Merged #40 'en40-dryrun'

diversicon-kb · Feb 3, 2017 · 9682890 · 9682890
2 parents 5b54d7f + 6474a3a
commit 9682890
Show file tree

Hide file tree

Showing 4 changed files with 179 additions and 48 deletions.
diff --git a/src/main/java/eu/kidf/diversicon/core/Diversicon.java b/src/main/java/eu/kidf/diversicon/core/Diversicon.java
@@ -415,8 +415,7 @@ public Iterator<Synset> getConnectedSynsets(
 
     /**
      * Validates, normalizes and augments the synsetRelation graph with edges to
-     * speed up
-     * searches.
+     * speed up searches.
      * 
      * @throws DivValidationException
      * 
@@ -1011,11 +1010,12 @@ public ImportJob importXml(String fileUrl) {
      * </p>
      * 
      * @throws DivValidationException
+     * @throws InterruptedImportException
      * 
      * @since 0.1.0
      */
-    public List<ImportJob> importFiles(ImportConfig importConfig) {
-
+    public List<ImportJob> importFiles(ImportConfig importConfig) {        
+        
         Date start = new Date();
 
         List<ImportJob> ret = new ArrayList<>();
@@ -1035,7 +1035,25 @@ public List<ImportJob> importFiles(ImportConfig importConfig) {
                                                   .size()
                 + " files by import author " + importConfig.getAuthor() + "...");
 
-        DbInfo oldDbInfo = prepareDbForImport();
+        if (importConfig.isDryRun() && importConfig.getFileUrls().size() > 1){
+            LOG.info("");
+            LOG.info("**********   DRY RUN: found many files to import!   ");
+            LOG.info("");
+            LOG.info("**********            If second file depends on first one there might be validation errors due to  ");
+            LOG.info("**********            data not actually being written to disk!");             
+            LOG.info("");
+        }
+
+        DbInfo oldDbInfo = null;
+
+        if (importConfig.isDryRun()){
+            LOG.info("");
+            LOG.info("**********   DRY RUN:  skipping DbInfo table update   **********");
+            LOG.info("");
+        } else {
+            oldDbInfo = prepareDbForImport();    
+        }
+
 
         try {
             for (String fileUrl : importConfig.getFileUrls()) {
@@ -1046,13 +1064,22 @@ public List<ImportJob> importFiles(ImportConfig importConfig) {
             if (ret.isEmpty()) {
                 LOG.error("Import failed, no LexicalResource data was written to disk. "
                         + "Aborting all imports.", ex);
-                setDbInfo(oldDbInfo);
+                if (importConfig.isDryRun()){
+                    LOG.info("********   DRY RUN: skipping restoring DbInfo table  (no change was done to it) *************");
+                } else {
+                    setDbInfo(oldDbInfo);    
+                }
+
             }
             throw ex;
         }
 
         if (importConfig.isSkipAugment()) {
             LOG.info("Skipping graph augmentation as requested by user.");
+        } else if (importConfig.isDryRun()) {
+            LOG.info("");
+            LOG.info("*************    DRY RUN: skipping graph augmentation   ************");
+            LOG.info("");            
         } else {
             try {
                 processGraph();
@@ -1062,6 +1089,13 @@ public List<ImportJob> importFiles(ImportConfig importConfig) {
         }
 
         logImportResult(importConfig, start, ret);
+
+        if (importConfig.isDryRun()){
+            LOG.info("");
+            LOG.info("*************    DRY RUN:  end of import,  no data was written to disk.   ************");
+            LOG.info("");                        
+        }
+
         return ret;
 
     }
@@ -1104,7 +1138,7 @@ private void logImportResult(ImportConfig importConfig, Date start, List<ImportJ
      * Imports a single LMF XML
      * 
      * @throws InvalidImportException
-     * @throws InterruptedImportException
+     * @throws InterruptedImportException If something goes wrong 
      * 
      * @since 0.1.0
      */
@@ -1134,20 +1168,31 @@ private ImportJob importFile(ImportConfig importConfig, String url) {
                 file,
                 pack);
 
