You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Marc Slemko <ma...@znep.com> on 1997/02/07 07:18:32 UTC
my version of the mod_include outbut buffering
Here is a cleaned up version of my way to add output buffering to
mod_include. Speeds up some test files by a factor of 2
or 3 in testing here, although real world effects are far less.
Most of it is just changing some functions to take a request_rec
instead of a pool, since I need the request_rec for rwrite. While
these may be longer than Ed's patches, I think they are simpler
in concept and practice.
I _think_ there are a few bugs in Ed's patches (that I have found...
could be some in mine to that I don't know about); the only one that
I can say looks odd for sure is the bit dealing with the return from
rwrite, but I don't have the time to figure out if some of the other
things just look odd... if people prefer his patch, I will find time
to either find the bugs or convince myself I'm crazy.
Index: mod_include.c
===================================================================
RCS file: /export/home/cvs/apache/src/mod_include.c,v
retrieving revision 1.21
diff -c -r1.21 mod_include.c
*** mod_include.c 1997/01/20 04:28:13 1.21
--- mod_include.c 1997/02/07 05:52:25
***************
*** 75,82 ****
#define SIZEFMT_KMG 1
static void decodehtml(char *s);
! static char *get_tag(pool *p, FILE *in, char *tag, int tag_len, int dodecode);
! static int get_directive(FILE *in, char *d, pool *p);
/* ------------------------ Environment function -------------------------- */
--- 75,82 ----
#define SIZEFMT_KMG 1
static void decodehtml(char *s);
! static char *get_tag(request_rec *r, FILE *in, char *tag, int tag_len, int dodecode);
! static int get_directive(FILE *in, char *d, request_rec *r);
/* ------------------------ Environment function -------------------------- */
***************
*** 113,157 ****
}
}
! #define GET_CHAR(f,c,r,p) \
{ \
int i = getc(f); \
! if(feof(f) || ferror(f) || (i == -1)) { \
! pfclose(p,f); \
! return r; \
} \
c = (char)i; \
}
/* --------------------------- Parser functions --------------------------- */
! /* Grrrr... rputc makes this slow as all-get-out. Elsewhere, it doesn't
! * matter much, but this is an inner loop...
*/
int find_string(FILE *in,char *str, request_rec *r, int printing) {
int x,l=strlen(str),p;
char c;
p=0;
while(1) {
! GET_CHAR(in,c,1,r->pool);
if(c == str[p]) {
! if((++p) == l)
return 0;
}
else {
if (printing) {
for(x=0;x<p;x++) {
! rputc(str[x],r);
}
! rputc(c,r);
}
p=0;
}
}
}
/*
* decodes a string containing html entities or numeric character references.
* 's' is overwritten with the decoded string.
--- 113,201 ----
}
}
! #define OUTBUFSIZE 4096
! /*
! * f: file handle being read from
! * c: character to read into
! * ret: return value to use if input fails
! * r: current request_rec
! *
! * Note that FLUSH_BUF() is redefined after find_string to do nothing
! * since the other routines do not support it. A lot of things in
! * this module need to be rethought.
! *
! */
! #define GET_CHAR(f,c,ret,r) \
{ \
int i = getc(f); \
! if(i == EOF) { /* either EOF or error -- needs error handling if latter */ \
! if (ferror(f)) \
! fprintf(stderr, "encountered error in GET_CHAR macro, mod_include.\n"); \
! FLUSH_BUF(r); \
! pfclose(r->pool,f); \
! return ret; \
} \
c = (char)i; \
}
+
/* --------------------------- Parser functions --------------------------- */
! /* PUT_CHAR and FLUSH_BUF currently only work within the scope of
! * find_string(); they are hacks to avoid calling rputc for each and
! * every character output. A common set of buffering calls for this
! * type of output SHOULD be implemented.
*/
+ #define PUT_CHAR(c,r) \
+ { \
+ outbuf[outind++] = c; \
+ if (outind == OUTBUFSIZE) { FLUSH_BUF(r) }; \
+ }
+
+ /* there SHOULD be some error checking on the return value of
+ * rwrite, however it is unclear what the API for rwrite returning
+ * errors is and little can really be done to help the error in
+ * any case.
