kpathsea: More changes to reduce diffs with W32TeX

git-svn-id: svn://tug.org/texlive/trunk@23300 c570f23f-e606-0410-a88d-b1316a301751

6 files changed, 88 insertions, 81 deletions

diff --git a/Build/source/texk/kpathsea/ChangeLog b/Build/source/texk/kpathsea/ChangeLog
index bfe6def546b..fe041db5dd4 100644
--- a/Build/source/texk/kpathsea/ChangeLog
+++ b/Build/source/texk/kpathsea/ChangeLog
@@ -1,3 +1,16 @@
+2011-07-31  Peter Breitenlohner  <peb@mppmu.mpg.de>
+
+	More diffs between TeX Live and W32TeX.
+	* c-pathch.h: IS_DIR_SEP_CH macro to be used with `slashified'
+	filenames without need to test for 2-Byte Kanji codes.
+
+	* elt-dirs.c (kpathsea_normalize_path): Slashify, reformulate,
+	and test for 2-Byte Kanji (CP 932, SJIS) codes.
+	(kpathsea_element_dirs): First call kpathsea_normalize_path()
+	which might modify `elt' and then test for cached answer.
+
+	* progname.c: Make sure SELFAUTO* is not just / only for WIN32.
+
 2011-07-30  Peter Breitenlohner  <peb@mppmu.mpg.de>
 
 	* dir.c (kpathsea_dir_p, dir_p), lib.h: Unconstify string parms.
@@ -5,6 +18,10 @@
 	expand_elt): Unconstify string parms and local vars.
 	* progname.c (kpathsea_selfdir): Don't uselessly initialize ret.
 
+	* mingw32.c: #include <kpathsea/c-pathch.h>.
+	* mingw32.h [IS_DIR_SEP, IS_DEVICE_SEP, IS_UNC_NAME]: Do not
+	duplicate them here, use definitions from c-pathch.h.
+
 2011-07-29  Akira Kakuto  <kakuto@fuk.kindai.ac.jp>
 
 	* tilde.c [WIN32]: Use kpathsea_getpwnam().
diff --git a/Build/source/texk/kpathsea/c-pathch.h b/Build/source/texk/kpathsea/c-pathch.h
index 97970f40c67..cc8b936abf8 100644
--- a/Build/source/texk/kpathsea/c-pathch.h
+++ b/Build/source/texk/kpathsea/c-pathch.h
@@ -24,46 +24,45 @@
 
 /* What separates filename components?  */
 #ifndef DIR_SEP
-#ifdef VMS
-#define DIR_SEP ':'
-#define DIR_SEP_STRING ":"
-#else
-#ifdef DOSISH
+# if defined(VMS)
+#  define DIR_SEP ':'
+#  define DIR_SEP_STRING ":"
+# elif defined(DOSISH) /* not VMS */
 /* Either \'s or 's work.  Wayne Sullivan's web2pc prefers /, so we'll
    go with that.  */
-#define DIR_SEP '/'
-#define DIR_SEP_STRING "/"
-#define IS_DEVICE_SEP(ch) ((ch) == ':')
-#define NAME_BEGINS_WITH_DEVICE(name) (*(name) && IS_DEVICE_SEP((name)[1]))
+#  define DIR_SEP '/'
+#  define DIR_SEP_STRING "/"
+#  define IS_DEVICE_SEP(ch) ((ch) == ':')
+#  define NAME_BEGINS_WITH_DEVICE(name) (*(name) && IS_DEVICE_SEP((name)[1]))
 /* On DOS, it's good to allow both \ and / between directories.  */
-#define IS_DIR_SEP(ch) ((ch) == '/' || (ch) == '\\')
-/* On win32, UNC names are authorized */
-#ifdef WIN32
-#define IS_UNC_NAME(name) (strlen(name)>=3 && IS_DIR_SEP(*name)  \
-                            && IS_DIR_SEP(*(name+1)) && isalnum(*(name+2)))
-#endif
-#else
-#ifdef AMIGA
-#define DIR_SEP '/'
-#define DIR_SEP_STRING "/"
-#define IS_DIR_SEP(ch) ((ch) == '/' || (ch) == ':')
-#define IS_DEVICE_SEP(ch) ((ch) == ':')
-#else
-#ifdef VMCMS
-#define DIR_SEP ' '
-#define DIR_SEP_STRING " "
-#else
-#define DIR_SEP '/'
-#define DIR_SEP_STRING "/"
-#endif /* not VM/CMS */
-#endif /* not AMIGA */
-#endif /* not DOSISH */
-#endif /* not VMS */
+#  define IS_DIR_SEP(ch) ((ch) == '/' || (ch) == '\\')
+#  ifdef WIN32
+/*  On win32, UNC names are authorized */
+#   define IS_UNC_NAME(name) (strlen(name)>=3 && IS_DIR_SEP(*name)  \
+                               && IS_DIR_SEP(*(name+1)) && isalnum(*(name+2)))
+/*  Used after converting '\\' into '/' */
+#   define IS_DIR_SEP_CH(ch) ((ch) == '/')
+#  endif
+# elif defined(AMIGA) /* not DOSISH */
+#  define DIR_SEP '/'
+#  define DIR_SEP_STRING "/"
+#  define IS_DIR_SEP(ch) ((ch) == '/' || (ch) == ':')
+#  define IS_DEVICE_SEP(ch) ((ch) == ':')
+# elif defined(VMCMS) /* not AMIGA */
+#  define DIR_SEP ' '
+#  define DIR_SEP_STRING " "
+# else /* not VMCMS */
+#  define DIR_SEP '/'
+#  define DIR_SEP_STRING "/"
+# endif /* not VMCMS */
 #endif /* not DIR_SEP */
 