-        LOG.info("");
-        LOG.info("Starting import...");
-        LOG.info("");
+        if (importConfig.isDryRun()){
+
+            LOG.info("");
+            LOG.info("*************    DRY RUN: skipping actual import!    *********");
+            LOG.info("");
+
+        } else {
+
+            LOG.info("");
+            LOG.info("Starting import...");
+            LOG.info("");
 
-        try {
-            DivXmlToDbTransformer trans = new DivXmlToDbTransformer(this);
-            trans.transform(file, null);
-            endImportJob(job);
+            try {
+                DivXmlToDbTransformer trans = new DivXmlToDbTransformer(this);
+                trans.transform(file, null);
+                endImportJob(job);
 
-        } catch (Exception ex) {
-            throw new InterruptedImportException("Error while loading lmf xml " + url, ex);
-        }
+            } catch (Exception ex) {
+                throw new InterruptedImportException("Error while loading lmf xml " + url, ex);
+            }
 
-        LOG.info("Done loading LMF : " + url + " .");
+            LOG.info("Done loading LMF : " + url + " .");
+
+        }
+
 
         return job;
     }
@@ -1475,14 +1520,7 @@ public ImportJob importResource(
      * on it instead.
      * </p>
      * 
-     * @param prefix
-     *            The prefix to use within the database
-     * @param namespaces
-     *            The namespaces used by the resource
-     * @param skipAugment
-     *            if true, after the import prevents the graph froom being
-     *            normalized
-     *            and augmented with transitive closure.
+     * @param importConfig a configuration for the import. Must have only one file url.
      * @param pack
      *            A LexicalResource doesn't contain stuff like namespaces, so we
      *            have to provide them by ourselves in pack.
@@ -1512,18 +1550,26 @@ public ImportJob importResource(
                     null,
                     pack);
 
-            prepareDbForImport();
+            if (importConfig.isDryRun()){
+                LOG.info("");
+                LOG.info("**********   DRY RUN:  skipping actual import     ***********");
+                LOG.info("");                                
+            } else {
+                prepareDbForImport();
 
-            setCurrentImportJob(job);
+                setCurrentImportJob(job);
 
-            new JavaToDbTransformer(this, lexRes).transform();
+                new JavaToDbTransformer(this, lexRes).transform();
 
-            if (!importConfig.isSkipAugment()) {
-                processGraph();
-            }
+                if (!importConfig.isSkipAugment()) {
+                    processGraph();
+                }
 
-            endImportJob(job);
+                endImportJob(job);
 
+            }
+
+
         } catch (InvalidImportException ex) {
 
             LOG.error("Import failed! Aborting all imports! (no LexicalResource data was written to disk) \n", ex);
@@ -1534,7 +1580,15 @@ public ImportJob importResource(
                     + lexRes.getName() + " !", ex);
         }
 
-        LOG.info("Done saving.");
+
+
+        if (importConfig.isDryRun()){
+            LOG.info("");
+            LOG.info("**********   DRY RUN:  end of import, no data was written to disk !     ***********");
+            LOG.info("");                                
+        } else {
+            LOG.info("Done saving.");
+        }                
 
         return job;
     }

diff --git a/src/main/java/eu/kidf/diversicon/core/ImportConfig.java b/src/main/java/eu/kidf/diversicon/core/ImportConfig.java
@@ -7,15 +7,14 @@
 import java.util.ArrayList;
 import java.util.List;
 
-import eu.kidf.diversicon.core.XmlValidationConfig.Builder;
 
 /**
  * Configuration for a whole {@link ImportJob} process.
  * 
  * @since 0.1.0
  * @see ImportJob
  */
-// TODO Why is it not immutable ??
+// TODO: make it immutable https://github.com/diversicon-kb/diversicon-core/issues/35
 public class ImportConfig {
 
     private List<String> fileUrls;
@@ -24,6 +23,7 @@ public class ImportConfig {
     private boolean skipAugment;
     private int logLimit;   
     private boolean force;
+    private boolean dryRun;
 
