closes bpo-38174: Update vendored expat library to 2.2.8. (GH-16346)
Fixes CVE-2019-15903. See full changelog at https://github.com/libexpat/libexpat/blob/R_2_2_8/expat/Changes.
diff --git a/Modules/expat/xmltok_impl.c b/Modules/expat/xmltok_impl.c
index 4d9ae7d..c209221 100644
--- a/Modules/expat/xmltok_impl.c
+++ b/Modules/expat/xmltok_impl.c
@@ -32,130 +32,124 @@
#ifdef XML_TOK_IMPL_C
-#ifndef IS_INVALID_CHAR
-#define IS_INVALID_CHAR(enc, ptr, n) (0)
-#endif
+# ifndef IS_INVALID_CHAR
+# define IS_INVALID_CHAR(enc, ptr, n) (0)
+# endif
-#define INVALID_LEAD_CASE(n, ptr, nextTokPtr) \
- case BT_LEAD ## n: \
- if (end - ptr < n) \
- return XML_TOK_PARTIAL_CHAR; \
- if (IS_INVALID_CHAR(enc, ptr, n)) { \
- *(nextTokPtr) = (ptr); \
- return XML_TOK_INVALID; \
- } \
- ptr += n; \
- break;
+# define INVALID_LEAD_CASE(n, ptr, nextTokPtr) \
+ case BT_LEAD##n: \
+ if (end - ptr < n) \
+ return XML_TOK_PARTIAL_CHAR; \
+ if (IS_INVALID_CHAR(enc, ptr, n)) { \
+ *(nextTokPtr) = (ptr); \
+ return XML_TOK_INVALID; \
+ } \
+ ptr += n; \
+ break;
-#define INVALID_CASES(ptr, nextTokPtr) \
- INVALID_LEAD_CASE(2, ptr, nextTokPtr) \
- INVALID_LEAD_CASE(3, ptr, nextTokPtr) \
- INVALID_LEAD_CASE(4, ptr, nextTokPtr) \
- case BT_NONXML: \
- case BT_MALFORM: \
- case BT_TRAIL: \
- *(nextTokPtr) = (ptr); \
+# define INVALID_CASES(ptr, nextTokPtr) \
+ INVALID_LEAD_CASE(2, ptr, nextTokPtr) \
+ INVALID_LEAD_CASE(3, ptr, nextTokPtr) \
+ INVALID_LEAD_CASE(4, ptr, nextTokPtr) \
+ case BT_NONXML: \
+ case BT_MALFORM: \
+ case BT_TRAIL: \
+ *(nextTokPtr) = (ptr); \
return XML_TOK_INVALID;
-#define CHECK_NAME_CASE(n, enc, ptr, end, nextTokPtr) \
- case BT_LEAD ## n: \
- if (end - ptr < n) \
- return XML_TOK_PARTIAL_CHAR; \
- if (!IS_NAME_CHAR(enc, ptr, n)) { \
- *nextTokPtr = ptr; \
- return XML_TOK_INVALID; \
- } \
- ptr += n; \
- break;
+# define CHECK_NAME_CASE(n, enc, ptr, end, nextTokPtr) \
+ case BT_LEAD##n: \
+ if (end - ptr < n) \
+ return XML_TOK_PARTIAL_CHAR; \
+ if (! IS_NAME_CHAR(enc, ptr, n)) { \
+ *nextTokPtr = ptr; \
+ return XML_TOK_INVALID; \
+ } \
+ ptr += n; \
+ break;
-#define CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) \
- case BT_NONASCII: \
- if (!IS_NAME_CHAR_MINBPC(enc, ptr)) { \
- *nextTokPtr = ptr; \
- return XML_TOK_INVALID; \
- } \
- /* fall through */ \
- case BT_NMSTRT: \
- case BT_HEX: \
- case BT_DIGIT: \
- case BT_NAME: \
- case BT_MINUS: \
- ptr += MINBPC(enc); \
- break; \
- CHECK_NAME_CASE(2, enc, ptr, end, nextTokPtr) \
- CHECK_NAME_CASE(3, enc, ptr, end, nextTokPtr) \
- CHECK_NAME_CASE(4, enc, ptr, end, nextTokPtr)
+# define CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) \
+ case BT_NONASCII: \
+ if (! IS_NAME_CHAR_MINBPC(enc, ptr)) { \
+ *nextTokPtr = ptr; \
+ return XML_TOK_INVALID; \
+ } \
+ /* fall through */ \
+ case BT_NMSTRT: \
+ case BT_HEX: \
+ case BT_DIGIT: \
+ case BT_NAME: \
+ case BT_MINUS: \
+ ptr += MINBPC(enc); \
+ break; \
+ CHECK_NAME_CASE(2, enc, ptr, end, nextTokPtr) \
+ CHECK_NAME_CASE(3, enc, ptr, end, nextTokPtr) \
+ CHECK_NAME_CASE(4, enc, ptr, end, nextTokPtr)
-#define CHECK_NMSTRT_CASE(n, enc, ptr, end, nextTokPtr) \
- case BT_LEAD ## n: \
- if (end - ptr < n) \
- return XML_TOK_PARTIAL_CHAR; \
- if (!IS_NMSTRT_CHAR(enc, ptr, n)) { \
- *nextTokPtr = ptr; \
- return XML_TOK_INVALID; \
- } \
- ptr += n; \
- break;
+# define CHECK_NMSTRT_CASE(n, enc, ptr, end, nextTokPtr) \
+ case BT_LEAD##n: \
+ if (end - ptr < n) \
+ return XML_TOK_PARTIAL_CHAR; \
+ if (! IS_NMSTRT_CHAR(enc, ptr, n)) { \
+ *nextTokPtr = ptr; \
+ return XML_TOK_INVALID; \
+ } \
+ ptr += n; \
+ break;
-#define CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) \
- case BT_NONASCII: \
- if (!IS_NMSTRT_CHAR_MINBPC(enc, ptr)) { \
- *nextTokPtr = ptr; \
- return XML_TOK_INVALID; \
- } \
- /* fall through */ \
- case BT_NMSTRT: \
- case BT_HEX: \
- ptr += MINBPC(enc); \
- break; \
- CHECK_NMSTRT_CASE(2, enc, ptr, end, nextTokPtr) \
- CHECK_NMSTRT_CASE(3, enc, ptr, end, nextTokPtr) \
- CHECK_NMSTRT_CASE(4, enc, ptr, end, nextTokPtr)
+# define CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) \
+ case BT_NONASCII: \
+ if (! IS_NMSTRT_CHAR_MINBPC(enc, ptr)) { \
+ *nextTokPtr = ptr; \
+ return XML_TOK_INVALID; \
+ } \
+ /* fall through */ \
+ case BT_NMSTRT: \
+ case BT_HEX: \
+ ptr += MINBPC(enc); \
+ break; \
+ CHECK_NMSTRT_CASE(2, enc, ptr, end, nextTokPtr) \
+ CHECK_NMSTRT_CASE(3, enc, ptr, end, nextTokPtr) \
+ CHECK_NMSTRT_CASE(4, enc, ptr, end, nextTokPtr)
-#ifndef PREFIX
-#define PREFIX(ident) ident
-#endif
+# ifndef PREFIX
+# define PREFIX(ident) ident
+# endif
+# define HAS_CHARS(enc, ptr, end, count) (end - ptr >= count * MINBPC(enc))
-#define HAS_CHARS(enc, ptr, end, count) \
- (end - ptr >= count * MINBPC(enc))
+# define HAS_CHAR(enc, ptr, end) HAS_CHARS(enc, ptr, end, 1)
-#define HAS_CHAR(enc, ptr, end) \
- HAS_CHARS(enc, ptr, end, 1)
-
