refactoring DBO_RDG__skill_graph out of Test_Domain_CRI

3 files changed, 171 insertions, 148 deletions

diff --git a/src/local/domain/cri.rhizi.net/domain_cri.py b/src/local/domain/cri.rhizi.net/domain_cri.py
deleted file mode 100644
index db6ed925..00000000
--- a/src/local/domain/cri.rhizi.net/domain_cri.py
+++ /dev/null
@@ -1,140 +0,0 @@
-from db_op import DB_op
-
-class DBO_random_data_generation__domain__CRI(DB_op):
-
-    def __init__(self, lim_n=50,
-                       lim_r=10000,
-                       prob_link_create=0.03):
-        """
-        generate random data: CRI domain: Person x Skill
-        """
-        super(DBO_random_data_generation__domain__CRI, self).__init__()
-
-        n_attr_set__skill_raw = ['Cryptocurrency',
-                             'Database architecture',
-                             'Web-development',
-                             'Mobile-development',
-                             'Machine learning',
-                             'Guitar Playing',
-                             'Image processing',
-                             'Algebra',
-                             'Calculus',
-                             'Molecular-Biology',
-                             'Graphic design',
-                             'Geo-location services',
-                             'Drone-building',
-                             'Artificial-intelligence',
-                             'Distributed information systems',
-                             'Wordpress',
-                             'Woodworking',
-                             'Quality-control',
-                             'Video-editing',
-                             'Soldering',
-                             'Network engineering',
-                             'GIT',
-                             'Electronic music',
-                             'Network administration']
-
-        n_attr_set__skill = []
-        for s in n_attr_set__skill_raw:
-            n_attr_set__skill.append(s)
-
-        n_attr_set__name = ['John',
-                            'William',
-                            'James',
-                            'Charles',
-                            'George',
-                            'Frank',
-                            'Joseph',
-                            'Thomas',
-                            'Henry',
-                            'Robert',
-                            'Edward',
-                            'Harry',
-                            'Walter',
-                            'Arthur',
-                            'Fred',
-                            'Albert',
-                            'Samuel',
-                            'David',
-                            'Louis',
-                            'Joe',
-                            'Charlie',
-                            'Clarence',
-                            'Richard',
-                            'Andrew',
-                            'Daniel',
-                            'Ernest',
-                            'Mary',
-                            'Anna',
-                            'Emma',
-                            'Elizabeth',
-                            'Minnie',
-                            'Margaret',
-                            'Ida',
-                            'Alice',
-                            'Bertha',
-                            'Sarah',
-                            'Annie',
-                            'Clara',
-                            'Ella',
-                            'Florence',
-                            'Cora',
-                            'Martha',
-                            'Laura',
-                            'Nellie',
-                            'Grace',
-                            'Carrie',
-                            'Maude',
-                            'Mabel',
-                            'Bessie',
-                            'Jennie',
-                            'Gertrude',
-                            'Julia']
-
-        l_attr_set__level = ['Novice', 'Intermediate', 'Expert']
-
-
-        min_id = 10000  # separate group ids with numerical perfix
-
-        # create persons
-        q_arr = ['with {n_attr_set__name} as n_attr_set',  # TODO clean: foreach triggers SyntaxException: otherwise
-                 'foreach (idx in range(0,%d)' % (lim_n - 1),
-                 '|',
-                 'create (n:%s' % ('Person'),
-                 '{id: \'test-id_\' + toString(%d + idx),' % (min_id),
-                  'name: n_attr_set[idx %% %d]}' % (len(n_attr_set__name)),
-                 '))',
-                ]
-        q_param = {'n_attr_set__name': n_attr_set__name}
-        self.add_statement(q_arr, q_param)
-        min_id *= 2
-
-        # create skills
-        q_arr = ['with {skill_set} as skill_set',  # TODO clean: foreach triggers SyntaxException: otherwise
-                 'foreach (idx in range(0,%d)' % (len(n_attr_set__skill) - 1),
-                 '|',
-                 'create (n:%s' % ('Skill'),
-                 '{id: \'test-id_\' + toString(idx + %d),' % (min_id),
-                  'name: skill_set[idx]}',
-                 '))',
-                ]
-        q_param = {'skill_set': n_attr_set__skill}
-        self.add_statement(q_arr, q_param)
-        min_id *= 2
-
-        # create links
-        for skill_level in l_attr_set__level:
-            q_arr = ['match (n:%s),(m:%s)' % ('Person', 'Skill'),
-                     'with n, m, rand() as rand',
-                     'order by rand',
-                     'limit %d' % (lim_r - 1),
-                     'where rand() < %.2f' % (prob_link_create),
-                     'create (n)-[r:%s' % (skill_level),
-                     '{id: \'test-id_\' + toString(%d + toInt(%d * rand())),' % (min_id, lim_r * 100000),  # aim for low id collision probability,
-                     'proficiency: \'%s\'}' % (skill_level),
-                     ']->(m)',
-                     'return collect(r.id)',
-                     ]
-
-            self.add_statement(q_arr, q_param)
diff --git a/src/local/domain/cri.rhizi.net/test_domain__CRI.py b/src/local/domain/cri.rhizi.net/test_domain__CRI.py
index 54131bbd..652c04a6 100644
--- a/src/local/domain/cri.rhizi.net/test_domain__CRI.py
+++ b/src/local/domain/cri.rhizi.net/test_domain__CRI.py
@@ -4,17 +4,20 @@ import tempfile
 import unittest
 
