diff --git a/.vs/rufus.vcxproj b/.vs/rufus.vcxproj
index 99adc9f0..f1047b25 100644
--- a/.vs/rufus.vcxproj
+++ b/.vs/rufus.vcxproj
@@ -410,6 +410,7 @@
+
@@ -446,6 +447,7 @@
+
diff --git a/.vs/rufus.vcxproj.filters b/.vs/rufus.vcxproj.filters
index cd02009c..7a980a63 100644
--- a/.vs/rufus.vcxproj.filters
+++ b/.vs/rufus.vcxproj.filters
@@ -93,6 +93,9 @@
Source Files
+
+ Source Files
+
@@ -197,6 +200,9 @@
Header Files
+
+ Header Files
+
diff --git a/src/Makefile.am b/src/Makefile.am
index 59cf87f7..ab03cd2b 100644
--- a/src/Makefile.am
+++ b/src/Makefile.am
@@ -18,7 +18,7 @@ AM_V_WINDRES = $(AM_V_WINDRES_$(V))
$(AM_V_WINDRES) $(AM_RCFLAGS) -i $< -o $@
rufus_SOURCES = badblocks.c dev.c dos.c dos_locale.c drive.c format.c format_ext.c format_fat32.c hash.c icon.c iso.c \
- localization.c net.c parser.c pki.c process.c re.c rufus.c smart.c stdfn.c stdio.c stdlg.c syslinux.c ui.c vhd.c wue.c
+ localization.c net.c parser.c pki.c process.c re.c rufus.c smart.c stdfn.c stdio.c stdlg.c syslinux.c ui.c vhd.c wue.c xml.c
rufus_CFLAGS = -I$(srcdir)/ms-sys/inc -I$(srcdir)/syslinux/libfat -I$(srcdir)/syslinux/libinstaller -I$(srcdir)/syslinux/win -I$(srcdir)/libcdio -I$(srcdir)/../res $(AM_CFLAGS) \
-DEXT2_FLAT_INCLUDES=0 -DSOLUTION=rufus
rufus_LDFLAGS = $(AM_LDFLAGS) -mwindows -L $(srcdir)/../.mingw
diff --git a/src/Makefile.in b/src/Makefile.in
index c0fa14e6..5f27da84 100644
--- a/src/Makefile.in
+++ b/src/Makefile.in
@@ -98,7 +98,8 @@ am_rufus_OBJECTS = rufus-badblocks.$(OBJEXT) rufus-dev.$(OBJEXT) \
rufus-rufus.$(OBJEXT) rufus-smart.$(OBJEXT) \
rufus-stdfn.$(OBJEXT) rufus-stdio.$(OBJEXT) \
rufus-stdlg.$(OBJEXT) rufus-syslinux.$(OBJEXT) \
- rufus-ui.$(OBJEXT) rufus-vhd.$(OBJEXT) rufus-wue.$(OBJEXT)
+ rufus-ui.$(OBJEXT) rufus-vhd.$(OBJEXT) rufus-wue.$(OBJEXT) \
+ rufus-xml.$(OBJEXT)
rufus_OBJECTS = $(am_rufus_OBJECTS)
am__DEPENDENCIES_1 =
rufus_DEPENDENCIES = rufus_rc.o bled/libbled.a ext2fs/libext2fs.a \
@@ -285,7 +286,7 @@ AM_V_WINDRES_1 = $(WINDRES)
AM_V_WINDRES_ = $(AM_V_WINDRES_$(AM_DEFAULT_VERBOSITY))
AM_V_WINDRES = $(AM_V_WINDRES_$(V))
rufus_SOURCES = badblocks.c dev.c dos.c dos_locale.c drive.c format.c format_ext.c format_fat32.c hash.c icon.c iso.c \
- localization.c net.c parser.c pki.c process.c re.c rufus.c smart.c stdfn.c stdio.c stdlg.c syslinux.c ui.c vhd.c wue.c
+ localization.c net.c parser.c pki.c process.c re.c rufus.c smart.c stdfn.c stdio.c stdlg.c syslinux.c ui.c vhd.c wue.c xml.c
rufus_CFLAGS = -I$(srcdir)/ms-sys/inc -I$(srcdir)/syslinux/libfat -I$(srcdir)/syslinux/libinstaller -I$(srcdir)/syslinux/win -I$(srcdir)/libcdio -I$(srcdir)/../res $(AM_CFLAGS) \
-DEXT2_FLAT_INCLUDES=0 -DSOLUTION=rufus
@@ -504,6 +505,12 @@ rufus-wue.o: wue.c
rufus-wue.obj: wue.c
$(AM_V_CC)$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(rufus_CFLAGS) $(CFLAGS) -c -o rufus-wue.obj `if test -f 'wue.c'; then $(CYGPATH_W) 'wue.c'; else $(CYGPATH_W) '$(srcdir)/wue.c'; fi`
+rufus-xml.o: xml.c
+ $(AM_V_CC)$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(rufus_CFLAGS) $(CFLAGS) -c -o rufus-xml.o `test -f 'xml.c' || echo '$(srcdir)/'`xml.c
+
+rufus-xml.obj: xml.c
+ $(AM_V_CC)$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(rufus_CFLAGS) $(CFLAGS) -c -o rufus-xml.obj `if test -f 'xml.c'; then $(CYGPATH_W) 'xml.c'; else $(CYGPATH_W) '$(srcdir)/xml.c'; fi`
