rewritten template parser for awx - uses @for, @if, @else, @end for basic flow control

SVN-Revision: 7276

package/busybox/patches/920-awx.patch

@@ -1,6 +1,6 @@
 diff -purN bb.old/editors/awk.c bb.dev/editors/awk.c
---- bb.old/editors/awk.c	2007-03-06 19:38:07.278092000 +0100
-+++ bb.dev/editors/awk.c	2007-03-11 05:14:11.776304544 +0100
+--- bb.old/editors/awk.c	2007-05-20 04:17:05.002197784 +0200
++++ bb.dev/editors/awk.c	2007-05-20 07:03:57.436076472 +0200
 @@ -30,6 +30,11 @@
  /* these flags are static, don't change them when value is changed */
  #define	VF_DONTTOUCH (VF_ARRAY | VF_SPECIAL | VF_WALK | VF_CHILD | VF_DIRTY)
@@ -97,8 +97,8 @@ diff -purN bb.old/editors/awk.c bb.dev/editors/awk.c
  		switch (c) {
 diff -purN bb.old/editors/awx.c bb.dev/editors/awx.c
 --- bb.old/editors/awx.c	1970-01-01 01:00:00.000000000 +0100
-+++ bb.dev/editors/awx.c	2007-03-14 02:03:50.566202928 +0100
-@@ -0,0 +1,590 @@
++++ bb.dev/editors/awx.c	2007-05-20 08:07:26.759971216 +0200
+@@ -0,0 +1,625 @@
 +/*
 + * awk web extension
 + *
@@ -109,14 +109,13 @@ diff -purN bb.old/editors/awx.c bb.dev/editors/awx.c
 +
 +#include <cgi.h>
 +#include <glob.h>
++#include "awx_parser.h"
 +
 +#define LINE_BUF 2048
 +#define HASH_MAX	1536
 +#define TR_START	"@TR<<"
 +#define TR_END		">>"
 +#define MAX_TR	32
-+#define SSI_START "<%"
-+#define SSI_END "%>"
 +
 +#undef fputs
 +
@@ -152,9 +151,17 @@ diff -purN bb.old/editors/awx.c bb.dev/editors/awx.c
 +{
 +	char *tok[MAX_TR * 3];
 +	char *l, *p, *p2, *res;
-+	int len = 0, _pos = 0, i;
++	int len = 0, _pos = 0, i, tr_abort = 0;
++	static char *backlog = NULL;
++
++	if (backlog) {
++		backlog = xrealloc(backlog, strlen(backlog) + strlen(line) + 1);
++		sprintf(backlog + strlen(backlog), line);
++		l = backlog;
++	} else {
++		l = line;
++	}
 +
-+	l = line;
 +	while (l != NULL) {
 +		if ((p = strstr(l, TR_START)) == NULL) {
 +			len += strlen((tok[_pos++] = l));
@@ -162,8 +169,14 @@ diff -purN bb.old/editors/awx.c bb.dev/editors/awx.c
 +		}
 +
 +		p2 = strstr(p, TR_END);
-+		if (p2 == NULL)
++		if (p2 == NULL) {
++			p2 = backlog;
++			backlog = xstrdup(l);
++			if (p2)
++				free(p2);
++			tr_abort = 1;
 +			break;
++		}
 +
 +		*p = 0;
 +		*p2 = 0;
@@ -182,7 +195,11 @@ diff -purN bb.old/editors/awx.c bb.dev/editors/awx.c
 +		strcat(p, tok[i]);
 +		p += strlen(tok[i]);
 +	}
-+
++	if (!tr_abort && backlog) {
++		free(backlog);
++		backlog = NULL;
++	}
++	
 +	return res;
 +}
 +
@@ -398,22 +415,25 @@ diff -purN bb.old/editors/awx.c bb.dev/editors/awx.c
 +	return res;
 +}
 +
-+
-+/* parse and evaluate an awk expression and return the result as string */
-+static char *render_lookup(char *fname, int lnr, char *str)
++/* parse an awk expression */
++static var *parse_awk(char *str)
 +{
 +	chain body;
-+	var tv;
++	node *n;
++	var *tv = xzalloc(sizeof(*tv));
 +
 +	memset(&body, 0, sizeof(body));
-+	zero_out_var(&tv);
 +	pos = str;
 +	seq = &body;
 +	
 +	/* end of expression, assume that there's going to be a free byte
 +	 * at the end of the string that can be used for the ')' */
-+	strcat(str + strlen(str), ")");
-+	return getvar_s(evaluate(parse_expr(TC_SEQTERM), &tv));
++	strcat(str + strlen(str), "}");
++	n = parse_expr(TC_GRPTERM);
++	if (!n)
++		return NULL;
++
++	return evaluate(n, tv);
 +}
 +
 +static inline void print_translate(char *s)
