updates lurching towards functionality

[feisty_meow.git] / nucleus / library / filesystem / filename.cpp

diff --git a/nucleus/library/filesystem/filename.cpp b/nucleus/library/filesystem/filename.cpp
index 4ea4269aabaff86631452a3212a8e788f1b410cc..d12d9e7e58825b735a2f93911591abc842ed5850 100644 (file)
--- a/nucleus/library/filesystem/filename.cpp
+++ b/nucleus/library/filesystem/filename.cpp
@@ -19,15 +19,25 @@
 
 #include <basis/byte_array.h>
 #include <basis/functions.h>
+#include <configuration/application_configuration.h>
+/*
+ hmmm: note that we are relying on forward declared code here.
+ the canonical ordering for feisty's nucleus has the filesystem code come before
+ the configuration code, because the configuratin library uses filesystem features.
+ not sure i want to resolve this bizarritude at this time, but it points to the
+ need for a virtual interface at lower level than either filesystem or configuration
+ libraries, so we can emplace the need for the virtual unix root as a low-level
+ dependency to be implemented later.
+ */
+
 #include <textual/parser_bits.h>
 
 #include <stdio.h>
 #include <sys/stat.h>
 #include <sys/types.h>
-#ifdef __UNIX__
+#if defined(__UNIX__) || defined(__GNU_WINDOWS__)
   #include <unistd.h>
-#endif
-#ifdef __WIN32__
+#else
   #include <io.h>
 #endif
 
@@ -43,13 +53,13 @@ class status_info : public stat
 
 namespace filesystem {
 
-#if defined(__WIN32__) || defined(__VMS__)
-  const char DEFAULT_SEPARATOR = '\\';
-#elif defined(__UNIX__)
+//#if defined(__WIN32__) || defined(__VMS__)
+//  const char DEFAULT_SEPARATOR = '\\';
+//#elif defined(__UNIX__)
   const char DEFAULT_SEPARATOR = '/';
-#else
-  #error "We have no idea what the default path separator is."
-#endif
+//#else
+//  #error "We have no idea what the default path separator is."
+//#endif
 
 const char *NO_PARENT_DEFAULT = ".";
   // used when no directory name can be popped off.
@@ -191,14 +201,24 @@ void filename::canonicalize()
   // on windows, we want to translate away from any cygwin or msys format into a more palatable
   // version that the rest of windows understands.
   // first, cygwin...
-  const astring CYGDRIVE_PATH = astring(astring(DEFAULT_SEPARATOR, 1) + "cygdrive"
-      + astring(DEFAULT_SEPARATOR, 1));
+//hmmm: make these into statics!
+  const astring CYGDRIVE_SENTINEL = "cygdrive";
+  const astring CYGDRIVE_PATH = astring(astring(DEFAULT_SEPARATOR, 1)
+      + CYGDRIVE_SENTINEL + astring(DEFAULT_SEPARATOR, 1));
+
   // must be at least as long as the string we're looking for, plus a drive letter afterwards.
-  if ( (length() > CYGDRIVE_PATH.length() + 1) && begins(CYGDRIVE_PATH) ) {
+  if ( (length() >= CYGDRIVE_PATH.length() + 1)
+      && separator(get(0))
+      && separator(get(CYGDRIVE_PATH.length() - 1))
+      && compare(CYGDRIVE_SENTINEL, 1, 
+          0, CYGDRIVE_SENTINEL.length(), true) ) {
     zap(0, CYGDRIVE_PATH.length() - 1);  // whack the cygdrive portion plus two slashes.
     insert(1, ":");  // add a colon after the imputed drive letter.
-LOG(astring("turned cygdrive string into: ") + *this);
+//LOG(astring("turned cygdrive path string into: ") + *this);
+  } else {
+//LOG(astring("path didn't match so left as: ") + *this);
   }
+
   // now we convert msys...
   if ( (length() >= 2) && (get(0) == DEFAULT_SEPARATOR)
         && textual::parser_bits::is_alpha(get(1)) ) {
@@ -216,9 +236,43 @@ LOG(astring("turned cygdrive string into: ") + *this);
       // then only on the near defunct windows platform.
       zap(0, 0);  // take off initial slash.
       insert(1, ":");  // add the obligatory colon.
-LOG(astring("turned msys string into: ") + *this);
+//LOG(astring("turned msys string into: ") + *this);
     }
   } 
+
+  // if we still have a unix style path here on windows, then there will be
+  // trouble when we pass that to the OS.  we are not using any cygwin or
+  // other virtualization libraries directly, so we can't rely on those to
+  // fix the path.  but if we built under something like cygwin, we should
+  // have stored the real dos-style location of the virtual unix root.  we
+  // will use that to replace the root '/' and this should fix most of that
+  // style of path.
+  bool inject_root = false;  // assume we don't need to do anything.
+
+LOG(astring("before root injection: ") + raw());
+
+  // condition here just checks if the path is only the root.
+  if ( (length() == 1) && separator(get(0)) ) { inject_root = true; }
+
+if (inject_root) LOG("decided to inject root since path is '/'.");
+
+  // condition is looking for first character being a slash, and second char as alphanumeric or dash or underscore.
+  // we will currently fail detecting freaky paths that don't start off with alphanumeric or one of that small set of special chars.
+  if ( (length() >= 2)
+      && separator(get(0)) 
+      && ( textual::parser_bits::is_alphanumeric(get(1)) || ('-' == get(1)) || ('_' == get(1)) ) ) { 
+    inject_root = true;
+if (inject_root) LOG(astring("decided to inject root since path is compatible: ") + *this);
+  }
+
+LOG(astring("after second phase root injection: ") + *this);
+
+  if (inject_root) {
+    // inject the actual path to the unix root in front, if we know it.
+    // if we don't know it, then a default path that's unlikely to work is idiotically plugged in.
+    insert(0, configuration::application_configuration::get_virtual_unix_root());
+LOG(astring("turned cygdrive path string into: ") + *this);
+  }
 #endif
 
   // we don't crop the last separator if the name's too small.  for msdos
@@ -383,11 +437,10 @@ bool filename::exists() const
 {
   if (is_directory())
     return true;
+  // if the file name is empty, that cannot exist.
   if (!length())
     return false;
   return is_readable();
-///  byte_filer opened(observe(), "rb");
-///  return opened.good();
 }
 
 bool filename::legal_character(char to_check)