+
# This directory's subdirectories are mostly independent; you can cd
# into them and run 'make' without going through this Makefile.
# To change the values of 'make' variables: instead of editing Makefiles,
diff --git a/src/license.h b/src/license.h
index 9a59e9f5..a14d7131 100644
--- a/src/license.h
+++ b/src/license.h
@@ -81,6 +81,10 @@ const char* additional_copyrights =
"http://www.ridgecrop.demon.co.uk/index.htm?fat32format.htm\\line\n"
"GNU General Public License (GPL) v2 or later\\line\n"
"\\line\n"
+"XML parsing support from ezxml by Aaron Voisine:\\line\n"
+"https://ezxml.sourceforge.net/\\line\n"
+"MIT license\\line\n"
+"\\line\n"
"fmifs.dll usage based on Formatx by Mark Russinovich:\\line\n"
"https://svn.reactos.org/reactos/trunk/reactos/dll/win32/fmifs\\line\n"
"Public Domain\\line\n"
diff --git a/src/msapi_utf8.h b/src/msapi_utf8.h
index e6ab9ad2..3daef720 100644
--- a/src/msapi_utf8.h
+++ b/src/msapi_utf8.h
@@ -55,9 +55,9 @@ extern "C" {
#define LTEXT(txt) _LTEXT(txt)
#define wchar_to_utf8_no_alloc(wsrc, dest, dest_size) \
- WideCharToMultiByte(CP_UTF8, 0, wsrc, -1, dest, dest_size, NULL, NULL)
+ WideCharToMultiByte(CP_UTF8, 0, wsrc, -1, dest, (int)(dest_size), NULL, NULL)
#define utf8_to_wchar_no_alloc(src, wdest, wdest_size) \
- MultiByteToWideChar(CP_UTF8, 0, src, -1, wdest, wdest_size)
+ MultiByteToWideChar(CP_UTF8, 0, src, -1, wdest, (int)(wdest_size))
#define Edit_ReplaceSelU(hCtrl, str) ((void)SendMessageLU(hCtrl, EM_REPLACESEL, (WPARAM)FALSE, str))
#define ComboBox_AddStringU(hCtrl, str) ((int)(DWORD)SendMessageLU(hCtrl, CB_ADDSTRING, (WPARAM)FALSE, str))
#define ComboBox_InsertStringU(hCtrl, index, str) ((int)(DWORD)SendMessageLU(hCtrl, CB_INSERTSTRING, (WPARAM)index, str))
@@ -83,7 +83,7 @@ extern "C" {
#define wfree(p) sfree(w ## p)
/*
- * Converts an UTF-16 string to UTF8 (allocate returned string)
+ * Converts an UTF-16 string to UTF8 (allocates returned string)
* Returns NULL on error
*/
static __inline char* wchar_to_utf8(const wchar_t* wstr)
@@ -115,7 +115,7 @@ static __inline char* wchar_to_utf8(const wchar_t* wstr)
}
/*
- * Converts an UTF8 string to UTF-16 (allocate returned string)
+ * Converts an UTF8 string to UTF-16 (allocates returned string)
* Returns NULL on error
*/
static __inline wchar_t* utf8_to_wchar(const char* str)
@@ -146,7 +146,7 @@ static __inline wchar_t* utf8_to_wchar(const char* str)
}
/*
-* Converts an non NUL-terminated UTF-16 string of length len to UTF8 (allocate returned string)
+* Converts a non NUL-terminated UTF-16 string of length wlen to NUL-terminated UTF8 (allocates returned string)
* Returns NULL on error
*/
static __inline char* wchar_len_to_utf8(const wchar_t* wstr, int wlen)
diff --git a/src/rufus.rc b/src/rufus.rc
index 53a60431..86fd5f39 100644
--- a/src/rufus.rc
+++ b/src/rufus.rc
@@ -33,7 +33,7 @@ LANGUAGE LANG_NEUTRAL, SUBLANG_NEUTRAL
IDD_DIALOG DIALOGEX 12, 12, 232, 326
STYLE DS_SETFONT | DS_MODALFRAME | DS_CENTER | WS_MINIMIZEBOX | WS_POPUP | WS_CAPTION | WS_SYSMENU
EXSTYLE WS_EX_ACCEPTFILES
-CAPTION "Rufus 4.8.2233"
+CAPTION "Rufus 4.8.2234"
FONT 9, "Segoe UI Symbol", 400, 0, 0x0
BEGIN
LTEXT "Drive Properties",IDS_DRIVE_PROPERTIES_TXT,8,6,53,12,NOT WS_GROUP
@@ -407,8 +407,8 @@ END
//
VS_VERSION_INFO VERSIONINFO
- FILEVERSION 4,8,2233,0
- PRODUCTVERSION 4,8,2233,0
+ FILEVERSION 4,8,2234,0
+ PRODUCTVERSION 4,8,2234,0
FILEFLAGSMASK 0x3fL
#ifdef _DEBUG
FILEFLAGS 0x1L
@@ -426,13 +426,13 @@ BEGIN
VALUE "Comments", "https://rufus.ie"
VALUE "CompanyName", "Akeo Consulting"
VALUE "FileDescription", "Rufus"
- VALUE "FileVersion", "4.8.2233"
+ VALUE "FileVersion", "4.8.2234"
VALUE "InternalName", "Rufus"
VALUE "LegalCopyright", "© 2011-2025 Pete Batard (GPL v3)"
VALUE "LegalTrademarks", "https://www.gnu.org/licenses/gpl-3.0.html"
VALUE "OriginalFilename", "rufus-4.8.exe"
VALUE "ProductName", "Rufus"
- VALUE "ProductVersion", "4.8.2233"
+ VALUE "ProductVersion", "4.8.2234"
END
END
BLOCK "VarFileInfo"
diff --git a/src/wue.c b/src/wue.c
index 8281a6f4..c2815abf 100644
--- a/src/wue.c
+++ b/src/wue.c
@@ -25,6 +25,7 @@
#include "rufus.h"
#include "vhd.h"
+#include "xml.h"
#include "drive.h"
#include "format.h"
#include "missing.h"
@@ -418,24 +419,24 @@ out:
///
/// Populate the img_report Window version from an install[.wim|.esd] XML index
///
-/// The path of the extracted index XML.
