From bf96408b4e9ba0192d35bd8750a2e92815b657db Mon Sep 17 00:00:00 2001
From: Daniel Stenberg <daniel@haxx.se>
Date: Wed, 26 Apr 2023 09:57:03 +0200
Subject: [PATCH] offer {puny:host} and {puny:url} for --get

Added two tests. Added docs.

Requires >= 7.88.0 at build time.

Fixes #64
---
 URL-QUIRKS.md |  6 ++++++
 tests.json    | 30 ++++++++++++++++++++++++++++++
 trurl.1       |  5 +++++
 trurl.c       | 24 +++++++++++++++++++++++-
 4 files changed, 64 insertions(+), 1 deletion(-)

diff --git a/URL-QUIRKS.md b/URL-QUIRKS.md
index 0c2270f2..e3437167 100644
--- a/URL-QUIRKS.md
+++ b/URL-QUIRKS.md
@@ -42,6 +42,12 @@ Implemented in libcurl 7.81.0. Before this, the source formatting was kept.
 
 Run-time requirement.
 
+## `CURLU_PUNYCODE`
+
+Added in libcurl 7.88.0.
+
+Build-time requirement.
+
 ## Accepting % in host names
 
 The host name parser has been made stricter over time, with the most recent
diff --git a/tests.json b/tests.json
index 794c3bdd..2f8ed4b5 100644
--- a/tests.json
+++ b/tests.json
@@ -1594,6 +1594,36 @@
             "returncode": 0,
             "stderr": ""
         }
+    },
+    {
+        "input": {
+            "arguments": [
+                "https://räksmörgås.se",
+                "-g",
+                "{puny:url}"
+            ]
+        },
+        "minbuildtime": "7.88.0",
+        "expected": {
+            "stderr": "",
+            "returncode": 0,
+            "stdout": "https://xn--rksmrgs-5wao1o.se:443/\n"
+        }
+    },
+    {
+        "input": {
+            "arguments": [
+                "https://räksmörgås.se",
+                "-g",
+                "{puny:host}"
+            ]
+        },
+        "minbuildtime": "7.88.0",
+        "expected": {
+            "stderr": "",
+            "returncode": 0,
+            "stdout": "xn--rksmrgs-5wao1o.se\n"
+        }
     }
 
             ]
diff --git a/trurl.1 b/trurl.1
index 47e1bc6d..357671fd 100644
--- a/trurl.1
+++ b/trurl.1
@@ -76,6 +76,11 @@ You can access specific keys in the query string and out all values using the
 format \fB{query-all:key}\fP. This looks for 'key' case sensitively and will
 output all values for that key space-separated.
 
+You can access the url and host components in their "punycoded" version, which
+is how International Domain Names are converted into plain ascii, by using the
+form \fB{puny:yrl}\fP and \fB{puny:host}\fP. If the host name is not using
+IDN, this option provides the regular ascii name.
+
 The "format" string supports the following backslash sequences:
 
 \&\\\\ - backslash
diff --git a/trurl.c b/trurl.c
index 5f0cdca8..41ac306f 100644
--- a/trurl.c
+++ b/trurl.c
@@ -29,6 +29,8 @@
 #include <curl/curl.h>
 #include <curl/mprintf.h>
 
+#include <locale.h> /* for setlocale() */
+
 #include "version.h"
 
 #ifdef _MSC_VER
@@ -48,6 +50,9 @@
 #else
 #define CURLU_ALLOW_SPACE 0
 #endif
+#if CURL_AT_LEAST_VERSION(7,88,0)
+#define SUPPORTS_PUNYCODE
+#endif
 
 #define OUTPUT_URL      0  /* default */
 #define OUTPUT_SCHEME   1
@@ -480,6 +485,8 @@ static void get(struct option *op, CURLU *uh)
         char *cl;
         size_t vlen;
         bool urldecode = true;
+        bool punycode = false;
+        bool handled = true;
         end = strchr(ptr, endbyte);
         ptr++; /* pass the { */
         if(!end) {
@@ -501,10 +508,21 @@ static void get(struct option *op, CURLU *uh)
             showqkey(&ptr[10], end - cl - 1, urldecode, true);
           else if(!strncmp(ptr, "query:", 6))
             showqkey(&ptr[6], end - cl - 1, urldecode, false);
+          else if(!strncmp(ptr, "puny:", 5)) {
+            punycode = true;
+#ifndef SUPPORTS_PUNYCODE
+            warnf("Built without punycode support");
+#endif
+            ptr = cl + 1;
+            vlen = end - ptr;
+            handled = false;
+          }
           else
             errorf(ERROR_GET, "Bad --get syntax: %s", ptr);
         }
-        else {
+        else
+          handled = false;
+        if(!handled) {
           const struct var *v = comp2var(ptr, vlen);
           if(v) {
             char *nurl;
@@ -512,6 +530,9 @@ static void get(struct option *op, CURLU *uh)
             rc = curl_url_get(uh, v->part, &nurl,
                               CURLU_DEFAULT_PORT|
                               CURLU_NO_DEFAULT_PORT|
+#ifdef SUPPORTS_PUNYCODE
+                              (punycode?CURLU_PUNYCODE:0)|
+#endif
                               (urldecode?CURLU_URLDECODE:0));
             switch(rc) {
             case CURLUE_OK:
@@ -1109,6 +1130,7 @@ int main(int argc, const char **argv)
   struct option o;
   struct curl_slist *node;
   memset(&o, 0, sizeof(o));
+  setlocale(LC_ALL, "");
   curl_global_init(CURL_GLOBAL_ALL);
 
   for(argc--, argv++; argc > 0; argc--, argv++) {