From e8a5ee327d0b5cfb7216f56b1017eefbd58ae3c5 Mon Sep 17 00:00:00 2001
From: Jean-Francois Dockes <jf@dockes.org>
Date: Wed, 30 Sep 2020 12:31:51 +0100
Subject: [PATCH] Windows wide path conversions: don't use SYSPATH outside
 pathut.cpp

---
 src/utils/pathut.cpp     | 33 +++++++++++++++++++++++++++++++++
 src/utils/pathut.h       | 12 ++++--------
 src/utils/readfile.cpp   | 15 ++++++++++++---
 src/windows/execmd_w.cpp |  4 ++--
 4 files changed, 51 insertions(+), 13 deletions(-)

diff --git a/src/utils/pathut.cpp b/src/utils/pathut.cpp
index 82cf7afd..b0df3f53 100644
--- a/src/utils/pathut.cpp
+++ b/src/utils/pathut.cpp
@@ -78,6 +78,7 @@
 #include <dirent.h>
 #endif // _MSC_VER
 
+
 #ifdef _WIN32
 
 #ifndef _MSC_VER
@@ -134,6 +135,10 @@
 #define RMDIR _wrmdir
 #define CHDIR _wchdir
 
+#define SYSPATH(PATH, SPATH) wchar_t PATH ## _buf[2048];      \
+    utf8towchar(PATH, PATH ## _buf, 2048);                    \
+    wchar_t *SPATH = PATH ## _buf;
+
 #define ftruncate _chsize_s
 
 #ifdef _MSC_VER
@@ -172,6 +177,9 @@
 #define RMDIR ::rmdir
 #define CHDIR ::chdir
 
+#define SYSPATH(PATH, SPATH) const char *SPATH = PATH.c_str()
+
+
 #endif /* !_WIN32 */
 
 using namespace std;
@@ -256,6 +264,31 @@ bool utf8towchar(const std::string& in, wchar_t *out, size_t obytescap)
     return true;
 }
 
+std::unique_ptr<wchar_t[]> utf8towchar(const std::string& in)
+{
+    // Note that as we supply in.size(), mbtowch computes the size
+    // without a terminating 0 (and won't write in the second call of
+    // course). We take this into account by allocating one more and
+    // terminating the output.
+    int wcharcnt = MultiByteToWideChar(
+        CP_UTF8, MB_ERR_INVALID_CHARS, in.c_str(), in.size(), nullptr, 0);
+    if (wcharcnt <= 0) {
+        LOGERR("utf8towchar: conversion error for [" << in << "]\n");
+        return std::unique_ptr<wchar_t[]>();
+    }
+    auto buf = unique_ptr<wchar_t[]>(new wchar_t[wcharcnt+1]);
+
+    wcharcnt = MultiByteToWideChar(
+        CP_UTF8, MB_ERR_INVALID_CHARS, in.c_str(), in.size(),
+        buf.get(), wcharcnt);
+    if (wcharcnt <= 0) {
+        LOGERR("utf8towchar: conversion error for [" << in << "]\n");
+        return std::unique_ptr<wchar_t[]>();
+    }
+    buf.get()[wcharcnt] = 0;
+    return buf;
+}
+
 /// Convert \ separators to /
 void path_slashize(string& s)
 {
diff --git a/src/utils/pathut.h b/src/utils/pathut.h
index ab1f33dc..f969d0d6 100644
--- a/src/utils/pathut.h
+++ b/src/utils/pathut.h
@@ -123,14 +123,10 @@ extern int path_fileprops(const std::string path, struct PathStat *stp,
 extern std::string path_PATHsep();
 
 #ifdef _WIN32
-extern bool wchartoutf8(const wchar_t *in, std::string& out, size_t len = 0);
-extern std::string wchartoutf8(const wchar_t *in, size_t len = 0);
-extern bool utf8towchar(const std::string& in, wchar_t *out, size_t obytescap);
-#define SYSPATH(PATH, SPATH) wchar_t PATH ## _buf[2048];      \
-    utf8towchar(PATH, PATH ## _buf, 2048);                    \
-    wchar_t *SPATH = PATH ## _buf;
-#else
-#define SYSPATH(PATH, SPATH) const char *SPATH = PATH.c_str()
+bool wchartoutf8(const wchar_t *in, std::string& out, size_t len = 0);
+std::string wchartoutf8(const wchar_t *in, size_t len = 0);
+bool utf8towchar(const std::string& in, wchar_t *out, size_t obytescap);
+std::unique_ptr<wchar_t[]> utf8towchar(const std::string& in);
 #endif
 
 /// Directory reading interface. UTF-8 on Windows.
diff --git a/src/utils/readfile.cpp b/src/utils/readfile.cpp
index 1cc1dac9..0ad5c2a9 100644
--- a/src/utils/readfile.cpp
+++ b/src/utils/readfile.cpp
@@ -32,7 +32,6 @@
 #include "safefcntl.h"
 #include "safesysstat.h"
 #include "safeunistd.h"
-#include "transcode.h"
 #define OPEN _wopen
 
 #else
@@ -333,7 +332,12 @@ public:
 
         // If we have a file name, open it, else use stdin.
         if (!m_fn.empty()) {
-            SYSPATH(m_fn, realpath);
+#ifdef _WIN32
+            auto buf = utf8towchar(m_fn);
+            auto realpath = buf.get();
+#else
+            auto realpath = m_fn.c_str();
+#endif
             fd = OPEN(realpath, O_RDONLY | O_BINARY);
             if (fd < 0 || fstat(fd, &st) < 0) {
                 catstrerror(m_reason, "open/stat", errno);
@@ -440,7 +444,12 @@ public:
         if (m_fn.empty()) {
             ret1 = mz_zip_reader_init_mem(&zip, m_data, m_cnt, 0);
         } else {
-            SYSPATH(m_fn, realpath);
+#ifdef _WIN32
+            auto buf = utf8towchar(m_fn);
+            auto realpath = buf.get();
+#else
+            auto realpath = m_fn.c_str();
+#endif
             ret1 = mz_zip_reader_init_file(&zip, realpath, 0);
         }
         if (!ret1) {
diff --git a/src/windows/execmd_w.cpp b/src/windows/execmd_w.cpp
index dd6711a7..c7412866 100644
--- a/src/windows/execmd_w.cpp
+++ b/src/windows/execmd_w.cpp
@@ -793,9 +793,9 @@ int ExecCmd::startExec(const string &cmd, const vector<string>& args,
     int flags = CREATE_NEW_PROCESS_GROUP | CREATE_UNICODE_ENVIRONMENT;
     // Create the child process. 
     LOGDEB("ExecCmd:startExec: cmdline [" << cmdline << "]\n");
-    SYSPATH(cmdline, wcmdline);
+    auto wcmdline = utf8towchar(cmdline);
     bSuccess = CreateProcessW(NULL,         // app name
-                              wcmdline,     // command line
+                              wcmdline.get(),     // command line
                               NULL,         // process security attributes
                               NULL,         // primary thread security attrs
                               TRUE,         // handles are inherited