+/// The index XML data.
+/// The length of the index XML data.
/// The index of the occurrence to look for.
-static void PopulateWindowsVersionFromXml(const char* xml_file, int index)
+static void PopulateWindowsVersionFromXml(const char* xml, size_t xml_len, int index)
{
char* val;
+ ezxml_t pxml = ezxml_parse_str((char*)xml, xml_len);
+ if (pxml == NULL)
+ return;
- val = get_token_data_file_indexed("MAJOR", xml_file, index);
+ val = ezxml_get_val(pxml, "IMAGE", index, "WINDOWS", 0, "VERSION", 0, "MAJOR", -1);
img_report.win_version.major = (uint16_t)safe_atoi(val);
- free(val);
- val = get_token_data_file_indexed("MINOR", xml_file, index);
+ val = ezxml_get_val(pxml, "IMAGE", index, "WINDOWS", 0, "VERSION", 0, "MINOR", -1);
img_report.win_version.minor = (uint16_t)safe_atoi(val);
- free(val);
- val = get_token_data_file_indexed("BUILD", xml_file, index);
+ val = ezxml_get_val(pxml, "IMAGE", index, "WINDOWS", 0, "VERSION", 0, "BUILD", -1);
img_report.win_version.build = (uint16_t)safe_atoi(val);
- free(val);
- val = get_token_data_file_indexed("SPBUILD", xml_file, index);
+ val = ezxml_get_val(pxml, "IMAGE", index, "WINDOWS", 0, "VERSION", 0, "SPBUILD", -1);
img_report.win_version.revision = (uint16_t)safe_atoi(val);
- free(val);
// Adjust versions so that we produce a more accurate report in the log
// (and yeah, I know we won't properly report Server, but I don't care)
if (img_report.win_version.major <= 5) {
@@ -463,10 +464,11 @@ static void PopulateWindowsVersionFromXml(const char* xml_file, int index)
if (img_report.win_version.build > 20000)
img_report.win_version.major = 11;
}
+ ezxml_free(pxml);
}
///
-/// Populate the img_report Window version from an an install[.wim|.esd], mounting the
+/// Populate the img_report Window version from an install[.wim|.esd].
/// ISO if needed. Requires Windows 8 or later.
///
/// (none)
@@ -474,7 +476,8 @@ static void PopulateWindowsVersionFromXml(const char* xml_file, int index)
BOOL PopulateWindowsVersion(void)
{
char *mounted_iso, mounted_image_path[128];
- char xml_file[MAX_PATH] = "";
+ char xml_file[MAX_PATH] = "", *xml;
+ size_t xml_len;
memset(&img_report.win_version, 0, sizeof(img_report.win_version));
@@ -506,7 +509,12 @@ BOOL PopulateWindowsVersion(void)
goto out;
}
- PopulateWindowsVersionFromXml(xml_file, 1);
+ xml_len = read_file(xml_file, (uint8_t**)&xml);
+ if (xml_len == 0) {
+ uprintf("Could not read WIM index XML");
+ goto out;
+ }
+ PopulateWindowsVersionFromXml(xml, xml_len, 1);
out:
DeleteFileU(xml_file);
@@ -549,12 +557,14 @@ BOOL CopySKUSiPolicy(const char* drive_name)
/// -2 on user cancel, -1 on other error, >=0 on success.
int SetWinToGoIndex(void)
{
- char* mounted_iso, mounted_image_path[128];
- char xml_file[MAX_PATH] = "";
- char* install_names[MAX_WININST];
- StrArray version_name, version_index;
int i;
+ char* mounted_iso, mounted_image_path[128];
+ char* install_names[MAX_WININST];
+ char xml_file[MAX_PATH] = "", *xml = NULL;
+ size_t xml_len;
+ StrArray version_name, version_index;
BOOL bNonStandard = FALSE;
+ ezxml_t index = NULL, image = NULL;
// Sanity checks
wintogo_index = -1;
@@ -600,22 +610,31 @@ int SetWinToGoIndex(void)
goto out;
}
+ xml_len = read_file(xml_file, (uint8_t**)&xml);
+ if (xml_len == 0) {
+ uprintf("Could not read WIM XML");
+ goto out;
+ }
+
StrArrayCreate(&version_name, 16);
StrArrayCreate(&version_index, 16);
- for (i = 0; StrArrayAdd(&version_index, get_token_data_file_indexed("IMAGE INDEX", xml_file, i + 1), FALSE) >= 0; i++) {
+ index = ezxml_parse_str((char*)xml, xml_len);
+ if (index == NULL) {
+ uprintf("Could not parse WIM XML");
+ goto out;
+ }
+
+ for (i = 0, image = ezxml_child(index, "IMAGE");
+ StrArrayAdd(&version_index, ezxml_attr(image, "INDEX"), TRUE) >= 0;
+ image = image->next, i++) {
// Some people are apparently creating *unofficial* Windows ISOs that don't have DISPLAYNAME elements.
- // If we are parsing such an ISO, try to fall back to using DESCRIPTION. Of course, since we don't use
- // a formal XML parser, if an ISO mixes entries with both DISPLAYNAME and DESCRIPTION and others with
- // only DESCRIPTION, the version names we report will be wrong.