 import db_controller as dbc
-from db_op import DBO_cypher_query
-from domain_cri import DBO_random_data_generation__domain__CRI
-from rz_server import Config
+from db_op import DBO_cypher_query, DBO_rzdoc__create
+from neo4j_test_util import rand_label
+from rz_config import RZ_Config
+from test_util import generate_random_RZDoc
+from test_util__pydev import debug__pydev_pd_arg
+from test_util__rzdoc import DBO_RDG__skill_graph
 
 
 class Test_Domain_CRI(unittest.TestCase):
 
     @classmethod
     def setUpClass(self):
-        cfg = Config.init_from_file('res/etc/rhizi-server.conf')
-        self.db_ctl = dbc.DB_Controller(cfg)
+        cfg = RZ_Config.init_from_file('res/etc/rhizi-server.conf')
+        self.db_ctl = dbc.DB_Controller(cfg.neo4j_url)
         self.log = logging.getLogger('rhizi')
         self.log.addHandler(logging.StreamHandler())
 
@@ -24,7 +27,95 @@ class Test_Domain_CRI(unittest.TestCase):
            - random CRI domain data generation
            - DB dump in cvs format: person x skill x skill-level
         """
-        op = DBO_random_data_generation__domain__CRI()
+
+        skill_set = ['Cryptocurrency',
+                     'Database architecture',
+                     'Web-development',
+                     'Mobile-development',
+                     'Machine learning',
+                     'Guitar Playing',
+                     'Image processing',
+                     'Algebra',
+                     'Calculus',
+                     'Molecular-Biology',
+                     'Graphic design',
+                     'Geo-location services',
+                     'Drone-building',
+                     'Artificial-intelligence',
+                     'Distributed information systems',
+                     'Wordpress',
+                     'Woodworking',
+                     'Quality-control',
+                     'Video-editing',
+                     'Soldering',
+                     'Network engineering',
+                     'GIT',
+                     'Electronic music',
+                     'Network administration']
+
+        name_set = ['John',
+                    'William',
+                    'James',
+                    'Charles',
+                    'George',
+                    'Frank',
+                    'Joseph',
+                    'Thomas',
+                    'Henry',
+                    'Robert',
+                    'Edward',
+                    'Harry',
+                    'Walter',
+                    'Arthur',
+                    'Fred',
+                    'Albert',
+                    'Samuel',
+                    'David',
+                    'Louis',
+                    'Joe',
+                    'Charlie',
+                    'Clarence',
+                    'Richard',
+                    'Andrew',
+                    'Daniel',
+                    'Ernest',
+                    'Mary',
+                    'Anna',
+                    'Emma',
+                    'Elizabeth',
+                    'Minnie',
+                    'Margaret',
+                    'Ida',
+                    'Alice',
+                    'Bertha',
+                    'Sarah',
+                    'Annie',
+                    'Clara',
+                    'Ella',
+                    'Florence',
+                    'Cora',
+                    'Martha',
+                    'Laura',
+                    'Nellie',
+                    'Grace',
+                    'Carrie',
+                    'Maude',
+                    'Mabel',
+                    'Bessie',
+                    'Jennie',
+                    'Gertrude',
+                    'Julia']
+
+        test_label = rand_label()
+        rzdoc = generate_random_RZDoc(test_label)
+
+        op = DBO_rzdoc__create(rzdoc)
+        self.db_ctl.exec_op(op)
+
+        op = DBO_RDG__skill_graph(rzdoc,
+                                  lim_n=10,
+                                  skill_set=skill_set,
+                                  name_set=name_set)
         self.db_ctl.exec_op(op)
 
         if False == export_as_csv:
@@ -37,10 +128,13 @@ class Test_Domain_CRI(unittest.TestCase):
                  ]
         op = DBO_cypher_query(q_arr)
 
