COUCHDB-760 - allow utf-8 in attachment names (patch by davisp/benoitc)

git-svn-id: https://svn.apache.org/repos/asf/couchdb/branches/1.1.x@1100254 13f79535-47bb-0310-9956-ffa450edef68

2 files changed, 27 insertions, 42 deletions

diff --git a/share/www/script/test/attachment_names.js b/share/www/script/test/attachment_names.js
index 988dd2d2..777b5ece 100644
--- a/share/www/script/test/attachment_names.js
+++ b/share/www/script/test/attachment_names.js
@@ -16,6 +16,24 @@ couchTests.attachment_names = function(debug) {
   db.createDb();
   if (debug) debugger;
 
+  var goodDoc = {
+    _id: "good_doc",
+    _attachments: {
+      "Колян.txt": {
+       content_type:"text/plain",
+       data: "VGhpcyBpcyBhIGJhc2U2NCBlbmNvZGVkIHRleHQ="
+      }
+    }
+  };
+
+  var save_response = db.save(goodDoc);
+  T(save_response.ok);
+
+  var xhr = CouchDB.request("GET", "/test_suite_db/good_doc/Колян.txt");
+  T(xhr.responseText == "This is a base64 encoded text");
+  T(xhr.getResponseHeader("Content-Type") == "text/plain");
+  T(xhr.getResponseHeader("Etag") == '"' + save_response.rev + '"');
+
   var binAttDoc = {
     _id: "bin_doc",
     _attachments:{
@@ -27,28 +45,23 @@ couchTests.attachment_names = function(debug) {
   };
 
   // inline attachments
-  try {
-    db.save(binAttDoc);
-    TEquals(1, 2, "Attachment name with non UTF-8 encoding saved. Should never show!");
-  } catch (e) {
-    TEquals("bad_request", e.error, "attachment_name: inline attachments");
-    TEquals("Attachment name is not UTF-8 encoded", e.reason, "attachment_name: inline attachments");
-  }
+  resp = db.save(binAttDoc);
+  TEquals(true, resp.ok, "attachment_name: inline attachment");
 
 
   // standalone docs
   var bin_data = "JHAPDO*AU£PN ){(3u[d 93DQ9¡€])}    ææøo'∂ƒæ≤çæππ•¥∫¶®#†π¶®¥π€ª®˙π8np";
 
+
   var xhr = (CouchDB.request("PUT", "/test_suite_db/bin_doc3/attachment\x80txt", {
     headers:{"Content-Type":"text/plain;charset=utf-8"},
     body:bin_data
   }));
 
   var resp = JSON.parse(xhr.responseText);
-  TEquals(400, xhr.status, "attachment_name: standalone API");
-  TEquals("bad_request", resp.error, "attachment_name: standalone API");
-  TEquals("Attachment name is not UTF-8 encoded", resp.reason, "attachment_name: standalone API");
-
+  TEquals(201, xhr.status, "attachment_name: standalone API");
+  TEquals("Created",  xhr.statusText, "attachment_name: standalone API");
+  TEquals(true, resp.ok, "attachment_name: standalone API");
 
   // bulk docs
   var docs = { docs: [binAttDoc] };
@@ -57,10 +70,8 @@ couchTests.attachment_names = function(debug) {
     body: JSON.stringify(docs)
   });
 
-  var resp = JSON.parse(xhr.responseText);
-  TEquals(400, xhr.status, "attachment_name: bulk docs");
-  TEquals("bad_request", resp.error, "attachment_name: bulk docs");
-  TEquals("Attachment name is not UTF-8 encoded", resp.reason, "attachment_name: bulk docs");
+  TEquals(201, xhr.status, "attachment_name: bulk docs");
+  TEquals("Created", xhr.statusText, "attachment_name: bulk docs");
 
 
   // leading underscores
diff --git a/src/couchdb/couch_httpd_db.erl b/src/couchdb/couch_httpd_db.erl
index 0b8f891b..0dbebb6e 100644
--- a/src/couchdb/couch_httpd_db.erl
+++ b/src/couchdb/couch_httpd_db.erl
@@ -1276,34 +1276,8 @@ validate_attachment_name(Name) when is_list(Name) ->
 validate_attachment_name(<<"_",_/binary>>) ->
     throw({bad_request, <<"Attachment name can't start with '_'">>});
 validate_attachment_name(Name) ->
-    case is_valid_utf8(Name) of
+    case couch_util:validate_utf8(Name) of
         true -> Name;
         false -> throw({bad_request, <<"Attachment name is not UTF-8 encoded">>})
     end.
 
-%% borrowed from mochijson2:json_bin_is_safe()
-is_valid_utf8(<<>>) ->
-    true;
-is_valid_utf8(<<C, Rest/binary>>) ->
-    case C of
-        $\" ->
-            false;
-        $\\ ->
-            false;
-        $\b ->
-            false;
-        $\f ->
-            false;
-        $\n ->
-            false;
-        $\r ->
-            false;
-        $\t ->
-            false;
-        C when C >= 0, C < $\s; C >= 16#7f, C =< 16#10FFFF ->
-            false;
-        C when C < 16#7f ->
-            is_valid_utf8(Rest);
-        _ ->
-            false
-    end.