kpathsea [win32]: Allow Unicode file name in kpsewhich, add new functions in knj.[ch]

git-svn-id: svn://tug.org/texlive/trunk@32783 c570f23f-e606-0410-a88d-b1316a301751

6 files changed, 152 insertions, 4 deletions

diff --git a/Build/source/texk/kpathsea/ChangeLog b/Build/source/texk/kpathsea/ChangeLog
index 0d8dfdf6697..18623145e16 100644
--- a/Build/source/texk/kpathsea/ChangeLog
+++ b/Build/source/texk/kpathsea/ChangeLog
@@ -1,3 +1,11 @@
+2014-01-25  TANAKA Takuji  <KXD02663@nifty.ne.jp>
+
+	* knj.[ch]: Add a new function win32_getc(), win32_ungetc(),
+	win32_fputs(), and win32_puts(). (Windows only)
+	* kpsewhich.c, pathsearch.c, line.c:
+	Allow Unicode file name in kpsewhich. (Windows only)
+	* knj.c: Fix typo in error message. (Windows only)
+
 2014-01-14  Peter Breitenlohner  <peb@mppmu.mpg.de>
 
 	* knj.c: Add #include <wchar.h> as required for MinGW32.
diff --git a/Build/source/texk/kpathsea/knj.c b/Build/source/texk/kpathsea/knj.c
index 52b4a43b9bc..cd9eb474129 100644
--- a/Build/source/texk/kpathsea/knj.c
+++ b/Build/source/texk/kpathsea/knj.c
@@ -84,7 +84,7 @@ get_mbstring_from_wstring(int cp, const wchar_t *wstr, char *mbstr)
 
   len = WideCharToMultiByte(cp, 0, wstr, -1, mbstr, 0, NULL, NULL);
   if (len==0) {
-    FATAL("cannot convert string to wide string");
+    FATAL("cannot convert string to multibyte string");
   }
   if (mbstr==NULL) {
     mbstr = xmalloc(len+1);
@@ -200,13 +200,13 @@ fsyscp_popen (const char *command, const char *mode)
     return f;
 }
 
-void
+int
 get_command_line_args_utf8 (const_string enc, int *p_ac, char ***p_av)
 {
     int argc;
     string *argv;
 
-    if (!enc || !strncmp(enc,"",1)) return;
+    if (!enc || !strncmp(enc,"",1)) return 0;
 
 #ifdef DEBUG
     fprintf(stderr, "command_line_encoding (%s)\n", enc);
@@ -237,8 +237,10 @@ get_command_line_args_utf8 (const_string enc, int *p_ac, char ***p_av)
       argv[argcw] = NULL;
       *p_ac = argc;
       *p_av = argv;
+      return file_system_codepage;
     } else {
       WARNING1("kpathsea: Ignoring unknown encoding `%s'", enc);
+      return 0;
     }
 }
 
@@ -331,3 +333,98 @@ fsyscp_system (const char *cmd)
     free (av[2]);
     return ret;
 }
+
+static int getc_len;
+static int getc_buff[4];
+
+int win32_getc(FILE *fp)
+{
+    const int fd = fileno(fp);
+    HANDLE hStdin;
+    DWORD ret;
+    wchar_t wc[3];
+    char mbc[5];
+    int j;
+    static wchar_t wcbuf = L'\0';
+
+    if (!(fd == fileno(stdin) && _isatty(fd) && file_system_codepage == CP_UTF8))
+        return getc(fp);
+
+    if (getc_len == 0)
+    {
+        hStdin = GetStdHandle(STD_INPUT_HANDLE);
+        if (wcbuf) {
+            wc[0] = wcbuf;
+            wcbuf = L'\0';
+        }
+        else if (ReadConsoleW(hStdin, wc, 1, &ret, NULL) == 0)
+            return EOF;
+        if (0xd800<=wc[0] && wc[0]<0xdc00) {
+            if (ReadConsoleW(hStdin, wc+1, 1, &ret, NULL) == 0)
+                return EOF;
+            if (0xdc00<=wc[1] && wc[1]<0xe000) {
+                wc[2]=L'\0';
+            } else {
+                wcbuf=wc[1];
+                wc[0]=0xfffd;    /* illegal surrogate pair */
+                wc[1]=L'\0';
+            }
+        } else if (0xdc00<=wc[0] && wc[0]<0xe000) {
+            wc[0]=0xfffd;        /* illegal surrogate pair */
+            wc[1]=L'\0';
+        } else {
+            wc[1]=L'\0';
+        }
+        get_utf8_from_wstring(wc,mbc);
+        j=strlen(mbc)-1;
+        while(j>=0) {
+            getc_buff[getc_len++]=(int)mbc[j--];
+        }
+    }
+    return getc_buff[--getc_len];
+}
+
+int win32_ungetc(int c, FILE *fp)
+{
+    const int fd = fileno(fp);
+
+    if (!(fd == fileno(stdin) && _isatty(fd) && file_system_codepage == CP_UTF8))
+        return ungetc(c, fp);
+
+    assert(getc_len < 4);
+    return getc_buff[getc_len++] = c;
+}
+
+int win32_fputs(const char *str, FILE *fp)
+{
+    const int fd = fileno(fp);
+    HANDLE hStdout;
+    DWORD ret;
+    wchar_t *wstr;
+    int j;
+
+    if (!((fd == fileno(stdout) || fd == fileno(stderr)) && _isatty(fd)
+        && file_system_codepage == CP_UTF8))
+        return fputs(str, fp);
+
+    hStdout = (fd == fileno(stdout)) ?
+        GetStdHandle(STD_OUTPUT_HANDLE) : GetStdHandle(STD_ERROR_HANDLE);
+
+    wstr = get_wstring_from_utf8(str, wstr=NULL);
+
+    if (WriteConsoleW(hStdout, wstr, wcslen(wstr), &ret, NULL) == 0) {
+        free(wstr);
+        return EOF;
+    }
+
+    free(wstr);
+    return ret;
+}
+
+int win32_puts(const char *str)
+{
+    if (win32_fputs(str, stdout)==EOF) {
+        return EOF;
+    }
+    return puts("");
+}
diff --git a/Build/source/texk/kpathsea/knj.h b/Build/source/texk/kpathsea/knj.h
index 0e942d518d6..df73074b8ad 100644
--- a/Build/source/texk/kpathsea/knj.h
+++ b/Build/source/texk/kpathsea/knj.h
@@ -36,7 +36,11 @@ extern KPSEDLL FILE* fsyscp_fopen(const char *filename, const char *mode);
 extern KPSEDLL FILE* fsyscp_popen(const char *command, const char *mode);
 extern KPSEDLL int fsyscp_spawnvp(int mode, const char *command, const char* const *argv);
 extern KPSEDLL int fsyscp_system(const char *cmd);
-extern KPSEDLL void get_command_line_args_utf8(const char *enc, int *p_ac, char ***p_av);
+extern KPSEDLL int get_command_line_args_utf8(const char *enc, int *p_ac, char ***p_av);
+extern KPSEDLL int win32_getc(FILE *fp);
+extern KPSEDLL int win32_ungetc(int c, FILE *fp);
+extern KPSEDLL int win32_fputs(const char *str, FILE *fp);
+extern KPSEDLL int win32_puts(const char *str);
 