- // But hey, there's only so far I'm willing to go to help people who, not content to have demonstrated
- // their utter ignorance on development matters, are also trying to lecture experienced developers
- // about specific "noob mistakes"... that don't exist in the code they are trying to criticize.
- if (StrArrayAdd(&version_name, get_token_data_file_indexed("DISPLAYNAME", xml_file, i + 1), FALSE) < 0) {
- bNonStandard = TRUE;
- if (StrArrayAdd(&version_name, get_token_data_file_indexed("DESCRIPTION", xml_file, i + 1), FALSE) < 0) {
+ // If we are parsing such an ISO, try to fall back to using DESCRIPTION.
+ if (StrArrayAdd(&version_name, ezxml_child_val(image, "DISPLAYNAME"), TRUE) < 0) {
+ if (StrArrayAdd(&version_name, ezxml_child_val(image, "DESCRIPTION"), TRUE) < 0) {
uprintf("Warning: Could not find a description for image index %d", i + 1);
StrArrayAdd(&version_name, "Unknown Windows Version", TRUE);
}
+ bNonStandard = TRUE;
}
}
if (bNonStandard)
@@ -632,7 +651,7 @@ int SetWinToGoIndex(void)
wintogo_index = atoi(version_index.String[i - 1]);
if (i > 0) {
// re-populate the version data from the selected XML index
- PopulateWindowsVersionFromXml(xml_file, i);
+ PopulateWindowsVersionFromXml(xml, xml_len, i);
// If we couldn't obtain the major and build, we have a problem
if (img_report.win_version.major == 0 || img_report.win_version.build == 0)
uprintf("Warning: Could not obtain version information from XML index (Nonstandard Windows image?)");
@@ -656,6 +675,8 @@ int SetWinToGoIndex(void)
StrArrayDestroy(&version_index);
out:
+ ezxml_free(index);
+ free(xml);
DeleteFileU(xml_file);
if (!img_report.is_windows_img)
VhdUnmountImage();
diff --git a/src/xml.c b/src/xml.c
index 82b11fb9..b344bb73 100644
--- a/src/xml.c
+++ b/src/xml.c
@@ -24,20 +24,31 @@
#include
#include
+#include
#include
#include
#include
#include
#include
-#ifndef EZXML_NOMMAP
-#include
-#endif // EZXML_NOMMAP
#include
-#include "ezxml.h"
+#include "xml.h"
+#include "rufus.h"
+#include "msapi_utf8.h"
+/* Memory leaks detection - define _CRTDBG_MAP_ALLOC as preprocessor macro */
+#ifdef _CRTDBG_MAP_ALLOC
+#include
+#include
+#endif
+
+#define EZXML_NOMMAP
#define EZXML_WS "\t\r\n " // whitespace
-#define EZXML_ERRL 128 // maximum error string length
+#define EZXML_ERRL 256 // maximum error string length
+#ifdef _MSC_VER
+#pragma warning(disable:6011)
+#pragma warning(disable:4267)
+#endif
typedef struct ezxml_root *ezxml_root_t;
struct ezxml_root { // additional data for the root tag
struct ezxml xml; // is a super-struct built on top of ezxml struct
@@ -56,6 +67,15 @@ struct ezxml_root { // additional data for the root tag
char *EZXML_NIL[] = { NULL }; // empty, null terminated array of strings
+// what realloc should be doing all along
+static inline void* _realloc(void* ptr, unsigned int size) {
+ void* old_ptr = ptr;
+ ptr = realloc(ptr, size);
+ if (ptr == NULL)
+ free(old_ptr);
+ return ptr;
+}
+
// returns the first child tag with the given name or NULL if not found
ezxml_t ezxml_child(ezxml_t xml, const char *name)
{
@@ -64,6 +84,14 @@ ezxml_t ezxml_child(ezxml_t xml, const char *name)
return xml;
}
+// returns the value of the first child tag with the given name or NULL if not found
+char* ezxml_child_val(ezxml_t xml, const char* name)
+{
+ xml = (xml) ? xml->child : NULL;
+ while (xml && strcmp(name, xml->name)) xml = xml->sibling;
+ return xml ? xml->txt : NULL;
+}
+
// returns the Nth tag with the same name in the same subsection or NULL if not
// found
ezxml_t ezxml_idx(ezxml_t xml, int idx)
@@ -119,6 +147,19 @@ ezxml_t ezxml_get(ezxml_t xml, ...)
return r;
}
+// Same as above but returns the text value or NULL if not found
+char* ezxml_get_val(ezxml_t xml, ...)
+{
+ va_list ap;
+ ezxml_t r;
+
+ va_start(ap, xml);
+ r = ezxml_vget(xml, ap);
+ va_end(ap);
+ return r ? r->txt : NULL;
+}
+
+
// returns a null terminated array of processing instructions for the given
// target
const char **ezxml_pi(ezxml_t xml, const char *target)
@@ -138,13 +179,14 @@ ezxml_t ezxml_err(ezxml_root_t root, char *s, const char *err, ...)
