From a5cc2cad133acc30cf5e042ad6848cf20935c78a Mon Sep 17 00:00:00 2001
From: Tom <info@tom-krueger.eu>
Date: Tue, 2 Apr 2019 16:55:34 +0200
Subject: [PATCH] data extraction

---
 test_data_extraction.py      | 60 +++++++++++++++++++++++++++++++++++
 test_sat_to_qubo_workflow.py |  2 +-
 util/queries.py              | 61 ++++++++++++++++++++++++++++++++++++
 util/script.py               | 12 +++++--
 4 files changed, 131 insertions(+), 4 deletions(-)
 create mode 100755 test_data_extraction.py

diff --git a/test_data_extraction.py b/test_data_extraction.py
new file mode 100755
index 0000000..5a9b30a
--- /dev/null
+++ b/test_data_extraction.py
@@ -0,0 +1,60 @@
+#!/usr/bin/env python3
+
+import util.script as script
+import util.queries as queries
+import dimod
+
+def main():
+    wmis_siman_results_alpha_num_of_assignments()
+    
+def wmis_siman_results():
+    db = script.connect_to_instance_pool()
+    
+    q = queries.WMIS_result_scope_query_raw(db)
+    q.query("c50_v[5, 50]_1", "wmis_siman_results")
+    
+    for i in range(30):
+        result = q.__next__()
+        
+        sample_set = queries.read_raw_wmis_sample_set(result["data"])
+        
+        data = script.analyze_wmis_sample(sample_set.first)
+        
+        print(data)
+        
+def wmis_siman_results_alpha_num_of_assignments():
+    edb = script.connect_to_experimetns_db()
+    edb_cursor = edb.cursor()
+    
+    idb = script.connect_to_instance_pool()
+    
+    q = queries.WMIS_result_scope_query_raw(idb)
+    q.query("c50_v[5, 50]_1", "wmis_siman_results")
+    
+    insert_row = ("INSERT INTO c50_v5to50_1_wmis_alpha_number_of_assignments "
+                  "(result_id, "
+                  " number_of_clauses, "
+                  " number_of_variables, "
+                  " number_of_found_assignments) "
+                  "VALUES (%s, %s, %s, %s) ")
+    
+    for result in q:
+        sample_set = queries.read_raw_wmis_sample_set(result["data"])
+        
+        data = script.analyze_wmis_sample(sample_set.first)
+        
+        sat = queries.get_instance_by_id(idb["instances"], result["instance"])
+        
+        edb_cursor.execute(insert_row, (str(result["_id"]),
+                                        int(sat.getNumberOfClauses()),
+                                        int(sat.getNumberOfVariables()),
+                                        int(data["number_of_assignments"])))
+        
+        print(data)
+        
+    edb.commit()
+    edb_cursor.close()
+    edb.close()
+
+if __name__ == "__main__":
+    main()
diff --git a/test_sat_to_qubo_workflow.py b/test_sat_to_qubo_workflow.py
index 972e4b9..d837338 100755
--- a/test_sat_to_qubo_workflow.py
+++ b/test_sat_to_qubo_workflow.py
@@ -78,7 +78,7 @@ def __wmis3():
         script.save_simulated_annealing_result(db["wmis_siman_results"],
                                                res,
                                                solver_input,
-                                               0)
+                                               0) 
     
     
 def __wmis():
diff --git a/util/queries.py b/util/queries.py
index 4e4fa75..d4577c2 100644
--- a/util/queries.py
+++ b/util/queries.py
@@ -1,5 +1,6 @@
 from .kSAT import kSAT
 import bson
+import dimod
 
 class Instance_scope_query:
     
@@ -195,6 +196,45 @@ class WMIS_solver_input_scope_query (WMIS_solver_input_scope_query_raw):
             data["embeddings"].append(read_raw_embedding(raw_emb))
             
         return data
+    
+class WMIS_result_scope_query_raw:
+    def __init__(self, database):
+        self.__database = database
+        self.__query = None
+        
+    def query(self, scope, collection):
+        self.__query = self.__database["experiment_scopes"].aggregate([
+            {
+                "$match": {"_id": scope}
+            },
+            {
+                "$unwind": "$instances"
+            },
+            {
+                "$project": {"instance_id": "$instances"}
+            },
+            {
+                "$lookup":
+                {
+                    "from": collection,
+                    "localField": "instance_id",  
+                    "foreignField": "instance",
+                    "as": "result"
+                }
+            },
+            {
+                "$unwind": "$result"
+            },
+            {
+                "$replaceRoot": {"newRoot": "$result"}
+            }
+        ])
+    
+    def __iter__(self):
+        return self
+    
+    def __next__(self):
+        return self.__query.next()
 
 def load_embedding(collection, qubo_id, solver_graph_id):
     doc = collection.find_one(
@@ -262,3 +302,24 @@ def read_raw_embedding(raw_embedding):
         emb[tuple(entry[0])] = entry[1]
         
     return emb
+
+def read_raw_wmis_sample_set(raw_sample_set):
+    sample_set_data = raw_sample_set.copy()
+    
+    sample_set_data["variable_labels"] = []
+    
+    for label in raw_sample_set["variable_labels"]:
+        sample_set_data["variable_labels"].append(tuple(label))
+        
+    return dimod.SampleSet.from_serializable(sample_set_data)
+
+def get_instance_by_id(collection, id):
+    doc = collection.find_one({"_id": bson.ObjectId(id)})
+    
+    sat = kSAT()
+    
+    for clause in doc["clauses"]:
+            sat.addClause(clause);
+        
+    return sat
+    
diff --git a/util/script.py b/util/script.py
index 78dad0e..796647e 100644
--- a/util/script.py
+++ b/util/script.py
@@ -9,7 +9,7 @@ from . import queries
 from . import graph
 import minorminer
 from tqdm import tqdm
-
+import numpy as np
 
 def readConfig(configFilePath):
     config = configparser.ConfigParser()
@@ -104,7 +104,7 @@ def getDBContext(dbConfigPath):
 
     return dbContext
 
-def connect_to_instance_pool(dbConfigPath):
+def connect_to_instance_pool(dbConfigPath = "database.config"):
     dbConf = readConfig(dbConfigPath)
     
     client = pymongo.MongoClient(
@@ -119,7 +119,7 @@ def connect_to_instance_pool(dbConfigPath):
         
     return client[dbConf["INSTANCE_POOL"]["database"]]
             
-def connect_to_experimetns_db(dbConfigPath):
+def connect_to_experimetns_db(dbConfigPath = "database.config"):
     dbConfig = readConfig(dbConfigPath)
     
     return mysql.connector.connect(
@@ -267,3 +267,9 @@ def save_simulated_annealing_result(collection, result, solver_input, emb_list_i
     
     collection.insert_one(doc)
     
+def analyze_wmis_sample(sample):
+    data = {}
+    
+    data["number_of_assignments"] = np.count_nonzero(list(sample.sample.values()))
+    
+    return data