urlapi: add CURLU_PUNYCODE

Allows curl_url_get() get the punycode version of host names for the
host name and URL parts.

Extend test 1560 to verify.

Closes #10109

10 files changed, 90 insertions, 14 deletions

diff --git a/.github/scripts/spellcheck.words b/.github/scripts/spellcheck.words
index 9f3906b80..ea5115be4 100644
--- a/.github/scripts/spellcheck.words
+++ b/.github/scripts/spellcheck.words
@@ -573,6 +573,7 @@ PSL
 pthreads
 PTR
 ptr
+punycode
 py
 pycurl
 QNX
diff --git a/docs/libcurl/curl_url_get.3 b/docs/libcurl/curl_url_get.3
index d86d93754..2ae21e4fd 100644
--- a/docs/libcurl/curl_url_get.3
+++ b/docs/libcurl/curl_url_get.3
@@ -76,6 +76,17 @@ typically using non-ASCII bytes that otherwise will be percent-encoded.
 
 Note that even when not asking for URL encoding, the '%' (byte 37) will be URL
 encoded to make sure the host name remains valid.
+.IP CURLU_PUNYCODE
+If set and \fICURLU_URLENCODE\fP is not set, and asked to retrieve the
+\fBCURLUPART_HOST\fP or \fBCURLUPART_URL\fP parts, libcurl returns the host
+name in its punycode version if it contains any non-ASCII octets (and is an
+IDN name).
+
+If libcurl is built without IDN capabilities, using this bit will make
+\fIcurl_url_get(3)\fP return \fICURLUE_LACKS_IDN\fP if the host name contains
+anything outside the ASCII range.
+
+(Added in curl 7.88.0)
 .SH PARTS
 .IP CURLUPART_URL
 When asked to return the full URL, \fIcurl_url_get(3)\fP will return a
diff --git a/docs/libcurl/symbols-in-versions b/docs/libcurl/symbols-in-versions
index 5ee245d36..6de01a08c 100644
--- a/docs/libcurl/symbols-in-versions
+++ b/docs/libcurl/symbols-in-versions
@@ -1055,6 +1055,7 @@ CURLU_NO_AUTHORITY              7.67.0
 CURLU_NO_DEFAULT_PORT           7.62.0
 CURLU_NON_SUPPORT_SCHEME        7.62.0
 CURLU_PATH_AS_IS                7.62.0
+CURLU_PUNYCODE                  7.88.0
 CURLU_URLDECODE                 7.62.0
 CURLU_URLENCODE                 7.62.0
 CURLUE_BAD_FILE_URL             7.81.0
@@ -1071,6 +1072,7 @@ CURLUE_BAD_QUERY                7.81.0
 CURLUE_BAD_SCHEME               7.81.0
 CURLUE_BAD_SLASHES              7.81.0
 CURLUE_BAD_USER                 7.81.0
+CURLUE_LACKS_IDN                7.88.0
 CURLUE_MALFORMED_INPUT          7.62.0
 CURLUE_NO_FRAGMENT              7.62.0
 CURLUE_NO_HOST                  7.62.0
diff --git a/include/curl/urlapi.h b/include/curl/urlapi.h
index e15c213cc..9e21df165 100644
--- a/include/curl/urlapi.h
+++ b/include/curl/urlapi.h
@@ -62,6 +62,7 @@ typedef enum {
   CURLUE_BAD_SCHEME,          /* 27 */
   CURLUE_BAD_SLASHES,         /* 28 */
   CURLUE_BAD_USER,            /* 29 */
+  CURLUE_LACKS_IDN,           /* 30 */
   CURLUE_LAST
 } CURLUcode;
 
@@ -95,6 +96,7 @@ typedef enum {
 #define CURLU_NO_AUTHORITY (1<<10)      /* Allow empty authority when the
                                            scheme is unknown. */
 #define CURLU_ALLOW_SPACE (1<<11)       /* Allow spaces in the URL */
+#define CURLU_PUNYCODE (1<<12)          /* get the host name in pynycode */
 
 typedef struct Curl_URL CURLU;
 
diff --git a/lib/idn.c b/lib/idn.c
index 6255221ae..758900bdd 100644
--- a/lib/idn.c
+++ b/lib/idn.c
@@ -116,7 +116,7 @@ bool Curl_is_ASCII_name(const char *hostname)
  * Curl_idn_decode() returns an allocated IDN decoded string if it was
  * possible. NULL on error.
  */
-static char *Curl_idn_decode(const char *input)
+static char *idn_decode(const char *input)
 {
   char *decoded = NULL;
 #ifdef USE_LIBIDN2
@@ -144,24 +144,29 @@ static char *Curl_idn_decode(const char *input)
   return decoded;
 }
 
