Better compliance to RFC7230/RFC3875 when building CGI headers

lpereira · Apr 29, 2024 · ce4161c · ce4161c
1 parent 2718dc2
commit ce4161c
Show file tree

Hide file tree

Showing 3 changed files with 84 additions and 64 deletions.
diff --git a/src/lib/lwan-private.h b/src/lib/lwan-private.h
@@ -133,6 +133,7 @@ uint8_t lwan_char_isxdigit(char ch) __attribute__((pure));
 uint8_t lwan_char_isdigit(char ch) __attribute__((pure));
 uint8_t lwan_char_isalpha(char ch) __attribute__((pure));
 uint8_t lwan_char_isalnum(char ch) __attribute__((pure));
+uint8_t lwan_char_iscgiheader(char ch) __attribute__((pure));
 
 static ALWAYS_INLINE __attribute__((pure)) size_t lwan_nextpow2(size_t number)
 {

diff --git a/src/lib/lwan-request.c b/src/lib/lwan-request.c
@@ -2254,11 +2254,10 @@ void lwan_request_foreach_header_for_cgi(struct lwan_request *request,
         if (r < 0 || r >= (int)sizeof(header_name))
             continue;
 
-        /* FIXME: RFC7230/RFC3875 compliance */
         for (char *p = header_name; *p; p++) {
             if (lwan_char_isalpha(*p))
                 *p &= ~0x20;
-            else if (!lwan_char_isdigit(*p))
+            else if (!lwan_char_iscgiheader(*p))
                 *p = '_';
         }
 

diff --git a/src/lib/lwan-tables.c b/src/lib/lwan-tables.c
@@ -196,82 +196,102 @@ enum {
     CHAR_PROP_HEX = 1 << 1,
     CHAR_PROP_DIG = 1 << 2,
     CHAR_PROP_ALPHA = 1 << 3,
+    CHAR_PROP_CGI_HEADER = 1 << 4,
 };
 
 static const uint8_t char_prop_tbl[256] = {
     [' '] = CHAR_PROP_SPACE,
     ['\t'] = CHAR_PROP_SPACE,
     ['\n'] = CHAR_PROP_SPACE,
     ['\r'] = CHAR_PROP_SPACE,
-    ['0'] = CHAR_PROP_HEX | CHAR_PROP_DIG,
-    ['1'] = CHAR_PROP_HEX | CHAR_PROP_DIG,
-    ['2'] = CHAR_PROP_HEX | CHAR_PROP_DIG,
-    ['3'] = CHAR_PROP_HEX | CHAR_PROP_DIG,
-    ['4'] = CHAR_PROP_HEX | CHAR_PROP_DIG,
-    ['5'] = CHAR_PROP_HEX | CHAR_PROP_DIG,
-    ['6'] = CHAR_PROP_HEX | CHAR_PROP_DIG,
-    ['7'] = CHAR_PROP_HEX | CHAR_PROP_DIG,
-    ['8'] = CHAR_PROP_HEX | CHAR_PROP_DIG,
-    ['9'] = CHAR_PROP_HEX | CHAR_PROP_DIG,
-    ['a'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['b'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['c'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['d'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['e'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['f'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['g'] = CHAR_PROP_ALPHA,
-    ['h'] = CHAR_PROP_ALPHA,
-    ['i'] = CHAR_PROP_ALPHA,
-    ['j'] = CHAR_PROP_ALPHA,
-    ['k'] = CHAR_PROP_ALPHA,
-    ['l'] = CHAR_PROP_ALPHA,
-    ['m'] = CHAR_PROP_ALPHA,
-    ['n'] = CHAR_PROP_ALPHA,
-    ['o'] = CHAR_PROP_ALPHA,
-    ['p'] = CHAR_PROP_ALPHA,
-    ['q'] = CHAR_PROP_ALPHA,
-    ['r'] = CHAR_PROP_ALPHA,
-    ['s'] = CHAR_PROP_ALPHA,
-    ['t'] = CHAR_PROP_ALPHA,
-    ['u'] = CHAR_PROP_ALPHA,
-    ['v'] = CHAR_PROP_ALPHA,
-    ['w'] = CHAR_PROP_ALPHA,
-    ['x'] = CHAR_PROP_ALPHA,
-    ['y'] = CHAR_PROP_ALPHA,
-    ['z'] = CHAR_PROP_ALPHA,
-    ['A'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['B'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['C'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['D'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['E'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['F'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA,
-    ['G'] = CHAR_PROP_ALPHA,
-    ['H'] = CHAR_PROP_ALPHA,
-    ['I'] = CHAR_PROP_ALPHA,
-    ['J'] = CHAR_PROP_ALPHA,
-    ['K'] = CHAR_PROP_ALPHA,
-    ['L'] = CHAR_PROP_ALPHA,
-    ['M'] = CHAR_PROP_ALPHA,
-    ['N'] = CHAR_PROP_ALPHA,
-    ['O'] = CHAR_PROP_ALPHA,
-    ['P'] = CHAR_PROP_ALPHA,
-    ['Q'] = CHAR_PROP_ALPHA,
-    ['R'] = CHAR_PROP_ALPHA,
-    ['S'] = CHAR_PROP_ALPHA,
-    ['T'] = CHAR_PROP_ALPHA,
-    ['U'] = CHAR_PROP_ALPHA,
-    ['V'] = CHAR_PROP_ALPHA,
-    ['W'] = CHAR_PROP_ALPHA,
-    ['X'] = CHAR_PROP_ALPHA,
-    ['Y'] = CHAR_PROP_ALPHA,
-    ['Z'] = CHAR_PROP_ALPHA,
+    ['0'] = CHAR_PROP_HEX | CHAR_PROP_DIG | CHAR_PROP_CGI_HEADER,
+    ['1'] = CHAR_PROP_HEX | CHAR_PROP_DIG | CHAR_PROP_CGI_HEADER,
+    ['2'] = CHAR_PROP_HEX | CHAR_PROP_DIG | CHAR_PROP_CGI_HEADER,
+    ['3'] = CHAR_PROP_HEX | CHAR_PROP_DIG | CHAR_PROP_CGI_HEADER,
+    ['4'] = CHAR_PROP_HEX | CHAR_PROP_DIG | CHAR_PROP_CGI_HEADER,
+    ['5'] = CHAR_PROP_HEX | CHAR_PROP_DIG | CHAR_PROP_CGI_HEADER,
+    ['6'] = CHAR_PROP_HEX | CHAR_PROP_DIG | CHAR_PROP_CGI_HEADER,
+    ['7'] = CHAR_PROP_HEX | CHAR_PROP_DIG | CHAR_PROP_CGI_HEADER,
+    ['8'] = CHAR_PROP_HEX | CHAR_PROP_DIG | CHAR_PROP_CGI_HEADER,
+    ['9'] = CHAR_PROP_HEX | CHAR_PROP_DIG | CHAR_PROP_CGI_HEADER,
+    ['a'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['b'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['c'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['d'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['e'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['f'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['g'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['h'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['i'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['j'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['k'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['l'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['m'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['n'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['o'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['p'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['q'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['r'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['s'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['t'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['u'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['v'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['w'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['x'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['y'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['z'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['A'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['B'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['C'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['D'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['E'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['F'] = CHAR_PROP_HEX | CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['G'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['H'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['I'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['J'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['K'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['L'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['M'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['N'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['O'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['P'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['Q'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['R'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['S'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['T'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['U'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['V'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['W'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['X'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['Y'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['Z'] = CHAR_PROP_ALPHA | CHAR_PROP_CGI_HEADER,
+    ['!'] = CHAR_PROP_CGI_HEADER,
+    ['#'] = CHAR_PROP_CGI_HEADER,
+    ['$'] = CHAR_PROP_CGI_HEADER,
+    ['%'] = CHAR_PROP_CGI_HEADER,
+    ['&'] = CHAR_PROP_CGI_HEADER,
+    ['\''] = CHAR_PROP_CGI_HEADER,
+    ['*'] = CHAR_PROP_CGI_HEADER,
+    ['+'] = CHAR_PROP_CGI_HEADER,
+    ['.'] = CHAR_PROP_CGI_HEADER,
+    ['^'] = CHAR_PROP_CGI_HEADER,
+    ['_'] = CHAR_PROP_CGI_HEADER,
+    ['`'] = CHAR_PROP_CGI_HEADER,
+    ['|'] = CHAR_PROP_CGI_HEADER,
+    ['~'] = CHAR_PROP_CGI_HEADER,
 };
 
 ALWAYS_INLINE uint8_t lwan_char_isspace(char ch)
 {
     return char_prop_tbl[(unsigned char)ch] & CHAR_PROP_SPACE;
 }
 
+ALWAYS_INLINE uint8_t lwan_char_iscgiheader(char ch)
+{
+    return char_prop_tbl[(unsigned char)ch] & CHAR_PROP_CGI_HEADER;
+}
+
 ALWAYS_INLINE uint8_t lwan_char_isxdigit(char ch)
 {
     return char_prop_tbl[(unsigned char)ch] & CHAR_PROP_HEX;