 #ifdef __cplusplus
 }
diff --git a/Build/source/texk/kpathsea/kpsewhich.c b/Build/source/texk/kpathsea/kpsewhich.c
index 4bcb77ac48b..59739e56c9c 100644
--- a/Build/source/texk/kpathsea/kpsewhich.c
+++ b/Build/source/texk/kpathsea/kpsewhich.c
@@ -30,6 +30,12 @@
 #include <kpathsea/variable.h>
 #include <kpathsea/version.h>
 
+#ifdef WIN32
+#undef fputs
+#undef puts
+#define fputs win32_fputs
+#define puts  win32_puts
+#endif
 
 /* For variable and path expansion.  (-expand-var, -expand-path,
    -show-path) */
@@ -672,6 +678,10 @@ kpathsea_set_program_enabled (kpse, fmt, false, kpse_src_cmdline - 1)
 int
 main (int argc,  string *argv)
 {
+#ifdef WIN32
+  string *av, enc;
+  int ac;
+#endif
   unsigned unfound = 0;
   kpathsea kpse = kpathsea_new();
 
@@ -679,6 +689,22 @@ main (int argc,  string *argv)
   read_command_line (kpse, argc, argv);
 
   kpathsea_set_program_name (kpse, argv[0], progname);
+#ifdef WIN32
+  if(strstr(progname,"xetex") || strstr(progname,"xelatex")
+     || strstr(progname,"uptex") || strstr(progname,"uplatex")
+     || strstr(progname,"dvipdfm") || strstr(progname,"extractbb")
+     || strstr(progname,"xbb") || strstr(progname,"ebb")
+     || strstr(progname,"dvips"))
+  {
+    enc = kpathsea_var_value (kpse, "command_line_encoding");
+    if (get_command_line_args_utf8(enc, &ac, &av)) {
+      optind = 0;
+      read_command_line (kpse, ac, av);
+      argv = av;
+      argc = ac;
+    }
+  }
+#endif
   init_more (kpse);
 
 
diff --git a/Build/source/texk/kpathsea/line.c b/Build/source/texk/kpathsea/line.c
index c26303a3012..b221a4032dc 100644
--- a/Build/source/texk/kpathsea/line.c
+++ b/Build/source/texk/kpathsea/line.c
@@ -18,6 +18,12 @@
 
 #include <kpathsea/config.h>
 #include <kpathsea/line.h>
+#ifdef WIN32
+#undef getc
+#undef ungetc
+#define getc   win32_getc
+#define ungetc win32_ungetc
+#endif
 
 /* Allocate in increments of this size.  */
 #define BLOCK_SIZE 75
diff --git a/Build/source/texk/kpathsea/pathsearch.c b/Build/source/texk/kpathsea/pathsearch.c
index f5eb216b2a4..0f59ac5ed95 100644
--- a/Build/source/texk/kpathsea/pathsearch.c
+++ b/Build/source/texk/kpathsea/pathsearch.c
@@ -34,6 +34,13 @@
 #include <sys/stat.h>   /* for stat bits */
 #endif
 
+#ifdef WIN32
+#undef fputs
+#undef puts
+#define fputs win32_fputs
+#define puts  win32_puts
+#endif
+
 /* The very first search is for texmf.cnf, called when someone tries to
    initialize the TFM path or whatever.  init_path calls kpse_cnf_get
    which calls kpse_all_path_search to find all the texmf.cnf's.  We