Define new ne_strparam() interface for encoding extended parameters

per RFC 5987.

* src/ne_string.c (ne_strparam): New function.

* test/string-tests.c (strparam): New test.

4 files changed, 111 insertions, 0 deletions

diff --git a/src/ne_string.c b/src/ne_string.c
index 613c9cd..187fa58 100644
--- a/src/ne_string.c
+++ b/src/ne_string.c
@@ -653,3 +653,72 @@ char *ne_vstrhash(unsigned int flags, va_list ap)
     return ne_strdup(ret);
 }
 #endif
+
+/* Determines whether a character is valid in a regular parameter (NQ)
+ * not (QT). Per https://tools.ietf.org/html/rfc5987#section-3.2.1
+ * every character in attr-char is NQ, everything else is QT. */
+#define QT 3
+#define NQ 1
+static const unsigned char ext_notation[256] = {
+/* 0xXX    x0      x2      x4      x6      x8      xA      xC      xE     */
+/*   0x */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT,
+/*   1x */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT,
+/*   2x */ QT, NQ, QT, NQ, NQ, QT, NQ, QT, QT, QT, QT, NQ, QT, NQ, NQ, QT,
+/*   3x */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT,
+/*   4x */ QT, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ,
+/*   5x */ NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, QT, QT, QT, NQ, NQ,
+/*   6x */ NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ,
+/*   7x */ NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, NQ, QT, NQ, QT, NQ, QT,
+/*   8x */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, 
+/*   9x */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, 
+/*   Ax */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, 
+/*   Bx */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, 
+/*   Cx */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, 
+/*   Dx */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, 
+/*   Ex */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, 
+/*   Fx */ QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT, QT
+};
+#undef QT
+#undef NQ
+
+char *ne_strparam(const char *charset, const char *lang,
+                  const unsigned char *value)
+{
+    const unsigned char *p;
+    size_t count = 0;
+    char *rv, *rp;
+
+    /* Determine length required for the value. */
+    for (p = value; *p; p++)
+        count += ext_notation[*p];
+
+    /* If length == input length, no encoding is required, return
+     * NULL. */
+    if (count == strlen((const char *)value)) return NULL;
+
+    /* +3 accounts for '' and trailing NUL */
+    rv = ne_malloc(strlen(charset) + (lang ? strlen(lang) : 0) + count + 3);
+    memcpy(rv, charset, strlen(charset));
+    rp = rv + strlen(charset);
+    *rp++ = '\'';
+    if (lang) {
+        memcpy(rp, lang, strlen(lang));
+        rp += strlen(lang);
+    }
+    *rp++ = '\'';
+
+    for (p = value; *p; p++) {
+        if (ext_notation[*p] == 1)  {
+            *rp++ = *p;
+        }
+        else {
+            *rp++ = '%';
+            *rp++ = NE_HEX2ASC(*p >> 4);
+            *rp++ = NE_HEX2ASC(*p & 0x0f);
+        }
+    }
+
+    *rp = '\0';
+
+    return rv;
+}
diff --git a/src/ne_string.h b/src/ne_string.h
index a818005..34edd06 100644
--- a/src/ne_string.h
+++ b/src/ne_string.h
@@ -199,6 +199,15 @@ const unsigned char *ne_tolower_array(void) ne_attribute((const));
  * ASCII hexadecimal equivalent character, in the range '0..9,'a..f' */
 #define NE_HEX2ASC(x) ((char) ((x) > 9 ? ((x) - 10 + 'a') : ((x) + '0')))
 
+/* Encodes a extended parameter value for HTTP headers, as defined in
+ * RFC 5987.  Returns a malloc-allocated string if the parameter
+ * 'value' needs to be encoded as an extended parameter, or NULL if it
+ * can be used as a regular parameter.  The charset of the string must
+ * be non-NULL, but the language value can be NULL. */
+char *ne_strparam(const char *charset, const char *lang,
+                  const unsigned char *value)
+    ne_attribute((nonnull (1, 3)));
+
 NE_END_DECLS
 
 #endif /* NE_STRING_H */
diff --git a/src/neon.vers b/src/neon.vers
index de4b327..b969360 100644
--- a/src/neon.vers
+++ b/src/neon.vers
@@ -29,4 +29,5 @@ NEON_0_31 {
 NEON_0_32 {
     ne_strhash;
     ne_vstrhash;
+    ne_strparam;
 };
diff --git a/test/string-tests.c b/test/string-tests.c
index 03911c4..0be11dd 100644
--- a/test/string-tests.c
+++ b/test/string-tests.c
@@ -685,6 +685,37 @@ static int strhash(void)
     return OK;
 }
 
+static int strparam(void)
+{
+    static const struct {
+        const char *charset, *lang;
+        const char *value;
+        const char *expect;
+    } ts[] = {
+        { "UTF-8", NULL, "foobar", NULL },
+        { "UTF-8", NULL, "foo@bar", "UTF-8''foo%40bar" },
+        { "UTF-8", NULL, "foo bar", "UTF-8''foo%20bar" },
+        { "iso-8859-1", "en", "\xA3 rates", "iso-8859-1'en'%a3%20rates" },
+        { "UTF-8", NULL, "£ and € rates", "UTF-8''%c2%a3%20and%20%e2%82%ac%20rates" },
+        { NULL }
+    };
+    unsigned n;
+
+    for (n = 0; ts[n].charset; n++) {
+        char *act = ne_strparam(ts[n].charset, ts[n].lang, (const unsigned char *)ts[n].value);
+
+        if (ts[n].expect == NULL) {
+            ONV(act != NULL, ("expected NULL output for '%s'", ts[n].value));
+        }
+        else {
+            ONCMP(act, ts[n].expect);
+        }
+    }
+
+    return OK;
+}
+
+
 ne_test tests[] = {
     T(simple),
     T(buf_concat),
@@ -714,6 +745,7 @@ ne_test tests[] = {
     T(buf_print),
     T(qappend),
     T(strhash),
+    T(strparam),
     T(NULL)
 };