Importer für RIS Format und Converter für für die RIS-Typen INPR, GEN, ABST, BOOK, JFULL, EBOOK, JOUR und EJOUR

git-svn-id: https://svn.libreccm.org/ccm/trunk@2029 8810af33-2d31-482b-a856-94f89814c4df

ccm-sci-publications/src/com/arsdigita/cms/scipublications/imexporter/ris/RisField.java

@@ -347,6 +347,8 @@ public enum RisField {
     C1,
     C2,
     C3,
+    C4,
+    C5,
     /**
      * Number of volumes
      */

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/RisDataset.java

@@ -16,15 +16,21 @@ public class RisDataset {
     
     private final RisType type;
     private final Map<RisField, List<String>> values = new EnumMap<RisField, List<String>>(RisField.class);
+    private final int firstLine;
     
-    public RisDataset(final RisType type) {
+    public RisDataset(final RisType type, final int firstLine) {
         this.type = type;
+        this.firstLine = firstLine;
     }
     
     public RisType getType() {
         return type;
     }
     
+    public int getFirstLine() {
+        return firstLine;
+    }
+    
     public Map<RisField, List<String>> getValues() {
         return Collections.unmodifiableMap(values);
     }

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/RisParser.java

@@ -34,7 +34,7 @@ public class RisParser {
                                           + "Aborting import.", i + 1));
                 } else {
                     try {
-                        entries.add(startDataset(field.getValue()));
+                        entries.add(startDataset(field.getValue(), i + 1));
                         openDataset = true;
                     } catch (IllegalArgumentException ex) {
                         throw new SciPublicationsImportException(String.format("Invalid type at line %d.", i + 1), ex);
@@ -58,9 +58,9 @@ public class RisParser {
         return Collections.unmodifiableList(entries);
     }
 
-    private RisDataset startDataset(final String type) {
+    private RisDataset startDataset(final String type, final int firstLine) {
         final RisType risType = RisType.valueOf(type);
-        return new RisDataset(risType);
+        return new RisDataset(risType, firstLine);
     }
 
     private RisFieldValue parseRisLine(final String line, final int index) throws SciPublicationsImportException {

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/converters/AbstConverter.java

@@ -0,0 +1,17 @@
+package com.arsdigita.cms.scipublications.importer.ris.converters;
+
+import com.arsdigita.cms.scipublications.imexporter.ris.RisType;
+
+/**
+ *
+ * @author Jens Pelzetter <jens@jp-digital.de>
+ * @version $Id$
+ */
+public class AbstConverter extends JourConverter {
+    
+    @Override
+    public RisType getRisType() {
+        return RisType.ABST;
+    }
+    
+}

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/converters/AbstractRisConverter.java

@@ -1,5 +1,6 @@
 package com.arsdigita.cms.scipublications.importer.ris.converters;
 
+import com.arsdigita.cms.contenttypes.ArticleInJournal;
 import com.arsdigita.cms.contenttypes.Publication;
 import com.arsdigita.cms.contenttypes.PublicationWithPublisher;
 import com.arsdigita.cms.scipublications.imexporter.ris.RisField;
@@ -19,6 +20,49 @@ import java.util.List;
  */
 public abstract class AbstractRisConverter implements RisConverter {
 