-#define REQUIRE_CHARS(enc, ptr, end, count) \
- { \
- if (! HAS_CHARS(enc, ptr, end, count)) { \
- return XML_TOK_PARTIAL; \
- } \
+# define REQUIRE_CHARS(enc, ptr, end, count) \
+ { \
+ if (! HAS_CHARS(enc, ptr, end, count)) { \
+ return XML_TOK_PARTIAL; \
+ } \
}
-#define REQUIRE_CHAR(enc, ptr, end) \
- REQUIRE_CHARS(enc, ptr, end, 1)
-
+# define REQUIRE_CHAR(enc, ptr, end) REQUIRE_CHARS(enc, ptr, end, 1)
/* ptr points to character following "<!-" */
static int PTRCALL
-PREFIX(scanComment)(const ENCODING *enc, const char *ptr,
- const char *end, const char **nextTokPtr)
-{
+PREFIX(scanComment)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
if (HAS_CHAR(enc, ptr, end)) {
- if (!CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
+ if (! CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
ptr += MINBPC(enc);
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- INVALID_CASES(ptr, nextTokPtr)
+ INVALID_CASES(ptr, nextTokPtr)
case BT_MINUS:
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
if (CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
- if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
+ if (! CHAR_MATCHES(enc, ptr, ASCII_GT)) {
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
@@ -175,9 +169,8 @@
/* ptr points to character following "<!" */
static int PTRCALL
-PREFIX(scanDecl)(const ENCODING *enc, const char *ptr,
- const char *end, const char **nextTokPtr)
-{
+PREFIX(scanDecl)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
REQUIRE_CHAR(enc, ptr, end);
switch (BYTE_TYPE(enc, ptr)) {
case BT_MINUS:
@@ -199,12 +192,17 @@
REQUIRE_CHARS(enc, ptr, end, 2);
/* don't allow <!ENTITY% foo "whatever"> */
switch (BYTE_TYPE(enc, ptr + MINBPC(enc))) {
- case BT_S: case BT_CR: case BT_LF: case BT_PERCNT:
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
+ case BT_PERCNT:
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
/* fall through */
- case BT_S: case BT_CR: case BT_LF:
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
*nextTokPtr = ptr;
return XML_TOK_DECL_OPEN;
case BT_NMSTRT:
@@ -220,12 +218,12 @@
}
static int PTRCALL
-PREFIX(checkPiTarget)(const ENCODING *UNUSED_P(enc), const char *ptr,
- const char *end, int *tokPtr)
-{
+PREFIX(checkPiTarget)(const ENCODING *enc, const char *ptr, const char *end,
+ int *tokPtr) {
int upper = 0;
+ UNUSED_P(enc);
*tokPtr = XML_TOK_PI;
- if (end - ptr != MINBPC(enc)*3)
+ if (end - ptr != MINBPC(enc) * 3)
return 1;
switch (BYTE_TO_ASCII(enc, ptr)) {
case ASCII_x:
@@ -265,30 +263,31 @@
/* ptr points to character following "<?" */
static int PTRCALL
-PREFIX(scanPi)(const ENCODING *enc, const char *ptr,
- const char *end, const char **nextTokPtr)
-{
+PREFIX(scanPi)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
int tok;
const char *target = ptr;
REQUIRE_CHAR(enc, ptr, end);
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
+ CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
default:
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
- case BT_S: case BT_CR: case BT_LF:
- if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
+ CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
+ if (! PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
ptr += MINBPC(enc);
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- INVALID_CASES(ptr, nextTokPtr)
+ INVALID_CASES(ptr, nextTokPtr)
case BT_QUEST:
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
@@ -304,7 +303,7 @@
}
return XML_TOK_PARTIAL;
case BT_QUEST:
- if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
+ if (! PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
@@ -324,16 +323,16 @@
}
static int PTRCALL
-PREFIX(scanCdataSection)(const ENCODING *UNUSED_P(enc), const char *ptr,
- const char *end, const char **nextTokPtr)
-{
- static const char CDATA_LSQB[] = { ASCII_C, ASCII_D, ASCII_A,
- ASCII_T, ASCII_A, ASCII_LSQB };
+PREFIX(scanCdataSection)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
+ static const char CDATA_LSQB[]
+ = {ASCII_C, ASCII_D, ASCII_A, ASCII_T, ASCII_A, ASCII_LSQB};
int i;
+ UNUSED_P(enc);
/* CDATA[ */
REQUIRE_CHARS(enc, ptr, end, 6);
for (i = 0; i < 6; i++, ptr += MINBPC(enc)) {
- if (!CHAR_MATCHES(enc, ptr, CDATA_LSQB[i])) {
+ if (! CHAR_MATCHES(enc, ptr, CDATA_LSQB[i])) {
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
@@ -343,9 +342,8 @@
}
static int PTRCALL
-PREFIX(cdataSectionTok)(const ENCODING *enc, const char *ptr,
- const char *end, const char **nextTokPtr)
-{
+PREFIX(cdataSectionTok)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
if (ptr >= end)
return XML_TOK_NONE;
if (MINBPC(enc) > 1) {
@@ -361,11 +359,11 @@
case BT_RSQB:
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
- if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB))
+ if (! CHAR_MATCHES(enc, ptr, ASCII_RSQB))
break;
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
