config file loading and checking (work in progress)

config/processor.yaml

@@ -2,6 +2,8 @@ global:
     debug: 1
     verbose: 1
 
+    tempdir: "/tmp"
+
     persistence: 1
     virtualization: 1
 
@@ -19,7 +21,8 @@ resources:
         - /home/resources/media/fr/unitex/dictionary/toponyms.bin
 
 # The 'options' section can contain any of the argument used by the unitex tools
-# functions.
+# functions. Note that some argument will be overriden to fit the 'tag' and 'extract'
+# behaviour. For intance, there is not point to define a font or a context for
+# 'concord'.
 options:
-    match-mode: longest
-    output-mode: merge
+    match_mode: longest

unitex/processor.py

@@ -2,13 +2,14 @@
 # -*- coding: utf-8 -*-
 
 import os
+import tempfile
 import yaml
 
-from unitex import UnitexException, LOGGER, DEFAULT_ENCODING
+from unitex import UnitexException, LOGGER, DEFAULT_ENCODING, VERBOSE, DEBUG
 
 
 
-class UnitexSettings(object):
+class UnitexConfig(object):
 
     def __init__(self):
         self.__settings = None
@@ -21,36 +22,63 @@ class UnitexSettings(object):
             raise UnitexException("Key '%s' not found!" % key)
         return self.__settings[key]
 
-    def set(self, key, value):
+    def __setitem__(self, key, value):
         self.__settings[key] = value
 
-    def load(self, f):
-        with open(f, 'r') as config:
-            self.__settings = yaml.load(config)
-        self.check()
-
-    def check(self):
-        resources = self.__settings.get("resources", None)
-        if resources is None:
-            raise UnitexException("You must provide the 'resources' config element.")
-
-        language = resources.get("language", None)
+    def __load_global(self, options):
+        verbose = options.get("verbose", VERBOSE)
+        if verbose not in (0, 1, 2):
+            raise UnitexException("Wrong value for the 'verbose' global option.")
+        self.__settings["verbose"] = verbose
+
+        debug = options.get("debug", DEBUG)
+        if debug not in (0, 1):
+            raise UnitexException("Wrong value for the 'debug' global option.")
+        self.__settings["debug"] = debug
+
+        for handler in LOGGER.handlers:
+            if debug == 1:
+                fh.setLevel(logging.DEBUG)
+            elif verbose == 1:
+                fh.setLevel(logging.WARNING)
+            elif verbose == 2:
+                fh.setLevel(logging.INFO)
+            else:
+                fh.setLevel(logging.ERROR)
+
+        persistence = options.get("persistence", 0)
+        if persistence not in (0, 1)
+            raise UnitexException("Wrong value for the 'persistence' global option.")
+
+        self.__settings["persistence"] = bool(persistence)
+
+        virtualization = options.get("virtualization", 0)
+        if virtualization not in (0, 1)
+            raise UnitexException("Wrong value for the 'virtualization' global option.")
+
+        self.__settings["virtualization"] = bool(virtualization)
+
+    def __load_resources(self, options):
+        language = options.get("language", None)
         if language is None:
             raise UnitexException("The 'resources' section must contain the 'language' element.")
+        self.__settings["language"] = language
 
-        alphabet = resources.get("alphabet", None)
+        alphabet = options.get("alphabet", None)
         if alphabet is None:
             LOGGER.warning("No alphabet file provided.")
         elif not os.path.exists(alphabet):
             raise UnitexException("Alphabet file '%s' doesn't exist." % alphabet)
+        self.__settings["alphabet"] = alphabet
 
-        alphabet_sort = resources.get("alphabet-sort", None)
+        alphabet_sort = options.get("alphabet-sort", None)
         if alphabet_sort is None:
             LOGGER.warning("No sorted alphabet file provided.")
         elif not os.path.exists(alphabet_sort):
             raise UnitexException("Sorted alphabet file '%s' doesn't exist." % alphabet_sort)
+        self.__settings["alphabet-sort"] = alphabet_sort
 
-        sentence = resources.get("sentence", None)
+        sentence = options.get("sentence", None)
         if sentence is None:
             LOGGER.warning("No sentence grammar provided.")
         else:
@@ -60,7 +88,9 @@ class UnitexSettings(object):
             if not os.path.exists(sentence):
                 raise UnitexException("Sentence grammar file '%s' doesn't exist." % sentence)
 
-        replace = resources.get("replace", None)
+            self.__settings["sentence"] = sentence
+
+        replace = options.get("replace", None)
         if replace is None:
             LOGGER.warning("No replace grammar provided.")
         else:
@@ -70,7 +100,9 @@ class UnitexSettings(object):
             if not os.path.exists(replace):
                 raise UnitexException("Replace grammar file '%s' doesn't exist." % replace)
 
-        dictionaries = resources.get("dictionaries", None)
+            self.__settings["replace"] = replace
+
+        dictionaries = options.get("dictionaries", None)
         if dictionaries is None:
             LOGGER.warning("No dictionaries provided.")
         else:
@@ -85,6 +117,31 @@ class UnitexSettings(object):
                 if not os.path.exists("%s.bin" % prefix):
                     raise UnitexException("Dictionary .inf file missing for '%s'." % dictionary)
 
+            self.__settings["dictionaries"] = dictionaries
+
+    def __load_normalize_options(self, options):
+        pass
+
+    def load(self, path):
+        self.__settings = {}
+        
+        settings = None
+        with open(path, 'r') as config:
+            settings = yaml.load(config)
+
+        if not settings:
+            return
+
+        if "global" in settings:
+            self.__load_global(settings["global"])
+
+        if "resources" not in settings:
+            raise UnitexException("You must provide the 'resources' config element.")
+        self.__load_resources(settings["resources"])
+
+        if "options" in settings:
+            pass
+
 
 
 class UnitexProcessor(object):
@@ -96,7 +153,7 @@ class UnitexProcessor(object):
             self.reset(config)
 
     def reset(self, config):
-        self.__settings = UnitexSettings()
+        self.__settings = UnitexConfig()
         self.__settings.load(config)
 
     def open(self, path, mode="srtlf", encoding=None, tagged=False, virtualize=False):