Merge pull request #5 from Zaryob/dev

Merge from "dev"

.github/workflows/iksemel_basic.yml

@@ -14,7 +14,7 @@ jobs:
     - name: Install build dependencies
       run: |
         sudo apt-get update
-        sudo apt-get install --yes -qq build-essential autoconf libtool git python3 python3-dev python3-pip python3-setuptools
+        sudo apt-get install --yes -qq build-essential autoconf libtool git python3 python3-dev python3-setuptools gnutls-dev gnutls-bin valgrind
     - name: Install meson via pip
       run: |
         sudo pip3 install meson
@@ -28,6 +28,10 @@ jobs:
     - name: Ninja tests
       run: |
         ninja test -C build
+    - name: Valgrind tests
+      run: |
+        cd build 
+        meson test --wrap='valgrind --leak-check=full --error-exitcode=1' --no-rebuild 
     - name: Prepare artifacts
       if: failure()
       run: |
@@ -38,5 +42,6 @@ jobs:
         name: Logs
         path: |
           /home/runner/work/iksemel/iksemel/build/meson-logs/meson-log.txt
+          /home/runner/work/iksemel/iksemel/build/meson-logs/testlog-valgrind.txt
         if-no-files-found: ignore
 

.gitignore

@@ -43,3 +43,8 @@ src/iksemel.egg-info/dependency_links.txt
 src/iksemel.egg-info/SOURCES.txt
 src/iksemel.egg-info/PKG-INFO
 src/iksemel.egg-info/top_level.txt
+
+*.patch
+
+# VSCode
+.vscode/

include/common.h

@@ -38,4 +38,4 @@ extern int errno;
 
 #include "finetune.h"
 
-#endif // __COMMON_H
+#endif // __COMMON_H

include/iksemel.h

@@ -102,7 +102,7 @@ enum ikserror {
 };
 
 enum ikstagtype {
-	IKS_OPEN,
+	IKS_OPEN = 0,
 	IKS_CLOSE,
 	IKS_SINGLE
 };

meson.build

@@ -70,7 +70,7 @@ pkgconf.set('VERSION', version)
 
 pkg_install_dir = '@0@/pkgconfig'.format(get_option('libdir'))
 
-configure_file(input : 'iksemel.pc.in'.format(version),
+configure_file(input : 'iksemel.pc.in',
   output : 'iksemel-@[email protected]'.format(version),
   configuration : pkgconf,
   install_dir : pkg_install_dir)

src/iks.c

@@ -4,6 +4,8 @@
 ** modify it under the terms of GNU Lesser General Public License.
 */
 
+#include <ctype.h>
+#include <stdbool.h>
 #include "common.h"
 #include "iksemel.h"
 
@@ -562,6 +564,11 @@ iks_has_attribs (iks *x)
 }
 
 /*****  Serializing  *****/
+static bool
+is_print(char c)
+{
+	return isprint(c) ||  c == '\t' || c == '\n' || c == '\r';
+}
 
 static size_t
 escape_size (char *src, size_t len)
@@ -573,6 +580,43 @@ escape_size (char *src, size_t len)
 	sz = 0;
 	for (i = 0; i < len; i++) {
 		c = src[i];
+
+
+		if( c < 0 || c > 127 ){
+
+			// Convert UTF-8 bytes to Unicode code point
+			unsigned char *ptr = (unsigned char *)(src+i);
+
+			if( *ptr >= 0x80 &&  *ptr <= 0x9F ){
+				continue;
+			}
+			else if ((*ptr & 0xE8) == 0xC0) {
+				sz += 2 * 1 + 4;
+				i++;
+				continue;
+			}
+			else if ((*ptr &  0xF0) == 0xE0) {
+				sz += 2 * 3 + 4;
+				i+=2;
+				continue;
+			}
+			else if ((*ptr & 0xF8) == 0xF0) {
+				sz += 2 * 4 + 4;
+				i+=3;
+				continue;
+			}
+			else {
+			    sz += 6;
+			    continue;
+			}
+
+		}
+
+		if (!is_print(c)) {
+			sz += 6;
+			continue;
+		}
+
 		switch (c) {
 			case '&': sz += 5; break;
 			case '\'': sz += 6; break;
@@ -597,11 +641,72 @@ static char *
 escape (char *dest, char *src, size_t len)
 {
 	char c;
-	int i;
-	int j = 0;
+
+	size_t i;
+	size_t j = 0;
+
 
