First set of algorithm to detect duplicate entries in the validation procedure.

modules/check_tools.py

@@ -7,7 +7,10 @@
 import re
 
 from gluon import current
-from plugin_dbui import UNDEF, UNDEF_ID
+from plugin_dbui import (UNDEF, 
+                         UNDEF_ID, 
+                         get_id, 
+                         get_where_query)
 
 
 # syntax for the submission date YYYY or YYYY-MM or YYYY-MM-DD
@@ -21,7 +24,7 @@ def check_publication(row):
     """Check the fields of a publication.
     
     @type row: gluon.dal.Row
-    @param row: record defineing a publication. Its contains the publications
+    @param row: record defining a publication. Its contains the publications
     table as well as its reference tables.
     
     @rtype: list
@@ -89,7 +92,12 @@ def check_publication(row):
         if not row.publications.preprint:
             text = T("Preprint number is not defined")
             li.append(text)
-            
+
+        ids = duplicate_article(row)
+        if ids:
+            text = T("Possible duplicate entries [%s]") % s
+            li.append(text)
+
     # specific fields for proceeding and talk
     if row.categories.usual in ('proceeding', 'talk'):
 
@@ -112,6 +120,11 @@ def check_publication(row):
         if not row.publications.conference_speaker:
             text = T("Conference speaker is missing")
             li.append(text)
+
+        ids = duplicate_conference(row)
+        if ids:
+            text = T("Possible duplicate entries [%s]") % s
+            li.append(text)
     
     # specific fields for report
     if row.categories.usual == 'report':
@@ -119,5 +132,124 @@ def check_publication(row):
         if not row.publications.report_numbers:
             text = T("Report number is missing")
             li.append(text)
+
+        ids = duplicate_report(row)
+        if ids:
+            text = T("Possible duplicate entries [%s]") % s
+            li.append(text)
+            
+    return li
+
+
+def duplicate_article(row):
+    """Look for duplicate article.
+    
+    @type row: gluon.dal.Row
+    @param row: record defining a publication. Its contains the publications
+    table as well as its reference tables.
+    
+    @rtype: list
+    @return: list of ids corresponding to duplicate entries
+    
+    """
+    ids = []
+    db = current.globalenv['db']
+    
+    qmain = get_where_query(db.publications)
+    qmain = ((qmain) & (db.publications.id != row.publications.id))
+    qmain = ((qmain) & (db.categories.code == 'ACL'))
+    qmain = ((qmain) & (db.teams.team == row.teams.team))
+    qmain = ((qmain) & (db.publications.id_publishers == row.publications.id_publishers))
+    
+    # look for publications with the same title, review and team
+    query = ((qmain) & (db.publications.title == row.publications.title))
+    query = ((query) & (db.publications.volume == row.publications.volume))
+    query = ((query) & (db.publications.pages == row.publications.pages))
+    
+    set = db(query)
+    if set.count():
+        for xrow in set.select():
+            ids.append(str(xrow.publications.id))
+            
+    # less stringent test same review and team
+    query = ((qmain) & (db.publications.volume == row.publications.volume))
+    query = ((query) & (db.publications.pages == row.publications.pages))
+
+    set = db(query)
+    if set.count():
+        for xrow in set.select():
+            xid = str(xrow.publications.id)
+            if xid not in ids:
+                ids.append(xid)
+
+    # less stringent test same title and team
+    query = ((qmain) & (db.publications.title == row.publications.title))
+
+    set = db(query)
+    if set.count():
+        for xrow in set.select():
+            xid = str(xrow.publications.id)
+            if xid not in ids:
+                ids.append(xid)
+
+    return ids
+
+
+def duplicate_conference(row):
+    """Look for duplicate talk / proceeding.
+    
+    @type row: gluon.dal.Row
+    @param row: record defining a publication. Its contains the publications
+    table as well as its reference tables.
+    
+    @rtype: list
+    @return: list of ids corresponding to duplicate entries
+    
+    """
+    ids = []
+    db = current.globalenv['db']
+    
+    query = get_where_query(db.publications)
+    query = ((query) & (db.publications.id != row.publications.id))
+    query = ((query) & ((db.categories.code == 'ACTI') | (db.categories.code == 'COM')))
+    query = ((query) & (db.teams.team == row.teams.team))
+    query = ((query) & (db.publications.conference_title == row.publications.conference_title))
+    query = ((query) & (db.publications.conference_dates == row.publications.conference_dates))
+    query = ((query) & (db.publications.conference_town == row.publications.conference_town))
+    query = ((query) & (db.publications.title == row.publications.title))
+    
+    set = db(query)
+    if set.count():
+        for xrow in set.select():
+            ids.append(str(xrow.publications.id))
+            
+    return ids
+
+
+def duplicate_report(row):
+    """Look for duplicate report.
+    
+    @type row: gluon.dal.Row
+    @param row: record defining a publication. Its contains the publications
+    table as well as its reference tables.
+    
+    @rtype: list
+    @return: list of ids corresponding to duplicate entries
+    
+    """
+    ids = []
+    db = current.globalenv['db']
+    
+    query = get_where_query(db.publications)
+    query = ((query) & (db.publications.id != row.publications.id))
+    query = ((query) & (db.categories.code == 'AP'))
+    query = ((query) & (db.teams.team == row.teams.team))
+    query = ((query) & (db.publications.title == row.publications.title))
+    
+    set = db(query)
+    if set.count():
+        for xrow in set.select():
+            ids.append(str(xrow.publications.id))
             
-    return li
\ No newline at end of file
+    return ids
+    
\ No newline at end of file

static/CHANGELOG

@@ -4,6 +4,7 @@ HEAD
   - Migrate to plugin_dbui 0.4.13.1
   - Add the module check_tools.
   - More stringent tests in the validation procedure
+    including duplicate entries.
 
 0.8.2 (Apr 2013)
   - Consolidation version.