+    protected void processField(final RisDataset dataset,
+                                final RisField field,
+                                final Publication publication,
+                                final String targetField,
+                                final PublicationImportReport report,
+                                final boolean pretend) {
+        final List<String> values = dataset.getValues().get(field);
+        if ((values != null) && !values.isEmpty()) {
+            publication.set(targetField, values.get(0));
+            report.addField(new FieldImportReport(targetField, values.get(0)));
+        }
+    }
+
+    protected void processIntField(final RisDataset dataset,
+                                   final RisField field,
+                                   final Publication publication,
+                                   final String targetField,
+                                   final PublicationImportReport report,
+                                   final boolean pretend) {
+        final List<String> values = dataset.getValues().get(field);
+        if ((values != null) && !values.isEmpty()) {
+            final String valueStr = values.get(0);
+            try {
+                final int value = Integer.parseInt(valueStr);
+                publication.set(targetField, value);
+                publication.set(targetField, valueStr);
+            } catch (NumberFormatException ex) {
+                report.addMessage(String.format("Failed to parse value of field '%s' into an integer for dataset "
+                                                + "starting on line %d.",
+                                                field,
+                                                dataset.getFirstLine()));
+            }
+        }
+    }
+
+    protected void processTitle(final RisDataset dataset,
+                                final Publication publication,
+                                final PublicationImportReport report,
+                                final boolean pretend) {
+        publication.setTitle(dataset.getValues().get(RisField.TI).get(0));
+        publication.setTitle(dataset.getValues().get(RisField.TI).get(0));
+    }
+
     protected void processAuthors(final RisDataset dataset,
                                   final RisField risField,
                                   final ImporterUtil importerUtil,
@@ -47,7 +91,13 @@ public abstract class AbstractRisConverter implements RisConverter {
         final List<String> authors = dataset.getValues().get(risField);
         if ((authors != null) && !authors.isEmpty()) {
             for (String authorStr : authors) {
-                processAuthorStr(authorStr, isEditors, importerUtil, publication, report, pretend);
+                processAuthorStr(authorStr,
+                                 isEditors,
+                                 importerUtil,
+                                 publication,
+                                 report,
+                                 dataset.getFirstLine(),
+                                 pretend);
             }
         }
     }
@@ -57,12 +107,14 @@ public abstract class AbstractRisConverter implements RisConverter {
                                   final ImporterUtil importerUtil,
                                   final Publication publication,
                                   final PublicationImportReport importReport,
+                                  final int firstLine,
                                   final boolean pretend) {
         final AuthorData authorData = new AuthorData();
 
         final String[] tokens = authorStr.split(",");
         if (tokens.length == 0) {
-            importReport.addMessage(String.format("Failed to parse author string '%s'.", authorStr));
+            importReport.addMessage(String.format("Failed to parse author string '%s' at dataset starting at line %d.",
+                                                  authorStr, firstLine));
             return;
         }
 
@@ -117,7 +169,8 @@ public abstract class AbstractRisConverter implements RisConverter {
                 publication.setNumberOfPages(value);
                 report.addField(new FieldImportReport("number of pages", numberOfPages.get(0)));
             } catch (NumberFormatException ex) {
-                report.addMessage("Failed to parse number of pages");
+                report.addMessage(String.format("Failed to parse number of pages at dataset starting at line %d",
+                                                dataset.getFirstLine()));
             }
         }
 
@@ -134,7 +187,8 @@ public abstract class AbstractRisConverter implements RisConverter {
                 publication.setNumberOfVolumes(value);
                 report.addField(new FieldImportReport("number of volumes", numberOfVols.get(0)));
             } catch (NumberFormatException ex) {
-                report.addMessage("Failed to parse number of volumes.");
+                report.addMessage(String.format("Failed to parse number of volumes at dataset starting at line %d.",
+                                                dataset.getFirstLine()));
             }
         }
     }
@@ -150,7 +204,8 @@ public abstract class AbstractRisConverter implements RisConverter {
                 publication.setVolume(value);
                 report.addField(new FieldImportReport("volume", volume.get(0)));
             } catch (NumberFormatException ex) {
-                report.addMessage("Failed to parse value of field 'volume'.");
+                report.addMessage(String.format("Failed to parse value of field 'volume' on dataset starting "
+                                                + "at line %d.", dataset.getFirstLine()));
             }
         }
     }
@@ -166,9 +221,77 @@ public abstract class AbstractRisConverter implements RisConverter {
             report.addField(new FieldImportReport("year", yearStr));
         } catch (NumberFormatException ex) {
             report.addMessage(String.format("Failed to convert year of publication value '%s' from RIS to"
-                                            + "integer value. Setting year of publication to 0"));
+                                            + "integer value on dataset starting at line %d. Setting year of "
+                                            + "publication to 0", dataset.getFirstLine()));
             publication.setYearOfPublication(0);
         }
     }
 
