pandoc-citeproc: fix multiple references in brackets

Closes greenelab#290

build/citations.py

@@ -18,19 +18,33 @@ def validate_reference(ref):
     return None
 
 
+bracketed_reference_pattern = re.compile(r'\[(@.+?)\]', flags=re.DOTALL)
+
+
 def get_references_from_text(text):
     """
     Extract the set of references in a text
     """
     refs = set()
-    for ref_text in re.findall(r'\[(@.+?)\]', text, flags=re.DOTALL):
+    for ref_text in bracketed_reference_pattern.findall(text):
         for ref in ref_text.split():
             if not ref:
                 continue
             refs.add(ref)
     return refs
 
 
+def semicolon_separate_references(text):
+    """
+    Separate multiple references inside the same brackets with a space and
+    semicolon for pandoc-citeproc.
+    """
+    return bracketed_reference_pattern.sub(
+        repl=lambda x: '; '.join(x.group().split()),
+        string=text
+    )
+
+
 def get_brackets_without_reference(text):
     """
     Find bracketed text that does not start with @. Does not match

build/references.py

@@ -19,6 +19,7 @@
     get_brackets_without_reference,
     get_references_from_text,
     get_text,
+    semicolon_separate_references,
     validate_reference,
 )
 
@@ -120,6 +121,8 @@ def get_standard_citatation(citation, cache):
     old = re.escape(old)
     new = f'@{new}'
     converted_text = re.sub(old + '(?=[\s\]])', new, converted_text)
+# Semicolon separate multiple refernces for pandoc-citeproc
+converted_text = semicolon_separate_references(converted_text)
 
 # Write manuscript for pandoc
 with gen_dir.joinpath('all-sections.md').open('wt') as write_file: