string_quote_removal: rm escaped and double quotes (#99)

Author: WillForan

bashlex/heredoc.py

@@ -14,15 +14,32 @@ def gatherheredocuments(tokenizer):
 
 
 def string_quote_removal(word):
-    """remove single quotes for heredoc
+    """
+    Remove surrounding quotes for heredoc token.
+    This function should also remove escaped quotes (\\", \\')
+    See definition in bash's source
+    https://github.com/bminor/bash/blob/master/subst.c#L11892
+
+    >>> string_quote_removal("EOF")
+    'EOF'
     >>> string_quote_removal("'EOF'")
     'EOF'
-    >>> string_quote_removal("EOF")
+    >>> string_quote_removal('"EOF"')
+    'EOF'
+    >>> string_quote_removal('\\"EOF\\"')
     'EOF'
+    
+    # TODO: confirm only first level of surrounding quotes should be removed
+    >>> string_quote_removal('""EOF""')
+    '"EOF"'
+    >>> string_quote_removal('"\\"EOF"\\"')
+    '\"EOF\"'
     """
-    quote_match = re.search("^'(.*)'$", word)
+    # match input starting with a single, double, or escaped quote (' " \" \')
+    #         and ending with the same (optionally escaped) quote
+    quote_match = re.search("^(\\\\*[\"'])(.*)\\1$", word)
     if quote_match:
-        word = quote_match.group(1)
+        word = quote_match.group(2)
     return word