scm: change ._get_content() to .raw_bytes attribute to file nodes to

allow getting raw bytes instead of text content of a file and added
.md5 and .is_binary attributes to file nodes

rhodecode/lib/vcs/backends/base.py

@@ -868,7 +868,7 @@ class BaseCommit(object):
             for f in files:
                 f_path = os.path.join(prefix, f.path)
                 file_info.append(
-                    (f_path, f.mode, f.is_link(), f._get_content()))
+                    (f_path, f.mode, f.is_link(), f.raw_bytes))
 
         if write_metadata:
             metadata = [

rhodecode/lib/vcs/nodes.py

@@ -28,6 +28,7 @@ import stat
 from zope.cachedescriptors.property import Lazy as LazyProperty
 
 from rhodecode.lib.utils import safe_unicode, safe_str
+from rhodecode.lib.utils2 import md5
 from rhodecode.lib.vcs import path as vcspath
 from rhodecode.lib.vcs.backends.base import EmptyCommit, FILEMODE_DEFAULT
 from rhodecode.lib.vcs.conf.mtypes import get_mimetypes_db
@@ -318,22 +319,35 @@ class FileNode(Node):
             mode = self._mode
         return mode
 
-    def _get_content(self):
+    @LazyProperty
+    def raw_bytes(self):
+        """
+        Returns lazily the raw bytes of the FileNode.
+        """
         if self.commit:
-            content = self.commit.get_file_content(self.path)
+            if self._content is None:
+                self._content = self.commit.get_file_content(self.path)
+            content = self._content
         else:
             content = self._content
         return content
 
-    @property
+    @LazyProperty
+    def md5(self):
+        """
+        Returns md5 of the file node.
+        """
+        return md5(self.raw_bytes)
+
+    @LazyProperty
     def content(self):
         """
         Returns lazily content of the FileNode. If possible, would try to
         decode content from UTF-8.
         """
-        content = self._get_content()
+        content = self.raw_bytes
 
-        if bool(content and '\0' in content):
+        if self.is_binary:
             return content
         return safe_unicode(content)
 
@@ -467,12 +481,12 @@ class FileNode(Node):
         else:
             return NodeState.NOT_CHANGED
 
-    @property
+    @LazyProperty
     def is_binary(self):
         """
         Returns True if file has binary content.
         """
-        _bin = '\0' in self._get_content()
+        _bin = self.raw_bytes and '\0' in self.raw_bytes
         return _bin
 
     @LazyProperty
@@ -502,7 +516,7 @@ class FileNode(Node):
         all_lines, empty_lines = 0, 0
 
         if not self.is_binary:
-            content = self._get_content()
+            content = self.content
             if count_empty:
                 all_lines = 0
                 empty_lines = 0
@@ -717,22 +731,10 @@ class LargeFileNode(FileNode):
         we override check since the LargeFileNode path is system absolute
         """
 
-    def _get_content(self):
+    def raw_bytes(self):
         if self.commit:
             with open(self.path, 'rb') as f:
                 content = f.read()
         else:
             content = self._content
-        return content
-
-    @property
-    def content(self):
-        """
-        Returns lazily content of the `FileNode`. If possible, would try to
-        decode content from UTF-8.
-        """
-        content = self._get_content()
-
-        if bool(content and '\0' in content):
-            return content
-        return safe_unicode(content)
+        return content

rhodecode/model/scm.py

@@ -486,7 +486,7 @@ class ScmModel(BaseModel):
         :param repo_name: name of repository
         :param commit_id: commit id for which to list nodes
         :param root_path: root path to list
-        :param flat: return as a list, if False returns a dict with decription
+        :param flat: return as a list, if False returns a dict with description
 
         """
         _files = list()
@@ -499,31 +499,29 @@ class ScmModel(BaseModel):
                 for f in files:
                     _content = None
                     _data = f.unicode_path
+
                     if not flat:
                         _data = {
                             "name": f.unicode_path,
                             "type": "file",
                             }
                         if extended_info:
-                            _content = safe_str(f.content)
                             _data.update({
-                                "md5": md5(_content),
+                                "md5": f.md5,
                                 "binary": f.is_binary,
                                 "size": f.size,
                                 "extension": f.extension,
-
                                 "mimetype": f.mimetype,
                                 "lines": f.lines()[0]
                             })
+
                         if content:
                             full_content = None
                             if not f.is_binary:
-                                # in case we loaded the _content already
-                                # re-use it, or load from f[ile]
-                                full_content = _content or safe_str(f.content)
+                                full_content = safe_str(f.content)
 
                             _data.update({
-                                "content": full_content
+                                "content": full_content,
                             })
                     _files.append(_data)
                 for d in dirs:
@@ -1098,4 +1096,4 @@ def _check_rhodecode_hook(hook_path):
 def _read_hook(hook_path):
     with open(hook_path, 'rb') as f:
         content = f.read()
-    return content
+    return content