va_list ap;
int line = 1;
char *t, fmt[EZXML_ERRL];
-
+
for (t = root->s; t < s; t++) if (*t == '\n') line++;
- snprintf(fmt, EZXML_ERRL, "[error near line %d]: %s", line, err);
+ snprintf(fmt, EZXML_ERRL, "xml parsing error near line %d: %s", line, err);
va_start(ap, err);
vsnprintf(root->err, EZXML_ERRL, fmt, ap);
va_end(ap);
+ uprintf("%s", root->err);
return &root->xml;
}
@@ -168,7 +210,7 @@ char *ezxml_decode(char *s, char **ent, char t)
}
for (s = r; ; ) {
- while (*s && *s != '&' && (*s != '%' || t != '%') && !isspace(*s)) s++;
+ while (*s && *s != '&' && (*s != '%' || t != '%') && !isspace((uint8_t)*s)) s++;
if (! *s) break;
else if (t != 'c' && ! strncmp(s, "&#", 2)) { // character reference
@@ -176,11 +218,13 @@ char *ezxml_decode(char *s, char **ent, char t)
else c = strtol(s + 2, &e, 10); // base 10
if (! c || *e != ';') { s++; continue; } // not a character ref
- if (c < 0x80) *(s++) = c; // US-ASCII subset
+ if (c < 0x80) *(s++) = (char)c; // US-ASCII subset
else { // multi-byte UTF-8 sequence
for (b = 0, d = c; d; d /= 2) b++; // number of bits in c
+ // UTF-8 can ecode max 36 bits (standard says 21) - noop on 32 bit.
+ if (b > 36) { s++; continue; }
b = (b - 2) / 5; // number of bytes in payload
- *(s++) = (0xFF << (7 - b)) | (c >> (6 * b)); // head
+ *(s++) = (char)(0xFF << (7 - b)) | (char)(c >> (6 * b)); // head
while (b) *(s++) = 0x80 | ((c >> (6 * --b)) & 0x3F); // payload
}
@@ -192,24 +236,24 @@ char *ezxml_decode(char *s, char **ent, char t)
b += 2); // find entity in entity list
if (ent[b++]) { // found a match
- if ((c = strlen(ent[b])) - 1 > (e = strchr(s, ';')) - s) {
- l = (d = (s - r)) + c + strlen(e); // new length
- r = (r == m) ? strcpy(malloc(l), r) : realloc(r, l);
+ if ((c = (long)strlen(ent[b])) - 1 > (e = strchr(s, ';')) - s) {
+ l = (d = (long)(s - r)) + c + (long)strlen(e); // new length
+ r = (r == m) ? strcpy(malloc(l), r) : _realloc(r, l);
e = strchr((s = r + d), ';'); // fix up pointers
}
memmove(s + c, e + 1, strlen(e)); // shift rest of string
- strncpy(s, ent[b], c); // copy in replacement text
+ strncpy_s(s, c, ent[b], _TRUNCATE); // copy in replacement text
}
else s++; // not a known entity
}
- else if ((t == ' ' || t == '*') && isspace(*s)) *(s++) = ' ';
+ else if ((t == ' ' || t == '*') && isspace((uint8_t)*s)) *(s++) = ' ';
else s++; // no decoding needed
}
if (t == '*') { // normalize spaces for non-cdata attributes
for (s = r; *s; s++) {
- if ((l = strspn(s, " "))) memmove(s, s + l, strlen(s + l) + 1);
+ if ((l = (long)strspn(s, " "))) memmove(s, s + l, strlen(s + l) + 1);
while (*s && *s != ' ') s++;
}
if (--s >= r && *s == ' ') *s = '\0'; // trim any trailing space
@@ -300,16 +344,16 @@ void ezxml_proc_inst(ezxml_root_t root, char *s, size_t len)
while (root->pi[i] && strcmp(target, root->pi[i][0])) i++; // find target
if (! root->pi[i]) { // new target
- root->pi = realloc(root->pi, sizeof(char **) * (i + 2));
+ root->pi = _realloc(root->pi, sizeof(char **) * (i + 2));
root->pi[i] = malloc(sizeof(char *) * 3);
root->pi[i][0] = target;
root->pi[i][1] = (char *)(root->pi[i + 1] = NULL); // terminate pi list
- root->pi[i][2] = strdup(""); // empty document position list
+ root->pi[i][2] = _strdup(""); // empty document position list
}
while (root->pi[i][j]) j++; // find end of instruction list for this target
- root->pi[i] = realloc(root->pi[i], sizeof(char *) * (j + 3));
- root->pi[i][j + 2] = realloc(root->pi[i][j + 1], j + 1);
+ root->pi[i] = _realloc(root->pi[i], sizeof(char *) * (j + 3));
+ root->pi[i][j + 2] = _realloc(root->pi[i][j + 1], j + 1);
strcpy(root->pi[i][j + 2] + j - 1, (root->xml.name) ? ">" : "<");
root->pi[i][j + 1] = NULL; // null terminate pi list for this target
root->pi[i][j] = s; // set instruction
@@ -320,6 +364,7 @@ short ezxml_internal_dtd(ezxml_root_t root, char *s, size_t len)
{
char q, *c, *t, *n = NULL, *v, **ent, **pe;
int i, j;
+ size_t n_len, n_off;
pe = memcpy(malloc(sizeof(EZXML_NIL)), EZXML_NIL, sizeof(EZXML_NIL));
@@ -330,7 +375,13 @@ short ezxml_internal_dtd(ezxml_root_t root, char *s, size_t len)
else if (! strncmp(s, "= n_len) {
+ ezxml_err(root, NULL, "write past buffer (ent; ent[i]; i++);
