api: allow uncached content fetching.

- this is mostly to optimize for memory of indexer

rhodecode/api/views/repo_api.py

@@ -504,7 +504,8 @@ def get_repo_nodes(request, apiuser, repoid, revision, root_path,
 
 @jsonrpc_method()
 def get_repo_file(request, apiuser, repoid, commit_id, file_path,
-                  max_file_bytes=Optional(None), details=Optional('basic')):
+                  max_file_bytes=Optional(None), details=Optional('basic'),
+                  cache=Optional(True)):
     """
     Returns a single file from repository at given revision.
 
@@ -523,8 +524,10 @@ def get_repo_file(request, apiuser, repoid, commit_id, file_path,
         The valid options are ``minimal`` ``basic`` and ``full``.
     :type details: Optional(str)
     :param max_file_bytes: Only return file content under this file size bytes
-    :type details: Optional(int)
-
+    :type max_file_bytes: Optional(int)
+    :param cache: Use internal caches for fetching files. If disabled fetching
+        files is slower but more memory efficient
+    :type cache: Optional(bool)
     Example output:
 
     .. code-block:: bash
@@ -549,6 +552,7 @@ def get_repo_file(request, apiuser, repoid, commit_id, file_path,
         _perms = ('repository.admin', 'repository.write', 'repository.read',)
         validate_repo_permissions(apiuser, repoid, repo, _perms)
 
+    cache = Optional.extract(cache, binary=True)
     details = Optional.extract(details)
     _extended_types = ['minimal', 'minimal+search', 'basic', 'full']
     if details not in _extended_types:
@@ -574,7 +578,7 @@ def get_repo_file(request, apiuser, repoid, commit_id, file_path,
 
         node = ScmModel().get_node(
             repo, commit_id, file_path, extended_info=extended_info,
-            content=content, max_file_bytes=max_file_bytes)
+            content=content, max_file_bytes=max_file_bytes, cache=cache)
 
     except Exception:
         log.exception("Exception occurred while trying to get repo node")

rhodecode/lib/vcs/nodes.py

@@ -381,16 +381,25 @@ class FileNode(Node):
         Returns md5, binary flag of the file node, without any cache usage.
         """
 
-        if self.commit:
-            content = self.commit.get_file_content(self.path)
-        else:
-            content = self._content
+        content = self.content_uncached()
 
         is_binary = content and '\0' in content
         size = 0
         if content:
             size = len(content)
-        return is_binary, md5(content), size
+
+        return is_binary, md5(content), size, content
+
+    def content_uncached(self):
+        """
+        Returns lazily content of the FileNode. If possible, would try to
+        decode content from UTF-8.
+        """
+        if self.commit:
+            content = self.commit.get_file_content(self.path)
+        else:
+            content = self._content
+        return content
 
     @LazyProperty
     def content(self):

rhodecode/model/scm.py

@@ -583,7 +583,7 @@ class ScmModel(BaseModel):
         return _dirs, _files
 
     def get_node(self, repo_name, commit_id, file_path,
-                 extended_info=False, content=False, max_file_bytes=None):
+                 extended_info=False, content=False, max_file_bytes=None, cache=True):
         """
         retrieve single node from commit
         """
@@ -606,12 +606,21 @@ class ScmModel(BaseModel):
 
             if extended_info:
                 file_data.update({
-                    "md5": file_node.md5,
-                    "binary": file_node.is_binary,
-                    "size": file_node.size,
                     "extension": file_node.extension,
                     "mimetype": file_node.mimetype,
-                    "lines": file_node.lines()[0]
+                })
+
+                if cache:
+                    md5 = file_node.md5
+                    is_binary = file_node.is_binary
+                    size = file_node.size
+                else:
+                    is_binary, md5, size, _content = file_node.metadata_uncached()
+
+                file_data.update({
+                    "md5": md5,
+                    "binary": is_binary,
+                    "size": size,
                 })
 
             if content:
@@ -619,7 +628,13 @@ class ScmModel(BaseModel):
                                    and file_node.size > max_file_bytes)
                 full_content = None
                 if not file_node.is_binary and not over_size_limit:
-                    full_content = safe_str(file_node.content)
+                    if cache:
+                        full_content = safe_str(file_node.content)
+                    else:
+                        if _content is None:
+                            is_binary, md5, size, _content = \
+                                file_node.metadata_uncached()
+                        full_content = safe_str(_content)
 
                 file_data.update({
                     "content": full_content,
@@ -647,7 +662,7 @@ class ScmModel(BaseModel):
                 for f in files:
                     _content = None
                     _data = f_name = f.unicode_path
-                    is_binary, md5, size = f.metadata_uncached()
+                    is_binary, md5, size, _content = f.metadata_uncached()
                     _data = {
                         "name": h.escape(f_name),
                         "md5": md5,