@@ -427,63 +447,78 @@ diff -purN bb.old/editors/awx.c bb.dev/editors/awx.c
 +		free(str);
 +}
 +
-+/* process awk calls in a template line and print the output to stdout */
-+static void render_line(char *fname, int lnr, char *line)
++static void render_element(struct template_cb *tcb, struct template_element *e)
 +{
-+	char *tok[MAX_TR * 3];
-+	char *l, *p, *p2, *res;
-+	int len = 0, _pos = 0, i;
-+
-+	l = line;
-+	while (l != NULL) {
-+		if ((p = strstr(l, SSI_START)) == NULL) {
-+			len += strlen((tok[_pos++] = l));
++	var *v;
++	char *s;
++	if (!e || !e->var)
++		return;
++	lineno = e->line;
++	switch (e->t) {
++		case T_TEXT:
++			s = strdup(e->var);
++			print_translate(s);
++			free(s);
 +			break;
-+		}
-+
-+		p2 = strstr(p, SSI_END);
-+		if (p2 == NULL) {
-+			fprintf(stderr, "Parse error in '%s', line '%d', unmatched %s\n", fname, lnr, SSI_END);
++		case T_CODE:
++			s = strdup(e->var);
++			v = parse_awk(s);
++			free(s);
++			s = strdup(getvar_s(v));
++			print_translate(s);
++			free(s);
++			break;
++		case T_IF:
++			s = strdup(e->var);
++			v = parse_awk(s);
++			free(s);
++			if (!v)
++				return;
++			if (istrue(v))
++				execute_template(tcb, e->sub);
++			else if (e->sub2)
++				execute_template(tcb, e->sub2);
++
++			clrvar(v);
++			break;
++		case T_FOR: {
++				v = newvar(e->var);
++				hashwalk_init(v, iamarray(findvar(vhash, e->in)));
++				while (hashwalk_next(v)) {
++					execute_template(tcb, e->sub);
++				}
++			}
++			break;
++		default:
 +			break;
-+		}
-+
-+		*p = 0;
-+		*p2 = 0;
-+		
-+		len += strlen((tok[_pos++] = l));
-+		len += strlen((tok[_pos++] = render_lookup(fname, lnr, p + strlen(SSI_START))));
-+
-+		l = p2;
-+		l += strlen(SSI_END);
-+	}
-+	len++;
-+
-+	p = xmalloc(len + 1);
-+	*p = 0;
-+	res = p;
-+	for (i = 0; i < _pos; i++) {
-+		strcat(p, tok[i]);
-+		p += strlen(tok[i]);
 +	}
-+	print_translate(res);
-+	free(res);
 +}
 +
 +/* awk method render(), which opens a template file and processes all awk ssi calls */
 +static void render_file(char *filename)
 +{
-+	int lnr = 0;
++	struct template_cb tcb;
++	struct template_element *e;
 +	FILE *f;
-+	char *buf1;
++	char *oldprg;
++	int oldlnr;
++	
++	oldlnr = lineno;
++	oldprg = programname;
++	programname = filename;
 +			
 +	f = fopen(filename, "r");
 +	if (!f)
 +		return;
 +	
-+	buf1 = xmalloc(LINE_BUF);
-+	while (!feof(f) && (fgets(buf1, LINE_BUF - 1, f) != NULL)) {
-+		render_line(filename, ++lnr, buf1);
-+	}
++	memset(&tcb, 0, sizeof(tcb));
++	tcb.handle_element = render_element;
++	e = parse_template(&tcb, f);
++	execute_template(&tcb, e);
++	free_template(&tcb, e);
++	fclose(f);
++	programname = oldprg;
++	lineno = oldlnr;
 +}
 +
 +static var *render(var *res, var *args, int nargs)
@@ -689,9 +724,357 @@ diff -purN bb.old/editors/awx.c bb.dev/editors/awx.c
 +	return awk_main(argc, argv);
 +}
 +
+diff -purN bb.old/editors/awx_parser.h bb.dev/editors/awx_parser.h
+--- bb.old/editors/awx_parser.h	1970-01-01 01:00:00.000000000 +0100
++++ bb.dev/editors/awx_parser.h	2007-05-20 08:30:30.010685144 +0200
+@@ -0,0 +1,38 @@
++#ifndef __TEMPLATE_PARSER_H
++#define __TEMPLATE_PARSER_H
++
++enum type {
++	T_TEXT,
++	T_FOR,
++	T_IF,
++	T_CODE
++};
++
++struct template_element;
++struct template_cb;
++
++struct template_cb {
++	void *(*prepare_code)(struct template_element *);
++	void (*handle_element)(struct template_cb *, struct template_element *);