+ */
+ #define FLUSH_BUF(r) \
+ { \
+ rwrite(outbuf, outind, r); \
+ outind = 0; \
+ }
int find_string(FILE *in,char *str, request_rec *r, int printing) {
int x,l=strlen(str),p;
+ char outbuf[OUTBUFSIZE];
+ int outind = 0;
char c;
p=0;
while(1) {
! GET_CHAR(in,c,1,r);
if(c == str[p]) {
! if((++p) == l) {
! FLUSH_BUF(r);
return 0;
+ }
}
else {
if (printing) {
for(x=0;x<p;x++) {
! PUT_CHAR(str[x],r);
}
! PUT_CHAR(c,r);
}
p=0;
}
}
}
+ #undef FLUSH_BUF
+ #undef PUT_CHAR
+ /* FLUSH_BUF is called from within GET_CHAR, so still needs to exist
+ * even if it does nothing */
+ #define FLUSH_BUF(r) {};
+
/*
* decodes a string containing html entities or numeric character references.
* 's' is overwritten with the decoded string.
***************
*** 243,264 ****
*/
static char *
! get_tag(pool *p, FILE *in, char *tag, int tagbuf_len, int dodecode) {
char *t = tag, *tag_val, c, term;
int n;
n = 0;
do { /* skip whitespace */
! GET_CHAR(in,c,NULL,p);
} while (isspace(c));
/* tags can't start with - */
if(c == '-') {
! GET_CHAR(in,c,NULL,p);
if(c == '-') {
do {
! GET_CHAR(in,c,NULL,p);
} while (isspace(c));
if(c == '>') {
strncpy(tag,"done", tagbuf_len-1);
--- 287,308 ----
*/
static char *
! get_tag(request_rec *r, FILE *in, char *tag, int tagbuf_len, int dodecode) {
char *t = tag, *tag_val, c, term;
int n;
n = 0;
do { /* skip whitespace */
! GET_CHAR(in,c,NULL,r);
} while (isspace(c));
/* tags can't start with - */
if(c == '-') {
! GET_CHAR(in,c,NULL,r);
if(c == '-') {
do {
! GET_CHAR(in,c,NULL,r);
} while (isspace(c));
if(c == '>') {
strncpy(tag,"done", tagbuf_len-1);
***************
*** 277,296 ****
}
if(c == '=' || isspace(c)) break;
*(t++) = tolower(c);
! GET_CHAR(in,c,NULL,p);
}
*t++ = '\0';
tag_val = t;
! while (isspace(c)) GET_CHAR(in, c, NULL,p); /* space before = */
if (c != '=') {
ungetc(c, in);
return NULL;
}
do {
! GET_CHAR(in,c,NULL,p); /* space after = */
} while (isspace(c));
/* we should allow a 'name' as a value */
--- 321,340 ----
}
if(c == '=' || isspace(c)) break;
*(t++) = tolower(c);
! GET_CHAR(in,c,NULL,r);
}
*t++ = '\0';
tag_val = t;
! while (isspace(c)) GET_CHAR(in, c, NULL,r); /* space before = */
if (c != '=') {
ungetc(c, in);
return NULL;
}
do {
! GET_CHAR(in,c,NULL,r); /* space after = */
} while (isspace(c));
/* we should allow a 'name' as a value */
***************
*** 298,304 ****
if (c != '"' && c != '\'') return NULL;
term = c;
while(1) {
! GET_CHAR(in,c,NULL,p);
if(++n == tagbuf_len) {
t[tagbuf_len - 1] = '\0';
return NULL;
--- 342,348 ----
if (c != '"' && c != '\'') return NULL;
term = c;
while(1) {
! GET_CHAR(in,c,NULL,r);
if(++n == tagbuf_len) {
t[tagbuf_len - 1] = '\0';
return NULL;
***************
*** 306,312 ****
/* Want to accept \" as a valid character within a string. */
if (c == '\\') {
*(t++) = c; /* Add backslash */
! GET_CHAR(in,c,NULL,p);
if (c == term) /* Only if */
*(--t) = c; /* Replace backslash ONLY for terminator */
} else if (c == term) break;
--- 350,356 ----
/* Want to accept \" as a valid character within a string. */
if (c == '\\') {
*(t++) = c; /* Add backslash */
! GET_CHAR(in,c,NULL,r);
if (c == term) /* Only if */
*(--t) = c; /* Replace backslash ONLY for terminator */
} else if (c == term) break;
***************
*** 314,336 ****
}
*t = '\0';
if (dodecode) decodehtml(tag_val);
! return pstrdup (p, tag_val);
}
static int
! get_directive(FILE *in, char *d, pool *p) {
char c;
/* skip initial whitespace */