 	for (i = 0; i < len; i++) {
 		c = src[i];
+
+		// handle non-ascii characters
+		if (!is_print(c)) {
+
+			if (i - j > 0) dest = my_strcat(dest, src + j, i - j);
+			j = i + 1;
+
+			char buf[13] = {0};
+
+			if( c < 0 || c > 127){
+				// Convert UTF-8 bytes to Unicode code point
+
+				unsigned int unicode_code_point = 0;
+				unsigned char *ptr = (unsigned char *)(src+i);
+
+				// Invalid characters
+				if( *ptr >= 0x80 &&  *ptr <= 0x9F ){
+					continue;
+				}
+				// 2 bytes
+				else if ((*ptr & 0xE8) == 0xC0) {
+					unicode_code_point = ((*ptr & 0x1F) << 6) | (*(ptr + 1) & 0x3F);
+					dest = my_strcat(dest, buf, snprintf(buf, sizeof buf, "&#x%02x;", unicode_code_point));
+					i++;
+					j = i + 1;
+					continue;
+				}
+				// 3 bytes
+				else if ((*ptr & 0xF0) == 0xE0) {
+					unicode_code_point = ((*ptr & 0x0F) << 12) | ((*(ptr + 1) & 0x3F) << 6) | (*(ptr + 2) & 0x3F);
+					dest = my_strcat(dest, buf, snprintf(buf, sizeof buf, "&#x%02x;", unicode_code_point));
+					i+=2;
+					j = i + 1;
+					continue;
+				}
+				// 4 bytes
+				else if ((*ptr & 0xF8) == 0xF0) {
+					unicode_code_point = ((*ptr & 0x07) << 18) | ((*(ptr + 1) & 0x3F) << 12) | ((*(ptr + 2) & 0x3F) << 6) | (*(ptr + 3) & 0x3F);
+					dest = my_strcat(dest, buf, snprintf(buf, sizeof buf, "&#x%02x;", unicode_code_point));
+					i+=3;
+					j = i + 1;
+					continue;
+				}
+				else {
+					dest = my_strcat(dest, buf, snprintf(buf, sizeof buf, "&#x%02x;", c));
+					continue;
+				}
+
+
+			}
+			else{
+				if(c != 0x00){
+					dest = my_strcat(dest, buf, snprintf(buf, sizeof buf, "&#x%02x;", c));
+					continue;
+				}
+			}
+
+		}
+
 		if ('&' == c || '<' == c || '>' == c || '\'' == c || '"' == c) {
 			if (i - j > 0) dest = my_strcat (dest, src + j, i - j);
 			j = i + 1;
@@ -632,7 +737,7 @@ iks_string (ikstack *s, iks *x)
 		if (s) {
 			return iks_stack_strdup (s, IKS_CDATA_CDATA (x), IKS_CDATA_LEN (x));
 		} else {
-			ret = iks_malloc (IKS_CDATA_LEN (x) + 1);
+			ret = iks_malloc (IKS_CDATA_LEN (x) + 1) ;
 			memcpy (ret, IKS_CDATA_CDATA (x), IKS_CDATA_LEN (x) + 1);
 			return ret;
 		}

src/ikstack.c

@@ -7,25 +7,28 @@
 #include "common.h"
 #include "iksemel.h"
 
-struct align_test { char a; double b; };
+typedef double align_type ;
+struct align_test { char a; align_type b; };
 #define DEFAULT_ALIGNMENT  ((size_t) ((char *) &((struct align_test *) 0)->b - (char *) 0))
 #define ALIGN_MASK ( DEFAULT_ALIGNMENT - 1 )
 #define MIN_CHUNK_SIZE ( DEFAULT_ALIGNMENT * 8 )
 #define MIN_ALLOC_SIZE DEFAULT_ALIGNMENT
-#define ALIGN(x) ( (x) + (DEFAULT_ALIGNMENT - ( (x) & ALIGN_MASK)) )
+#define ALIGN(x) ( ((x) + DEFAULT_ALIGNMENT - 1) & ~ALIGN_MASK )
 
 typedef struct ikschunk_struct {
 	struct ikschunk_struct *next;
 	size_t size;
 	size_t used;
 	size_t last;
+	align_type align[0] ;	// Align data, and ensure struct size matches alignment
 	char data[4];
 } ikschunk;
 
 struct ikstack_struct {
 	size_t allocated;
 	ikschunk *meta;
 	ikschunk *data;
+	align_type align[0] ;	// Ensure struct size matches alignment
 };
 
 static ikschunk *

src/meson.build

@@ -83,7 +83,7 @@ if backends.length() != 0
 
   pkg_install_dir = '@0@/pkgconfig'.format(get_option('libdir'))
 
