Handle pandoc logging returning two unknown log levels (#272)

* Handle pandoc logging returning two unknown log levels

* Improve the function description

* Replace redundant yield with return

* Revert "Replace redundant yield with return"

This reverts commit f271d73.

* Make level_map represent pandoc logging levels

Author: H0R5E

pypandoc/__init__.py

@@ -390,16 +390,27 @@ def _convert_input(source, format, input_type, to, extra_args=(),
 
 
 def _classify_pandoc_logging(raw, default_level="WARNING"):
-    # Process raw and yeild the contained logging levels and messages.
+    # Process raw and yield the contained logging levels and messages.
     # Assumes that the messages are formatted like "[LEVEL] message". If the 
-    # first message does not have a level, use the default_level value instead.
+    # first message does not have a level or any other message has a level 
+    # that does not conform to the pandoc standard, use the default_level 
+    # value instead.
 
-    level_map = {"CRITICAL": 50,
-                 "ERROR": 40,
-                 "WARNING": 30,
-                 "INFO": 20,
-                 "DEBUG": 10,
-                 "NOTSET": 0}
+    # Available pandoc logging levels adapted from:
+    # https://github.com/jgm/pandoc/blob/5e1249481b2e3fc27e845245a0c96c3687a23c3d/src/Text/Pandoc/Logging.hs#L44
+    def get_python_level(pandoc_level):
+        
+        level_map = {"ERROR": 40,
+                     "WARNING": 30,
+                     "INFO": 20,
+                     "DEBUG": 10}
+        
+        if pandoc_level not in level_map:
+            level = level_map[default_level]
+        else:
+            level = level_map[pandoc_level]
+        
+        return level
 
     msgs = raw.split("\n")
     first = msgs.pop(0)
@@ -408,29 +419,25 @@ def _classify_pandoc_logging(raw, default_level="WARNING"):
 
     # Use the default if the first message doesn't have a level
     if search is None:
-        level = default_level
+        pandoc_level = default_level
     else:
-        level = first[search.start(1):search.end(1)]
-    
-    log_msgs = [first.replace('[{}] '.format(level), '')]
-    
-    if level not in level_map:
-        level = default_level
+        pandoc_level = first[search.start(1):search.end(1)]
 
+    log_msgs = [first.replace('[{}] '.format(pandoc_level), '')]
 
     for msg in msgs:
 
         search = re.search(r"\[(.*?)\]", msg)
 
         if search is not None:
-            yield level_map[level], "\n".join(log_msgs)
-            level = msg[search.start(1):search.end(1)]
-            log_msgs = [msg.replace('[{}] '.format(level), '')]
+            yield get_python_level(pandoc_level), "\n".join(log_msgs)
+            pandoc_level = msg[search.start(1):search.end(1)]
+            log_msgs = [msg.replace('[{}] '.format(pandoc_level), '')]
             continue
 
         log_msgs.append(msg)
 
-    yield level_map[level], "\n".join(log_msgs)
+    yield get_python_level(pandoc_level), "\n".join(log_msgs)
 
 
 def _get_base_format(format):

tests.py

@@ -337,10 +337,12 @@ def test_classify_pandoc_logging_default(self):
     def test_classify_pandoc_logging_invalid_level(self):
 
         test = ("[WARN] This is some message on\ntwo lines\n"
-                "[ERROR] This is a second message.")
-        expected_levels = [30, 40]
+                "[ERR] This is a second message.\n"
+                "[ERROR] This is a third message.")
+        expected_levels = [30, 30, 40]
         expected_msgs = ["This is some message on\ntwo lines",
-                         "This is a second message."]
+                         "This is a second message.",
+                         "This is a third message."]
 
         for i, (l, m) in enumerate(pypandoc._classify_pandoc_logging(test)):
             self.assertEqual(expected_levels[i], l)