar/g' (global) */
-#else
- unsigned int sub_flags:2; /* sed -e 's/foo/bar/gp' (global/pattern) */
-#endif
+ unsigned int sub_g:1; /* sed -e 's/foo/bar/g' (global) */
+ unsigned int sub_p:2; /* sed -e 's/foo/bar/p' (print substitution) */
+
+
+ /* EDIT COMMAND (a,i,c) SPEICIFIC FIELDS */
- /* edit command (a,i,c) speicific field */
char *editline;
};
@@ -278,7 +279,7 @@ static int parse_subst_cmd(struct sed_cmd *sed_cmd, const char *substr)
/* sed_cmd->num_backrefs = 0; */ /* XXX: not needed? --apparently not */
for (j = 0; match[j]; j++) {
/* GNU/POSIX sed does not save more than nine backrefs */
- if (match[j] == '\\' && match[j+1] == '(' && sed_cmd->num_backrefs < 9)
+ if (match[j] == '\\' && match[j+1] == '(' && sed_cmd->num_backrefs <= 9)
sed_cmd->num_backrefs++;
}
@@ -293,16 +294,14 @@ static int parse_subst_cmd(struct sed_cmd *sed_cmd, const char *substr)
while (substr[++idx]) {
switch (substr[idx]) {
case 'g':
- sed_cmd->sub_flags = SUB_G;
+ sed_cmd->sub_g = 1;
break;
case 'I':
cflags |= REG_ICASE;
break;
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
case 'p':
- sed_cmd->sub_flags = SUB_P;
+ sed_cmd->sub_p = 1;
break;
-#endif
default:
/* any whitespace or semicolon trailing after a s/// is ok */
if (strchr("; \t\v\n\r", substr[idx]))
@@ -534,59 +533,53 @@ static void print_subst_w_backrefs(const char *line, const char *replace, regmat
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 substitute if the substitution 'search' expression matches */
- if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == 0) {
- regmatch_t *regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
- int i;
- char *ptr = (char *)line;
-
- while (*ptr) {
- /* if we can match the search string... */
- if (regexec(sed_cmd->sub_match, ptr, sed_cmd->num_backrefs+1, regmatch, 0) == 0) {
- /* print everything before the match, */
- for (i = 0; i < regmatch[0].rm_so; i++) {
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
- if(!be_quiet || (sed_cmd->sub_flags & SUB_P))
-#endif
- fputc(ptr[i], stdout);
- }
+ /* if the user specified that they didn't want anything printed (i.e. a -n
+ * flag and no 'p' flag after the s///), then there's really no point doing
+ * anything here. */
+ if (be_quiet && !sed_cmd->sub_p)
+ return 0;
- /* then print the substitution in its place */
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
- if(!be_quiet || (sed_cmd->sub_flags & SUB_P))
-#endif
- print_subst_w_backrefs(ptr, sed_cmd->replace, regmatch);
+ /* we only proceed if the substitution 'search' expression matches */
+ if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == REG_NOMATCH)
+ return 0;
- /* then advance past the match */
- ptr += regmatch[0].rm_eo;
+ /* whaddaya know, it matched. get the number of back references */
+ regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
- /* and flag that something has changed */
- altered++;
+ /* 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,
+ sed_cmd->num_backrefs+1, regmatch, 0) == 0) ) {
+ int i;
- /* if we're not doing this globally... */
- if (!sed_cmd->sub_flags & SUB_G)
- break;
- }
- /* if we COULD NOT match the search string (meaning we've gone past
- * all previous instances), get out */
- else
- break;
- }
+ /* print everything before the match */
+ for (i = 0; i < regmatch[0].rm_so; i++)
+ fputc(hackline[i], stdout);
- /* is there anything left to print? */
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
- if (*ptr && (!be_quiet || sed_cmds->sub_flags & SUB_P))
-#else
- if (*ptr)
-#endif
- fputs(ptr, stdout);
+ /* then print the substitution string */
+ print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch);
+
+ /* advance past the match */
+ hackline += regmatch[0].rm_eo;
+ /* flag that something has changed */
+ altered++;
- /* cleanup */
- free(regmatch);
+ /* if we're not doing this globally, get out now */
+ if (!sed_cmd->sub_g)
+ break;
}
+ /* if there's anything left of the line, print it */
+ if (*hackline)
+ fputs(hackline, stdout);
+
+ /* cleanup */
+ free(regmatch);
+
return altered;
}
@@ -605,7 +598,32 @@ static int do_sed_command(const struct sed_cmd *sed_cmd, const char *line)
break;
case 's':
- altered = do_subst_command(sed_cmd, line);
+
+ /*
+ * Some special cases for 's' printing to make it compliant with
+ * GNU sed printing behavior (aka "The -n | s///p Matrix"):
+ *
+ * -n ONLY = never print anything regardless of any successful
+ * substitution
+ *
+ * s///p ONLY = always print successful substitutions, even if
+ * the line is going to be printed anyway (line will be printed
+ * twice).
+ *
+ * -n AND s///p = print ONLY a successful substitution ONE TIME;
+ * no other lines are printed - this is the reason why the 'p'
+ * flag exists in the first place.
