Rename corpus info file

eea · Aug 21, 2017 · a5c107e · a5c107e
1 parent dc7f2d6
commit a5c107e
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 11 deletions.
diff --git a/src/eea.corpus/eea/corpus/corpus.py b/src/eea.corpus/eea/corpus/corpus.py
@@ -15,7 +15,7 @@
 
 def save_corpus_metadata(stats, file_name, corpus_id, text_column, **kw):
     cpath = corpus_base_path(file_name)      # corpus_id
-    meta_name = "{0}_eea.json".format(corpus_id)
+    meta_name = "{0}_info.json".format(corpus_id)
     meta_path = os.path.join(cpath, meta_name)
 
     title = kw.pop('title')
@@ -107,6 +107,7 @@ def get_corpus(request, doc=None, corpus_id=None):
     if not (doc and corpus_id):
         doc, corpus_id = extract_corpus_id(request)
 
+    assert doc and corpus_id
     corpus = load_corpus(file_name=doc, corpus_id=corpus_id)
     return corpus
 

diff --git a/src/eea.corpus/eea/corpus/tests/test_corpus.py b/src/eea.corpus/eea/corpus/tests/test_corpus.py
@@ -27,7 +27,7 @@ def test_build_corpus(self, build_pipeline, corpus_base_path, tmpdir):
 
         build_corpus(pipeline, corpus_id, file_name, text_column, **kw)
 
-        assert path.join('test_eea.json').exists()
+        assert path.join('test_info.json').exists()
         assert path.join('test_docs.json').exists()
 
         docs = []
@@ -39,7 +39,7 @@ def test_build_corpus(self, build_pipeline, corpus_base_path, tmpdir):
         assert docs[1] == {'text': 'Second time', 'metadata': {'3': 4}}
         assert len(docs) == 2
 
-        with path.join('test_eea.json').open() as f:
+        with path.join('test_info.json').open() as f:
             meta = json.load(f)
             assert meta == {
                 'description': 'something else',

diff --git a/src/eea.corpus/eea/corpus/utils.py b/src/eea.corpus/eea/corpus/utils.py
@@ -70,7 +70,7 @@ def available_corpus(file_name):
         files[base].append(spec)
 
         for corpus, cfs in files.items():
-            if len(cfs) != 4:
+            if len(cfs) != len(('docs', 'info')):
                 logger.warning("Not a valid corpus: %s (%s)",
                                file_name, corpus)
                 continue
@@ -79,19 +79,19 @@ def available_corpus(file_name):
     return res
 
 
-def corpus_metadata_path(file_name, corpus_id):
-    """ Returns the <corpusid>_eea.json file path for a given doc/corpus
+def corpus_info_path(file_name, corpus_id):
+    """ Returns the <corpusid>_info.json file path for a given doc/corpus
     """
     cpath = corpus_base_path(file_name)      # corpus_id
-    meta_name = "{0}_eea.json".format(corpus_id)
+    meta_name = "{0}_info.json".format(corpus_id)
     meta_path = os.path.join(cpath, meta_name)
     return meta_path
 
 
 def load_corpus_metadata(file_name, corpus_id):
     """ Returns the EEA specific metadata saved for a doc/corpus
     """
-    meta_path = corpus_metadata_path(file_name, corpus_id)
+    meta_path = corpus_info_path(file_name, corpus_id)
 
     res = None
 
@@ -120,9 +120,8 @@ def available_documents(request):
                 files[base].append(spec)
 
             for corpus, cfs in files.items():
-                if len(cfs) != 4:
-                    # logger.warning("Not a valid corpus: %s (%s)", name,
-                    # corpus)
+                if len(cfs) != len(('docs', 'info')):
+                    logger.warning("Not a valid corpus: %s (%s)", name, corpus)
                     continue
                 meta = load_corpus_metadata(name, corpus)
                 corpuses.append((corpus, meta))