[rset] Use RQL cache in syntax_tree method
This avoid several unnecessary parsings per HTTP request, since in most case
rset's rql is in the rql cache (since it has just been executed), while yapps
based parsing is known to be slow.
"""This module compare the Schema on the file system to the one in the database"""
from cStringIO import StringIO
from cubicweb.web.schemaviewer import SchemaViewer
from logilab.common.ureports import TextWriter
import difflib
viewer = SchemaViewer()
layout_db = viewer.visit_schema(schema, display_relations=True)
layout_fs = viewer.visit_schema(fsschema, display_relations=True)
writer = TextWriter()
stream_db = StringIO()
stream_fs = StringIO()
writer.format(layout_db, stream=stream_db)
writer.format(layout_fs, stream=stream_fs)
stream_db.seek(0)
stream_fs.seek(0)
db = stream_db.getvalue().splitlines()
fs = stream_fs.getvalue().splitlines()
open('db_schema.txt', 'w').write(stream_db.getvalue())
open('fs_schema.txt', 'w').write(stream_fs.getvalue())
#for diff in difflib.ndiff(fs, db):
# print diff