- ent = realloc(ent, (i + 3) * sizeof(char *)); // space for next ent
+ ent = _realloc(ent, (i + 3) * sizeof(char *)); // space for next ent
if (*c == '%') pe = ent;
else root->ent = ent;
@@ -359,9 +410,10 @@ short ezxml_internal_dtd(ezxml_root_t root, char *s, size_t len)
if (! *t) { ezxml_err(root, t, "unclosed ")) == '>') continue;
else *s = '\0'; // null terminate tag name
- for (i = 0; root->attr[i] && strcmp(n, root->attr[i][0]); i++);
+ for (i = 0; n && root->attr[i] && strcmp(n, root->attr[i][0]); i++);
- while (*(n = ++s + strspn(s, EZXML_WS)) && *n != '>') {
+ ++s;
+ while (*(n = s + strspn(s, EZXML_WS)) && *n != '>') {
if (*(s = n + strcspn(n, EZXML_WS))) *s = '\0'; // attr name
else { ezxml_err(root, t, "malformed attr[i]) { // new tag name
root->attr = (! i) ? malloc(2 * sizeof(char **))
- : realloc(root->attr,
+ : _realloc(root->attr,
(i + 2) * sizeof(char **));
root->attr[i] = malloc(2 * sizeof(char *));
root->attr[i][0] = t; // set tag name
@@ -394,7 +446,7 @@ short ezxml_internal_dtd(ezxml_root_t root, char *s, size_t len)
}
for (j = 1; root->attr[i][j]; j += 3); // find end of list
- root->attr[i] = realloc(root->attr[i],
+ root->attr[i] = _realloc(root->attr[i],
(j + 4) * sizeof(char *));
root->attr[i][j + 3] = NULL; // null terminate list
@@ -438,23 +490,23 @@ char *ezxml_str2utf8(char **s, size_t *len)
c = (((c & 0x3FF) << 10) | (d & 0x3FF)) + 0x10000;
}
- while (l + 6 > max) u = realloc(u, max += EZXML_BUFSIZE);
- if (c < 0x80) u[l++] = c; // US-ASCII subset
+ while (l + 6 > max) u = _realloc(u, max += EZXML_BUFSIZE);
+ if (c < 0x80) u[l++] = (char)c; // US-ASCII subset
else { // multi-byte UTF-8 sequence
for (b = 0, d = c; d; d /= 2) b++; // bits in c
b = (b - 2) / 5; // bytes in payload
- u[l++] = (0xFF << (7 - b)) | (c >> (6 * b)); // head
+ u[l++] = (char)(0xFF << (7 - b)) | (char)(c >> (6 * b)); // head
while (b) u[l++] = 0x80 | ((c >> (6 * --b)) & 0x3F); // payload
}
}
- return *s = realloc(u, *len = l);
+ return *s = _realloc(u, *len = l);
}
// frees a tag attribute list
void ezxml_free_attr(char **attr) {
int i = 0;
char *m;
-
+
if (! attr || attr == EZXML_NIL) return; // nothing to free
while (attr[i]) i += 2; // find end of attribute list
m = attr[i + 1]; // list of which names and values are malloced
@@ -473,6 +525,7 @@ ezxml_t ezxml_parse_str(char *s, size_t len)
char q, e, *d, **attr, **a = NULL; // initialize a to avoid compile warning
int l, i, j;
+ if (!root) return NULL;
root->m = s;
if (! len) return ezxml_err(root, NULL, "root tag missing");
root->u = ezxml_str2utf8(&s, &len); // convert utf-16 to utf-8
@@ -493,7 +546,7 @@ ezxml_t ezxml_parse_str(char *s, size_t len)
return ezxml_err(root, d, "markup outside of root element");
s += strcspn(s, EZXML_WS "/>");
- while (isspace(*s)) *(s++) = '\0'; // null terminate tag name
+ while (isspace((uint8_t)*s)) *(s++) = '\0'; // null terminate tag name
if (*s && *s != '/' && *s != '>') // find tag in default attr list
for (i = 0; (a = root->attr[i]) && strcmp(a[0], d); i++);
@@ -509,7 +562,7 @@ ezxml_t ezxml_parse_str(char *s, size_t len)
attr[l] = s; // set attribute name
s += strcspn(s, EZXML_WS "=/>");
- if (*s == '=' || isspace(*s)) {
+ if (*s == '=' || isspace((uint8_t)*s)) {
*(s++) = '\0'; // null terminate tag attribute name
q = *(s += strspn(s, EZXML_WS "="));
if (q == '"' || q == '\'') { // attribute value
@@ -528,7 +581,7 @@ ezxml_t ezxml_parse_str(char *s, size_t len)
attr[l + 3][l / 2] = EZXML_TXTM; // value malloced
}
}
- while (isspace(*s)) s++;
+ while (isspace((uint8_t)*s)) s++;
}
if (*s == '/') { // self closing tag
@@ -555,7 +608,7 @@ ezxml_t ezxml_parse_str(char *s, size_t len)
if (! (q = *s) && e != '>') return ezxml_err(root, d, "missing >");
*s = '\0'; // temporarily null terminate tag name
if (ezxml_close_tag(root, d, s)) return &root->xml;
- if (isspace(*s = q)) s += strspn(s, EZXML_WS);
+ if (isspace((uint8_t)(*s = q))) s += strspn(s, EZXML_WS);
}
else if (! strncmp(s, "!--", 3)) { // xml comment
if (! (s = strstr(s + 3, "--")) || (*(s += 2) != '>' && *s) ||
@@ -611,11 +664,11 @@ ezxml_t ezxml_parse_fp(FILE *fp)
if (! (s = malloc(EZXML_BUFSIZE))) return NULL;