++	void (*free_code)(struct template_element *);
++};
++
++struct template_element {
++	enum type t;
++	char *var;
++	char *in;
++	int line;
++	void *priv;
++	struct template_element *parent;
++	struct template_element *sub;
++	struct template_element *sub2;
++	struct template_element *prev;
++	struct template_element *next;
++};
++
++
++struct template_element *parse_template(struct template_cb *cb, FILE *in);
++void execute_template(struct template_cb *cb, struct template_element *e);
++void free_template(struct template_cb *cb, struct template_element *e);
++
++#endif
+diff -purN bb.old/editors/awx_parser.l bb.dev/editors/awx_parser.l
+--- bb.old/editors/awx_parser.l	1970-01-01 01:00:00.000000000 +0100
++++ bb.dev/editors/awx_parser.l	2007-05-20 08:30:55.698779960 +0200
+@@ -0,0 +1,302 @@
++%{
++#include <stdio.h>
++#include <string.h>
++#include <stdlib.h>
++#include "busybox.h"
++#include "awx_parser.h"
++
++enum {
++	S_INIT,
++	S_TEXT,
++	S_CODE,
++	S_IF_START,
++	S_FOR_START,
++	S_FOR_IN,
++	S_END,
++	S_ELSE,
++	S_EOF
++}; 
++int state;
++
++#undef DEBUG
++#ifdef DEBUG
++char *statestr[] = {
++	[S_INIT] = "S_INIT",
++	[S_TEXT] = "S_TEXT",
++	[S_CODE] = "S_CODE",
++	[S_IF_START] = "S_IF_START",
++	[S_FOR_START] = "S_FOR_START",
++	[S_FOR_IN] = "S_FOR_IN",
++	[S_EOF] = "S_EOF"
++};
++
++char *typestr[] = {
++	[T_TEXT] = "T_TEXT",
++	[T_FOR] = "T_FOR",
++	[T_IF] = "T_IF",
++	[T_CODE] = "T_CODE"
++};
++#endif
++
++static struct template_cb *parse_cb;
++static struct template_element *cur, *head;
++static char *textbuf;
++static unsigned int buflen;
++static unsigned int buf_offset;
++static int _lnr = 0;
++
++static void buf_realloc(void)
++{
++	buflen *= 2;
++	textbuf = xrealloc(textbuf, buflen);
++}
++
++static void parse_error(char *str)
++{
++	fprintf(stderr, "Parse error%s%s\n", (str ? ": " : "!"), (str ?: ""));
++	exit(255);
++}
++
++
++static struct template_element *new_template_element(struct template_element *parent)
++{
++	struct template_element *ptr;
++	
++	ptr = xzalloc(sizeof(*ptr));
++	ptr->parent = parent;
++	return ptr;
++}
++
++static inline void next_template_element(void)
++{
++	cur->next = new_template_element(cur->parent);
++	cur->next->prev = cur;
++	cur = cur->next;
++}
++
++static void addtext(char *text)
++{
++	while(buf_offset + strlen(text) + 1 > buflen)
++		buf_realloc();
++
++	buf_offset += sprintf(&textbuf[buf_offset], "%s", text);
++}
++
++static void set_state(int newstate)
++{
++	char *ptr;
++
++#ifdef DEBUG
++	static int _rec = 0;
++	fprintf(stderr, "DEBUG(%d): %s => %s: %s\n", _rec, statestr[state], statestr[newstate], textbuf);
++#endif
++	ptr = xstrdup(textbuf);
++	if (state == S_FOR_IN)
++		cur->in = ptr;
++	else
++		cur->var = ptr;
++
++	if (parse_cb && (cur->t == T_CODE) && parse_cb->prepare_code)
++		parse_cb->prepare_code(cur);
++
++	buf_offset = 0;
++	*textbuf = 0;
++
++	switch(newstate) {
++#if 0
++		case S_EOF:
++			if (cur->parent)
++				parse_error();
++			break;
++#endif
++		case S_FOR_START:
++			if (ptr || !cur->prev)
++				next_template_element();
++			cur->t = T_FOR;
++			break;
++		case S_IF_START:
++			if (ptr || !cur->prev)
++				next_template_element();
++			cur->t = T_IF;
++			break;
++		case S_ELSE:
++			cur = cur->parent;
++			if (!cur)
++				parse_error("'@else' without parent element");
++			cur->sub2 = new_template_element(cur);