+ */
+
+ /* we print the line once, unless we were told to be quiet */
+ if (!be_quiet)
+ altered = do_subst_command(sed_cmd, line);
+
+ /* we also print the line if we were given the 'p' flag
+ * (this is quite possibly the second printing) */
+ if (sed_cmd->sub_p)
+ altered = do_subst_command(sed_cmd, line);
+
break;
case 'a':
@@ -662,7 +680,8 @@ static void process_file(FILE *file)
/* are we acting on a range of line numbers? */
else if (sed_cmds[i].beg_line > 0 && sed_cmds[i].end_line != 0) {
- if (linenum >= sed_cmds[i].beg_line && (sed_cmds[i].end_line == -1 || linenum <= sed_cmds[i].end_line))
+ if (linenum >= sed_cmds[i].beg_line &&
+ (sed_cmds[i].end_line == -1 || linenum <= sed_cmds[i].end_line))
line_altered += do_sed_command(&sed_cmds[i], line);
}
@@ -678,14 +697,10 @@ static void process_file(FILE *file)
}
- /* we will print the line unless we were told to be quiet or if
- * the line was altered (via a 'd'elete or 's'ubstitution) */
-#ifndef BB_FEATURE_SED_PATTERN_SPACE
- if (!be_quiet &&!line_altered)
-#else
- /* we where specificly requested to print the output */
- if ((!be_quiet || (sed_cmds[i].sub_flags & SUB_P)) && !line_altered)
-#endif
+ /* we will print the line unless we were told to be quiet or if the
+ * line was altered (via a 'd'elete or 's'ubstitution), in which case
+ * the altered line was already printed */
+ if (!be_quiet && !line_altered)
fputs(line, stdout);
free(line);
diff --git a/sed.c b/sed.c
index 75435f67a..eaca9ada5 100644
--- a/sed.c
+++ b/sed.c
@@ -62,13 +62,12 @@ extern char *optarg; /* ditto */
/* options */
static int be_quiet = 0;
-static const int SUB_G = 1 << 0;
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
-static const int SUB_P = 1 << 1;
-#endif
struct sed_cmd {
+
+ /* GENERAL FIELDS */
+
/* address storage */
int beg_line; /* 'sed 1p' 0 == no begining line, apply commands to all lines */
int end_line; /* 'sed 1,3p' 0 == no end line, use only beginning. -1 == $ */
@@ -78,19 +77,21 @@ struct sed_cmd {
/* the command */
char cmd; /* p,d,s (add more at your leisure :-) */
- /* substitution command specific fields */
- regex_t *sub_match; /* sed -e 's/sub_match/replace/' */
- char *replace; /* sed -e 's/sub_match/replace/' XXX: who will hold the \1 \2 \3s? */
+
+ /* SUBSTITUTION COMMAND SPECIFIC FIELDS */
+
+ /* sed -e 's/sub_match/replace/' */
+ regex_t *sub_match;
+ char *replace;
unsigned int num_backrefs:4; /* how many back references (\1..\9) */
/* Note: GNU/POSIX sed does not save more than nine backrefs, so
* we only use 4 bits to hold the number */
-#ifndef BB_FEATURE_SED_PATTERN_SPACE
- unsigned int sub_flags:1; /* sed -e 's/foo/bar/g' (global) */
-#else
- unsigned int sub_flags:2; /* sed -e 's/foo/bar/gp' (global/pattern) */
-#endif
+ unsigned int sub_g:1; /* sed -e 's/foo/bar/g' (global) */
+ unsigned int sub_p:2; /* sed -e 's/foo/bar/p' (print substitution) */
+
+
+ /* EDIT COMMAND (a,i,c) SPEICIFIC FIELDS */
- /* edit command (a,i,c) speicific field */
char *editline;
};
@@ -278,7 +279,7 @@ static int parse_subst_cmd(struct sed_cmd *sed_cmd, const char *substr)
/* sed_cmd->num_backrefs = 0; */ /* XXX: not needed? --apparently not */
for (j = 0; match[j]; j++) {
/* GNU/POSIX sed does not save more than nine backrefs */
- if (match[j] == '\\' && match[j+1] == '(' && sed_cmd->num_backrefs < 9)
+ if (match[j] == '\\' && match[j+1] == '(' && sed_cmd->num_backrefs <= 9)
sed_cmd->num_backrefs++;
}
@@ -293,16 +294,14 @@ static int parse_subst_cmd(struct sed_cmd *sed_cmd, const char *substr)
while (substr[++idx]) {
switch (substr[idx]) {
case 'g':
- sed_cmd->sub_flags = SUB_G;
+ sed_cmd->sub_g = 1;
break;
case 'I':
cflags |= REG_ICASE;
break;
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
case 'p':
- sed_cmd->sub_flags = SUB_P;
+ sed_cmd->sub_p = 1;
break;
-#endif
default:
/* any whitespace or semicolon trailing after a s/// is ok */
if (strchr("; \t\v\n\r", substr[idx]))
@@ -534,59 +533,53 @@ static void print_subst_w_backrefs(const char *line, const char *replace, regmat
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 substitute if the substitution 'search' expression matches */
- if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == 0) {
- regmatch_t *regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
- int i;
- char *ptr = (char *)line;
-
- while (*ptr) {
- /* if we can match the search string... */