+    protected void processSeries(final RisDataset dataset,
+                                 final RisField field,
+                                 final Publication publication,
+                                 final ImporterUtil importerUtil,
+                                 final boolean pretend,
+                                 final PublicationImportReport report) {
+        final List<String> series = dataset.getValues().get(field);
+        if ((series != null) && !series.isEmpty()) {
+            report.setSeries(importerUtil.processSeries(publication, series.get(0), pretend));
+        }
+    }
+
+    protected void processJournal(final RisDataset dataset,
+                                  final RisField field,
+                                  final ArticleInJournal article,
+                                  final ImporterUtil importerUtil,
+                                  final boolean pretend,
+                                  final PublicationImportReport report) {
+        final List<String> journal = dataset.getValues().get(field);
+        if ((journal != null) && !journal.isEmpty()) {
+            report.setJournal(importerUtil.processJournal(article, journal.get(0), pretend));
+        }
+    }
+
+    protected void processPages(final RisDataset dataset,
+                                final RisField field,
+                                final Publication publication,
+                                final boolean pretend,
+                                final PublicationImportReport report) {
+        final List<String> values = dataset.getValues().get(field);
+        final String pages = values.get(0);
+
+        final String[] tokens = pages.split("-");
+        if (tokens.length == 2) {
+            try {
+                final int pagesFrom = Integer.parseInt(tokens[0]);
+                final int pagesTo = Integer.parseInt(tokens[1]);
+
+                publication.set("pagesFrom", pagesFrom);
+                publication.set("pagesTo", pagesTo);
+
+                report.addField(new FieldImportReport("pagesFrom", Integer.toString(pagesFrom)));
+                report.addField(new FieldImportReport("pagesTo", Integer.toString(pagesTo)));
+            } catch (NumberFormatException ex) {
+                report.addMessage(String.format("Failed to parse pages value in dataset starting at line %d. "
+                                                + "On of the values given is not an integer.",
+                                                dataset.getFirstLine()));
+            }
+        } else if (tokens.length == 1) {
+            try {
+                final int pagesFrom = Integer.parseInt(tokens[0]);
+
+                publication.set("pagesFrom", pagesFrom);
+
+                report.addField(new FieldImportReport("pagesFrom", Integer.toString(pagesFrom)));
+            } catch (NumberFormatException ex) {
+                report.addMessage(String.format("Failed to parse pages value in dataset starting at line %d. "
+                                                + "Value is not an integer.",
+                                                dataset.getFirstLine()));
+            }
+        } else if (tokens.length > 2) {
+            report.addMessage(String.format("Failed to parse pages value in dataset starting at line %d. "
+                                            + "Invalid format",
+                                            dataset.getFirstLine()));
+        }
+    }
+
 }

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/converters/BookConverter.java

@@ -26,8 +26,7 @@ public class BookConverter extends AbstractRisConverter {
 
         final Monograph monograph = new Monograph();
 
-        monograph.setTitle(dataset.getValues().get(RisField.TI).get(0));
+        processTitle(dataset, monograph, report, pretend);
-        report.setTitle(dataset.getValues().get(RisField.TI).get(0));
 
         processYear(dataset, pretend, monograph, report);
 
@@ -37,34 +36,19 @@ public class BookConverter extends AbstractRisConverter {
 
         processPublisher(dataset, pretend, monograph, importerUtil, report);
 
-        final List<String> abstractList = dataset.getValues().get(RisField.AB);
+        processField(dataset, RisField.AB, monograph, "abstract", report, pretend);
-        if ((abstractList != null) && (!abstractList.isEmpty())) {
+
-            monograph.setAbstract(abstractList.get(0));
+        processField(dataset, RisField.ET, monograph, "edition", report, pretend);
-            report.addField(new FieldImportReport("abstract", abstractList.get(0)));
-        }
 
-        final List<String> edition = dataset.getValues().get(RisField.ET);
-        if ((edition != null) && !edition.isEmpty()) {
-            monograph.setEdition(edition.get(0));
-            report.addField(new FieldImportReport("edition", edition.get(0)));
-        }
-        
         processNumberOfVolumes(dataset, pretend, monograph, report);
 
