Rework RDFox simulation to be more modular

This will help with faster testing and debugging.

6 files changed, 93 insertions, 40 deletions

diff --git a/README.md b/README.md
index 87b721f..9cd3029 100644
--- a/README.md
+++ b/README.md
@@ -187,12 +187,11 @@ java -cp lib/JRDFox.jar:target/scala-2.13/RSAComb-assembly-0.2.0.jar uk.ac.ox.cs
 This will answers all the queries in `tests/lubm/queries.sparql` and generate debug information in a new folder in the current working directory (let's say, `rsacomb-20211005120845/`).
 You can run the provided RDFox script as follows
 ```{.sh}
-./lib/RDFox-linux-x86_64-5.2.1/RDFox sandbox . "simulate <debug-folder> <data> <query-id>"
+path/to/RDFox sandbox <debug-folder> "simulate <query-id>"
 ```
 where
 - `debug-folder` is the newly generated folder (`rsacomb-20211005120845` in this example)
-- `<data>` is the path to the data file used with RSAComb (`tests/lubm/data/lubm1.ttl` in this example)
-- `query-id` is the identifier of the query we want to simulate (if we want to simulate query 16 we will pass `16` as an argument)
+- `query-id` is the identifier of the query we want to simulate (if we want to simulate query 16 we will pass `16` as an argument). We can pass `all` to simulate all queries.
 
 This will launch a sandboxed RDFox console, where you will be able to explore a simulation of the datastore used by RSAComb.
 You can also access the same datastore from the web interface at [http://localhost:12110/console/](http://localhost:12110/console/).
diff --git a/simulate.rdfox b/simulate.rdfox
deleted file mode 100644
index 80dbf7c..0000000
--- a/simulate.rdfox
+++ /dev/null
@@ -1,21 +0,0 @@
-
-echo "\n[Start endpoint]"
-endpoint start
-
-echo "\n[Create new datastore]"
-dstore create rsacomb
-active rsacomb
-prefix rsacomb: <http://www.cs.ox.ac.uk/isg/RSAComb#>
-tupletable create rsacomb:CanonicalModel type "named-graph"
-tupletable create rsacomb:Filter$(3) type "named-graph"
-
-echo "\n[Import data]"
-import > rsacomb:CanonicalModel "$(2)"
-import "$(1)/axiomatisation.dlog"
-insert { graph rsacomb:CanonicalModel { ?x a rsacomb:Named } } where { graph rsacomb:CanonicalModel { ?x a owl:Thing } }
-
-echo "\n[Load canonical model program]"
-import "$(1)/canonical_model.dlog"
-
-echo "\n[Load filtering program for query $(3)]"
-import "$(1)/filter_query$(3).dlog"
diff --git a/src/main/scala/uk/ac/ox/cs/rsacomb/Main.scala b/src/main/scala/uk/ac/ox/cs/rsacomb/Main.scala
index 302db75..fe7a6db 100644
--- a/src/main/scala/uk/ac/ox/cs/rsacomb/Main.scala
+++ b/src/main/scala/uk/ac/ox/cs/rsacomb/Main.scala
@@ -42,10 +42,9 @@ object RSAComb extends App {
     Logger.level = config('logger).get[Logger.Level]
 
   /* Load original ontology and normalize it */
-  val ontology = Ontology(
-    config('ontology).get[os.Path],
-    config('data).get[List[os.Path]]
-  ).normalize(new Normalizer)
+  val ontopath = config('ontology).get[os.Path]
+  val data = config('data).get[List[os.Path]]
+  val ontology = Ontology(ontopath, data).normalize(new Normalizer)
 
   //ontology.axioms foreach println
 
@@ -62,12 +61,14 @@ object RSAComb extends App {
     val answers = rsa ask queries
 
     /* Write answers to output file */
-    if (config.contains('answers))
-      os.write(
-        config('answers).get[os.Path],
-        ujson.write(ujson.Arr(answers.map(_.toJSON)), indent = 2),
-        createFolders = true
-      )
+    os.write(
+      config('answers).get[os.Path],
+      ujson.write(ujson.Arr(answers.map(_.toJSON)), indent = 2),
+      createFolders = true
+    )
+
+    /* Generate simulation script */
+    Logger.generateSimulationScripts(data, queries)
 
     // Logger.print(s"$answers", Logger.VERBOSE)
     // Logger print s"Number of answers: ${answers.length} (${answers.lengthWithMultiplicity})"
diff --git a/src/main/scala/uk/ac/ox/cs/rsacomb/RSAConfig.scala b/src/main/scala/uk/ac/ox/cs/rsacomb/RSAConfig.scala
index 4d96850..f3039b3 100644
--- a/src/main/scala/uk/ac/ox/cs/rsacomb/RSAConfig.scala
+++ b/src/main/scala/uk/ac/ox/cs/rsacomb/RSAConfig.scala
@@ -149,6 +149,8 @@ object RSAConfig {
       exit("The following flag is mandatory: '-o' or '--ontology'.")
     if (!config.contains('data))
       config += ('data -> List.empty[os.Path])
+    if (!config.contains('answers))
+      config += ('answers -> Logger.dir / "answers.json")
     config
   }
 }
diff --git a/src/main/scala/uk/ac/ox/cs/rsacomb/RSAOntology.scala b/src/main/scala/uk/ac/ox/cs/rsacomb/RSAOntology.scala
index 4b8e015..5e864bb 100644
--- a/src/main/scala/uk/ac/ox/cs/rsacomb/RSAOntology.scala
+++ b/src/main/scala/uk/ac/ox/cs/rsacomb/RSAOntology.scala
@@ -623,7 +623,10 @@ class RSAOntology(
 