- if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
+ if (! CHAR_MATCHES(enc, ptr, ASCII_GT)) {
ptr -= MINBPC(enc);
break;
}
@@ -381,23 +379,25 @@
case BT_LF:
*nextTokPtr = ptr + MINBPC(enc);
return XML_TOK_DATA_NEWLINE;
- INVALID_CASES(ptr, nextTokPtr)
+ INVALID_CASES(ptr, nextTokPtr)
default:
ptr += MINBPC(enc);
break;
}
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
-#define LEAD_CASE(n) \
- case BT_LEAD ## n: \
- if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
- *nextTokPtr = ptr; \
- return XML_TOK_DATA_CHARS; \
- } \
- ptr += n; \
- break;
- LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
-#undef LEAD_CASE
+# define LEAD_CASE(n) \
+ case BT_LEAD##n: \
+ if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
+ *nextTokPtr = ptr; \
+ return XML_TOK_DATA_CHARS; \
+ } \
+ ptr += n; \
+ break;
+ LEAD_CASE(2)
+ LEAD_CASE(3)
+ LEAD_CASE(4)
+# undef LEAD_CASE
case BT_NONXML:
case BT_MALFORM:
case BT_TRAIL:
@@ -418,23 +418,26 @@
/* ptr points to character following "</" */
static int PTRCALL
-PREFIX(scanEndTag)(const ENCODING *enc, const char *ptr,
- const char *end, const char **nextTokPtr)
-{
+PREFIX(scanEndTag)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
REQUIRE_CHAR(enc, ptr, end);
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
+ CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
default:
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
- case BT_S: case BT_CR: case BT_LF:
+ CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
for (ptr += MINBPC(enc); HAS_CHAR(enc, ptr, end); ptr += MINBPC(enc)) {
switch (BYTE_TYPE(enc, ptr)) {
- case BT_S: case BT_CR: case BT_LF:
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
break;
case BT_GT:
*nextTokPtr = ptr + MINBPC(enc);
@@ -445,13 +448,13 @@
}
}
return XML_TOK_PARTIAL;
-#ifdef XML_NS
+# ifdef XML_NS
case BT_COLON:
/* no need to check qname syntax here,
since end-tag must match exactly */
ptr += MINBPC(enc);
break;
-#endif
+# endif
case BT_GT:
*nextTokPtr = ptr + MINBPC(enc);
return XML_TOK_END_TAG;
@@ -466,9 +469,8 @@
/* ptr points to character following "&#X" */
static int PTRCALL
-PREFIX(scanHexCharRef)(const ENCODING *enc, const char *ptr,
- const char *end, const char **nextTokPtr)
-{
+PREFIX(scanHexCharRef)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
if (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
case BT_DIGIT:
@@ -498,9 +500,8 @@
/* ptr points to character following "&#" */
static int PTRCALL
-PREFIX(scanCharRef)(const ENCODING *enc, const char *ptr,
- const char *end, const char **nextTokPtr)
-{
+PREFIX(scanCharRef)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
if (HAS_CHAR(enc, ptr, end)) {
if (CHAR_MATCHES(enc, ptr, ASCII_x))
return PREFIX(scanHexCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
@@ -531,11 +532,10 @@
static int PTRCALL
PREFIX(scanRef)(const ENCODING *enc, const char *ptr, const char *end,
- const char **nextTokPtr)
-{
+ const char **nextTokPtr) {
REQUIRE_CHAR(enc, ptr, end);
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
+ CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
case BT_NUM:
return PREFIX(scanCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
default:
@@ -544,7 +544,7 @@
}
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
+ CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
case BT_SEMI:
*nextTokPtr = ptr + MINBPC(enc);
return XML_TOK_ENTITY_REF;
@@ -560,15 +560,14 @@
static int PTRCALL
PREFIX(scanAtts)(const ENCODING *enc, const char *ptr, const char *end,
- const char **nextTokPtr)
-{
-#ifdef XML_NS
+ const char **nextTokPtr) {
+# ifdef XML_NS
int hadColon = 0;
-#endif
+# endif
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
-#ifdef XML_NS
+ CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
+# ifdef XML_NS
case BT_COLON:
if (hadColon) {
*nextTokPtr = ptr;
@@ -578,14 +577,16 @@
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
+ CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
default:
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
break;
-#endif
- case BT_S: case BT_CR: case BT_LF:
+# endif
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
for (;;) {
int t;
@@ -605,101 +606,101 @@
}
}
/* fall through */
- case BT_EQUALS:
- {
- int open;
-#ifdef XML_NS
- hadColon = 0;
-#endif
- for (;;) {
- ptr += MINBPC(enc);
- REQUIRE_CHAR(enc, ptr, end);
- open = BYTE_TYPE(enc, ptr);
- if (open == BT_QUOT || open == BT_APOS)
- break;
- switch (open) {
- case BT_S:
- case BT_LF:
- case BT_CR:
- break;
- default:
- *nextTokPtr = ptr;
- return XML_TOK_INVALID;
- }
- }
- ptr += MINBPC(enc);
- /* in attribute value */
- for (;;) {
- int t;
- REQUIRE_CHAR(enc, ptr, end);
- t = BYTE_TYPE(enc, ptr);
- if (t == open)
- break;
- switch (t) {
- INVALID_CASES(ptr, nextTokPtr)
- case BT_AMP:
- {
- int tok = PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, &ptr);
- if (tok <= 0) {