-        final List<String> isbn = dataset.getValues().get(RisField.SN);
+        processField(dataset, RisField.SN, monograph, "isbn", report, pretend);
-        if ((isbn != null) && !isbn.isEmpty()) {
-            monograph.setISBN(isbn.get(0));
-            report.addField(new FieldImportReport("isbn", isbn.get(0)));
-        }
 
         processNumberOfPages(dataset, pretend, monograph, report);
 
         processVolume(dataset, pretend, monograph, report);
 
-        final List<String> series = dataset.getValues().get(RisField.T2);
+        processSeries(dataset, RisField.T2, monograph, importerUtil, pretend, report);
-        if ((series != null) && !series.isEmpty()) {
-            report.setSeries(importerUtil.processSeries(monograph, series.get(0), pretend));
-        }
 
         return report;
     }

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/converters/EbookConverter.java

@@ -3,11 +3,9 @@ package com.arsdigita.cms.scipublications.importer.ris.converters;
 import com.arsdigita.cms.contenttypes.Monograph;
 import com.arsdigita.cms.scipublications.imexporter.ris.RisField;
 import com.arsdigita.cms.scipublications.imexporter.ris.RisType;
-import com.arsdigita.cms.scipublications.importer.report.FieldImportReport;
 import com.arsdigita.cms.scipublications.importer.report.PublicationImportReport;
 import com.arsdigita.cms.scipublications.importer.ris.RisDataset;
 import com.arsdigita.cms.scipublications.importer.util.ImporterUtil;
-import java.util.List;
 
 /**
  * Converter for the RIS type {@code EBOOK} to the SciPublications type {@link Monograph}.
@@ -26,9 +24,8 @@ public class EbookConverter extends AbstractRisConverter {
 
         final Monograph monograph = new Monograph();
 
-        monograph.setTitle(dataset.getValues().get(RisField.TI).get(0));
+        processTitle(dataset, monograph, report, pretend);
-        report.setTitle(dataset.getValues().get(RisField.TI).get(0));
+        
-
         processYear(dataset, pretend, monograph, report);
 
         processAuthors(dataset, RisField.AU, importerUtil, monograph, report, pretend);
@@ -37,32 +34,17 @@ public class EbookConverter extends AbstractRisConverter {
 
         processPublisher(dataset, pretend, monograph, importerUtil, report);
 
-        final List<String> abstractList = dataset.getValues().get(RisField.AB);
+        processField(dataset, RisField.AB, monograph, "abstract", report, pretend);
-        if ((abstractList != null) && (!abstractList.isEmpty())) {
+        
-            monograph.setAbstract(abstractList.get(0));
+        processField(dataset, RisField.ET, monograph, "edition", report, pretend);
-            report.addField(new FieldImportReport("abstract", abstractList.get(0)));
-        }
 
-        final List<String> edition = dataset.getValues().get(RisField.ET);
+        processField(dataset, RisField.SN, monograph, "isbn", report, pretend);
-        if ((edition != null) && !edition.isEmpty()) {
-            monograph.setEdition(edition.get(0));
-            report.addField(new FieldImportReport("edition", edition.get(0)));
-        }
-
-        final List<String> isbn = dataset.getValues().get(RisField.SN);
-        if ((isbn != null) && !isbn.isEmpty()) {
-            monograph.setISBN(isbn.get(0));
-            report.addField(new FieldImportReport("isbn", isbn.get(0)));
-        }
 
         processNumberOfPages(dataset, pretend, monograph, report);
 
         processVolume(dataset, pretend, monograph, report);
 
-        final List<String> series = dataset.getValues().get(RisField.T2);
+        processSeries(dataset, RisField.T2, monograph, importerUtil, pretend, report);
-        if ((series != null) && !series.isEmpty()) {
-            report.setSeries(importerUtil.processSeries(monograph, series.get(0), pretend));
-        }
 
         return report;
     }

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/converters/EjourConverter.java