+char *Curl_idn_decode(const char *input)
+{
+  char *d = idn_decode(input);
+#ifdef USE_LIBIDN2
+  if(d) {
+    char *c = strdup(d);
+    idn2_free(d);
+    d = c;
+  }
+#endif
+  return d;
+}
+
 /*
  * Frees data allocated by idnconvert_hostname()
  */
 void Curl_free_idnconverted_hostname(struct hostname *host)
 {
-#if defined(USE_LIBIDN2)
   if(host->encalloc) {
-    idn2_free(host->encalloc); /* must be freed with idn2_free() since this was
-                                  allocated by libidn */
+    /* must be freed with idn2_free() if allocated by libidn */
+    Curl_idn_free(host->encalloc);
     host->encalloc = NULL;
   }
-#elif defined(USE_WIN32_IDN)
-  free(host->encalloc); /* must be freed with free() since this was
-                           allocated by Curl_win32_idn_to_ascii */
-  host->encalloc = NULL;
-#else
-  (void)host;
-#endif
 }
 
 #endif /* USE_IDN */
@@ -177,7 +182,7 @@ CURLcode Curl_idnconvert_hostname(struct hostname *host)
 #ifdef USE_IDN
   /* Check name for non-ASCII and convert hostname if we can */
   if(!Curl_is_ASCII_name(host->name)) {
-    char *decoded = Curl_idn_decode(host->name);
+    char *decoded = idn_decode(host->name);
     if(decoded) {
       /* successful */
       host->encalloc = decoded;
@@ -190,4 +195,3 @@ CURLcode Curl_idnconvert_hostname(struct hostname *host)
 #endif
   return CURLE_OK;
 }
-
diff --git a/lib/idn.h b/lib/idn.h
index 2d04efcdf..c7adf97e6 100644
--- a/lib/idn.h
+++ b/lib/idn.h
@@ -32,7 +32,15 @@ CURLcode Curl_idnconvert_hostname(struct hostname *host);
 #if defined(USE_LIBIDN2) || defined(USE_WIN32_IDN)
 #define USE_IDN
 void Curl_free_idnconverted_hostname(struct hostname *host);
+char *Curl_idn_decode(const char *input);
+#ifdef USE_LIBIDN2
+#define Curl_idn_free(x) idn2_free(x)
+#else
+#define Curl_idn_free(x) free(x)
+#endif
+
 #else
 #define Curl_free_idnconverted_hostname(x)
+#define Curl_idn_decode(x) NULL
 #endif
 #endif /* HEADER_CURL_IDN_H */
diff --git a/lib/strerror.c b/lib/strerror.c
index b9a51e26b..3d7193509 100644
--- a/lib/strerror.c
+++ b/lib/strerror.c
@@ -550,6 +550,9 @@ curl_url_strerror(CURLUcode error)
   case CURLUE_BAD_USER:
     return "Bad user";
 
+  case CURLUE_LACKS_IDN:
+    return "libcurl lacks IDN support";
+
   case CURLUE_LAST:
     break;
   }
diff --git a/lib/urlapi.c b/lib/urlapi.c
index b96af35ad..29238c7e3 100644
--- a/lib/urlapi.c
+++ b/lib/urlapi.c
@@ -33,6 +33,7 @@
 #include "inet_pton.h"
 #include "inet_ntop.h"
 #include "strdup.h"
+#include "idn.h"
 