+        #
+        # write csv file
+        #
         cur_f_name = inspect.stack()[0][3]
         def q_process_result_set():
             with tempfile.NamedTemporaryFile(prefix='rz_%s' % (cur_f_name), dir='/tmp', suffix='.csv', delete=False) as f_out:
-                for _, _, r_set in op:
+                for _, _, r_set in op.iter__r_set():
                     for row in r_set:
                         person_name, skill_dict_set = row  # person to {sname: skill, s:pro: skill_level{ dict set
                         cvs_line_arr = [person_name]
@@ -54,5 +148,9 @@ class Test_Domain_CRI(unittest.TestCase):
         op.process_result_set = q_process_result_set
         self.db_ctl.exec_op(op)
 
-if __name__ == "__main__":
+@debug__pydev_pd_arg
+def main():
     unittest.main()
+
+if __name__ == "__main__":
+    main()
diff --git a/src/server-tests/test_util__rzdoc.py b/src/server-tests/test_util__rzdoc.py
new file mode 100644
index 00000000..ffcc2423
--- /dev/null
+++ b/src/server-tests/test_util__rzdoc.py
@@ -0,0 +1,65 @@
+from db_op import DBO_cypher_query
+import neo4j_util
+
+
+class DBO_RDG__skill_graph(DBO_cypher_query):
+
+    def __init__(self, rzdoc,
+                       lim_n=50,
+                       lim_r=10000,
+                       prob_link_create=0.03,
+                       name_set=['Bruce Lee', 'Kesuke Miyagi'],
+                       skill_set=['Kung Foo', 'Karate'],
+                       skill_level_set=['Novice', 'Intermediate', 'Expert']
+                      ):
+        """
+        Random data generator:
+           - Name x Skill[proficiency: {Novice|Intermediate|Expert}]
+        """
+        super(DBO_RDG__skill_graph, self).__init__()
+
+        rzdoc_label = neo4j_util.rzdoc__ns_label(rzdoc)
+        min_id = 10000  # separate group ids with numerical perfix
+
+        # create persons
+        q_arr = ['with {name_set} as n_attr_set',  # TODO clean: foreach triggers SyntaxException: otherwise
+                 'foreach (idx in range(0,%d)' % (lim_n - 1),
+                 '|',
+                 'create (n:%s:%s' % (rzdoc_label, 'Person'),
+                 '{id: \'test-id_\' + toString(%d + idx),' % (min_id),
+                  'name: n_attr_set[idx %% %d]}' % (len(name_set)),
+                 '))',
+                ]
+        q_param = {'name_set': name_set}
+        self.add_statement(q_arr, q_param)
+        min_id *= 2
+
+        # create skills
+        q_arr = ['with {skill_set} as skill_set',  # TODO clean: foreach triggers SyntaxException: otherwise
+                 'foreach (idx in range(0,%d)' % (len(skill_set) - 1),
+                 '|',
+                 'create (n:%s:%s' % (rzdoc_label, 'Skill'),
+                 '{id: \'test-id_\' + toString(idx + %d),' % (min_id),
+                  'name: skill_set[idx]}',
+                 '))',
+                ]
+        q_param = {'skill_set': skill_set}
+        self.add_statement(q_arr, q_param)
+        min_id *= 2
+
+        # create links
+        for skill_level in skill_level_set:
+            q_arr = ['match (n:%s),(m:%s)' % ('Person', 'Skill'),
+                     'with n, m, rand() as rand',
+                     'order by rand',
+                     'limit %d' % (lim_r - 1),
+                     'where rand() < %.2f' % (prob_link_create),
+                     'create (n)-[r:%s' % (skill_level),
+                     '{id: \'test-id_\' + toString(%d + toInt(%d * rand())),' % (min_id, lim_r * 100000),  # aim for low id collision probability,
+                     'proficiency: \'%s\'}' % (skill_level),
+                     ']->(m)',
+                     'return collect(r.id)',
+                     ]
+
+            self.add_statement(q_arr, q_param)
+