This closes #85. -- automatically load digipres defaults for csv output.

tballison · Nov 29, 2022 · ee25dff · ee25dff
1 parent 47edbb0
commit ee25dff
Show file tree

Hide file tree

Showing 10 changed files with 76 additions and 57 deletions.
diff --git a/tika-gui-app/assemblies/assembly-linux_x64.xml b/tika-gui-app/assemblies/assembly-linux_x64.xml
@@ -40,6 +40,11 @@
     </fileSet>
   </fileSets>
   <files>
+    <file>
+      <source>${project.basedir}/examples/example-digipres-metadata-mappings-jdbc.csv</source>
+      <destName>default-metadata-mappings.csv</destName>
+      <outputDirectory>config</outputDirectory>
+    </file>
     <file>
       <source>${project.basedir}/target/jres/linux_x64/zulu17.38.21-ca-fx-jre17.0.5-linux_x64.tar.gz</source>
       <outputDirectory>jre</outputDirectory>

diff --git a/tika-gui-app/assemblies/assembly-macosx_aarch64.xml b/tika-gui-app/assemblies/assembly-macosx_aarch64.xml
@@ -40,6 +40,11 @@
     </fileSet>
   </fileSets>
   <files>
+    <file>
+      <source>${project.basedir}/examples/example-digipres-metadata-mappings-jdbc.csv</source>
+      <destName>default-metadata-mappings.csv</destName>
+      <outputDirectory>config</outputDirectory>
+    </file>
     <file>
       <source>${project.basedir}/target/jres/macosx_aarch64/zulu17.38.21-ca-fx-jre17.0.5-macosx_aarch64.zip</source>
       <outputDirectory>jre</outputDirectory>

diff --git a/tika-gui-app/assemblies/assembly-macosx_x64.xml b/tika-gui-app/assemblies/assembly-macosx_x64.xml
@@ -40,6 +40,11 @@
     </fileSet>
   </fileSets>
   <files>
+    <file>
+      <source>${project.basedir}/examples/example-digipres-metadata-mappings-jdbc.csv</source>
+      <destName>default-metadata-mappings.csv</destName>
+      <outputDirectory>config</outputDirectory>
+    </file>
     <file>
       <source>${project.basedir}/target/jres/macosx_x64/zulu17.38.21-ca-fx-jre17.0.5-macosx_x64.zip</source>
       <outputDirectory>jre</outputDirectory>

diff --git a/tika-gui-app/assemblies/assembly-win_x64.xml b/tika-gui-app/assemblies/assembly-win_x64.xml
@@ -44,6 +44,11 @@
     </fileSet>
   </fileSets>
   <files>
+    <file>
+      <source>${project.basedir}/examples/example-digipres-metadata-mappings-jdbc.csv</source>
+      <destName>default-metadata-mappings.csv</destName>
+      <outputDirectory>config</outputDirectory>
+    </file>
     <file>
       <source>${project.basedir}/bin/tika-gui.bat</source>
       <outputDirectory></outputDirectory>

diff --git a/tika-gui-app/examples/example-digipres-metadata-mappings-csv.csv b/tika-gui-app/examples/example-digipres-metadata-mappings-csv.csv
diff --git a/tika-gui-app/examples/example-digipres-metadata-mappings-jdbc.csv b/tika-gui-app/examples/example-digipres-metadata-mappings-jdbc.csv
@@ -37,4 +37,6 @@ pdf:producer,pdf_producer,VARCHAR(512)
 pdfa:PDFVersion,pdfa_version,VARCHAR(32)
 pdfaid:conformance,pdfaid_conformance,VARCHAR(12)
 pdfx:conformance,pdfx_conformance,VARCHAR(12)
-xmp:CreatorTool,xmp_creator_tool,VARCHAR(128)
+xmp:CreatorTool,xmp_creator_tool,VARCHAR(128)
+csv:num_rows,csv_rows,INTEGER
+csv:num_columns,csv_columns,INTEGER
diff --git a/tika-gui-app/src/main/java/org/tallison/tika/app/fx/Constants.java b/tika-gui-app/src/main/java/org/tallison/tika/app/fx/Constants.java
@@ -27,5 +27,7 @@ public class Constants {
     //used as working directory for csv emitter
     public static final String BASE_PATH = "basePath";
 
+    public static final String DEFAULT_METADATA_MAPPINGS = "default-metadata-mappings.csv";
+
 
 }
diff --git a/tika-gui-app/src/main/java/org/tallison/tika/app/fx/emitters/AbstractEmitterController.java b/tika-gui-app/src/main/java/org/tallison/tika/app/fx/emitters/AbstractEmitterController.java
@@ -41,6 +41,7 @@
 import org.apache.commons.csv.CSVRecord;
 import org.apache.logging.log4j.LogManager;
 import org.apache.logging.log4j.Logger;
+import org.tallison.tika.app.fx.Constants;
 import org.tallison.tika.app.fx.ControllerBase;
 import org.tallison.tika.app.fx.batch.BatchProcessConfig;
 import org.tallison.tika.app.fx.ctx.AppContext;
@@ -86,6 +87,24 @@ protected void safelySetCsvMetadataPath(String csvMetadataFilePath) {
         }
     }
 