- if (tok == XML_TOK_INVALID)
- *nextTokPtr = ptr;
- return tok;
- }
- break;
- }
- case BT_LT:
- *nextTokPtr = ptr;
- return XML_TOK_INVALID;
- default:
- ptr += MINBPC(enc);
- break;
- }
- }
+ case BT_EQUALS: {
+ int open;
+# ifdef XML_NS
+ hadColon = 0;
+# endif
+ for (;;) {
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
- switch (BYTE_TYPE(enc, ptr)) {
- case BT_S:
- case BT_CR:
- case BT_LF:
+ open = BYTE_TYPE(enc, ptr);
+ if (open == BT_QUOT || open == BT_APOS)
break;
- case BT_SOL:
- goto sol;
- case BT_GT:
- goto gt;
+ switch (open) {
+ case BT_S:
+ case BT_LF:
+ case BT_CR:
+ break;
default:
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
- /* ptr points to closing quote */
- for (;;) {
- ptr += MINBPC(enc);
- REQUIRE_CHAR(enc, ptr, end);
- switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
- case BT_S: case BT_CR: case BT_LF:
- continue;
- case BT_GT:
- gt:
- *nextTokPtr = ptr + MINBPC(enc);
- return XML_TOK_START_TAG_WITH_ATTS;
- case BT_SOL:
- sol:
- ptr += MINBPC(enc);
- REQUIRE_CHAR(enc, ptr, end);
- if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
+ }
+ ptr += MINBPC(enc);
+ /* in attribute value */
+ for (;;) {
+ int t;
+ REQUIRE_CHAR(enc, ptr, end);
+ t = BYTE_TYPE(enc, ptr);
+ if (t == open)
+ break;
+ switch (t) {
+ INVALID_CASES(ptr, nextTokPtr)
+ case BT_AMP: {
+ int tok = PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, &ptr);
+ if (tok <= 0) {
+ if (tok == XML_TOK_INVALID)
*nextTokPtr = ptr;
- return XML_TOK_INVALID;
- }
- *nextTokPtr = ptr + MINBPC(enc);
- return XML_TOK_EMPTY_ELEMENT_WITH_ATTS;
- default:
- *nextTokPtr = ptr;
- return XML_TOK_INVALID;
+ return tok;
}
break;
}
+ case BT_LT:
+ *nextTokPtr = ptr;
+ return XML_TOK_INVALID;
+ default:
+ ptr += MINBPC(enc);
+ break;
+ }
+ }
+ ptr += MINBPC(enc);
+ REQUIRE_CHAR(enc, ptr, end);
+ switch (BYTE_TYPE(enc, ptr)) {
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
+ break;
+ case BT_SOL:
+ goto sol;
+ case BT_GT:
+ goto gt;
+ default:
+ *nextTokPtr = ptr;
+ return XML_TOK_INVALID;
+ }
+ /* ptr points to closing quote */
+ for (;;) {
+ ptr += MINBPC(enc);
+ REQUIRE_CHAR(enc, ptr, end);
+ switch (BYTE_TYPE(enc, ptr)) {
+ CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
+ continue;
+ case BT_GT:
+ gt:
+ *nextTokPtr = ptr + MINBPC(enc);
+ return XML_TOK_START_TAG_WITH_ATTS;
+ case BT_SOL:
+ sol:
+ ptr += MINBPC(enc);
+ REQUIRE_CHAR(enc, ptr, end);
+ if (! CHAR_MATCHES(enc, ptr, ASCII_GT)) {
+ *nextTokPtr = ptr;
+ return XML_TOK_INVALID;
+ }
+ *nextTokPtr = ptr + MINBPC(enc);
+ return XML_TOK_EMPTY_ELEMENT_WITH_ATTS;
+ default:
+ *nextTokPtr = ptr;
+ return XML_TOK_INVALID;
+ }
break;
}
+ break;
+ }
default:
*nextTokPtr = ptr;
return XML_TOK_INVALID;
@@ -712,14 +713,13 @@
static int PTRCALL
PREFIX(scanLt)(const ENCODING *enc, const char *ptr, const char *end,
- const char **nextTokPtr)
-{
-#ifdef XML_NS
+ const char **nextTokPtr) {
+# ifdef XML_NS
int hadColon;
-#endif
+# endif
REQUIRE_CHAR(enc, ptr, end);
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
+ CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
case BT_EXCL:
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
@@ -727,8 +727,7 @@
case BT_MINUS:
return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr);
case BT_LSQB:
- return PREFIX(scanCdataSection)(enc, ptr + MINBPC(enc),
- end, nextTokPtr);
+ return PREFIX(scanCdataSection)(enc, ptr + MINBPC(enc), end, nextTokPtr);
}
*nextTokPtr = ptr;
return XML_TOK_INVALID;
@@ -740,14 +739,14 @@
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
-#ifdef XML_NS
+# ifdef XML_NS
hadColon = 0;
-#endif
+# endif
/* we have a start-tag */
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
-#ifdef XML_NS
+ CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
+# ifdef XML_NS
case BT_COLON:
if (hadColon) {
*nextTokPtr = ptr;
@@ -757,34 +756,37 @@
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
+ CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
default:
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
break;
-#endif
- case BT_S: case BT_CR: case BT_LF:
- {
- ptr += MINBPC(enc);
- while (HAS_CHAR(enc, ptr, end)) {
- switch (BYTE_TYPE(enc, ptr)) {
+# endif
+ case BT_S:
+ case BT_CR:
+ case BT_LF: {
+ ptr += MINBPC(enc);
+ while (HAS_CHAR(enc, ptr, end)) {
+ switch (BYTE_TYPE(enc, ptr)) {
CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
- case BT_GT:
- goto gt;
- case BT_SOL:
- goto sol;
- case BT_S: case BT_CR: case BT_LF:
- ptr += MINBPC(enc);
- continue;
- default:
- *nextTokPtr = ptr;
- return XML_TOK_INVALID;
- }
- return PREFIX(scanAtts)(enc, ptr, end, nextTokPtr);
+ case BT_GT:
+ goto gt;
+ case BT_SOL:
+ goto sol;
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
+ ptr += MINBPC(enc);
+ continue;
+ default:
+ *nextTokPtr = ptr;
+ return XML_TOK_INVALID;
}
- return XML_TOK_PARTIAL;
