speed up big config parsing by 2x using gperf

rofl0r · rofl0r · commit 45323584a0b2 · 2020-09-16T21:01:02.000+01:00
diff --git a/src/Makefile.am b/src/Makefile.am
@@ -29,6 +29,7 @@ tinyproxy_SOURCES = \
 	buffer.c buffer.h \
 	child.c child.h \
 	common.h \
+	conf-tokens.c conf-tokens.h \
 	conf.c conf.h \
 	conns.c conns.h \
 	daemon.c daemon.h \
@@ -58,3 +59,6 @@ EXTRA_tinyproxy_SOURCES = filter.c filter.h \
 	transparent-proxy.c transparent-proxy.h
 tinyproxy_DEPENDENCIES = @ADDITIONAL_OBJECTS@
 tinyproxy_LDADD = @ADDITIONAL_OBJECTS@ -lpthread
+
+conf-tokens.c: conf-tokens.gperf
+	gperf $< > $@
diff --git a/src/conf-tokens.gperf b/src/conf-tokens.gperf
@@ -0,0 +1,61 @@
+%{
+#include <string.h>
+#include <stdlib.h>
+#include "conf-tokens.h"
+%}
+
+struct config_directive_entry { const char* name; enum config_directive value; };
+
+%struct-type
+%define slot-name name
+%define initializer-suffix ,CD_NIL
+%define lookup-function-name config_directive_find
+%ignore-case
+%7bit
+%compare-lengths
+%readonly-tables
+%define constants-prefix CDS_
+%omit-struct-type
+
+%%
+logfile, CD_logfile
+pidfile, CD_pidfile
+anonymous, CD_anonymous
+viaproxyname, CD_viaproxyname
+defaulterrorfile, CD_defaulterrorfile
+statfile, CD_statfile
+stathost, CD_stathost
+xtinyproxy, CD_xtinyproxy
+syslog, CD_syslog
+bindsame, CD_bindsame
+disableviaheader, CD_disableviaheader
+port, CD_port
+maxclients, CD_maxclients
+maxspareservers, CD_maxspareservers
+minspareservers, CD_minspareservers
+startservers, CD_startservers
+maxrequestsperchild, CD_maxrequestsperchild
+timeout, CD_timeout
+connectport, CD_connectport
+user, CD_user
+group, CD_group
+listen, CD_listen
+allow, CD_allow
+deny, CD_deny
+bind, CD_bind
+basicauth, CD_basicauth
+errorfile, CD_errorfile
+addheader, CD_addheader
+filter, CD_filter
+filterurls, CD_filterurls
+filterextended, CD_filterextended
+filterdefaultdeny, CD_filterdefaultdeny
+filtercasesensitive, CD_filtercasesensitive
+reversebaseurl, CD_reversebaseurl
+reverseonly, CD_reverseonly
+reversemagic, CD_reversemagic
+reversepath, CD_reversepath
+upstream, CD_upstream
+loglevel, CD_loglevel
+%%
+
diff --git a/src/conf-tokens.h b/src/conf-tokens.h
@@ -0,0 +1,53 @@
+#ifndef CONF_TOKENS_H
+#define CONF_TOKENS_H
+
+enum config_directive {
+CD_NIL = 0,
+CD_logfile,
+CD_pidfile,
+CD_anonymous,
+CD_viaproxyname,
+CD_defaulterrorfile,
+CD_statfile,
+CD_stathost,
+CD_xtinyproxy,
+CD_syslog,
+CD_bindsame,
+CD_disableviaheader,
+CD_port,
+CD_maxclients,
+CD_maxspareservers,
+CD_minspareservers,
+CD_startservers,
+CD_maxrequestsperchild,
+CD_timeout,
+CD_connectport,
+CD_user,
+CD_group,
+CD_listen,
+CD_allow,
+CD_deny,
+CD_bind,
+CD_basicauth,
+CD_errorfile,
+CD_addheader,
+CD_filter,
+CD_filterurls,
+CD_filterextended,
+CD_filterdefaultdeny,
+CD_filtercasesensitive,
+CD_reversebaseurl,
+CD_reverseonly,
+CD_reversemagic,
+CD_reversepath,
+CD_upstream,
+CD_loglevel,
+};
+
+struct config_directive_entry { const char* name; enum config_directive value; };
+
+const struct config_directive_entry *
+config_directive_find (register const char *str, register size_t len);
+
+#endif
+
diff --git a/src/conf.c b/src/conf.c
@@ -37,6 +37,7 @@
 #include "upstream.h"
 #include "connect-ports.h"
 #include "basicauth.h"
+#include "conf-tokens.h"
 
 /*
  * The configuration directives are defined in the structure below.  Each
@@ -98,10 +99,13 @@ typedef int (*CONFFILE_HANDLER) (struct config_s *, const char *,
  * List all the handling functions.  These are defined later, but they need
  * to be in-scope before the big structure below.
  */
-static HANDLE_FUNC (handle_nop)
+static HANDLE_FUNC (handle_disabled_feature)
 {
-        return 0;
-}                               /* do nothing function */
+        fprintf (stderr, "ERROR: accessing feature that was disabled at compiletime on line %lu\n",
+                 lineno);
+
+        return -1;
+}
 
 static HANDLE_FUNC (handle_allow);
 static HANDLE_FUNC (handle_basicauth);
@@ -161,7 +165,7 @@ static void config_free_regex (void);
  * do not follow the pattern above.  This macro is for convenience
  * only.
  */
-#define STDCONF(d, re, func) { BEGIN "(" #d ")" WS re END, func, NULL }
+#define STDCONF(d, re, func) [CD_ ## d] = { BEGIN "()" WS re END, func, NULL }
 
