From 64cf30a54589f8dd9010097e02caa4d1bf8ceca0 Mon Sep 17 00:00:00 2001
From: Matthew Flatt <mflatt@racket-lang.org>
Date: Tue, 20 Feb 2024 07:53:00 -0700
Subject: [PATCH] adjustment and workaround for macOS iconv (#805)

* work around issues with macOS iconv library

* add arm64osx to CI

Closes #797
---
 .github/workflows/ci.yml |  4 ++++
 c/prim5.c                | 47 +++++++++++++++++++++++++++++++++++++++-
 c/version.h              |  2 ++
 mats/io.ms               |  8 +++++--
 4 files changed, 58 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 2892b53b7..4a3a627e2 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -15,6 +15,10 @@ jobs:
             os: macos-12
           - machine: ta6osx
             os: macos-12
+          - machine: arm64osx
+            os: macos-14
+          - machine: tarm64osx
+            os: macos-14
           - machine: i3le
             os: ubuntu-22.04
           - machine: ti3le
diff --git a/c/prim5.c b/c/prim5.c
index 90b087f12..78eaf0cf9 100644
--- a/c/prim5.c
+++ b/c/prim5.c
@@ -2273,6 +2273,50 @@ static void s_iconv_close(uptr cd) {
   ICONV_CLOSE((iconv_t)cd);
 }
 
+#ifdef DISTRUST_ICONV_PROGRESS
+# define ICONV_FROM iconv_fixup
+static size_t iconv_fixup(iconv_t cd, char **src, size_t *srcleft, char **dst, size_t *dstleft) {
+  size_t r;
+  char *orig_src = *src, *orig_dst = *dst;
+  size_t orig_srcleft = *srcleft, orig_dstleft = *dstleft, srcuntried = 0;
+
+  while (1) {
+    r = iconv((iconv_t)cd, src, srcleft, dst, dstleft);
+    if ((r == (size_t)-1)
+        && (errno == E2BIG)
+        && ((*srcleft < orig_srcleft) || (*dstleft < orig_dstleft))) {
+      /* Avoid a macOS (as of 14.2.1 and 14.3.1) iconv bug in this
+         case, where we don't trust that consumed input characters are
+         reflected in the output pointer. Reverting progress should be
+         ok for a correct iconv, too, since a -1 result means that no
+         irreversible progress was made. */
+      *src = orig_src;
+      *dst = orig_dst;
+      *srcleft = orig_srcleft;
+      *dstleft = orig_dstleft;
+
+      /* We need to make progress, if possible, to satify normal iconv
+         behavior and "io.ss" expectations. Try converting fewer
+         characters. */
+      if (orig_srcleft > sizeof(string_char)) {
+        size_t try_chars = (orig_srcleft / sizeof(string_char)) / 2;
+        srcuntried += orig_srcleft - (try_chars * sizeof(string_char));
+        orig_srcleft = try_chars * sizeof(string_char);
+        *srcleft = orig_srcleft;
+      } else
+        break;
+    } else
+      break;
+  }
+
+  *srcleft += srcuntried;
+
+  return r;
+}
+#else
+# define ICONV_FROM ICONV
+#endif
+
 #define ICONV_BUFSIZ 400
 
 static ptr s_iconv_from_string(uptr cd, ptr in, uptr i, uptr iend, ptr out, uptr o, uptr oend) {
@@ -2298,7 +2342,8 @@ static ptr s_iconv_from_string(uptr cd, ptr in, uptr i, uptr iend, ptr out, uptr
     under Windows, the iconv dll might have been linked against a different C runtime
     and might therefore set a different errno */
   errno = 0;
-  ICONV((iconv_t)cd, (ICONV_INBUF_TYPE)&inbuf, &inbytesleft, &outbuf, &outbytesleft);
+  ICONV_FROM((iconv_t)cd, (ICONV_INBUF_TYPE)&inbuf, &inbytesleft, &outbuf, &outbytesleft);
+
   new_i = i + inmax - inbytesleft / sizeof(string_char);
   new_o = oend - outbytesleft;
   if (new_i != i || new_o != o) return Scons(Sinteger(new_i), Sinteger(new_o));
diff --git a/c/version.h b/c/version.h
index e62922065..6eed74e1f 100644
--- a/c/version.h
+++ b/c/version.h
@@ -328,6 +328,8 @@ typedef int tputsputcchar;
 #define NSECCTIME(sb) (sb).st_ctimespec.tv_nsec
 #define NSECMTIME(sb) (sb).st_mtimespec.tv_nsec
 #define ICONV_INBUF_TYPE char **
+/* workaround issue in macOS 14.2.1 iconv: */
+#define DISTRUST_ICONV_PROGRESS
 #endif
 
 #if defined(__QNX__)
diff --git a/mats/io.ms b/mats/io.ms
index 98f68719c..06087b506 100644
--- a/mats/io.ms
+++ b/mats/io.ms
@@ -1012,7 +1012,9 @@
                 '()
                 (if (fx= i #xD800)
                     (f #xE000)
-                    (cons i (f (fx+ i 1)))))))
+                    (if (fx= i #xFEFF) ; avoid BOM, which an encoder is arguably justified in dropping
+                        (f (fx+ i 1))
+                        (cons i (f (fx+ i 1))))))))
         (define ls2
           (let f ([n 1000000])
             (if (fx= n 0)
@@ -1021,7 +1023,9 @@
                   (let ([n (random (- #x110000 (- #xE000 #xD800)))])
                     (if (<= #xD800 n #xDFFF)
                         (+ n (- #xE000 #xD800))
-                        n))
+                        (if (fx= n #xFEFF) ; avoid BOM
+                            #xFEFE
+                            n)))
                   (f (fx- n 1))))))
         (define s (apply string (map integer->char (append ls1 ls2))))
         #;(define s (apply string (map integer->char ls1)))