+ return PREFIX(scanAtts)(enc, ptr, end, nextTokPtr);
}
+ return XML_TOK_PARTIAL;
+ }
case BT_GT:
gt:
*nextTokPtr = ptr + MINBPC(enc);
@@ -793,7 +795,7 @@
sol:
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
- if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
+ if (! CHAR_MATCHES(enc, ptr, ASCII_GT)) {
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
@@ -809,8 +811,7 @@
static int PTRCALL
PREFIX(contentTok)(const ENCODING *enc, const char *ptr, const char *end,
- const char **nextTokPtr)
-{
+ const char **nextTokPtr) {
if (ptr >= end)
return XML_TOK_NONE;
if (MINBPC(enc) > 1) {
@@ -842,48 +843,50 @@
ptr += MINBPC(enc);
if (! HAS_CHAR(enc, ptr, end))
return XML_TOK_TRAILING_RSQB;
- if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB))
+ if (! CHAR_MATCHES(enc, ptr, ASCII_RSQB))
break;
ptr += MINBPC(enc);
if (! HAS_CHAR(enc, ptr, end))
return XML_TOK_TRAILING_RSQB;
- if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
+ if (! CHAR_MATCHES(enc, ptr, ASCII_GT)) {
ptr -= MINBPC(enc);
break;
}
*nextTokPtr = ptr;
return XML_TOK_INVALID;
- INVALID_CASES(ptr, nextTokPtr)
+ INVALID_CASES(ptr, nextTokPtr)
default:
ptr += MINBPC(enc);
break;
}
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
-#define LEAD_CASE(n) \
- case BT_LEAD ## n: \
- if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
- *nextTokPtr = ptr; \
- return XML_TOK_DATA_CHARS; \
- } \
- ptr += n; \
- break;
- LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
-#undef LEAD_CASE
+# define LEAD_CASE(n) \
+ case BT_LEAD##n: \
+ if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
+ *nextTokPtr = ptr; \
+ return XML_TOK_DATA_CHARS; \
+ } \
+ ptr += n; \
+ break;
+ LEAD_CASE(2)
+ LEAD_CASE(3)
+ LEAD_CASE(4)
+# undef LEAD_CASE
case BT_RSQB:
if (HAS_CHARS(enc, ptr, end, 2)) {
- if (!CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_RSQB)) {
- ptr += MINBPC(enc);
- break;
- }
- if (HAS_CHARS(enc, ptr, end, 3)) {
- if (!CHAR_MATCHES(enc, ptr + 2*MINBPC(enc), ASCII_GT)) {
- ptr += MINBPC(enc);
- break;
- }
- *nextTokPtr = ptr + 2*MINBPC(enc);
- return XML_TOK_INVALID;
- }
+ if (! CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_RSQB)) {
+ ptr += MINBPC(enc);
+ break;
+ }
+ if (HAS_CHARS(enc, ptr, end, 3)) {
+ if (! CHAR_MATCHES(enc, ptr + 2 * MINBPC(enc), ASCII_GT)) {
+ ptr += MINBPC(enc);
+ break;
+ }
+ *nextTokPtr = ptr + 2 * MINBPC(enc);
+ return XML_TOK_INVALID;
+ }
}
/* fall through */
case BT_AMP:
@@ -908,12 +911,14 @@
static int PTRCALL
PREFIX(scanPercent)(const ENCODING *enc, const char *ptr, const char *end,
- const char **nextTokPtr)
-{
+ const char **nextTokPtr) {
REQUIRE_CHAR(enc, ptr, end);
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
- case BT_S: case BT_LF: case BT_CR: case BT_PERCNT:
+ CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
+ case BT_S:
+ case BT_LF:
+ case BT_CR:
+ case BT_PERCNT:
*nextTokPtr = ptr;
return XML_TOK_PERCENT;
default:
@@ -922,7 +927,7 @@
}
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
+ CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
case BT_SEMI:
*nextTokPtr = ptr + MINBPC(enc);
return XML_TOK_PARAM_ENTITY_REF;
@@ -936,20 +941,24 @@
static int PTRCALL
PREFIX(scanPoundName)(const ENCODING *enc, const char *ptr, const char *end,
- const char **nextTokPtr)
-{
+ const char **nextTokPtr) {
REQUIRE_CHAR(enc, ptr, end);
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
+ CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
default:
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
- case BT_CR: case BT_LF: case BT_S:
- case BT_RPAR: case BT_GT: case BT_PERCNT: case BT_VERBAR:
+ CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
+ case BT_CR:
+ case BT_LF:
+ case BT_S:
+ case BT_RPAR:
+ case BT_GT:
+ case BT_PERCNT:
+ case BT_VERBAR:
*nextTokPtr = ptr;
return XML_TOK_POUND_NAME;
default:
@@ -961,14 +970,12 @@
}
static int PTRCALL
-PREFIX(scanLit)(int open, const ENCODING *enc,
- const char *ptr, const char *end,
- const char **nextTokPtr)
-{
+PREFIX(scanLit)(int open, const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
while (HAS_CHAR(enc, ptr, end)) {
int t = BYTE_TYPE(enc, ptr);
switch (t) {
- INVALID_CASES(ptr, nextTokPtr)
+ INVALID_CASES(ptr, nextTokPtr)
case BT_QUOT:
case BT_APOS:
ptr += MINBPC(enc);
@@ -978,8 +985,12 @@
return -XML_TOK_LITERAL;
*nextTokPtr = ptr;
switch (BYTE_TYPE(enc, ptr)) {
- case BT_S: case BT_CR: case BT_LF:
- case BT_GT: case BT_PERCNT: case BT_LSQB:
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
+ case BT_GT:
+ case BT_PERCNT:
+ case BT_LSQB:
return XML_TOK_LITERAL;
default:
return XML_TOK_INVALID;
@@ -994,8 +1005,7 @@
static int PTRCALL
PREFIX(prologTok)(const ENCODING *enc, const char *ptr, const char *end,
- const char **nextTokPtr)
-{
+ const char **nextTokPtr) {
int tok;
if (ptr >= end)
return XML_TOK_NONE;
@@ -1013,27 +1023,26 @@
return PREFIX(scanLit)(BT_QUOT, enc, ptr + MINBPC(enc), end, nextTokPtr);