 /*
  * Holds the regular expression used to match the configuration directive,
@@ -174,14 +178,6 @@ struct {
         CONFFILE_HANDLER handler;
         regex_t *cre;
 } directives[] = {
-        /* comments */
-        {
-                BEGIN "#", handle_nop, NULL
-        },
-        /* blank lines */
-        {
-                "^[[:space:]]+$", handle_nop, NULL
-        },
         /* string arguments */
         STDCONF (logfile, STR, handle_logfile),
         STDCONF (pidfile, STR, handle_pidfile),
@@ -326,20 +322,25 @@ void free_config (struct config_s *conf)
 }
 
 /*
+ * Initializes Config parser. Currently this means:
  * Compiles the regular expressions used by the configuration file.  This
  * routine MUST be called before trying to parse the configuration file.
  *
  * Returns 0 on success; negative upon failure.
  */
 int
-config_compile_regex (void)
+config_init (void)
 {
         unsigned int i, r;
 
         for (i = 0; i != ndirectives; ++i) {
-                assert (directives[i].handler);
                 assert (!directives[i].cre);
 
+                if (!directives[i].handler) {
+                        directives[i].handler = handle_disabled_feature;
+                        continue;
+                }
+
                 directives[i].cre = (regex_t *) safemalloc (sizeof (regex_t));
                 if (!directives[i].cre)
                         return -1;
@@ -383,20 +384,17 @@ config_free_regex (void)
  * a negative number is returned.
  */
 static int check_match (struct config_s *conf, const char *line,
-                        unsigned long lineno)
+                        unsigned long lineno, enum config_directive cd)
 {
         regmatch_t match[RE_MAX_MATCHES];
-        unsigned int i;
-
-        assert (ndirectives > 0);
+        unsigned int i = cd;
 
-        for (i = 0; i != ndirectives; ++i) {
-                assert (directives[i].cre);
-                if (!regexec
-                    (directives[i].cre, line, RE_MAX_MATCHES, match, 0))
-                        return (*directives[i].handler) (conf, line, lineno, match);
-        }
+        if (!directives[i].cre)
+                return (*directives[i].handler) (conf, line, lineno, match);
 
+        if (!regexec
+            (directives[i].cre, line, RE_MAX_MATCHES, match, 0))
+                return (*directives[i].handler) (conf, line, lineno, match);
         return -1;
 }
 
@@ -405,15 +403,25 @@ static int check_match (struct config_s *conf, const char *line,
  */
 static int config_parse (struct config_s *conf, FILE * f)
 {
-        char buffer[LINE_MAX];
+        char buffer[LINE_MAX], *p, *q, c;
+        const struct config_directive_entry *e;
         unsigned long lineno = 1;
 
-        while (fgets (buffer, sizeof (buffer), f)) {
-                if (check_match (conf, buffer, lineno)) {
-                        printf ("Syntax error on line %ld\n", lineno);
+        for (;fgets (buffer, sizeof (buffer), f);++lineno) {
+                if(buffer[0] == '#') continue;
+                p = buffer;
+                while(isspace(*p))p++;
+                if(!*p) continue;
+                q = p;
+                while(!isspace(*q))q++;
+                c = *q;
+                *q = 0;
+                e = config_directive_find(p, strlen(p));
+                *q = c;
+                if (!e || e->value == CD_NIL || check_match (conf, q, lineno, e->value)) {
+                        fprintf (stderr, "ERROR: Syntax error on line %lu\n", lineno);
                         return 1;
                 }
-                ++lineno;
         }
         return 0;
 }
diff --git a/src/conf.h b/src/conf.h
@@ -115,7 +115,7 @@ struct config_s {
 
 extern int reload_config_file (const char *config_fname, struct config_s *conf);
 
-int config_compile_regex (void);
+int config_init (void);
 void free_config (struct config_s *conf);
 
 #endif
diff --git a/src/main.c b/src/main.c
@@ -300,7 +300,7 @@ main (int argc, char **argv)
 
         log_message (LOG_NOTICE, "Initializing " PACKAGE " ...");
 
-        if (config_compile_regex()) {
+        if (config_init()) {
                 exit (EX_SOFTWARE);
         }
 

Original file line number	Diff line number	Diff line change
`@@ -300,7 +300,7 @@ main (int argc, char **argv)`
`300`	`300`
`301`	`301`	`log_message (LOG_NOTICE, "Initializing " PACKAGE " ...");`
`302`	`302`
`303`		`- if (config_compile_regex()) {`
	`303`	`+ if (config_init()) {`
`304`	`304`	`exit (EX_SOFTWARE);`
`305`	`305`	`}`
`306`	`306`