do {
len += (l = fread((s + len), 1, EZXML_BUFSIZE, fp));
- if (l == EZXML_BUFSIZE) s = realloc(s, len + EZXML_BUFSIZE);
+ if (l == EZXML_BUFSIZE) s = _realloc(s, len + EZXML_BUFSIZE);
} while (s && l == EZXML_BUFSIZE);
if (! s) return NULL;
- root = (ezxml_root_t)ezxml_parse_str(s, len);
+ if (!(root = (ezxml_root_t)ezxml_parse_str(s, len))) { free(s); return NULL; };
root->len = -1; // so we know to free s in ezxml_free()
return &root->xml;
}
@@ -639,12 +692,14 @@ ezxml_t ezxml_parse_fd(int fd)
MAP_FAILED) {
madvise(m, l, MADV_SEQUENTIAL); // optimize for sequential access
root = (ezxml_root_t)ezxml_parse_str(m, st.st_size);
+ if (!root) { munmap(m,l); return NULL; };
madvise(m, root->len = l, MADV_NORMAL); // put it back to normal
}
else { // mmap failed, read file into memory
#endif // EZXML_NOMMAP
- l = read(fd, m = malloc(st.st_size), st.st_size);
+ l = (long)_read(fd, m = malloc((size_t)st.st_size), (unsigned int)st.st_size);
root = (ezxml_root_t)ezxml_parse_str(m, l);
+ if (!root) { free(m); return NULL; };
root->len = -1; // so we know to free s in ezxml_free()
#ifndef EZXML_NOMMAP
}
@@ -655,10 +710,10 @@ ezxml_t ezxml_parse_fd(int fd)
// a wrapper for ezxml_parse_fd that accepts a file name
ezxml_t ezxml_parse_file(const char *file)
{
- int fd = open(file, O_RDONLY, 0);
+ int fd = _openU(file, O_RDONLY, 0);
ezxml_t xml = ezxml_parse_fd(fd);
- if (fd >= 0) close(fd);
+ if (fd >= 0) _close(fd);
return xml;
}
@@ -670,7 +725,7 @@ char *ezxml_ampencode(const char *s, size_t len, char **dst, size_t *dlen,
const char *e;
for (e = s + len; s != e; s++) {
- while (*dlen + 10 > *max) *dst = realloc(*dst, *max += EZXML_BUFSIZE);
+ while (*dlen + 10 > *max) *dst = _realloc(*dst, *max += EZXML_BUFSIZE);
switch (*s) {
case '\0': return *dst;
@@ -701,13 +756,13 @@ char *ezxml_toxml_r(ezxml_t xml, char **s, size_t *len, size_t *max,
*s = ezxml_ampencode(txt + start, xml->off - start, s, len, max, 0);
while (*len + strlen(xml->name) + 4 > *max) // reallocate s
- *s = realloc(*s, *max += EZXML_BUFSIZE);
+ *s = _realloc(*s, *max += EZXML_BUFSIZE);
*len += sprintf(*s + *len, "<%s", xml->name); // open tag
for (i = 0; xml->attr[i]; i += 2) { // tag attributes
if (ezxml_attr(xml, xml->attr[i]) != xml->attr[i + 1]) continue;
while (*len + strlen(xml->attr[i]) + 7 > *max) // reallocate s
- *s = realloc(*s, *max += EZXML_BUFSIZE);
+ *s = _realloc(*s, *max += EZXML_BUFSIZE);
*len += sprintf(*s + *len, " %s=\"", xml->attr[i]);
ezxml_ampencode(xml->attr[i + 1], -1, s, len, max, 1);
@@ -719,7 +774,7 @@ char *ezxml_toxml_r(ezxml_t xml, char **s, size_t *len, size_t *max,
if (! attr[i][j + 1] || ezxml_attr(xml, attr[i][j]) != attr[i][j + 1])
continue; // skip duplicates and non-values
while (*len + strlen(attr[i][j]) + 7 > *max) // reallocate s
- *s = realloc(*s, *max += EZXML_BUFSIZE);
+ *s = _realloc(*s, *max += EZXML_BUFSIZE);
*len += sprintf(*s + *len, " %s=\"", attr[i][j]);
ezxml_ampencode(attr[i][j + 1], -1, s, len, max, 1);
@@ -731,7 +786,7 @@ char *ezxml_toxml_r(ezxml_t xml, char **s, size_t *len, size_t *max,
: ezxml_ampencode(xml->txt, -1, s, len, max, 0); //data
while (*len + strlen(xml->name) + 4 > *max) // reallocate s
- *s = realloc(*s, *max += EZXML_BUFSIZE);
+ *s = _realloc(*s, *max += EZXML_BUFSIZE);
*len += sprintf(*s + *len, "", xml->name); // close tag
@@ -747,10 +802,12 @@ char *ezxml_toxml(ezxml_t xml)
ezxml_t p = (xml) ? xml->parent : NULL, o = (xml) ? xml->ordered : NULL;
ezxml_root_t root = (ezxml_root_t)xml;
size_t len = 0, max = EZXML_BUFSIZE;
- char *s = strcpy(malloc(max), ""), *t, *n;
+ char *s = malloc(max), *t, *n;
int i, j, k;
- if (! xml || ! xml->name) return realloc(s, len + 1);
+ if (!s) return (NULL);
+ s = strcpy(s, "");
+ if (! xml || ! xml->name) return _realloc(s, len + 1);
while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; // root tag
for (i = 0; ! p && root->pi[i]; i++) { // pre-root processing instructions
@@ -758,7 +815,7 @@ char *ezxml_toxml(ezxml_t xml)
for (j = 1; (n = root->pi[i][j]); j++) {
if (root->pi[i][k][j - 1] == '>') continue; // not pre-root
while (len + strlen(t = root->pi[i][0]) + strlen(n) + 7 > max)