case BT_APOS:
return PREFIX(scanLit)(BT_APOS, enc, ptr + MINBPC(enc), end, nextTokPtr);
- case BT_LT:
- {
- ptr += MINBPC(enc);
- REQUIRE_CHAR(enc, ptr, end);
- switch (BYTE_TYPE(enc, ptr)) {
- case BT_EXCL:
- return PREFIX(scanDecl)(enc, ptr + MINBPC(enc), end, nextTokPtr);
- case BT_QUEST:
- return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
- case BT_NMSTRT:
- case BT_HEX:
- case BT_NONASCII:
- case BT_LEAD2:
- case BT_LEAD3:
- case BT_LEAD4:
- *nextTokPtr = ptr - MINBPC(enc);
- return XML_TOK_INSTANCE_START;
- }
- *nextTokPtr = ptr;
- return XML_TOK_INVALID;
+ case BT_LT: {
+ ptr += MINBPC(enc);
+ REQUIRE_CHAR(enc, ptr, end);
+ switch (BYTE_TYPE(enc, ptr)) {
+ case BT_EXCL:
+ return PREFIX(scanDecl)(enc, ptr + MINBPC(enc), end, nextTokPtr);
+ case BT_QUEST:
+ return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
+ case BT_NMSTRT:
+ case BT_HEX:
+ case BT_NONASCII:
+ case BT_LEAD2:
+ case BT_LEAD3:
+ case BT_LEAD4:
+ *nextTokPtr = ptr - MINBPC(enc);
+ return XML_TOK_INSTANCE_START;
}
+ *nextTokPtr = ptr;
+ return XML_TOK_INVALID;
+ }
case BT_CR:
if (ptr + MINBPC(enc) == end) {
*nextTokPtr = end;
@@ -1041,13 +1050,15 @@
return -XML_TOK_PROLOG_S;
}
/* fall through */
- case BT_S: case BT_LF:
+ case BT_S:
+ case BT_LF:
for (;;) {
ptr += MINBPC(enc);
if (! HAS_CHAR(enc, ptr, end))
break;
switch (BYTE_TYPE(enc, ptr)) {
- case BT_S: case BT_LF:
+ case BT_S:
+ case BT_LF:
break;
case BT_CR:
/* don't split CR/LF pair */
@@ -1076,7 +1087,7 @@
if (CHAR_MATCHES(enc, ptr, ASCII_RSQB)) {
REQUIRE_CHARS(enc, ptr, end, 2);
if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_GT)) {
- *nextTokPtr = ptr + 2*MINBPC(enc);
+ *nextTokPtr = ptr + 2 * MINBPC(enc);
return XML_TOK_COND_SECT_CLOSE;
}
}
@@ -1099,8 +1110,12 @@
case BT_PLUS:
*nextTokPtr = ptr + MINBPC(enc);
return XML_TOK_CLOSE_PAREN_PLUS;
- case BT_CR: case BT_LF: case BT_S:
- case BT_GT: case BT_COMMA: case BT_VERBAR:
+ case BT_CR:
+ case BT_LF:
+ case BT_S:
+ case BT_GT:
+ case BT_COMMA:
+ case BT_VERBAR:
case BT_RPAR:
*nextTokPtr = ptr;
return XML_TOK_CLOSE_PAREN;
@@ -1115,24 +1130,26 @@
return XML_TOK_DECL_CLOSE;
case BT_NUM:
return PREFIX(scanPoundName)(enc, ptr + MINBPC(enc), end, nextTokPtr);
-#define LEAD_CASE(n) \
- case BT_LEAD ## n: \
- if (end - ptr < n) \
- return XML_TOK_PARTIAL_CHAR; \
- if (IS_NMSTRT_CHAR(enc, ptr, n)) { \
- ptr += n; \
- tok = XML_TOK_NAME; \
- break; \
- } \
- if (IS_NAME_CHAR(enc, ptr, n)) { \
- ptr += n; \
- tok = XML_TOK_NMTOKEN; \
- break; \
- } \
- *nextTokPtr = ptr; \
+# define LEAD_CASE(n) \
+ case BT_LEAD##n: \
+ if (end - ptr < n) \
+ return XML_TOK_PARTIAL_CHAR; \
+ if (IS_NMSTRT_CHAR(enc, ptr, n)) { \
+ ptr += n; \
+ tok = XML_TOK_NAME; \
+ break; \
+ } \
+ if (IS_NAME_CHAR(enc, ptr, n)) { \
+ ptr += n; \
+ tok = XML_TOK_NMTOKEN; \
+ break; \
+ } \
+ *nextTokPtr = ptr; \
return XML_TOK_INVALID;
- LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
-#undef LEAD_CASE
+ LEAD_CASE(2)
+ LEAD_CASE(3)
+ LEAD_CASE(4)
+# undef LEAD_CASE
case BT_NMSTRT:
case BT_HEX:
tok = XML_TOK_NAME;
@@ -1141,9 +1158,9 @@
case BT_DIGIT:
case BT_NAME:
case BT_MINUS:
-#ifdef XML_NS
+# ifdef XML_NS
case BT_COLON:
-#endif
+# endif
tok = XML_TOK_NMTOKEN;
ptr += MINBPC(enc);
break;
@@ -1165,13 +1182,19 @@
}
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
- case BT_GT: case BT_RPAR: case BT_COMMA:
- case BT_VERBAR: case BT_LSQB: case BT_PERCNT:
- case BT_S: case BT_CR: case BT_LF:
+ CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
+ case BT_GT:
+ case BT_RPAR:
+ case BT_COMMA:
+ case BT_VERBAR:
+ case BT_LSQB:
+ case BT_PERCNT:
+ case BT_S:
+ case BT_CR:
+ case BT_LF:
*nextTokPtr = ptr;
return tok;
-#ifdef XML_NS
+# ifdef XML_NS
case BT_COLON:
ptr += MINBPC(enc);
switch (tok) {
@@ -1179,7 +1202,7 @@
REQUIRE_CHAR(enc, ptr, end);
tok = XML_TOK_PREFIXED_NAME;
switch (BYTE_TYPE(enc, ptr)) {
- CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
+ CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
default:
tok = XML_TOK_NMTOKEN;
break;
@@ -1190,23 +1213,23 @@
break;
}
break;
-#endif
+# endif
case BT_PLUS:
- if (tok == XML_TOK_NMTOKEN) {
+ if (tok == XML_TOK_NMTOKEN) {
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
*nextTokPtr = ptr + MINBPC(enc);
return XML_TOK_NAME_PLUS;
case BT_AST:
- if (tok == XML_TOK_NMTOKEN) {
+ if (tok == XML_TOK_NMTOKEN) {
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
*nextTokPtr = ptr + MINBPC(enc);
return XML_TOK_NAME_ASTERISK;
case BT_QUEST:
- if (tok == XML_TOK_NMTOKEN) {
+ if (tok == XML_TOK_NMTOKEN) {
*nextTokPtr = ptr;
return XML_TOK_INVALID;
}
@@ -1221,9 +1244,8 @@
}
static int PTRCALL
-PREFIX(attributeValueTok)(const ENCODING *enc, const char *ptr,
- const char *end, const char **nextTokPtr)
-{
+PREFIX(attributeValueTok)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
const char *start;