-  configure_file(input : 'jabber.pc.in'.format(version),
+  configure_file(input : 'jabber.pc.in',
     output : 'jabber-@[email protected]'.format(version),
     configuration : jabber_pkgconf,
     install_dir : pkg_install_dir)

src/sax.c

@@ -162,7 +162,7 @@ stack_expand (iksparser *prs, int len)
 	prs->stack_max = need;
 	prs->tag_name += diff;
 	if (prs->attflag != 0) {
-		int i = 0;
+		unsigned int i = 0;
 		while (i < (prs->attmax * 2)) {
 			if (prs->atts[i]) prs->atts[i] += diff;
 			i++;

src/utility.c

@@ -100,7 +100,7 @@ char *
 iks_escape (ikstack *s, char *src, size_t len)
 {
 	char *ret;
-	int i, j, nlen;
+	size_t i, j, nlen;
 
 	if (!src || !s) return NULL;
 	if (len == -1) len = strlen (src);
@@ -138,7 +138,7 @@ iks_escape (ikstack *s, char *src, size_t len)
 char *
 iks_unescape (ikstack *s, char *src, size_t len)
 {
-	int i,j;
+	size_t i,j;
 	char *ret;
 
 	if (!s || !src) return NULL;

test/meson.build

@@ -6,6 +6,10 @@ iks_test = executable('test-iks', 'tst-iks.c',
                     include_directories : include_dir,
                     link_with : libiksemel)
 
+iks_test_utf8 = executable('test-iks-utf8', 'tst-iks-utf8.c',
+                           include_directories : include_dir,
+                           link_with : libiksemel)
+
 ikstack_test = executable('test-ikstack', 'tst-ikstack.c',
                     include_directories : include_dir,
                     link_with : libiksemel)
@@ -24,6 +28,7 @@ sha_test = executable('test-sha', 'tst-sha.c',
 
 test('Iksemel Test Suite - DOM Test', dom_test, is_parallel : true)
 test('Iksemel Test Suite - IKS Test', iks_test, is_parallel : true)
+test('Iksemel Test Suite - IKS UTF-8 Test', iks_test_utf8, is_parallel : true)
 test('Iksemel Test Suite - IkStack Test', ikstack_test, is_parallel : true)
 test('Iksemel Test Suite - MD5 Test', md5_test, is_parallel : true)
 test('Iksemel Test Suite - SAX Test', sax_test, is_parallel : true)

test/tst-iks-utf8.c

@@ -0,0 +1,48 @@
+/* iksemel (XML parser for Jabber)
+** Copyright (C) 2000-2003 Gurer Ozen
+** This code is free software; you can redistribute it and/or
+** modify it under the terms of GNU Lesser General Public License.
+*/
+
+#include <stdio.h>
+#include <string.h>
+#include <locale.h>
+
+#include "iksemel.h"
+
+int main (int argc, char *argv[])
+{
+	setlocale (LC_ALL, "");
+	static char xml[] = "<test>"
+						"<text>Hello, &#x4e16;&#x754c;</text>"
+						"<emoji>&#x1f4a9;</emoji>"
+						"<invalid></invalid>"
+						"<null></null>"
+						"<twoun>&#xc5;</twoun>"
+						"<threeun>&#x100;</threeun>"
+						"<katana>&#x30bb;</katana>"
+						"<wideunicode>&#x26007;</wideunicode>"
+						"<nonprint>&#x01;&#x07;&#x0b;&#x7f;</nonprint>"
+						"</test>";
+
+	iks *x = iks_new ("test");
+	iks_insert_cdata (iks_insert (x, "text"), "Hello, 世界", 13);
+	iks_insert_cdata (iks_insert (x, "emoji"), "\U0001F4A9", 4);
+	iks_insert_cdata (iks_insert (x, "invalid"), "\x80\x81", 2);
+	iks_insert_cdata (iks_insert (x, "null"), "\0", 1);
+	iks_insert_cdata (iks_insert (x, "twoun"), "Å", 3);
+	iks_insert_cdata (iks_insert (x, "threeun"), "Ā", 3);
+	iks_insert_cdata (iks_insert (x, "katana"), "セ", 4);
+	iks_insert_cdata (iks_insert (x, "wideunicode"), "\U00026007", 4);
+	iks_insert_cdata (iks_insert (x, "nonprint"), "\x1\a\v\x7F", 4);
+
+	char *t = iks_string (iks_stack (x), x);
+	if(!t || strcmp(t, xml) != 0) {
+		printf("Result:   %s\n", t);
+		printf("Expected: %s\n", xml);
+		return 1;
+	}
+	iks_delete(x);
+
+	return 0;
+}

test/tst-sax.c

@@ -152,7 +152,7 @@ tagHook (void *udata, char *name, char **atts, int type)
 
 	if (!tester.cur) TAG_FAIL;
 	if (tester.cur->type != IKS_TAG) TAG_FAIL;
-	if (tester.cur->tag != type) TAG_FAIL;
+	if ((int)tester.cur->tag != type) TAG_FAIL;
 	if (iks_strcmp (tester.cur->name, name) != 0) TAG_FAIL;
 	if (!atts && tester.cur->nr_atts > 0) TAG_FAIL;
 	if (atts && tester.cur->nr_atts == 0) TAG_FAIL;
@@ -177,7 +177,7 @@ tagHook (void *udata, char *name, char **atts, int type)
 void
 debug_cdata (char *data, size_t len, int pos)
 {
-	int i;
+	size_t i;
 
 	PRINT_TEST;
 	if (tester.cur && tester.cur->type == IKS_CDATA)