while(1) {
! GET_CHAR(in,c,1,p);
if(!isspace(c))
break;
}
/* now get directive */
while(1) {
*d++ = tolower(c);
! GET_CHAR(in,c,1,p);
if(isspace(c))
break;
}
--- 358,380 ----
}
*t = '\0';
if (dodecode) decodehtml(tag_val);
! return pstrdup (r->pool, tag_val);
}
static int
! get_directive(FILE *in, char *d, request_rec *r) {
char c;
/* skip initial whitespace */
while(1) {
! GET_CHAR(in,c,1,r);
if(!isspace(c))
break;
}
/* now get directive */
while(1) {
*d++ = tolower(c);
! GET_CHAR(in,c,1,r);
if(isspace(c))
break;
}
***************
*** 453,459 ****
char *tag_val;
while(1) {
! if(!(tag_val = get_tag(r->pool, in, tag, MAX_STRING_LEN, 1)))
return 1;
if(!strcmp(tag,"file") || !strcmp (tag, "virtual")) {
request_rec *rr=NULL;
--- 497,503 ----
char *tag_val;
while(1) {
! if(!(tag_val = get_tag(r, in, tag, MAX_STRING_LEN, 1)))
return 1;
if(!strcmp(tag,"file") || !strcmp (tag, "virtual")) {
request_rec *rr=NULL;
***************
*** 601,607 ****
char parsed_string[MAX_STRING_LEN];
while(1) {
! if(!(tag_val = get_tag (r->pool, in, tag, MAX_STRING_LEN, 1)))
return 1;
if(!strcmp(tag,"cmd")) {
parse_string(r, tag_val, parsed_string, MAX_STRING_LEN, 1);
--- 645,651 ----
char parsed_string[MAX_STRING_LEN];
while(1) {
! if(!(tag_val = get_tag (r, in, tag, MAX_STRING_LEN, 1)))
return 1;
if(!strcmp(tag,"cmd")) {
parse_string(r, tag_val, parsed_string, MAX_STRING_LEN, 1);
***************
*** 638,644 ****
char *tag_val;
while(1) {
! if(!(tag_val = get_tag (r->pool, in, tag, MAX_STRING_LEN, 1)))
return 1;
if(!strcmp(tag,"var")) {
char *val = table_get (r->subprocess_env, tag_val);
--- 682,688 ----
char *tag_val;
while(1) {
! if(!(tag_val = get_tag (r, in, tag, MAX_STRING_LEN, 1)))
return 1;
if(!strcmp(tag,"var")) {
char *val = table_get (r->subprocess_env, tag_val);
***************
*** 666,672 ****
table *env = r->subprocess_env;
while(1) {
! if(!(tag_val = get_tag(r->pool, in, tag, MAX_STRING_LEN, 0)))
return 1;
if(!strcmp(tag,"errmsg")) {
parse_string(r, tag_val, parsed_string, MAX_STRING_LEN, 0);
--- 710,716 ----
table *env = r->subprocess_env;
while(1) {
! if(!(tag_val = get_tag(r, in, tag, MAX_STRING_LEN, 0)))
return 1;
if(!strcmp(tag,"errmsg")) {
parse_string(r, tag_val, parsed_string, MAX_STRING_LEN, 0);
***************
*** 753,759 ****
char parsed_string[MAX_STRING_LEN];
while(1) {
! if(!(tag_val = get_tag(r->pool, in, tag, MAX_STRING_LEN, 1)))
return 1;
else if(!strcmp(tag,"done"))
return 0;
--- 797,803 ----
char parsed_string[MAX_STRING_LEN];
while(1) {
! if(!(tag_val = get_tag(r, in, tag, MAX_STRING_LEN, 1)))
return 1;
else if(!strcmp(tag,"done"))
return 0;
***************
*** 792,798 ****
char parsed_string[MAX_STRING_LEN];
while(1) {
! if(!(tag_val = get_tag(r->pool, in, tag, MAX_STRING_LEN, 1)))
return 1;
else if(!strcmp(tag,"done"))
return 0;
--- 836,842 ----
char parsed_string[MAX_STRING_LEN];
while(1) {
! if(!(tag_val = get_tag(r, in, tag, MAX_STRING_LEN, 1)))
return 1;
else if(!strcmp(tag,"done"))
return 0;
***************
*** 1386,1392 ****
char *expr = '\0';
while(1) {
! tag_val = get_tag(r->pool, in, tag, MAX_STRING_LEN, 0);
if(*tag == '\0')
return 1;
else if(!strcmp(tag,"done")) {
--- 1430,1436 ----
char *expr = '\0';
while(1) {
! tag_val = get_tag(r, in, tag, MAX_STRING_LEN, 0);
if(*tag == '\0')
return 1;
else if(!strcmp(tag,"done")) {
***************
*** 1416,1422 ****
char *expr = '\0';
while(1) {
! tag_val = get_tag(r->pool, in, tag, MAX_STRING_LEN, 0);