++			cur = cur->sub2;
++			newstate = S_TEXT;
++			break;
++		case S_END:
++#ifdef DEBUG
++			_rec--;
++#endif
++			cur = cur->parent;
++			if (!cur) 
++				parse_error("'@end' without parent element");
++
++			next_template_element();
++			cur->t = T_TEXT;
++			newstate = S_TEXT;
++			break;
++		case S_TEXT:
++			switch (cur->t) {
++				case T_CODE:
++					next_template_element();
++					break;
++				case T_IF:
++				case T_FOR:
++#ifdef DEBUG
++					_rec++;
++#endif
++					cur->sub = new_template_element(cur);
++					cur = cur->sub;
++					break;
++				default:
++					break;
++			}
++			cur->t = T_TEXT;
++			break;
++		case S_CODE:
++			if (ptr || !cur->prev)
++				next_template_element();
++			cur->t = T_CODE;
++			break;
++		default:
++			break;
++	}
++	cur->line = _lnr;
++	state = newstate;
++}
++
++%}
++
++%%
++"<%"[ \n\t]*"@if"[ \n\t]+ {
++	if (state == S_TEXT) 
++		set_state(S_IF_START);
++	else
++		REJECT;
++}
++
++"<%"[ \n\t]*"@for"[ \n\t]+ {
++	if (state == S_TEXT) 
++		set_state(S_FOR_START);
++	else
++		REJECT;
++}
++
++[ \n\t]+"in"[ \n\t]+ {
++	if (state == S_FOR_START)
++		set_state(S_FOR_IN);
++	else
++		REJECT;
++}
++
++"<%"[ \n\t]*"@end"[ \n\t]*%> {
++	if (state != S_TEXT)
++		REJECT;
++	set_state(S_END);
++}
++
++"<%"[ \n\t]*"@else"[ \n\t]*%> {
++	if (state != S_TEXT)
++		REJECT;
++	set_state(S_ELSE);
++}
++
++"<%" {
++	if (state != S_TEXT) 
++		parse_error("'<%' cannot be nested");
++	set_state(S_CODE);
++}
++
++[ \n\t]"%>" {
++	if (state == S_TEXT)
++		REJECT;
++	set_state(S_TEXT);
++}
++
++\n {
++	_lnr++;
++	if (state == S_TEXT)
++		addtext(yytext);
++}
++.	{
++	addtext(yytext);
++}
++
++
++%%
++
++
++void execute_template(struct template_cb *cb, struct template_element *e)
++{
++	static int rec = 0;
++
++	while (e) {
++#ifdef DEBUG
++		fprintf(stderr, "DEBUG: execute(%d)\t%s\n", rec, typestr[e->t]);
++#endif
++		rec++;
++		if (cb->handle_element)
++			cb->handle_element(cb, e);
++		rec--;
++		e = e->next;
++	}
++}
++
++int yywrap()
++{
++	set_state(S_EOF);
++	return 1;
++}
++
++struct template_element *parse_template(struct template_cb *cb, FILE *in)
++{
++	_lnr = 1;
++	buf_offset = 0;
++	state = S_TEXT;
++	parse_cb = cb;
++	
++	buflen = 4096;
++	textbuf = xzalloc(buflen);
++
++	head = xzalloc(sizeof(*head));
++	head->t = T_TEXT;
++	cur = head;
++
++	yyin = in;
++	yylex();
++
++	return head;
++}
++
++void free_template(struct template_cb *cb, struct template_element *e)
++{
++	struct template_element *next;
++
++	if (!e)
++		return;
++		
++	switch (e->t) {
++		case T_CODE:
++			if (cb->free_code)
++				cb->free_code(e);
++			break;
++		case T_FOR:
++		case T_IF:
++			free_template(cb, e->sub);
++			break;
++		default:
++			break;
++	}
++	if (e->var)
++		free(e->var);
++	if (e->in)
++		free(e->in);
++		
++	next = e->next;
++	free(e);
++	return free_template(cb, next);
++}
 diff -purN bb.old/editors/Config.in bb.dev/editors/Config.in
---- bb.old/editors/Config.in	2007-01-24 22:34:50.000000000 +0100
-+++ bb.dev/editors/Config.in	2007-03-11 06:19:51.469380160 +0100
+--- bb.old/editors/Config.in	2007-05-20 04:17:05.003197632 +0200
++++ bb.dev/editors/Config.in	2007-05-20 04:16:47.180907032 +0200
 @@ -12,6 +12,13 @@ config AWK
  	  Awk is used as a pattern scanning and processing language.  This is
  	  the BusyBox implementation of that programming language.