 #ifndef IS_DIR_SEP
 #define IS_DIR_SEP(ch) ((ch) == DIR_SEP)
 #endif
+#ifndef IS_DIR_SEP_CH
+#define IS_DIR_SEP_CH(ch) IS_DIR_SEP(ch)
+#endif
 #ifndef IS_DEVICE_SEP /* No `devices' on, e.g., Unix.  */
 #define IS_DEVICE_SEP(ch) 0
 #endif
diff --git a/Build/source/texk/kpathsea/elt-dirs.c b/Build/source/texk/kpathsea/elt-dirs.c
index cb658207d03..2fe3139cf90 100644
--- a/Build/source/texk/kpathsea/elt-dirs.c
+++ b/Build/source/texk/kpathsea/elt-dirs.c
@@ -332,6 +332,10 @@ expand_elt (kpathsea kpse, str_llist_type * str_list_ptr, string elt,
   checked_dir_list_add (kpse, str_list_ptr, elt);
 }
 
+/*  On win32 we slashify ELT, i.e., change '\\' to '/', and then can use
+   IS_DIR_SEP_CH instead of IS_DIR_SEP and need not test for the presence
+   of 2-Byte Kanji (CP 932, SJIS) codes.  */
+
 /* The first bits of a path element can be problematic because they
    look like a request to expand a whole disk, rather than a subtree.
    - It can contain a drive specification.
@@ -346,8 +350,8 @@ expand_elt (kpathsea kpse, str_llist_type * str_list_ptr, string elt,
      handle well filenames like c://dir/foo. So canonicalize the names.
      The resulting name will always be shorter than the one passed, so no
      problem.
-   - If possible, we merely skip multiple leading slashes to prevent
-     expanding from the root of a UNIX filesystem tree.  */
+   - Remove multiple leading slashes to prevent expanding from the root
+     of a UNIX filesystem tree.  */
 
 unsigned
 kpathsea_normalize_path (kpathsea kpse, string elt)
@@ -355,33 +359,38 @@ kpathsea_normalize_path (kpathsea kpse, string elt)
   unsigned ret;
   unsigned i;
 
+#if defined(WIN32)
+  for (i = 0; elt[i]; i++) {
+    if (elt[i] == '\\')
+      elt[i] = '/';
+    else if (IS_KANJI(elt + i))
+      i++;
+  }
+#endif
+
   if (NAME_BEGINS_WITH_DEVICE(elt)) {
     if (*elt >= 'A' && *elt <= 'Z')
       *elt += 'a' - 'A';
-    for (i = 2; IS_DIR_SEP(elt[i]); ++i)
-      ;
-    if (i > 3)
-      memmove (elt+3, elt+i, strlen(elt+i) + 1);
     ret = 2;
 
   } else if (IS_UNC_NAME(elt)) {
-    for (ret = 2; elt[ret] && !IS_DIR_SEP(elt[ret]); ++ret)
-      ;
-    for (i = ret; elt[i] && IS_DIR_SEP(elt[i]); ++i)
+    for (ret = 2; elt[ret] && !IS_DIR_SEP_CH(elt[ret]); ret++)
       ;
-    if (i > ret+1)
-      memmove (elt+ret+1, elt+i, strlen(elt+i) + 1);
 
-  } else {
-    for (ret = 0; IS_DIR_SEP(elt[ret]); ++ret)
-      ;
-  }
+  } else
+    ret = 0;
 
+  for (i = ret; IS_DIR_SEP_CH(elt[i]); ++i)
+    ;
+  if (i > ret + 1) {
 #ifdef KPSE_DEBUG
-  if (KPATHSEA_DEBUG_P (KPSE_DEBUG_STAT) && ret != 1)
+  if (KPATHSEA_DEBUG_P (KPSE_DEBUG_STAT))
     DEBUGF2 ("kpse_normalize_path (%s) => %u\n", elt, ret);
 #endif /* KPSE_DEBUG */
 
+    memmove (elt + ret + 1, elt + i, strlen (elt + i) + 1);
+  }
+
   return ret;
 }
 
@@ -399,11 +408,15 @@ str_llist_type *
 kpathsea_element_dirs (kpathsea kpse, string elt)
 {
   str_llist_type *ret;
+  unsigned i;
 
   /* If given nothing, return nothing.  */
   if (!elt || !*elt)
     return NULL;
 
+  /* Normalize ELT before looking for a cached value.  */
+  i = kpathsea_normalize_path (kpse, elt);
+
   /* If we've already cached the answer for ELT, return it.  */
   ret = cached (kpse, elt);
   if (ret)
@@ -414,7 +427,7 @@ kpathsea_element_dirs (kpathsea kpse, string elt)
   *ret = NULL;
 
   /* We handle the hard case in a subroutine.  */
-  expand_elt (kpse, ret, elt, kpathsea_normalize_path (kpse, elt));
+  expand_elt (kpse, ret, elt, i);
 
   /* Remember the directory list we just found, in case future calls are
      made with the same ELT.  */
diff --git a/Build/source/texk/kpathsea/mingw32.c b/Build/source/texk/kpathsea/mingw32.c
index 913df2deba5..9da8d774074 100644
--- a/Build/source/texk/kpathsea/mingw32.c
+++ b/Build/source/texk/kpathsea/mingw32.c
@@ -26,6 +26,7 @@
 #ifdef __MINGW32__
 
 #include <kpathsea/config.h>
+#include <kpathsea/c-pathch.h>
 #include <kpathsea/c-proto.h>
 #include <kpathsea/mingw32.h>
 #include <kpathsea/lib.h>
diff --git a/Build/source/texk/kpathsea/mingw32.h b/Build/source/texk/kpathsea/mingw32.h
index 7bc15cc9c56..c32312c2fce 100644
--- a/Build/source/texk/kpathsea/mingw32.h
+++ b/Build/source/texk/kpathsea/mingw32.h
@@ -26,22 +26,6 @@
 #define MAX_PIPES 128
 #endif
 
-/* On DOS, it's good to allow both \ and / between directories.  */
-#ifndef IS_DIR_SEP
-#define IS_DIR_SEP(ch) ((ch) == '/' || (ch) == '\\')
-#endif
-#ifndef IS_DEVICE_SEP
-#define IS_DEVICE_SEP(ch) ((ch) == ':')
-#endif
-#ifndef NAME_BEGINS_WITH_DEVICE
-#define NAME_BEGINS_WITH_DEVICE(name) (*(name) && IS_DEVICE_SEP((name)[1]))
-#endif
-/* On win32, UNC names are authorized */
-#ifndef IS_UNC_NAME
-#define IS_UNC_NAME(name) (strlen(name)>=3 && IS_DIR_SEP(*name)  \
-                            && IS_DIR_SEP(*(name+1)) && isalnum(*(name+2)))
-#endif
-
 #ifdef MAKE_KPSE_DLL /* libkpathsea internal only */
 
 extern void init_user_info (void);
diff --git a/Build/source/texk/kpathsea/progname.c b/Build/source/texk/kpathsea/progname.c
index a4b1272fe34..eccf31028a4 100644
--- a/Build/source/texk/kpathsea/progname.c
+++ b/Build/source/texk/kpathsea/progname.c
@@ -473,21 +473,15 @@ mk_suffixlist (kpathsea kpse)
 }
 #endif /* WIN32 || __CYGWIN__ */
 
-/* Append a dot if SELFAUTO{LOC,DIR,PARENT} is just `/'.  Otherwise,
-   e.g., $SELFAUTODIR/texmf would search the entire filesystem.  */
+/* On win32 SELFAUTO{LOC,DIR,PARENT} must not be just `/', otherwise,
+   e.g., $SELFAUTODIR/texmf/tex would be mistaken as UNC name.  */
 static inline string
 fix_selfdir (string dir)
 {
 #if defined(WIN32)
-  if (dir[strlen (dir) - 1] == '/') {
-#else
-  if (IS_DIR_SEP (dir[strlen (dir) - 1])) {
+  if (IS_DIR_SEP (*dir) && dir[1] == 0)
+    *dir = 0;
 #endif
-    string ret = concat (dir, ".");
-    free (dir);
-    return ret;
-  }
-
   return dir;
 }
 
@@ -673,12 +667,11 @@ kpathsea_set_program_name (kpathsea kpse,  const_string argv0,
 #endif
   /* SELFAUTODIR is actually the parent of the invocation directory,
      and SELFAUTOPARENT the grandparent.  This is how teTeX did it.  */
-  sdir = fix_selfdir (sdir);
-  kpathsea_xputenv (kpse, "SELFAUTOLOC", sdir);
-  sdir_parent = fix_selfdir (xdirname (sdir));
-  kpathsea_xputenv (kpse, "SELFAUTODIR", sdir_parent);
-  sdir_grandparent = fix_selfdir (xdirname (sdir_parent));
-  kpathsea_xputenv (kpse, "SELFAUTOPARENT", sdir_grandparent);
+  sdir_parent = xdirname (sdir);
+  sdir_grandparent = xdirname (sdir_parent);
+  kpathsea_xputenv (kpse, "SELFAUTOLOC", fix_selfdir (sdir));
+  kpathsea_xputenv (kpse, "SELFAUTODIR", fix_selfdir (sdir_parent));
+  kpathsea_xputenv (kpse, "SELFAUTOPARENT", fix_selfdir (sdir_grandparent));
 
 #if defined(WIN32) || defined(__CYGWIN__)
   mk_suffixlist(kpse);