Adding io::CleanPath in path.h to clean up paths. Also made TranslateName call CleanPath so that all paths are clean before doing file system ops.

Change: 133140241

1 files changed, 81 insertions, 0 deletions

diff --git a/tensorflow/core/lib/io/path.cc b/tensorflow/core/lib/io/path.cc
index 107bd5e56a..c59d6d1893 100644
--- a/tensorflow/core/lib/io/path.cc
+++ b/tensorflow/core/lib/io/path.cc
@@ -101,5 +101,86 @@ StringPiece Extension(StringPiece path) {
   return internal::SplitBasename(path).second;
 }
 
+string CleanPath(StringPiece unclean_path) {
+  string path = unclean_path.ToString();
+  const char *src = path.c_str();
+  string::iterator dst = path.begin();
+
+  // Check for absolute path and determine initial backtrack limit.
+  const bool is_absolute_path = *src == '/';
+  if (is_absolute_path) {
+    *dst++ = *src++;
+    while (*src == '/') ++src;
+  }
+  string::const_iterator backtrack_limit = dst;
+
+  // Process all parts
+  while (*src) {
+    bool parsed = false;
+
+    if (src[0] == '.') {
+      //  1dot ".<whateverisnext>", check for END or SEP.
+      if (src[1] == '/' || !src[1]) {
+        if (*++src) {
+          ++src;
+        }
+        parsed = true;
+      } else if (src[1] == '.' && (src[2] == '/' || !src[2])) {
+        // 2dot END or SEP (".." | "../<whateverisnext>").
+        src += 2;
+        if (dst != backtrack_limit) {
+          // We can backtrack the previous part
+          for (--dst; dst != backtrack_limit && dst[-1] != '/'; --dst) {
+            // Empty.
+          }
+        } else if (!is_absolute_path) {
+          // Failed to backtrack and we can't skip it either. Rewind and copy.
+          src -= 2;
+          *dst++ = *src++;
+          *dst++ = *src++;
+          if (*src) {
+            *dst++ = *src;
+          }
+          // We can never backtrack over a copied "../" part so set new limit.
+          backtrack_limit = dst;
+        }
+        if (*src) {
+          ++src;
+        }
+        parsed = true;
+      }
+    }
+
+    // If not parsed, copy entire part until the next SEP or EOS.
+    if (!parsed) {
+      while (*src && *src != '/') {
+        *dst++ = *src++;
+      }
+      if (*src) {
+        *dst++ = *src++;
+      }
+    }
+
+    // Skip consecutive SEP occurrences
+    while (*src == '/') {
+      ++src;
+    }
+  }
+
+  // Calculate and check the length of the cleaned path.
+  int path_length = dst - path.begin();
+  if (path_length != 0) {
+    // Remove trailing '/' except if it is root path ("/" ==> path_length := 1)
+    if (path_length > 1 && path[path_length - 1] == '/') {
+      --path_length;
+    }
+    path.resize(path_length);
+  } else {
+    // The cleaned path is empty; assign "." as per the spec.
+    path.assign(1, '.');
+  }
+  return path;
+}
+
 }  // namespace io
 }  // namespace tensorflow