if(*tag == '\0')
return 1;
else if(!strcmp(tag,"done")) {
--- 1460,1466 ----
char *expr = '\0';
while(1) {
! tag_val = get_tag(r, in, tag, MAX_STRING_LEN, 0);
if(*tag == '\0')
return 1;
else if(!strcmp(tag,"done")) {
***************
*** 1451,1457 ****
char tag[MAX_STRING_LEN];
char *tag_val;
! if(!(tag_val = get_tag(r->pool, in, tag, MAX_STRING_LEN, 1)))
return 1;
else if(!strcmp(tag,"done")) {
#ifdef DEBUG_INCLUDE
--- 1495,1501 ----
char tag[MAX_STRING_LEN];
char *tag_val;
! if(!(tag_val = get_tag(r, in, tag, MAX_STRING_LEN, 1)))
return 1;
else if(!strcmp(tag,"done")) {
#ifdef DEBUG_INCLUDE
***************
*** 1473,1479 ****
char tag[MAX_STRING_LEN];
char *tag_val;
! if(!(tag_val = get_tag(r->pool, in, tag, MAX_STRING_LEN, 1))) {
return 1;
} else if(!strcmp(tag,"done")) {
#ifdef DEBUG_INCLUDE
--- 1517,1523 ----
char tag[MAX_STRING_LEN];
char *tag_val;
! if(!(tag_val = get_tag(r, in, tag, MAX_STRING_LEN, 1))) {
return 1;
} else if(!strcmp(tag,"done")) {
#ifdef DEBUG_INCLUDE
***************
*** 1497,1503 ****
var = (char *)NULL;
while (1) {
! if(!(tag_val = get_tag(r->pool, in, tag, MAX_STRING_LEN, 1)))
return 1;
else if(!strcmp(tag,"done"))
return 0;
--- 1541,1547 ----
var = (char *)NULL;
while (1) {
! if(!(tag_val = get_tag(r, in, tag, MAX_STRING_LEN, 1)))
return 1;
else if(!strcmp(tag,"done"))
return 0;
***************
*** 1523,1529 ****
table_entry *elts = (table_entry *) r->subprocess_env->elts;
int i;
! if(!(tag_val = get_tag(r->pool, in, tag, MAX_STRING_LEN, 1)))
return 1;
else if(!strcmp(tag,"done")) {
for (i = 0; i < r->subprocess_env->nelts; ++i)
--- 1567,1573 ----
table_entry *elts = (table_entry *) r->subprocess_env->elts;
int i;
! if(!(tag_val = get_tag(r, in, tag, MAX_STRING_LEN, 1)))
return 1;
else if(!strcmp(tag,"done")) {
for (i = 0; i < r->subprocess_env->nelts; ++i)
***************
*** 1572,1578 ****
while(1) {
if(!find_string(f,STARTING_SEQUENCE,r,printing)) {
! if(get_directive(f,directive,r->pool))
return;
if(!strcmp(directive,"if")) {
if (!printing) {
--- 1616,1622 ----
while(1) {
if(!find_string(f,STARTING_SEQUENCE,r,printing)) {
! if(get_directive(f,directive,r))
return;
if(!strcmp(directive,"if")) {
if (!printing) {
Re: my version of the mod_include outbut buffering
Posted by Marc Slemko <ma...@znep.com>.
On Fri, 7 Feb 1997, Ed Korthof wrote:
> I'm not particularly invested in my patch for mod_include (and I just realized,
> so I'll remove it from the running. However, I would suggest a few changes to
> Marc's patch, specifically to avoid having the change the other functions which
> use GET_CHAR. IMO, changing the arguments they take is something we should
> avoid.
>
> Marc, why don't we use the GET_CHAR, PUT_CHAR, and FLUSH_BUFFER macros you
> posted -- define them above find_string -- and undef them below find_string;
> then define the old GET_CHAR macro for use by the other functions? Not the
> cleanest code, but it should do the job efficiently.
That's ugly. Oh yes, so is my patch... Hey, so is the whole module.
Heck, the entire _concept_ is ugly. <sigh>
I agree. I thought about doing it before but didn't. My new patch. I am
just guessing at a size of 4k for the output buffer; seems reasonable to
me...
In general, functions that pass fields of a request_rec instead of just
passing the request_rec annoy me. Yes, yes, it is cheaper to not pass the
whole thing but it sure saves changes later...