+    protected void tryToLoadDefaultMetadataMappings() {
+        Path defaultMetadataMappingCSV =
+                AppContext.CONFIG_PATH.resolve(Constants.DEFAULT_METADATA_MAPPINGS);
+        if (! Files.exists(defaultMetadataMappingCSV)) {
+            LOGGER.debug("Can't find default metadata mappings path: {}",
+                    defaultMetadataMappingCSV.toAbsolutePath().toString());
+            return;
+        }
+        try {
+            LOGGER.debug("loading default csv");
+            loadMetadataCSV(defaultMetadataMappingCSV.toFile());
+        } catch (IOException e) {
+            LOGGER.warn("couldn't load default metadata mappings file "
+                    + defaultMetadataMappingCSV.toAbsolutePath(), e);
+        }
+
+    }
+
     protected Optional<Path> getCsvMetadataPath() {
         return csvMetadataPath;
     }

diff --git a/tika-gui-app/src/main/java/org/tallison/tika/app/fx/emitters/CSVEmitterController.java b/tika-gui-app/src/main/java/org/tallison/tika/app/fx/emitters/CSVEmitterController.java
@@ -144,6 +144,7 @@ public void selectCSVOutputDirectory(ActionEvent actionEvent) {
         }
         this.csvWorkingDirectory = Optional.of(directory.toPath());
         this.csvDirectory.setText(directory.getName());
+        saveState(false);
     }
 
 
@@ -204,9 +205,15 @@ public void validateCSV(ActionEvent actionEvent) {
         }
 
         if (getMetadataRows().size() == 0) {
-            alert(ALERT_TITLE, "Metadata Not Configured", "Need to configure metadata");
-            csvAccordion.setExpandedPane(csvAccordion.getPanes().get(1));
-            return;
+            LOGGER.debug("metadata rows size == 0");
+            tryToLoadDefaultMetadataMappings();
+            LOGGER.debug("after trying default metadata mappings: {}", getMetadataRows().size());
+            if (getMetadataRows().size() == 0) {
+                LOGGER.debug("metadata rows size still == 0");
+                alert(ALERT_TITLE, "Metadata Not Configured", "Need to configure metadata");
+                csvAccordion.setExpandedPane(csvAccordion.getPanes().get(1));
+                return;
+            }
         }
         ValidationResult validationResult = validateMetadataRows();
         if (validationResult != ValidationResult.OK) {

diff --git a/tika-gui-app/src/main/java/org/tallison/tika/app/fx/emitters/CSVEmitterSpec.java b/tika-gui-app/src/main/java/org/tallison/tika/app/fx/emitters/CSVEmitterSpec.java
@@ -50,26 +50,18 @@ public class CSVEmitterSpec extends JDBCEmitterSpec {
     private Optional<Path> tmpDbDirectory = Optional.empty();
     private Optional<Path> csvDirectory = Optional.empty();
     private Optional<String> csvFileName = Optional.empty();
+    private volatile boolean closed = false;
 
     public CSVEmitterSpec(@JsonProperty("metadataTuples") List<MetadataTuple> metadataTuples) {
         super(metadataTuples);
         setTableName(CSV_DB_TABLE_NAME);
     }
 
-    private static void writeRow(ResultSet rs, CSVPrinter printer, List<String> cells,
-                                 int columnCount) throws SQLException, IOException {
-        for (int i = 1; i <= columnCount; i++) {
-            String val = rs.getString(i);
-            if (rs.wasNull()) {
-                val = StringUtils.EMPTY;
-            }
-            cells.add(val);
-        }
-        printer.printRecord(cells);
-    }
-
     @Override
     public ValidationResult initialize() throws IOException {
+        if (closed) {
+            throw new IOException("This csv emitter has been closed");
+        }
         tmpDbDirectory = Optional.of(Files.createTempDirectory("tika-app-csv-tmp"));
         LOGGER.debug("tmp db directory: {}", tmpDbDirectory.get().toAbsolutePath());
         setConnectionString("jdbc:sqlite:" + tmpDbDirectory.get().toAbsolutePath() +
@@ -92,7 +84,7 @@ public Optional<Path> getCsvDirectory() {
     }
 
     public void setCsvDirectory(Path csvDirectory) {
-        this.csvDirectory = Optional.of(csvDirectory);
+        this.csvDirectory = Optional.ofNullable(csvDirectory);
     }
 
     private void createTable() throws SQLException {
@@ -122,12 +114,17 @@ private void createTable() throws SQLException {
 
     @Override
     public void close() throws IOException {
+        //avoid double closures -- TODO figure out a more elegant way of handling this
+        if (closed) {
+            return;
+        }
         try {
             writeCSV();
         } catch (IOException e) {
             LOGGER.warn("problem writing csv", e);
         } finally {
             cleanCSVTempResources();
+            closed = true;
         }
     }
 
@@ -235,4 +232,16 @@ private void cleanCSVTempResources() throws IOException {
         }
         FileUtils.deleteDirectory(tmpDbDirectory.get().toFile());
     }
+
+    private static void writeRow(ResultSet rs, CSVPrinter printer, List<String> cells,
+                                 int columnCount) throws SQLException, IOException {
+        for (int i = 1; i <= columnCount; i++) {
+            String val = rs.getString(i);
+            if (rs.wasNull()) {
+                val = StringUtils.EMPTY;
+            }
+            cells.add(val);
+        }
+        printer.printRecord(cells);
+    }
 }