@@ -1,6 +1,7 @@
 package com.arsdigita.cms.scipublications.importer.ris.converters;
 
 import com.arsdigita.cms.contenttypes.ArticleInJournal;
+import com.arsdigita.cms.scipublications.imexporter.ris.RisField;
 import com.arsdigita.cms.scipublications.imexporter.ris.RisType;
 import com.arsdigita.cms.scipublications.importer.report.PublicationImportReport;
 import com.arsdigita.cms.scipublications.importer.ris.RisConverter;
@@ -13,7 +14,7 @@ import com.arsdigita.cms.scipublications.importer.util.ImporterUtil;
  * @author Jens Pelzetter <jens@jp-digital.de>
  * @version $Id$
  */
-public class EjourConverter implements RisConverter {
+public class EjourConverter extends AbstractRisConverter implements RisConverter {
 
     public PublicationImportReport convert(final RisDataset dataset, 
                                            final ImporterUtil importerUtil, 
@@ -24,7 +25,21 @@ public class EjourConverter implements RisConverter {
         
         final ArticleInJournal article = new ArticleInJournal();
         
+        processTitle(dataset, article, report, pretend);
         
+        processYear(dataset, pretend, article, report);
+        
+        processAuthors(dataset, RisField.AU, importerUtil, article, report, pretend);
+        
+        processField(dataset, RisField.AB, article, "abstract", report, pretend);
+        
+        processJournal(dataset, RisField.T2, article, importerUtil, pretend, report);               
+        
+        processField(dataset, RisField.M1, article, "issue", report, pretend);
+        
+        processPages(dataset, RisField.SP, article, pretend, report);
+        
+        processField(dataset, RisField.VL, article, "volume", report, pretend);
         
         return report;        
     }

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/converters/GenConverter.java

@@ -4,12 +4,9 @@ import com.arsdigita.cms.contenttypes.GreyLiterature;
 import com.arsdigita.cms.contenttypes.Monograph;
 import com.arsdigita.cms.scipublications.imexporter.ris.RisField;
 import com.arsdigita.cms.scipublications.imexporter.ris.RisType;
-import com.arsdigita.cms.scipublications.importer.report.FieldImportReport;
 import com.arsdigita.cms.scipublications.importer.report.PublicationImportReport;
-import com.arsdigita.cms.scipublications.importer.ris.RisConverter;
 import com.arsdigita.cms.scipublications.importer.ris.RisDataset;
 import com.arsdigita.cms.scipublications.importer.util.ImporterUtil;
-import java.util.List;
 
 /**
  * Converter for the RIS type {@code GEN} to the SciPublications {@link GreyLiterature}
@@ -28,19 +25,9 @@ public class GenConverter extends AbstractRisConverter {
 
         final Monograph publication = new Monograph();
 
-        publication.setTitle(dataset.getValues().get(RisField.TI).get(0));
+        processTitle(dataset, publication, report, pretend);
-        report.setTitle(dataset.getValues().get(RisField.TI).get(0));
 
-        final String yearStr = dataset.getValues().get(RisField.PY).get(0);
+        processYear(dataset, pretend, publication, report);
-        try {
-            final int year = Integer.parseInt(yearStr);
-            publication.setYearOfPublication(year);
-            report.addField(new FieldImportReport("year", yearStr));
-        } catch (NumberFormatException ex) {
-            report.addMessage(String.format("Failed to convert year of publication value '%s' from RIS to"
-                                            + "integer value. Setting year of publication to 0"));
-            publication.setYearOfPublication(0);
-        }
 
         processAuthors(dataset, RisField.AU, importerUtil, publication, report, pretend);
         processAuthors(dataset, RisField.A2, importerUtil, publication, report, pretend);
@@ -49,26 +36,16 @@ public class GenConverter extends AbstractRisConverter {
 
         processPublisher(dataset, pretend, publication, importerUtil, report);
 
-        final List<String> abstractList = dataset.getValues().get(RisField.AB);
+        processField(dataset, RisField.AB, publication, "abstract", report, pretend);
-        if ((abstractList != null) && (!abstractList.isEmpty())) {
-            publication.setAbstract(abstractList.get(0));
-            report.addField(new FieldImportReport("abstract", abstractList.get(0)));
-        }
 
