Fix for self-colliding ids on $import (#102)

* Fix for self-colliding ids on $import

Author: tetron

MANIFEST.in

@@ -4,4 +4,5 @@ include schema_salad/tests/test_schema/*.md
 include schema_salad/tests/test_schema/*.yml
 include schema_salad/tests/test_schema/*.cwl
 include schema_salad/metaschema/*
+global-exclude *~
 global-exclude *.pyc

schema_salad/ref_resolver.py

@@ -233,7 +233,7 @@ def __init__(self,
         self.url_fields = set()         # type: Set[unicode]
         self.scoped_ref_fields = {}     # type: Dict[unicode, int]
         self.vocab_fields = set()       # type: Set[unicode]
-        self.identifiers = set()        # type: Set[unicode]
+        self.identifiers = []           # type: List[unicode]
         self.identity_links = set()     # type: Set[unicode]
         self.standalone = None          # type: Optional[Set[unicode]]
         self.nolinkcheck = set()        # type: Set[unicode]
@@ -345,7 +345,7 @@ def add_context(self, newcontext, baseuri=""):
         self.url_fields = set(("$schemas",))
         self.scoped_ref_fields = {}
         self.vocab_fields = set()
-        self.identifiers = set()
+        self.identifiers = []
         self.identity_links = set()
         self.standalone = set()
         self.nolinkcheck = set()
@@ -361,7 +361,7 @@ def add_context(self, newcontext, baseuri=""):
 
         for key, value in self.ctx.items():
             if value == u"@id":
-                self.identifiers.add(key)
+                self.identifiers.append(key)
                 self.identity_links.add(key)
             elif isinstance(value, dict) and value.get(u"@type") == u"@id":
                 self.url_fields.add(key)
@@ -393,6 +393,8 @@ def add_context(self, newcontext, baseuri=""):
         for k, v in self.vocab.items():
             self.rvocab[self.expand_url(v, u"", scoped_id=False)] = k
 
+        self.identifiers.sort()
+
         _logger.debug("identifiers is %s", self.identifiers)
         _logger.debug("identity_links is %s", self.identity_links)
         _logger.debug("url_fields is %s", self.url_fields)
@@ -949,6 +951,7 @@ def validate_links(self, document, base_url, all_doc_ids):
                                 "%s object %s `%s` previously defined" % (all_doc_ids[document[identifier]], identifier, relname(document[identifier]), ))
                         else:
                             all_doc_ids[document[identifier]] = sl.makeLead()
+                            break
             except validate.ValidationException as v:
                 errors.append(sl.makeError(unicode(v)))
             if hasattr(document, "iteritems"):

schema_salad/validate.py

@@ -21,7 +21,7 @@ class ClassValidationException(ValidationException):
 
 def validate(expected_schema,           # type: Schema
              datum,                     # type: Any
-             identifiers=set(),         # type: Set[unicode]
+             identifiers=[],            # type: List[unicode]
              strict=False,              # type: bool
              foreign_properties=set()   # type: Set[unicode]
              ):
@@ -59,7 +59,7 @@ def vpformat(datum):  # type: (Any) -> str
 
 def validate_ex(expected_schema,                  # type: Schema
                 datum,                            # type: Any
-                identifiers=None,                 # type: Set[unicode]
+                identifiers=None,                 # type: List[unicode]
                 strict=False,                     # type: bool
                 foreign_properties=None,          # type: Set[unicode]
                 raise_ex=True,                    # type: bool
@@ -70,7 +70,7 @@ def validate_ex(expected_schema,                  # type: Schema
     """Determine if a python datum is an instance of a schema."""
 
     if not identifiers:
-        identifiers = set()
+        identifiers = []
 
     if not foreign_properties:
         foreign_properties = set()