New patch:
Index: mod_include.c
===================================================================
RCS file: /export/home/cvs/apache/src/mod_include.c,v
retrieving revision 1.21
diff -c -r1.21 mod_include.c
*** mod_include.c 1997/01/20 04:28:13 1.21
--- mod_include.c 1997/02/08 02:49:56
***************
*** 113,156 ****
}
}
! #define GET_CHAR(f,c,r,p) \
{ \
int i = getc(f); \
! if(feof(f) || ferror(f) || (i == -1)) { \
! pfclose(p,f); \
! return r; \
} \
c = (char)i; \
}
- /* --------------------------- Parser functions --------------------------- */
-
- /* Grrrr... rputc makes this slow as all-get-out. Elsewhere, it doesn't
- * matter much, but this is an inner loop...
- */
-
int find_string(FILE *in,char *str, request_rec *r, int printing) {
int x,l=strlen(str),p;
char c;
p=0;
while(1) {
! GET_CHAR(in,c,1,r->pool);
if(c == str[p]) {
! if((++p) == l)
return 0;
}
else {
if (printing) {
for(x=0;x<p;x++) {
! rputc(str[x],r);
}
! rputc(c,r);
}
p=0;
}
}
}
/*
* decodes a string containing html entities or numeric character references.
--- 113,209 ----
}
}
!
!
! /* --------------------------- Parser functions --------------------------- */
!
! #define OUTBUFSIZE 4096
! /* PUT_CHAR and FLUSH_BUF currently only work within the scope of
! * find_string(); they are hacks to avoid calling rputc for each and
! * every character output. A common set of buffering calls for this
! * type of output SHOULD be implemented.
! */
! #define PUT_CHAR(c,r) \
! { \
! outbuf[outind++] = c; \
! if (outind == OUTBUFSIZE) { FLUSH_BUF(r) }; \
! }
!
! /* there SHOULD be some error checking on the return value of
! * rwrite, however it is unclear what the API for rwrite returning
! * errors is and little can really be done to help the error in
! * any case.
! */
! #define FLUSH_BUF(r) \
! { \
! rwrite(outbuf, outind, r); \
! outind = 0; \
! }
!
! /*
! * f: file handle being read from
! * c: character to read into
! * ret: return value to use if input fails
! * r: current request_rec
! *
! * This macro is redefined after find_string() for historical reasons
! * to avoid too many code changes. This is one of the many things
! * that should be fixed.
! */
! #define GET_CHAR(f,c,ret,r) \
{ \
int i = getc(f); \
! if(i == EOF) { /* either EOF or error -- needs error handling if latter */ \
! if (ferror(f)) \
! fprintf(stderr, "encountered error in GET_CHAR macro, mod_include.\n"); \
! FLUSH_BUF(r); \
! pfclose(r->pool,f); \
! return ret; \
} \
c = (char)i; \
}
int find_string(FILE *in,char *str, request_rec *r, int printing) {
int x,l=strlen(str),p;
+ char outbuf[OUTBUFSIZE];
+ int outind = 0;
char c;
p=0;
while(1) {
! GET_CHAR(in,c,1,r);
if(c == str[p]) {
! if((++p) == l) {
! FLUSH_BUF(r);
return 0;
+ }
}
else {
if (printing) {
for(x=0;x<p;x++) {
! PUT_CHAR(str[x],r);
}
! PUT_CHAR(c,r);
}
p=0;
}
}
}
+
+ #undef FLUSH_BUF
+ #undef PUT_CHAR
+ #undef GET_CHAR
+ #define GET_CHAR(f,c,r,p) \
+ { \
+ int i = getc(f); \
+ if(i == EOF) { /* either EOF or error -- needs error handling if latter */ \
+ if (ferror(f)) \
+ fprintf(stderr, "encountered error in GET_CHAR macro, mod_include.\n"); \
+ pfclose(p,f); \
+ return r; \
+ } \
+ c = (char)i; \
+ }
/*
* decodes a string containing html entities or numeric character references.
Re: my version of the mod_include outbut buffering
Posted by Ed Korthof <ed...@organic.com>.
I'm not particularly invested in my patch for mod_include (and I just realized,
so I'll remove it from the running. However, I would suggest a few changes to
Marc's patch, specifically to avoid having the change the other functions which
use GET_CHAR. IMO, changing the arguments they take is something we should
avoid.
Marc, why don't we use the GET_CHAR, PUT_CHAR, and FLUSH_BUFFER macros you
posted -- define them above find_string -- and undef them below find_string;
then define the old GET_CHAR macro for use by the other functions? Not the
cleanest code, but it should do the job efficiently.
--
-- Ed Korthof | Web Server Engineer --
-- ed@organic.com | Organic Online, Inc --
-- (415) 278-5676 | Fax: (415) 284-6891 --