-        final List<String> edition = dataset.getValues().get(RisField.ET);
+        processField(dataset, RisField.ET, publication, "edition", report, pretend);
-        if ((edition != null) && !edition.isEmpty()) {
-            publication.setEdition(edition.get(0));
-            report.addField(new FieldImportReport("edition", edition.get(0)));
-        }
 
-        final List<String> isbn = dataset.getValues().get(RisField.SN);
+        processField(dataset, RisField.SN, publication, "isbn", report, pretend);
-        if ((isbn != null) && !isbn.isEmpty()) {
-            publication.setISBN(isbn.get(0));
-            report.addField(new FieldImportReport("isbn", isbn.get(0)));
-        }
 
         processNumberOfPages(dataset, pretend, publication, report);
 
+        processNumberOfVolumes(dataset, pretend, publication, report);
+        
         processVolume(dataset, pretend, publication, report);
 
         return report;

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/converters/InprConverter.java

@@ -0,0 +1,17 @@
+package com.arsdigita.cms.scipublications.importer.ris.converters;
+
+import com.arsdigita.cms.scipublications.imexporter.ris.RisType;
+
+/**
+ *
+ * @author Jens Pelzetter <jens@jp-digital.de>
+ * @version $Id$
+ */
+public class InprConverter extends JourConverter {
+    
+    @Override
+    public RisType getRisType() {
+        return RisType.INPR;
+    }
+    
+}

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/converters/JfullConverter.java

@@ -0,0 +1,17 @@
+package com.arsdigita.cms.scipublications.importer.ris.converters;
+
+import com.arsdigita.cms.scipublications.imexporter.ris.RisType;
+
+/**
+ *
+ * @author Jens Pelzetter <jens@jp-digital.de>
+ * @version $Id$
+ */
+public class JfullConverter extends JourConverter {
+
+    @Override
+    public RisType getRisType() {
+        return RisType.JFULL;
+    }
+
+}

ccm-sci-publications/src/com/arsdigita/cms/scipublications/importer/ris/converters/JourConverter.java

@@ -0,0 +1,53 @@
+package com.arsdigita.cms.scipublications.importer.ris.converters;
+
+import com.arsdigita.cms.contenttypes.ArticleInJournal;
+import com.arsdigita.cms.scipublications.imexporter.ris.RisField;
+import com.arsdigita.cms.scipublications.imexporter.ris.RisType;
+import com.arsdigita.cms.scipublications.importer.report.PublicationImportReport;
+import com.arsdigita.cms.scipublications.importer.ris.RisDataset;
+import com.arsdigita.cms.scipublications.importer.util.ImporterUtil;
+
+/**
+ *
+ * @author Jens Pelzetter <jens@jp-digital.de>
+ * @version $Id$
+ */
+public class JourConverter extends AbstractRisConverter {
+
+    @Override
+    public PublicationImportReport convert(final RisDataset dataset,
+                                           final ImporterUtil importerUtil,
+                                           final boolean pretend,
+                                           final boolean publishNewItems) {
+        final PublicationImportReport report = new PublicationImportReport();
+        report.setType(ArticleInJournal.BASE_DATA_OBJECT_TYPE);
+
+        final ArticleInJournal article = new ArticleInJournal();
+
+        processTitle(dataset, article, report, pretend);
+
+        processYear(dataset, pretend, article, report);
+
+        processAuthors(dataset, RisField.AU, importerUtil, article, report, pretend);
+
+        processField(dataset, RisField.AB, article, "abstract", report, pretend);
+
+        processJournal(dataset, RisField.T2, article, importerUtil, pretend, report);
+
+        processField(dataset, RisField.M1, article, "issue", report, pretend);
+
+        processPages(dataset, RisField.SP, article, pretend, report);
+
+        processField(dataset, RisField.VL, article, "volume", report, pretend);
+
+        processIntField(dataset, RisField.M2, article, "pagesFrom", report, pretend);
+                
+        return report;
+    }
+
+    @Override
+    public RisType getRisType() {
+        return RisType.JOUR;
+    }
+
+}