Add the script fix_acti2com_cppm.py

scripts/fix_acti2com_cppm.py

+# -*- coding: utf-8 -*-
+""" NAME
+        fix_acti2com_cppm
+
+    SYNOPSIS
+        Copy ACTI to COM for the CPPM database.
+
+    DESCRIPTION
+        Up to the end of 2014 a talk to a conference (COM) is transformed
+        into a proceeding (ACTI) when the later is published.
+
+        In 2015, the policy changes, COM and ACTI are kept as separated
+        publications
+
+        This script implement the new policy for the publications registered
+        before 2015. It mainly copy the ACTI into COM for the year from
+        2009 up to 2014.
+
+    OPTIONS
+
+        -h, --help
+            Display the help and exit.
+
+    EXAMPLE
+
+        > cd ...track_publications/scripts
+        > ./run -S test_publications script fix_acti2com_cppm.py
+        > ./run -S track_publications script fix_acti2com_cppm.py
+
+
+    AUTHOR
+        R. Le Gac -- Jan 2016
+
+"""
+import re
+
+
+REG_ORIGIN = re.compile("https?://([a-z\.]+)/record/(\d+)")
+
+if __name__ == "__main__":
+
+    import sys
+
+    from argparse import ArgumentParser, FileType
+    from invenio_tools import load_record, OAI_URL
+    from plugin_dbui import CALLBACK_ERRORS, get_id
+
+    # command line options
+    parser = ArgumentParser()
+    args = parser.parse_args()
+
+    # get the ACTI / COM identifier
+    id_acti = get_id(db.categories, code="ACTI")
+    id_com = get_id(db.categories, code="COM")
+
+    # build the query
+    query = db.publications.id_categories == id_acti
+    query &= db.publications.year >= 2009
+    query &= db.publications.year <= 2014
+
+    # scan the publications table
+    for row in db(query).select():
+        data = row.as_dict()
+
+        # skip if the the speaker is not from CPPM
+        if data["conference_speaker"] not in data["authors_institute"]:
+            continue
+
+        # remove publisher information
+        data["id_publishers"] = 1
+        data["pages"] = ""
+        data["volume"] = ""
+        data["submitted"] = ""
+        data["publication_url"] = ""
+        data["preprint"] = ""
+
+        # the year is the one of the conference
+        conference_dates = data["conference_dates"]
+        if conference_dates:
+            data["year"] = int(data["conference_dates"][-4:])
+
+        else:
+            print "No conferences dates", data["id"]
+
+        # change the category and the status
+        data["id_categories"] = id_com
+        data["id_status"] = 1
+
+        # change the origin
+        val = data["origin"]
+        if val:
+            origin = val.split(',')[0].strip()
+
+            match = REG_ORIGIN.match(origin)
+            host, rec_id = match.group(1), match.group(2)
+
+            proceeding = load_record(host, rec_id)
+
+            talk_id = proceeding.reference_conference_talk()
+            data["origin"] = (OAI_URL % (host, talk_id) if talk_id else "")
+
+        # insert the new record in the database
+        id_rec = data["id"],
+
+        del data["id"]
+        id_new = db.publications.insert(**data)
+
+        print "Copy", id_rec, "→",
+        if id_new:
+            print id_new
+
+        elif CALLBACK_ERRORS in db.publications:
+
+            print " ".join(db.publications._callback_errors)
+
+        else:
+            print "???"
+
+    # commit change
+    rep = raw_input("Commit change in the database? [y/N]:")
+    if rep == 'y':
+        print "The database is modified."
+        db.commit()
+
+    # close
+    sys.exit(0)