unify platform encoding.c, support all

Closes GH #1120. Get support for other platform encodings for free.
parrot · Nov 5, 2014 · a31fbc5 · a31fbc5
1 parent 3a6375c
commit a31fbc5
Show file tree

Hide file tree

Showing 10 changed files with 33 additions and 495 deletions.
diff --git a/ChangeLog b/ChangeLog
@@ -3,6 +3,8 @@
         + Add imcc -d2 flag for MKCONST tracing.
         + Fix darwin --m=32 Parrot_sysmem_amount() #1113
         + Honor rlimit settings on all non-windows platforms. #935
+        + Unify code for platform encodings, now supports all. #1120
+          e.g. unicode filenames or UTF-8 term output on cygwin, solaris, dragonfly
     - Build
         + Set -mabi=64 for gcc --m=64 on mips, -m64 on powerpc #1110
         + Add --{en,dis}able-static #1104

diff --git a/MANIFEST b/MANIFEST
@@ -1325,10 +1325,8 @@ src/platform/ansi/dl.c                                      []
 src/platform/ansi/exec.c                                    []
 src/platform/ansi/time.c                                    []
 src/platform/cygwin/math.c                                  []
-src/platform/darwin/encoding.c                              []
 src/platform/darwin/hires_timer.c                           []
 src/platform/darwin/sysmem.c                                []
-src/platform/freebsd/encoding.c                             []
 src/platform/generic/cpu_type.c                             []
 src/platform/generic/dl.c                                   []
 src/platform/generic/encoding.c                             []
@@ -1348,10 +1346,7 @@ src/platform/generic/sysmem.c                               []
 src/platform/generic/time.c                                 []
 src/platform/generic/uid.c                                  []
 src/platform/ia64/asm.s                                     []
-src/platform/linux/encoding.c                               []
-src/platform/netbsd/encoding.c                              []
 src/platform/netbsd/misc.c                                  []
-src/platform/openbsd/encoding.c                             []
 src/platform/openbsd/math.c                                 []
 src/platform/solaris/math.c                                 []
 src/platform/solaris/time.c                                 []

diff --git a/config/gen/makefiles/root.in b/config/gen/makefiles/root.in
@@ -1840,14 +1840,10 @@ src/platform/ansi/time$(O) : src/platform/ansi/time.c $(PARROT_H_HEADERS)
 
 src/platform/cygwin/math$(O) : src/platform/cygwin/math.c $(PARROT_H_HEADERS)
 
-src/platform/darwin/encoding$(O) : src/platform/darwin/encoding.c $(PARROT_H_HEADERS)
-
 src/platform/darwin/hires_timer$(O) : src/platform/darwin/hires_timer.c $(PARROT_H_HEADERS)
 
 src/platform/darwin/sysmem$(O) : src/platform/darwin/sysmem.c $(PARROT_H_HEADERS)
 
-src/platform/freebsd/encoding$(O) : src/platform/freebsd/encoding.c $(PARROT_H_HEADERS)
-
 src/platform/generic/cpu_type$(O) : src/platform/generic/cpu_type.c $(PARROT_H_HEADERS)
 
 src/platform/generic/dl$(O) : src/platform/generic/dl.c $(PARROT_H_HEADERS)
@@ -1888,14 +1884,8 @@ src/platform/generic/entropy$(O) : src/platform/generic/entropy.c $(PARROT_H_HEA
 
 src/platform/win32/entropy$(O) : src/platform/win32/entropy.c $(PARROT_H_HEADERS)
 
-src/platform/linux/encoding$(O) : src/platform/linux/encoding.c $(PARROT_H_HEADERS)
-
-src/platform/netbsd/encoding$(O) : src/platform/netbsd/encoding.c $(PARROT_H_HEADERS)
-
 src/platform/netbsd/misc$(O) : src/platform/netbsd/misc.c $(PARROT_H_HEADERS)
 
-src/platform/openbsd/encoding$(O) : src/platform/openbsd/encoding.c $(PARROT_H_HEADERS)
-
 src/platform/openbsd/math$(O) : src/platform/openbsd/math.c $(PARROT_H_HEADERS)
 
 src/platform/solaris/math$(O) : src/platform/solaris/math.c $(PARROT_H_HEADERS)

diff --git a/src/platform/darwin/encoding.c b/src/platform/darwin/encoding.c
diff --git a/src/platform/freebsd/encoding.c b/src/platform/freebsd/encoding.c
diff --git a/src/platform/generic/encoding.c b/src/platform/generic/encoding.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2012, Parrot Foundation.
+ * Copyright (C) 2012-2014, Parrot Foundation.
  */
 
 /*
@@ -22,6 +22,11 @@ Platform C string encoding management
 
 #include "parrot/parrot.h"
 
+#if defined(PARROT_HAS_HEADER_LOCALE) && defined(PARROT_HAS_HEADER_LANGINFO)
+#  include <locale.h>
+#  include <langinfo.h>
+#endif
+
 /* HEADERIZER HFILE: none */
 
 /*
@@ -38,7 +43,32 @@ the OS, to an appropriate value.
 void
 Parrot_init_platform_encoding(SHIM_INTERP)
 {
+#if defined(PARROT_HAS_HEADER_LOCALE) && defined(PARROT_HAS_HEADER_LANGINFO)
+    const char *orig  = setlocale(LC_CTYPE, NULL); /* store original locale */
+    setlocale(LC_CTYPE, "");                       /* set locale to environment specification */
+    {
+        const char *codeset = nl_langinfo(CODESET);
+        if (STREQ(codeset, "UTF-8")
+         || STREQ(codeset, "utf-8"))          /* darwin 10.4 ppc */
+            Parrot_platform_encoding_ptr = Parrot_utf8_encoding_ptr;
+        else if (STREQ(codeset, "ISO-8859-1") /* linux */
+              || STREQ(codeset, "ISO8859-1")) /* bsd */
+            Parrot_platform_encoding_ptr = Parrot_latin1_encoding_ptr;
+        else if (STREQ(codeset, "ANSI_X3.4-1968") /* linux */
+              || STREQ(codeset, "US-ASCII")   /* freebsd */
+              || STREQ(codeset, "646"))       /* bsd */
+            Parrot_platform_encoding_ptr = Parrot_ascii_encoding_ptr;
+        else {
+            /* Can't use Parrot_warn here, the interpreter is not ready */
+            fprintf(stderr,
+                "Unknown codeset `%s', defaulting to ASCII\n", codeset);
+            Parrot_platform_encoding_ptr = Parrot_ascii_encoding_ptr;
+        }
+    }
+    setlocale(LC_CTYPE, orig); /* restore original locale */
+#else
     Parrot_platform_encoding_ptr = Parrot_ascii_encoding_ptr;
+#endif
 }
 
 

diff --git a/src/platform/linux/encoding.c b/src/platform/linux/encoding.c