 /* The last 3 #include files should be in this order */
 #include "curl_printf.h"
@@ -1379,6 +1380,7 @@ CURLUcode curl_url_get(CURLU *u, CURLUPart what,
   char portbuf[7];
   bool urldecode = (flags & CURLU_URLDECODE)?1:0;
   bool urlencode = (flags & CURLU_URLENCODE)?1:0;
+  bool punycode = FALSE;
   bool plusdecode = FALSE;
   (void)flags;
   if(!u)
@@ -1408,6 +1410,7 @@ CURLUcode curl_url_get(CURLU *u, CURLUPart what,
   case CURLUPART_HOST:
     ptr = u->host;
     ifmissing = CURLUE_NO_HOST;
+    punycode = (flags & CURLU_PUNYCODE)?1:0;
     break;
   case CURLUPART_ZONEID:
     ptr = u->zoneid;
@@ -1460,6 +1463,7 @@ CURLUcode curl_url_get(CURLU *u, CURLUPart what,
     char *options = u->options;
     char *port = u->port;
     char *allochost = NULL;
+    punycode = (flags & CURLU_PUNYCODE)?1:0;
     if(u->scheme && strcasecompare("file", u->scheme)) {
       url = aprintf("file://%s%s%s",
                     u->path,
@@ -1514,6 +1518,17 @@ CURLUcode curl_url_get(CURLU *u, CURLUPart what,
         if(!allochost)
           return CURLUE_OUT_OF_MEMORY;
       }
+      else if(punycode) {
+        if(!Curl_is_ASCII_name(u->host)) {
+#ifndef USE_IDN
+          return CURLUE_LACKS_IDN;
+#else
+          allochost = Curl_idn_decode(u->host);
+          if(!allochost)
+            return CURLUE_OUT_OF_MEMORY;
+#endif
+        }
+      }
       else {
         /* only encode '%' in output host name */
         char *host = u->host;
@@ -1611,6 +1626,19 @@ CURLUcode curl_url_get(CURLU *u, CURLUPart what,
       free(*part);
       *part = Curl_dyn_ptr(&enc);
     }
+    else if(punycode) {
+      if(!Curl_is_ASCII_name(u->host)) {
+#ifndef USE_IDN
+        return CURLUE_LACKS_IDN;
+#else
+        char *allochost = Curl_idn_decode(*part);
+        if(!allochost)
+          return CURLUE_OUT_OF_MEMORY;
+        free(*part);
+        *part = allochost;
+#endif
+      }
+    }
 
     return CURLUE_OK;
   }
diff --git a/tests/data/test1538 b/tests/data/test1538
index ba5cf29ce..59cd1628e 100644
--- a/tests/data/test1538
+++ b/tests/data/test1538
@@ -185,7 +185,8 @@ u26: Bad query
 u27: Bad scheme
 u28: Unsupported number of slashes following scheme
 u29: Bad user
-u30: CURLUcode unknown
+u30: libcurl lacks IDN support
+u31: CURLUcode unknown
 </stdout>
 </verify>
 
diff --git a/tests/libtest/lib1560.c b/tests/libtest/lib1560.c
index 570e5082d..f28bdf111 100644
--- a/tests/libtest/lib1560.c
+++ b/tests/libtest/lib1560.c
@@ -31,6 +31,9 @@
  */
 
 #include "test.h"
+#if defined(USE_LIBIDN2) || defined(USE_WIN32_IDN)
+#define USE_IDN
+#endif
 
 #include "testutil.h"
 #include "warnless.h"
@@ -138,6 +141,15 @@ struct clearurlcase {
 };
 
 static const struct testcase get_parts_list[] ={
+#ifdef USE_IDN
+  {"https://räksmörgås.se",
+   "https | [11] | [12] | [13] | xn--rksmrgs-5wao1o.se | "
+   "[15] | / | [16] | [17]", 0, CURLU_PUNYCODE, CURLUE_OK},
+#else
+  {"https://räksmörgås.se",
+   "https | [11] | [12] | [13] | [30] | [15] | / | [16] | [17]",
+   0, CURLU_PUNYCODE, CURLUE_OK},
+#endif
   /* https://ℂᵤⓇℒ。𝐒🄴 */
   {"https://"
    "%e2%84%82%e1%b5%a4%e2%93%87%e2%84%92%e3%80%82%f0%9d%90%92%f0%9f%84%b4",
@@ -454,6 +466,10 @@ static const struct testcase get_parts_list[] ={
 };
 
 static const struct urltestcase get_url_list[] = {
+#ifdef USE_IDN
+  {"https://räksmörgås.se/path?q#frag",
+   "https://xn--rksmrgs-5wao1o.se/path?q#frag", 0, CURLU_PUNYCODE, CURLUE_OK},
+#endif
   /* unsupported schemes with no guessing enabled */
   {"data:text/html;charset=utf-8;base64,PCFET0NUWVBFIEhUTUw+PG1ldGEgY",
    "", 0, 0, CURLUE_UNSUPPORTED_SCHEME},