Update checkAndFix to add a debug mode.

modules/harvest_tools/automaton.py

@@ -129,7 +129,7 @@ class Automaton(object):
         if not id_category:
             raise ToolException(MSG_NO_CAT)
 
-        self.check = CheckAndFix()
+        self.check = CheckAndFix(debug)
         self.collection_logs = []
         self.controller = automaton
         self.db = db

modules/harvest_tools/checkandfix.py

@@ -91,13 +91,17 @@ UNIVERSITY = "University"
 
 
 class CheckAndFix(object):
-    """A collection of tools to check and repair the content
-    of the Marc12 record.
+    """A collection of tools to check and repair the content of record.
+
+    Args:
+        debug (bool):
+            activate the debug mode.
 
     """
-    def __init__(self):
+    def __init__(self, debug=False):
 
         self.db = current.db
+        self.dbg = debug
         self.reg_institute = self._get_reg_institute()
 
         # private cache for my_author rescue list
@@ -339,8 +343,7 @@ class CheckAndFix(object):
 
         return val
 
-    @staticmethod
-    def authors(record):
+    def authors(self, record):
         """Check that author fields are defined.
 
         Args:
@@ -352,6 +355,8 @@ class CheckAndFix(object):
                 when there is no authors.
 
         """
+        if self.dbg:
+            print "\t\tCheck authors"
 
         if not record.is_authors():
             raise CheckException(MSG_NO_AUTHOR)
@@ -370,6 +375,9 @@ class CheckAndFix(object):
                 nor entered as a synonym.
 
         """
+        if self.dbg:
+            print "\t\tCheck collaboration"
+
         val = record.collaboration()
         if not val:
             return
@@ -393,6 +401,9 @@ class CheckAndFix(object):
                 the country is not defined nor entered as a synonym.
 
         """
+        if self.dbg:
+            print "\t\tCheck country"
+
         if not isinstance(record, RecordConf):
             return
 
@@ -416,6 +427,9 @@ class CheckAndFix(object):
                 dates are not found.
 
         """
+        if self.dbg:
+            print "\t\tCheck conference date"
+
         # conference information are available, i.e proceeding
         if not isinstance(record, RecordConf):
             return
@@ -465,6 +479,9 @@ class CheckAndFix(object):
                 the bad OAI.
 
         """
+        if self.dbg:
+            print "\t\tCheck is bad oai used"
+
         value = record.oai()
         match = REG_OAI.match(value)
 
@@ -478,8 +495,7 @@ class CheckAndFix(object):
 
         return False
 
-    @staticmethod
-    def is_oai(record):
+    def is_oai(self, record):
         """Raise exception when the OAI is not defined.
 
         Args:
@@ -489,6 +505,9 @@ class CheckAndFix(object):
             ToolException:
                 OAI is not defined
         """
+        if self.dbg:
+            print "\t\tCheck is oai"
+
         # field / subfield depends on the store
         test =  (u"oai" in record and u"value" in record[u"oai"]) or \
                 (u"FIXME_OAI" in record and u"id" in record[u"FIXME_OAI"])
@@ -496,8 +515,7 @@ class CheckAndFix(object):
         if not test:
             raise ToolException(MSG_NO_OAI)
 
-    @staticmethod
-    def format_authors(record, fmt="Last, First"):
+    def format_authors(self, record, fmt="Last, First"):
         """Format the author names.
 
         Args:
@@ -508,10 +526,12 @@ class CheckAndFix(object):
                 "Last, First" and "Last F."
 
         """
+        if self.dbg:
+            print "\t\tFormat authors"
+
         record.reformat_authors(fmt)
 
-    @staticmethod
-    def format_editor(record):
+    def format_editor(self, record):
         """Format the editor abbreviation.
         The encoding depends on the store::
 
@@ -529,6 +549,9 @@ class CheckAndFix(object):
                 when the editor is not well formed.
 
         """
+        if self.dbg:
+            print "\t\tFormat editor"
+
         if not record.is_published():
             return
 
@@ -562,6 +585,9 @@ class CheckAndFix(object):
                 record describing a thesis.
 
         """
+        if self.dbg:
+            print "\t\tFormat university"
+
         # protection
         if not isinstance(record, RecordThesis):
             return
@@ -611,6 +637,9 @@ class CheckAndFix(object):
                 when the list is empty
 
         """
+        if self.dbg:
+            print "\t\tGet my authors"
+
         # might have been computed when affiliation is checked
         rec_id = record.id()
         if rec_id in self._my_authors:
@@ -628,8 +657,7 @@ class CheckAndFix(object):
 
         record.my_authors = value
 
-    @staticmethod
-    def is_conference(record):
+    def is_conference(self, record):
         """Check that the record described a conference talk / proceeding.
 
         Args:
@@ -641,11 +669,13 @@ class CheckAndFix(object):
                 the record is not associated to a conference.
 
         """
+        if self.dbg:
+            print "\t\tIs conference"
+
         if not isinstance(record, RecordConf):
             raise CheckException(MSG_NO_CONF)
 
-    @staticmethod
-    def is_thesis(record):
+    def is_thesis(self, record):
         """Check that the record described a thesis.
 
         Args:
@@ -657,6 +687,9 @@ class CheckAndFix(object):
                 the record does not describe a thesis.
 
         """
+        if self.dbg:
+            print "\t\tIs thesis"
+
         if not isinstance(record, RecordThesis):
             raise CheckException(MSG_NO_THESIS)
 
@@ -701,6 +734,9 @@ class CheckAndFix(object):
                 and the author is null.
 
         """
+        if self.dbg:
+            print "\t\tCheck my affiliation"
+
         value = record.find_affiliation(self.reg_institute)
         if len(value) > 0:
             return value
@@ -738,8 +774,7 @@ class CheckAndFix(object):
 
         return ""
 
-    @staticmethod
-    def paper_reference(record):
+    def paper_reference(self, record):
         """Check that editor, page, volume and paper year are defined
         for a published paper. Repair it from doi when possible.
 
@@ -752,6 +787,9 @@ class CheckAndFix(object):
                 when the paper reference is not well formed.
 
         """
+        if self.dbg:
+            print "\t\tCheck paper reference"
+
         if record.is_published():
             return
 
@@ -811,6 +849,9 @@ class CheckAndFix(object):
                 when the publisher is not defined nor entered as a synonym.
 
         """
+        if self.dbg:
+            print "\t\tCheck publisher"
+
         val = record.paper_editor()
         if len(val) == 0:
             return
@@ -840,6 +881,9 @@ class CheckAndFix(object):
                 than one date are found.
 
         """
+        if self.dbg:
+            print "\t\tCheck submitted"
+
         date = record.submitted()
 
         # recover missing date using conference, preprint, thesis information
@@ -876,8 +920,7 @@ class CheckAndFix(object):
         else:
             record[u"prepublication"] = {u"date": date}
 
-    @staticmethod
-    def temporary_record(record):
+    def temporary_record(self, record):
         """Some records are marked temporary.
 
         Args:
@@ -887,6 +930,9 @@ class CheckAndFix(object):
             CheckException: when the record is marked temporary
 
         """
+        if self.dbg:
+            print "\t\tCheck temporary record"
+
         # INSPIREHEP
         # Can be find by using the XML syntax:
         #    http://inspirehep.net/search?500__a="*Temporary record*"