       /* Add filtering program */
       Logger print s"Filtering program rules: ${filter.rules.length}"
-      Logger.write(filter.rules.mkString("\n"), s"filter_query${query.id}.dlog")
+      Logger.write(
+        filter.rules.mkString("\n"),
+        s"filter_query_${query.id}.dlog"
+      )
       RDFoxUtil.addRules(data, filter.rules)
 
       // TODO: We remove the rules, should we drop the tuple table as well?
diff --git a/src/main/scala/uk/ac/ox/cs/rsacomb/util/Logger.scala b/src/main/scala/uk/ac/ox/cs/rsacomb/util/Logger.scala
index b86230e..0fcde53 100644
--- a/src/main/scala/uk/ac/ox/cs/rsacomb/util/Logger.scala
+++ b/src/main/scala/uk/ac/ox/cs/rsacomb/util/Logger.scala
@@ -19,14 +19,12 @@ package uk.ac.ox.cs.rsacomb.util
 import java.util.Calendar
 import java.text.SimpleDateFormat
 import java.io.PrintStream
+import uk.ac.ox.cs.rsacomb.sparql.ConjunctiveQuery
 
-/** Rough implementation of a logger.
-  *
-  * This is a WIP class for debugging and benchmarking.
-  */
+/** Simple logger */
 object Logger {
 
-  private lazy val dir = {
+  lazy val dir = {
     val timestamp = (new SimpleDateFormat("yyyyMMddHHmmss")).format(
       Calendar.getInstance().getTime
     )
@@ -52,10 +50,21 @@ object Logger {
   /** Currend logger level */
   var level: Level = DEBUG
 
+  /** Print a line padded with logger level and timestamp.
+    *
+    * @param str object to be printed.
+    * @param lvl minimum logger level required to print.
+    */
   def print(str: Any, lvl: Level = NORMAL): Unit =
     if (lvl <= level)
       output println s"[$lvl][${Calendar.getInstance().getTime}] $str"
 
+  /** Write provided content to file.
+    *
+    * @param content content to append to the file.
+    * @param file name of the file to append the content to.
+    * @param lvl minimum logger level required to write.
+    */
   def write(content: => os.Source, file: String, lvl: Level = VERBOSE): Unit =
     if (lvl <= level)
       os.write.append(dir / file, content)
@@ -69,4 +78,64 @@ object Logger {
     result
   }
 
+  /** Generate simulation scripts for current run
+    *
+    * @param data data files to be imported.
+    * @param queries collection of executed queries.
+    * @param lvl minimum logger level required.
+    */
+  def generateSimulationScripts(
+      data: Seq[os.Path],
+      queries: Seq[ConjunctiveQuery],
+      lvl: Level = VERBOSE
+  ): Unit =
+    if (lvl <= level) {
+      /* Create script folder */
+      val sim = os.rel / 'sim
+      os.makeDir(dir / sim)
+      /* Generate main script */
+      os.write.append(
+        dir / "simulate.rdfox",
+        """
+echo "\n[Start endpoint]"
+endpoint start
+
+echo "\n[Create new datastore]"
+dstore create rsacomb
+active rsacomb
+prefix rsacomb: <http://www.cs.ox.ac.uk/isg/RSAComb#>
+tupletable create rsacomb:CanonicalModel type "named-graph"
+
+echo "\n[Import data]"
+""" ++
+          data
+            .map(d => s"""import > rsacomb:CanonicalModel \"$d\"""")
+            .mkString("\n")
+          ++ s"""
+import "axiomatisation.dlog"
+insert { graph rsacomb:CanonicalModel { ?x a rsacomb:Named } } where { graph rsacomb:CanonicalModel { ?x a owl:Thing } }
+
+echo "\\n[Load canonical model program]"
+import "canonical_model.dlog"
+
+exec "$sim/filter_query_$$(1).rdfox"
+"""
+      )
+      /* Generate query scripts */
+      queries.map(q => {
+        val id = q.id
+        os.write.append(
+          dir / sim / "filter_query_all.rdfox",
+          s"exec $sim/filter_query_$id.rdfox\n"
+        )
+        os.write.append(
+          dir / sim / s"filter_query_$id.rdfox",
+          s"""
+echo "\\n[Load filtering program for query $id]"
+tupletable create rsacomb:Filter$id type "named-graph"
+import "filter_query_$id.dlog"
+"""
+        )
+      })
+    }
 }