- s = realloc(s, max += EZXML_BUFSIZE);
+ s = _realloc(s, max += EZXML_BUFSIZE);
len += sprintf(s + len, "\n", t, *n ? " " : "", n);
}
}
@@ -773,11 +830,11 @@ char *ezxml_toxml(ezxml_t xml)
for (j = 1; (n = root->pi[i][j]); j++) {
if (root->pi[i][k][j - 1] == '<') continue; // not post-root
while (len + strlen(t = root->pi[i][0]) + strlen(n) + 7 > max)
- s = realloc(s, max += EZXML_BUFSIZE);
+ s = _realloc(s, max += EZXML_BUFSIZE);
len += sprintf(s + len, "\n", t, *n ? " " : "", n);
}
}
- return realloc(s, len + 1);
+ return _realloc(s, len + 1);
}
// free the memory allocated for the ezxml structure
@@ -835,12 +892,15 @@ ezxml_t ezxml_new(const char *name)
{
static char *ent[] = { "lt;", "<", "gt;", ">", "quot;", """,
"apos;", "'", "amp;", "&", NULL };
- ezxml_root_t root = (ezxml_root_t)memset(malloc(sizeof(struct ezxml_root)),
- '\0', sizeof(struct ezxml_root));
+ ezxml_root_t root;
+ char **p_ent;
+ if (!(root = malloc(sizeof(struct ezxml_root)))) return NULL;
+ if (!(p_ent = malloc(sizeof(ent)))) { free(root); return NULL; };
+ root = (ezxml_root_t)memset(root, '\0', sizeof(struct ezxml_root));
root->xml.name = (char *)name;
root->cur = &root->xml;
strcpy(root->err, root->xml.txt = "");
- root->ent = memcpy(malloc(sizeof(ent)), ent, sizeof(ent));
+ root->ent = memcpy(p_ent, ent, sizeof(ent));
root->attr = root->pi = (char ***)(root->xml.attr = EZXML_NIL);
return &root->xml;
}
@@ -924,15 +984,17 @@ ezxml_t ezxml_set_attr(ezxml_t xml, const char *name, const char *value)
if (! xml->attr[l]) { // not found, add as new attribute
if (! value) return xml; // nothing to do
if (xml->attr == EZXML_NIL) { // first attribute
+ if (l >= 1) return NULL;
xml->attr = malloc(4 * sizeof(char *));
- xml->attr[1] = strdup(""); // empty list of malloced names/vals
+ xml->attr[1] = _strdup(""); // empty list of malloced names/vals
}
- else xml->attr = realloc(xml->attr, (l + 4) * sizeof(char *));
+ else xml->attr = _realloc(xml->attr, (l + 4) * sizeof(char *));
xml->attr[l] = (char *)name; // set attribute name
xml->attr[l + 2] = NULL; // null terminate attribute list
xml->attr[l + 3] = realloc(xml->attr[l + 1],
- (c = strlen(xml->attr[l + 1])) + 2);
+ (c =(long)strlen(xml->attr[l + 1])) + 2);
+ if (!xml->attr[l + 3]) return NULL;
strcpy(xml->attr[l + 3] + c, " "); // set name/value as not malloced
if (xml->flags & EZXML_DUP) xml->attr[l + 3][c] = EZXML_NAMEM;
}
@@ -947,7 +1009,7 @@ ezxml_t ezxml_set_attr(ezxml_t xml, const char *name, const char *value)
else { // remove attribute
if (xml->attr[c + 1][l / 2] & EZXML_NAMEM) free(xml->attr[l]);
memmove(xml->attr + l, xml->attr + l + 2, (c - l + 2) * sizeof(char*));
- xml->attr = realloc(xml->attr, (c + 2) * sizeof(char *));
+ xml->attr = _realloc(xml->attr, (c + 2) * sizeof(char *));
memmove(xml->attr[c + 1] + (l / 2), xml->attr[c + 1] + (l / 2) + 1,
(c / 2) - (l / 2)); // fix list of which name/vals are malloced
}
@@ -1006,8 +1068,8 @@ int main(int argc, char **argv)
if (argc != 2) return fprintf(stderr, "usage: %s xmlfile\n", argv[0]);
xml = ezxml_parse_file(argv[1]);
- printf("%s\n", (s = ezxml_toxml(xml)));
- free(s);
+ s = ezxml_toxml(xml);
+ if (s) { printf("%s\n", s); free(s); }
i = fprintf(stderr, "%s", ezxml_error(xml));
ezxml_free(xml);
return (i) ? 1 : 0;
diff --git a/src/xml.h b/src/xml.h
index 3e020788..1b5bf131 100644
--- a/src/xml.h
+++ b/src/xml.h
@@ -76,6 +76,10 @@ ezxml_t ezxml_parse_fp(FILE *fp);
// if not found
ezxml_t ezxml_child(ezxml_t xml, const char *name);
+// returns the value of the first child tag (one level deeper) with the given
+// name or NULL if not found
+char* ezxml_child_val(ezxml_t xml, const char* name);
+
// returns the next tag of the same name in the same section and depth or NULL
// if not found
#define ezxml_next(xml) ((xml) ? xml->next : NULL)
@@ -101,6 +105,9 @@ const char *ezxml_attr(ezxml_t xml, const char *attr);
// Returns NULL if not found.
ezxml_t ezxml_get(ezxml_t xml, ...);
+// Same as above but returns the text value or NULL if not found
+char* ezxml_get_val(ezxml_t xml, ...);
+
// Converts an ezxml structure back to xml. Returns a string of xml data that
// must be freed.
char *ezxml_toxml(ezxml_t xml);