@@ -706,9 +1089,25 @@ diff -purN bb.old/editors/Config.in bb.dev/editors/Config.in
  config FEATURE_AWK_MATH
  	bool "Enable math functions (requires libm)"
  	default y
+diff -purN bb.old/editors/Kbuild bb.dev/editors/Kbuild
+--- bb.old/editors/Kbuild	2007-03-18 17:59:37.000000000 +0100
++++ bb.dev/editors/Kbuild	2007-05-20 05:13:45.363264328 +0200
+@@ -10,3 +10,12 @@ lib-$(CONFIG_ED)        += ed.o
+ lib-$(CONFIG_PATCH)     += patch.o
+ lib-$(CONFIG_SED)       += sed.o
+ lib-$(CONFIG_VI)        += vi.o
++lib-$(CONFIG_AWX)		+= awx_parser.o
++
++editors/awx_parser.c: editors/awx_parser.l editors/awx_parser.h
++	@flex $<
++	@mv lex.yy.c $@
++
++editors/awx_parser.o: editors/awx_parser.c FORCE
++	$(call cmd,force_checksrc)
++	$(call if_changed_rule,cc_o_c)
 diff -purN bb.old/include/applets.h bb.dev/include/applets.h
---- bb.old/include/applets.h	2007-03-06 19:38:07.355081000 +0100
-+++ bb.dev/include/applets.h	2007-03-07 02:12:24.280681880 +0100
+--- bb.old/include/applets.h	2007-05-20 04:17:05.003197632 +0200
++++ bb.dev/include/applets.h	2007-05-20 04:16:47.180907032 +0200
 @@ -60,6 +60,7 @@ USE_ARP(APPLET(arp, _BB_DIR_SBIN, _BB_SU
  USE_ARPING(APPLET(arping, _BB_DIR_USR_BIN, _BB_SUID_NEVER))
  USE_ASH(APPLET_NOUSAGE(ash, ash, _BB_DIR_BIN, _BB_SUID_NEVER))
@@ -719,7 +1118,7 @@ diff -purN bb.old/include/applets.h bb.dev/include/applets.h
  //USE_BBSH(APPLET(bbsh, _BB_DIR_BIN, _BB_SUID_NEVER))
 diff -purN bb.old/include/cgi.h bb.dev/include/cgi.h
 --- bb.old/include/cgi.h	1970-01-01 01:00:00.000000000 +0100
-+++ bb.dev/include/cgi.h	2007-03-11 18:58:10.708444448 +0100
++++ bb.dev/include/cgi.h	2007-05-20 04:16:47.180907032 +0200
 @@ -0,0 +1,8 @@
 +#ifndef CGI_H
 +#define CGI_H
@@ -731,7 +1130,7 @@ diff -purN bb.old/include/cgi.h bb.dev/include/cgi.h
 +#endif
 diff -purN bb.old/libbb/cgi.c bb.dev/libbb/cgi.c
 --- bb.old/libbb/cgi.c	1970-01-01 01:00:00.000000000 +0100
-+++ bb.dev/libbb/cgi.c	2007-03-11 19:02:04.691873560 +0100
++++ bb.dev/libbb/cgi.c	2007-05-20 04:16:47.181906880 +0200
 @@ -0,0 +1,457 @@
 +/* --------------------------------------------------------------------------
 + * functions for processing cgi form data
@@ -1191,8 +1590,8 @@ diff -purN bb.old/libbb/cgi.c bb.dev/libbb/cgi.c
 +	return retval;
 +} 
 diff -purN bb.old/libbb/Kbuild bb.dev/libbb/Kbuild
---- bb.old/libbb/Kbuild	2007-03-06 19:38:07.361080000 +0100
-+++ bb.dev/libbb/Kbuild	2007-03-11 18:40:51.384445712 +0100
+--- bb.old/libbb/Kbuild	2007-05-20 04:17:05.004197480 +0200
++++ bb.dev/libbb/Kbuild	2007-05-20 04:16:47.181906880 +0200
 @@ -118,3 +118,6 @@ lib-$(CONFIG_GREP) += xregcomp.o
  lib-$(CONFIG_MDEV) += xregcomp.o
  lib-$(CONFIG_LESS) += xregcomp.o