#!/usr/bin/env python3

#

# Print differencies between data in tables.

#

# Requirements:

#  * Source table must have defined PRIMARY KEY.

#  * Destination table must contain all columns from source table.

#    Order is not important.

#

from pgtoolkit import toolbase, pgmanager, pgdatadiff

from pycolib.ansicolor import highlight, BOLD, YELLOW

class TableDiffTool(toolbase.SrcDstTablesTool):

    def __init__(self):

        toolbase.SrcDstTablesTool.__init__(self, name='tablediff', desc='Table diff.', allow_reverse = True)

        self.parser.add_argument('--sql', action='store_true', help='Output is SQL script.')

        self.parser.add_argument('--rowcount', action='store_true', help='Compare number of rows.')

        self.init()

    def main(self):

        srcconn = self.pgm.get_conn('src')

        dstconn = self.pgm.get_conn('dst')

        dd = pgdatadiff.PgDataDiff(srcconn, dstconn)

        for srcschema, srctable, dstschema, dsttable in self.tables():

            print('-- Diff from [%s] %s.%s to [%s] %s.%s' % (

                self.args.src, srcschema, srctable,

                self.args.dst, dstschema, dsttable))

            if self.args.rowcount:

                with self.pgm.cursor('src') as curs:

                    curs.execute('''SELECT count(*) FROM "%s"."%s"''' % (srcschema, srctable))

                    srccount = curs.fetchone()[0]

                with self.pgm.cursor('dst') as curs:

                    curs.execute('''SELECT count(*) FROM "%s"."%s"''' % (dstschema, dsttable))

                    dstcount = curs.fetchone()[0]

                if srccount != dstcount:

                    print(highlight(1, BOLD | YELLOW), "Row count differs: src=%s dst=%s" % (srccount, dstcount), highlight(0), sep='')

                continue

            dd.settable1(srctable, srcschema)

            dd.settable2(dsttable, dstschema)

            if self.args.sql:

                dd.print_patch()

            else:

                dd.print_diff()

tool = TableDiffTool()

tool.main()