Update CheckAndFix to use logger.

modules/harvest_tools/automaton.py

@@ -586,7 +586,7 @@ class Automaton(object):
         # check that the record is well formed
         # repair non-conformity as far as possible
         if not self.check_record(record):
-            logger.debug(f"{T4}rejected {logs[-1].txt}")
+            logger.debug(f"{T4}{logs[-1].txt}")
             return
 
         txt = ("(dry run)" if self.dry_run else "")

modules/harvest_tools/checkandfix.py

 """ harvest_tools.checkandfix
 
 """
+import logging
 import numpy as np
 import re
 
@@ -69,6 +70,7 @@ REG_WELL_FORMED_CONF_DATES_1 = \
 REG_WELL_FORMED_CONF_DATES_2 = \
     re.compile("\d{1,2} [A-Z][a-z]{2} - \d{1,2} [A-Z][a-z]{2} \d{4}")
 
+T6 = " "*6
 UNIVERSITY = "University"
 
 
@@ -84,6 +86,7 @@ class CheckAndFix(object):
 
         self.db = current.db
         self.dbg = debug
+        self.logger = logging.getLogger("web2py.app.limbra")
         self.reg_institute = self._get_reg_institute()
 
         # private cache for my_author rescue list
@@ -340,8 +343,7 @@ class CheckAndFix(object):
                 when there is no authors.
 
         """
-        if self.dbg:
-            print("\t\tCheck authors")
+        self.logger.debug(f"{T6}check authors")
 
         if not record.is_authors():
             raise CheckException(MSG_NO_AUTHOR)
@@ -361,8 +363,7 @@ class CheckAndFix(object):
                 * the collaboration is unknown (neither collaborationnor synonym)
                 * more than one synonym found.
         """
-        if self.dbg:
-            print("\t\tCheck collaboration")
+        self.logger.debug(f"{T6}check collaboration")
 
         val = record.collaboration()
         if not val:
@@ -404,8 +405,7 @@ class CheckAndFix(object):
                 * more than one synonym found.
 
         """
-        if self.dbg:
-            print("\t\tCheck country")
+        self.logger.debug(f"{T6}check country")
 
         if not isinstance(record, RecordConf):
             return
@@ -451,8 +451,7 @@ class CheckAndFix(object):
                 dates are not found.
 
         """
-        if self.dbg:
-            print("\t\tCheck conference date")
+        self.logger.debug(f"{T6}check conference date")
 
         # conference information are available, i.e proceeding
         if not isinstance(record, RecordConf):
@@ -503,8 +502,7 @@ class CheckAndFix(object):
                 the bad OAI.
 
         """
-        if self.dbg:
-            print("\t\tCheck is bad oai used")
+        self.logger.debug(f"{T6}check is bad oai used")
 
         value = record.oai()
         match = REG_OAI.match(value)
@@ -529,8 +527,7 @@ class CheckAndFix(object):
             bool:
                 ``True`` when the OAI is not defined in the record.
         """
-        if self.dbg:
-            print("\t\tCheck is oai")
+        self.logger.debug(f"{T6}check is oai")
 
         # field / subfield depends on the store
         test = ("oai" in record and "value" in record["oai"]) or \
@@ -551,8 +548,7 @@ class CheckAndFix(object):
                 ``Last, First`` and ``Last F.``
 
         """
-        if self.dbg:
-            print("\t\tFormat authors")
+        self.logger.debug(f"{T6}format authors")
 
         record.reformat_authors(fmt)
 
@@ -574,8 +570,7 @@ class CheckAndFix(object):
                 when the editor is not well formed.
 
         """
-        if self.dbg:
-            print("\t\tFormat editor")
+        self.logger.debug(f"{T6}format editor")
 
         if not record.is_published():
             return
@@ -610,8 +605,7 @@ class CheckAndFix(object):
                 record describing a thesis.
 
         """
-        if self.dbg:
-            print("\t\tFormat university")
+        self.logger.debug(f"{T6}format university")
 
         # protection
         if not isinstance(record, RecordThesis):
@@ -675,8 +669,7 @@ class CheckAndFix(object):
                 the list is empty
 
         """
-        if self.dbg:
-            print("\t\tGet my authors")
+        self.logger.debug(f"{T6}get my authors")
 
         # might have been computed when affiliation is checked
         rec_id = record.id()
@@ -707,8 +700,7 @@ class CheckAndFix(object):
                 the record is not associated to a conference.
 
         """
-        if self.dbg:
-            print("\t\tIs conference")
+        self.logger.debug(f"{T6}is conference")
 
         if not isinstance(record, RecordConf):
             raise CheckException(MSG_NO_CONF)
@@ -728,8 +720,7 @@ class CheckAndFix(object):
                 the record does not describe a thesis.
 
         """
-        if self.dbg:
-            print("\t\tIs thesis")
+        self.logger.debug(f"{T6}is thesis")
 
         if not isinstance(record, RecordThesis):
             raise CheckException(MSG_NO_THESIS)
@@ -775,8 +766,7 @@ class CheckAndFix(object):
                 and the author is null.
 
         """
-        if self.dbg:
-            print("\t\tCheck my affiliation")
+        self.logger.debug(f"{T6}check my affiliation")
 
         value = record.find_affiliation(self.reg_institute)
         if len(value) > 0:
@@ -828,8 +818,7 @@ class CheckAndFix(object):
                 the paper reference is not well formed.
 
         """
-        if self.dbg:
-            print("\t\tCheck paper reference")
+        self.logger.debug(f"{T6}check paper reference")
 
         if record.is_published():
             return
@@ -891,8 +880,7 @@ class CheckAndFix(object):
                 * more than one synonym found.
 
         """
-        if self.dbg:
-            print("\t\tCheck publisher")
+        self.logger.debug(f"{T6}check publisher")
 
         val = record.paper_editor()
         if len(val) == 0:
@@ -931,8 +919,7 @@ class CheckAndFix(object):
                 * more than one date are found.
 
         """
-        if self.dbg:
-            print("\t\tCheck submitted")
+        self.logger.debug(f"{T6}check submitted")
 
         date = record.submitted()
 
@@ -982,8 +969,7 @@ class CheckAndFix(object):
                 the record is marked temporary
 
         """
-        if self.dbg:
-            print("\t\tCheck temporary record")
+        self.logger.debug(f"{T6}check temporary record")
 
         # INSPIREHEP
         # Can be find by using the XML syntax: