Added a separate regex service, added a common base class for objects that...

Added a separate regex service, added a common base class for objects that hold commonly used SimpleMatcher elements.

common/python/nomadcore/baseclasses.py

@@ -14,6 +14,7 @@ from nomadcore.simple_parser import mainFunction
 from nomadcore.local_backend import LocalBackend
 from nomadcore.local_meta_info import load_metainfo
 from nomadcore.caching_backend import CachingLevel
+from nomadcore.simple_parser import extractOnCloseTriggers
 import nomadcore.ActivateLogging
 from future.utils import with_metaclass
 logger = logging.getLogger("nomad")
@@ -252,6 +253,48 @@ class FileService(object):
             return path
 
 
+#===============================================================================
+class RegexService(object):
+    """
+    Stores basic regex definitions that can be reused on multiple parsers.
+    """
+    def __init__(self):
+        self.regex_f = "-?\d+\.\d+(?:E(?:\+|-)\d+)?"  # Regex for a floating point value
+        self.regex_i = "-?\d+"  # Regex for an integer
+        self.regex_word = "[\S]+"  # Regex for a single word. Can contain anything else but whitespace
+        self.regex_letter = "[^\W\d_]"  # Regex for a single alphabetical letter
+        self.regex_eol = "[^\n]+"  # Regex for a single alphabetical letter
+
+
+#===============================================================================
+class CommonMatcher(object):
+    """
+    This class is used as a base class for objects that store and instantiate
+    common parts of the hierarchical SimpleMatcher structure. The object can be
+    shared for many HierarchicalParsers.
+    """
+    def __init__(self, parser_context):
+
+        # Repeating regex definitions
+        self.parser_context = parser_context
+        self.backend = parser_context.caching_backend
+        self.file_service = parser_context.file_service
+        self.cache_service = parser_context.cache_service
+        self.caching_levels = {}
+        self.regexs = RegexService()
+
+    def getOnCloseTriggers(self):
+        """
+        Returns:
+            A dictionary containing a section name as a key, and a list of
+            trigger functions associated with closing that section.
+        """
+        onClose = {}
+        for attr, callback in extractOnCloseTriggers(self).items():
+            onClose[attr] = [callback]
+        return onClose
+
+
 #===============================================================================
 class BasicParser(with_metaclass(ABCMeta, object)):
     """A base class for all objects that parse contents from files.
@@ -295,36 +338,45 @@ class HierarchicalParser(BasicParser):
 
     Attributes:
         root_matcher: The SimpleMatcher object at the root of this parser.
-            caching_level_for_metaname: A dicionary containing the caching options
+            caching_levels: A dicionary containing the caching options
             that the ActiveBackend will use for individual metanames.
 
             Example:
-                self.caching_level_for_metaname = {
+                self.caching_levels = {
                     'section_XC_functionals': CachingLevel.ForwardAndCache,
                 }
+        caching_levels: Dictionary that stores the caching levels for
+            individual metanames.
         default_data_caching_level: A default caching level for data, i.e.
             metainfo with kindStr=type_document_content or type_dimension
         default_section_caching_level: A default caching level for sections.
         onClose: a dictionary of onClose triggers. These are added to the
             triggers that are already present in the parser context object.
+        cm: An optional CommonMatcher object that is used to store common
+            onClose triggers, SimpleMatchers, caching levels, etc.
+
     """
     def __init__(self, file_path, parser_context):
         super(HierarchicalParser, self).__init__(file_path, parser_context)
         self.root_matcher = None
-        self.caching_level_for_metaname = {}
+        self.caching_levels = {}
         self.default_data_caching_level = CachingLevel.ForwardAndCache
         self.default_section_caching_level = CachingLevel.Forward
         self.onClose = {}
         self.cm = None
+        self.regexs = RegexService()
 
     def setup_common_matcher(self, common_matcher):
         """ Used to setup the CommonMatcher object. This object will contain
         SimpleMatchers, onClose functions, caching levels, etc. that are common
         for many different HierarchicalParsers.
+
+        Args:
+            common_matcher: A CommonMatcher object from which to exctract stuff.
         """
         self.cm = common_matcher
         self.onClose.update(common_matcher.getOnCloseTriggers())
-        self.caching_level_for_metaname.update(common_matcher.caching_levels)
+        self.caching_levels.update(common_matcher.caching_levels)
 
 
 #===============================================================================
@@ -352,7 +404,7 @@ class MainHierarchicalParser(HierarchicalParser):
                 metaInfoEnv=self.parser_context.metainfo_env,
                 parserInfo=self.parser_context.parser_info,
                 outF=self.parser_context.super_backend.fileOut,
-                cachingLevelForMetaName=self.caching_level_for_metaname,
+                cachingLevelForMetaName=self.caching_levels,
                 superContext=self,
                 onClose=self.onClose,
                 default_units=self.parser_context.default_units,
@@ -374,7 +426,8 @@ class MainHierarchicalParser(HierarchicalParser):
         self.parser_context.super_backend = parser.backend.superBackend
         self.backend = self.parser_context.caching_backend
         self.super_backend = self.parser_context.super_backend
-        self.cm.backend = parser.backend
+        if self.cm is not None:
+            self.cm.backend = parser.backend
 
 
 #===============================================================================
@@ -605,7 +658,7 @@ class CacheService(object):
         # # Setup the backend that caches ond handles triggers
         # self.caching_backend = ActiveBackend.activeBackend(
             # metaInfoEnv=self.backend.metaInfoEnv(),
-            # cachingLevelForMetaName=self.caching_level_for_metaname,
+            # cachingLevelForMetaName=self.caching_levels,
             # defaultDataCachingLevel=self.default_data_caching_level,
             # defaultSectionCachingLevel=self.default_section_caching_level,
             # onClose=onClose,