if (ptr >= end)
return XML_TOK_NONE;
@@ -1238,10 +1260,14 @@
start = ptr;
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
-#define LEAD_CASE(n) \
- case BT_LEAD ## n: ptr += n; break;
- LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
-#undef LEAD_CASE
+# define LEAD_CASE(n) \
+ case BT_LEAD##n: \
+ ptr += n; \
+ break;
+ LEAD_CASE(2)
+ LEAD_CASE(3)
+ LEAD_CASE(4)
+# undef LEAD_CASE
case BT_AMP:
if (ptr == start)
return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
@@ -1287,9 +1313,8 @@
}
static int PTRCALL
-PREFIX(entityValueTok)(const ENCODING *enc, const char *ptr,
- const char *end, const char **nextTokPtr)
-{
+PREFIX(entityValueTok)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
const char *start;
if (ptr >= end)
return XML_TOK_NONE;
@@ -1304,10 +1329,14 @@
start = ptr;
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
-#define LEAD_CASE(n) \
- case BT_LEAD ## n: ptr += n; break;
- LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
-#undef LEAD_CASE
+# define LEAD_CASE(n) \
+ case BT_LEAD##n: \
+ ptr += n; \
+ break;
+ LEAD_CASE(2)
+ LEAD_CASE(3)
+ LEAD_CASE(4)
+# undef LEAD_CASE
case BT_AMP:
if (ptr == start)
return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
@@ -1315,8 +1344,7 @@
return XML_TOK_DATA_CHARS;
case BT_PERCNT:
if (ptr == start) {
- int tok = PREFIX(scanPercent)(enc, ptr + MINBPC(enc),
- end, nextTokPtr);
+ int tok = PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr);
return (tok == XML_TOK_PERCENT) ? XML_TOK_INVALID : tok;
}
*nextTokPtr = ptr;
@@ -1349,12 +1377,11 @@
return XML_TOK_DATA_CHARS;
}
-#ifdef XML_DTD
+# ifdef XML_DTD
static int PTRCALL
-PREFIX(ignoreSectionTok)(const ENCODING *enc, const char *ptr,
- const char *end, const char **nextTokPtr)
-{
+PREFIX(ignoreSectionTok)(const ENCODING *enc, const char *ptr, const char *end,
+ const char **nextTokPtr) {
int level = 0;
if (MINBPC(enc) > 1) {
size_t n = end - ptr;
@@ -1365,7 +1392,7 @@
}
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
- INVALID_CASES(ptr, nextTokPtr)
+ INVALID_CASES(ptr, nextTokPtr)
case BT_LT:
ptr += MINBPC(enc);
REQUIRE_CHAR(enc, ptr, end);
@@ -1402,12 +1429,11 @@
return XML_TOK_PARTIAL;
}
-#endif /* XML_DTD */
+# endif /* XML_DTD */
static int PTRCALL
PREFIX(isPublicId)(const ENCODING *enc, const char *ptr, const char *end,
- const char **badPtr)
-{
+ const char **badPtr) {
ptr += MINBPC(enc);
end -= MINBPC(enc);
for (; HAS_CHAR(enc, ptr, end); ptr += MINBPC(enc)) {
@@ -1430,9 +1456,9 @@
case BT_AST:
case BT_PERCNT:
case BT_NUM:
-#ifdef XML_NS
+# ifdef XML_NS
case BT_COLON:
-#endif
+# endif
break;
case BT_S:
if (CHAR_MATCHES(enc, ptr, ASCII_TAB)) {
@@ -1442,7 +1468,7 @@
break;
case BT_NAME:
case BT_NMSTRT:
- if (!(BYTE_TO_ASCII(enc, ptr) & ~0x7f))
+ if (! (BYTE_TO_ASCII(enc, ptr) & ~0x7f))
break;
/* fall through */
default:
@@ -1466,9 +1492,8 @@
*/
static int PTRCALL
-PREFIX(getAtts)(const ENCODING *enc, const char *ptr,
- int attsMax, ATTRIBUTE *atts)
-{
+PREFIX(getAtts)(const ENCODING *enc, const char *ptr, int attsMax,
+ ATTRIBUTE *atts) {
enum { other, inName, inValue } state = inName;
int nAtts = 0;
int open = 0; /* defined when state == inValue;
@@ -1476,32 +1501,35 @@
for (ptr += MINBPC(enc);; ptr += MINBPC(enc)) {
switch (BYTE_TYPE(enc, ptr)) {
-#define START_NAME \
- if (state == other) { \
- if (nAtts < attsMax) { \
- atts[nAtts].name = ptr; \
- atts[nAtts].normalized = 1; \
- } \
- state = inName; \
- }
-#define LEAD_CASE(n) \
- case BT_LEAD ## n: START_NAME ptr += (n - MINBPC(enc)); break;
- LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
-#undef LEAD_CASE
+# define START_NAME \
+ if (state == other) { \
+ if (nAtts < attsMax) { \
+ atts[nAtts].name = ptr; \
+ atts[nAtts].normalized = 1; \
+ } \
+ state = inName; \
+ }
+# define LEAD_CASE(n) \
+ case BT_LEAD##n: \
+ START_NAME ptr += (n - MINBPC(enc)); \
+ break;
+ LEAD_CASE(2)
+ LEAD_CASE(3)
+ LEAD_CASE(4)
+# undef LEAD_CASE
case BT_NONASCII:
case BT_NMSTRT:
case BT_HEX:
START_NAME
break;
-#undef START_NAME
+# undef START_NAME
case BT_QUOT:
if (state != inValue) {
if (nAtts < attsMax)
atts[nAtts].valuePtr = ptr + MINBPC(enc);
state = inValue;
open = BT_QUOT;
- }
- else if (open == BT_QUOT) {
+ } else if (open == BT_QUOT) {
state = other;
if (nAtts < attsMax)
atts[nAtts].valueEnd = ptr;
@@ -1514,8 +1542,7 @@
atts[nAtts].valuePtr = ptr + MINBPC(enc);
state = inValue;
open = BT_APOS;
- }
- else if (open == BT_APOS) {
+ } else if (open == BT_APOS) {
state = other;
if (nAtts < attsMax)
atts[nAtts].valueEnd = ptr;
@@ -1529,16 +1556,15 @@
case BT_S:
if (state == inName)
state = other;
- else if (state == inValue
- && nAtts < attsMax
- && atts[nAtts].normalized
+ else if (state == inValue && nAtts < attsMax && atts[nAtts].normalized
&& (ptr == atts[nAtts].valuePtr
|| BYTE_TO_ASCII(enc, ptr) != ASCII_SPACE
|| BYTE_TO_ASCII(enc, ptr + MINBPC(enc)) == ASCII_SPACE
|| BYTE_TYPE(enc, ptr + MINBPC(enc)) == open))
atts[nAtts].normalized = 0;
break;
- case BT_CR: case BT_LF:
+ case BT_CR:
+ case BT_LF:
/* This case ensures that the first attribute name is counted
Apart from that we could just change state on the quote. */
if (state == inName)
@@ -1559,29 +1585,44 @@
}
static int PTRFASTCALL
-PREFIX(charRefNumber)(const ENCODING *UNUSED_P(enc), const char *ptr)
-{
+PREFIX(charRefNumber)(const ENCODING *enc, const char *ptr) {
int result = 0;
/* skip &# */
- ptr += 2*MINBPC(enc);
+ UNUSED_P(enc);
+ ptr += 2 * MINBPC(enc);
if (CHAR_MATCHES(enc, ptr, ASCII_x)) {
- for (ptr += MINBPC(enc);
- !CHAR_MATCHES(enc, ptr, ASCII_SEMI);
+ for (ptr += MINBPC(enc); ! CHAR_MATCHES(enc, ptr, ASCII_SEMI);
ptr += MINBPC(enc)) {
int c = BYTE_TO_ASCII(enc, ptr);
switch (c) {
- case ASCII_0: case ASCII_1: case ASCII_2: case ASCII_3: case ASCII_4:
- case ASCII_5: case ASCII_6: case ASCII_7: case ASCII_8: case ASCII_9:
+ case ASCII_0:
+ case ASCII_1:
+ case ASCII_2:
+ case ASCII_3:
+ case ASCII_4:
+ case ASCII_5:
+ case ASCII_6:
+ case ASCII_7:
+ case ASCII_8:
+ case ASCII_9:
result <<= 4;
result |= (c - ASCII_0);
break;
- case ASCII_A: case ASCII_B: case ASCII_C:
- case ASCII_D: case ASCII_E: case ASCII_F:
+ case ASCII_A:
+ case ASCII_B:
+ case ASCII_C:
+ case ASCII_D:
+ case ASCII_E:
+ case ASCII_F:
result <<= 4;
result += 10 + (c - ASCII_A);
break;
- case ASCII_a: case ASCII_b: case ASCII_c:
- case ASCII_d: case ASCII_e: case ASCII_f:
+ case ASCII_a:
+ case ASCII_b:
+ case ASCII_c:
+ case ASCII_d:
+ case ASCII_e:
+ case ASCII_f:
result <<= 4;
result += 10 + (c - ASCII_a);
break;
@@ -1589,9 +1630,8 @@
if (result >= 0x110000)
return -1;
}
- }
- else {
- for (; !CHAR_MATCHES(enc, ptr, ASCII_SEMI); ptr += MINBPC(enc)) {
+ } else {
+ for (; ! CHAR_MATCHES(enc, ptr, ASCII_SEMI); ptr += MINBPC(enc)) {
int c = BYTE_TO_ASCII(enc, ptr);
result *= 10;
result += (c - ASCII_0);
@@ -1603,10 +1643,10 @@
}
static int PTRCALL
-PREFIX(predefinedEntityName)(const ENCODING *UNUSED_P(enc), const char *ptr,
- const char *end)
-{
- switch ((end - ptr)/MINBPC(enc)) {
+PREFIX(predefinedEntityName)(const ENCODING *enc, const char *ptr,
+ const char *end) {
+ UNUSED_P(enc);
+ switch ((end - ptr) / MINBPC(enc)) {
case 2:
if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_t)) {
switch (BYTE_TO_ASCII(enc, ptr)) {
@@ -1657,9 +1697,9 @@
}
static int PTRCALL
-PREFIX(nameMatchesAscii)(const ENCODING *UNUSED_P(enc), const char *ptr1,
- const char *end1, const char *ptr2)
-{
+PREFIX(nameMatchesAscii)(const ENCODING *enc, const char *ptr1,
+ const char *end1, const char *ptr2) {
+ UNUSED_P(enc);
for (; *ptr2; ptr1 += MINBPC(enc), ptr2++) {
if (end1 - ptr1 < MINBPC(enc)) {
/* This line cannot be executed. The incoming data has already
@@ -1669,27 +1709,30 @@
*/
return 0; /* LCOV_EXCL_LINE */
}
- if (!CHAR_MATCHES(enc, ptr1, *ptr2))
+ if (! CHAR_MATCHES(enc, ptr1, *ptr2))
return 0;
}
return ptr1 == end1;
}
static int PTRFASTCALL
-PREFIX(nameLength)(const ENCODING *enc, const char *ptr)
-{
+PREFIX(nameLength)(const ENCODING *enc, const char *ptr) {
const char *start = ptr;
for (;;) {
switch (BYTE_TYPE(enc, ptr)) {
-#define LEAD_CASE(n) \
- case BT_LEAD ## n: ptr += n; break;
- LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
-#undef LEAD_CASE
+# define LEAD_CASE(n) \
+ case BT_LEAD##n: \
+ ptr += n; \
+ break;
+ LEAD_CASE(2)
+ LEAD_CASE(3)
+ LEAD_CASE(4)
+# undef LEAD_CASE
case BT_NONASCII:
case BT_NMSTRT:
-#ifdef XML_NS
+# ifdef XML_NS
case BT_COLON:
-#endif
+# endif
case BT_HEX:
case BT_DIGIT:
case BT_NAME:
@@ -1702,9 +1745,8 @@
}
}
-static const char * PTRFASTCALL
-PREFIX(skipS)(const ENCODING *enc, const char *ptr)
-{
+static const char *PTRFASTCALL
+PREFIX(skipS)(const ENCODING *enc, const char *ptr) {
for (;;) {
switch (BYTE_TYPE(enc, ptr)) {
case BT_LF:
@@ -1719,19 +1761,18 @@
}
static void PTRCALL
-PREFIX(updatePosition)(const ENCODING *enc,
- const char *ptr,
- const char *end,
- POSITION *pos)
-{
+PREFIX(updatePosition)(const ENCODING *enc, const char *ptr, const char *end,
+ POSITION *pos) {
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
-#define LEAD_CASE(n) \
- case BT_LEAD ## n: \
- ptr += n; \
- break;
- LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
-#undef LEAD_CASE
+# define LEAD_CASE(n) \
+ case BT_LEAD##n: \
+ ptr += n; \
+ break;
+ LEAD_CASE(2)
+ LEAD_CASE(3)
+ LEAD_CASE(4)
+# undef LEAD_CASE
case BT_LF:
pos->columnNumber = (XML_Size)-1;
pos->lineNumber++;
@@ -1752,12 +1793,12 @@
}
}
-#undef DO_LEAD_CASE
-#undef MULTIBYTE_CASES
-#undef INVALID_CASES
-#undef CHECK_NAME_CASE
-#undef CHECK_NAME_CASES
-#undef CHECK_NMSTRT_CASE
-#undef CHECK_NMSTRT_CASES
+# undef DO_LEAD_CASE
+# undef MULTIBYTE_CASES
+# undef INVALID_CASES
+# undef CHECK_NAME_CASE
+# undef CHECK_NAME_CASES
+# undef CHECK_NMSTRT_CASE
+# undef CHECK_NMSTRT_CASES
#endif /* XML_TOK_IMPL_C */