- if (regexec(sed_cmd->sub_match, ptr, sed_cmd->num_backrefs+1, regmatch, 0) == 0) {
- /* print everything before the match, */
- for (i = 0; i < regmatch[0].rm_so; i++) {
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
- if(!be_quiet || (sed_cmd->sub_flags & SUB_P))
-#endif
- fputc(ptr[i], stdout);
- }
+ /* if the user specified that they didn't want anything printed (i.e. a -n
+ * flag and no 'p' flag after the s///), then there's really no point doing
+ * anything here. */
+ if (be_quiet && !sed_cmd->sub_p)
+ return 0;
- /* then print the substitution in its place */
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
- if(!be_quiet || (sed_cmd->sub_flags & SUB_P))
-#endif
- print_subst_w_backrefs(ptr, sed_cmd->replace, regmatch);
+ /* we only proceed if the substitution 'search' expression matches */
+ if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == REG_NOMATCH)
+ return 0;
- /* then advance past the match */
- ptr += regmatch[0].rm_eo;
+ /* whaddaya know, it matched. get the number of back references */
+ regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
- /* and flag that something has changed */
- altered++;
+ /* 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,
+ sed_cmd->num_backrefs+1, regmatch, 0) == 0) ) {
+ int i;
- /* if we're not doing this globally... */
- if (!sed_cmd->sub_flags & SUB_G)
- break;
- }
- /* if we COULD NOT match the search string (meaning we've gone past
- * all previous instances), get out */
- else
- break;
- }
+ /* print everything before the match */
+ for (i = 0; i < regmatch[0].rm_so; i++)
+ fputc(hackline[i], stdout);
- /* is there anything left to print? */
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
- if (*ptr && (!be_quiet || sed_cmds->sub_flags & SUB_P))
-#else
- if (*ptr)
-#endif
- fputs(ptr, stdout);
+ /* then print the substitution string */
+ print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch);
+
+ /* advance past the match */
+ hackline += regmatch[0].rm_eo;
+ /* flag that something has changed */
+ altered++;
- /* cleanup */
- free(regmatch);
+ /* if we're not doing this globally, get out now */
+ if (!sed_cmd->sub_g)
+ break;
}
+ /* if there's anything left of the line, print it */
+ if (*hackline)
+ fputs(hackline, stdout);
+
+ /* cleanup */
+ free(regmatch);
+
return altered;
}
@@ -605,7 +598,32 @@ static int do_sed_command(const struct sed_cmd *sed_cmd, const char *line)
break;
case 's':
- altered = do_subst_command(sed_cmd, line);
+
+ /*
+ * Some special cases for 's' printing to make it compliant with
+ * GNU sed printing behavior (aka "The -n | s///p Matrix"):
+ *
+ * -n ONLY = never print anything regardless of any successful
+ * substitution
+ *
+ * s///p ONLY = always print successful substitutions, even if
+ * the line is going to be printed anyway (line will be printed
+ * twice).
+ *
+ * -n AND s///p = print ONLY a successful substitution ONE TIME;
+ * no other lines are printed - this is the reason why the 'p'
+ * flag exists in the first place.
+ */
+
+ /* we print the line once, unless we were told to be quiet */
+ if (!be_quiet)
+ altered = do_subst_command(sed_cmd, line);
+
+ /* we also print the line if we were given the 'p' flag
+ * (this is quite possibly the second printing) */
+ if (sed_cmd->sub_p)
+ altered = do_subst_command(sed_cmd, line);
+
break;
case 'a':
@@ -662,7 +680,8 @@ static void process_file(FILE *file)
/* are we acting on a range of line numbers? */
else if (sed_cmds[i].beg_line > 0 && sed_cmds[i].end_line != 0) {
- if (linenum >= sed_cmds[i].beg_line && (sed_cmds[i].end_line == -1 || linenum <= sed_cmds[i].end_line))
+ if (linenum >= sed_cmds[i].beg_line &&
+ (sed_cmds[i].end_line == -1 || linenum <= sed_cmds[i].end_line))
line_altered += do_sed_command(&sed_cmds[i], line);
}
@@ -678,14 +697,10 @@ static void process_file(FILE *file)
}
- /* we will print the line unless we were told to be quiet or if
- * the line was altered (via a 'd'elete or 's'ubstitution) */
-#ifndef BB_FEATURE_SED_PATTERN_SPACE
- if (!be_quiet &&!line_altered)
-#else
- /* we where specificly requested to print the output */
- if ((!be_quiet || (sed_cmds[i].sub_flags & SUB_P)) && !line_altered)
-#endif
+ /* we will print the line unless we were told to be quiet or if the
+ * line was altered (via a 'd'elete or 's'ubstitution), in which case
+ * the altered line was already printed */
+ if (!be_quiet && !line_altered)
fputs(line, stdout);
free(line);