     /**
      * Default constructor.
@@ -39,6 +39,7 @@ public ImportConfig() {
         this.skipAugment = false;
         this.logLimit = Diversicons.DEFAULT_LOG_LIMIT;
         this.force = false;
+        this.dryRun = false;
     }
 
     /**
@@ -57,9 +58,10 @@ public List<String> getFileUrls() {
      * 
      * @since 0.1.0
      */
-    public void setFileUrls(List<String> fileUrls) {
+    public ImportConfig setFileUrls(List<String> fileUrls) {
         checkNotNull(fileUrls);
         this.fileUrls = fileUrls;
+        return this;
     }
 
     /**
@@ -76,9 +78,10 @@ public String getAuthor() {
      * 
      * @since 0.1.0
      */
-    public void setAuthor(String author) {
+    public ImportConfig setAuthor(String author) {
         checkNotNull(author);
         this.author = author;
+        return this;
     }
 
     /**
@@ -98,9 +101,10 @@ public String getDescription() {
      * 
      * @since 0.1.0
      */
-    public void setDescription(String description) {
+    public ImportConfig setDescription(String description) {
         checkNotNull(description);
         this.description = description;
+        return this;
     }
 
     /**
@@ -119,17 +123,19 @@ public boolean isSkipAugment() {
      * 
      * @since 0.1.0
      */
-    public void setSkipAugment(boolean skipAugment) {
+    public ImportConfig setSkipAugment(boolean skipAugment) {
         this.skipAugment = skipAugment;
+        return this;
     }
 
     /**
      * 
      * @since 0.1.0
      */
-    public void addLexResFileUrl(String fileUrl) {
+    public ImportConfig addLexResFileUrl(String fileUrl) {
         checkNotEmpty(fileUrl, "Invalid lexical resource file URL!");
         this.fileUrls.add(fileUrl);
+        return this;
     }
 
     /**
@@ -173,9 +179,10 @@ public int getLogLimit() {
      * 
      * @since 0.1.0
      */
-    public void setLogLimit(int logLimit) {
+    public ImportConfig setLogLimit(int logLimit) {
         checkArgument(logLimit >= -1, "Log limit must be >= -1, found instead %s", logLimit);
         this.logLimit = logLimit;
+        return this;
     }
 
     /**
@@ -192,7 +199,27 @@ public boolean isForce() {
      * 
      * @since 0.1.0
      */
-    public void setForce(boolean force) {        
+    public ImportConfig setForce(boolean force) {        
         this.force = force;
-    }        
+        return this;
+    }
+
+    /**
+     * A dry run simply simulates the import without writing anything into the database.
+     * 
+     * @since 0.1.0
+     */
+    public boolean isDryRun() {
+        return dryRun;
+    }
+
+    /**
+     * See {@link #isDryRun()}
+     * 
+     * @since 0.1.0
+     */
+    public ImportConfig setDryRun(boolean dryRun) {
+        this.dryRun = dryRun;
+        return this;
+    }
 }
diff --git a/src/main/java/eu/kidf/diversicon/core/internal/Internals.java b/src/main/java/eu/kidf/diversicon/core/internal/Internals.java
@@ -2041,9 +2041,9 @@ public static String longestCommonPrefix(String a, String b) {
     public static ImportConfig createImportConfig(LexicalResource lexRes){
         checkNotNull(lexRes);
 
-        ImportConfig importConfig = new ImportConfig();
-        importConfig.setAuthor(Diversicons.DEFAULT_AUTHOR);        
-        importConfig.addLexResFileUrl(Diversicons.MEMORY_PROTOCOL + ":" + lexRes.hashCode());
+        ImportConfig importConfig = new ImportConfig()
+                .setAuthor(Diversicons.DEFAULT_AUTHOR)        
+                .addLexResFileUrl(Diversicons.MEMORY_PROTOCOL + ":" + lexRes.hashCode());
         return importConfig;
 
     }