Merge pull request #800 from btut/feature/bibtexHeaderAndFulltext

Accept application/x-bibtex for processHeaderDocument
kermitt2 · Aug 1, 2021 · 1a6b103 · 1a6b103
2 parents e4ca196 + 2a69fa4
commit 1a6b103
Show file tree

Hide file tree

Showing 4 changed files with 60 additions and 13 deletions.
diff --git a/grobid-core/src/main/java/org/grobid/core/data/BiblioItem.java b/grobid-core/src/main/java/org/grobid/core/data/BiblioItem.java
@@ -1922,12 +1922,19 @@ public String toBibTeX(String id, GrobidAnalysisConfig config) {
                     fullAuthors.stream()
                                .filter(person -> person != null)
                                .forEachOrdered(person -> {
-                                   String author = person.getLastName();
+                                   String author = "";
+                                   if (person.getLastName() != null) {
+                                       author = person.getLastName();
+                                   }
                                    if (person.getFirstName() != null) {
-                                       author += ", ";
+                                       if (author.length() > 0) {
+                                           author += ", ";
+                                       }
                                        author += person.getFirstName();
                                    }
-                                   authors.add(author);
+                                   if (author.length() > 0 ) {
+                                       authors.add(author);
+                                   }
                                });
                 } else if (this.authors != null) {
                     StringTokenizer st = new StringTokenizer(this.authors, ";");
@@ -1973,9 +1980,12 @@ public String toBibTeX(String id, GrobidAnalysisConfig config) {
             }
             // fullEditors has to be used instead
 
-            // year
-            if (publication_date != null) {
-                bibtex.add("  year = {" + publication_date + "}");
+            // dates
+            if (normalized_publication_date != null) {
+                String isoDate = TEIFormatter.toISOString(normalized_publication_date);
+                if (isoDate != null) {
+                    bibtex.add("  date = {" + isoDate + "}");
+                }
             }
 
             // address

diff --git a/grobid-core/src/main/java/org/grobid/core/data/Date.java b/grobid-core/src/main/java/org/grobid/core/data/Date.java
@@ -211,4 +211,4 @@ public String toXML() {
         return theDate;
     }
 
-}
+}
diff --git a/grobid-service/src/main/java/org/grobid/service/GrobidRestService.java b/grobid-service/src/main/java/org/grobid/service/GrobidRestService.java
@@ -152,26 +152,54 @@ public Response getAdmin_htmlGet(@QueryParam(SHA1) String sha1) {
     @Consumes(MediaType.MULTIPART_FORM_DATA)
     @Produces(MediaType.APPLICATION_XML)
     @POST
-    public Response processHeaderDocument_post(
+    public Response processHeaderDocumentReturnXml_post(
         @FormDataParam(INPUT) InputStream inputStream,
         @DefaultValue("0") @FormDataParam(CONSOLIDATE_HEADER) String consolidate,
         @DefaultValue("0") @FormDataParam(INCLUDE_RAW_AFFILIATIONS) String includeRawAffiliations) {
         int consol = validateConsolidationParam(consolidate);
         return restProcessFiles.processStatelessHeaderDocument(
             inputStream, consol,
-            validateIncludeRawParam(includeRawAffiliations)
+            validateIncludeRawParam(includeRawAffiliations),
+            ExpectedResponseType.XML
         );
     }
 
     @Path(PATH_HEADER)
     @Consumes(MediaType.MULTIPART_FORM_DATA)
     @Produces(MediaType.APPLICATION_XML)
     @PUT
-    public Response processStatelessHeaderDocument(
+    public Response processStatelessHeaderDocumentReturnXml(
+        @FormDataParam(INPUT) InputStream inputStream,
+        @DefaultValue("0") @FormDataParam(CONSOLIDATE_HEADER) String consolidate,
+        @DefaultValue("0") @FormDataParam(INCLUDE_RAW_AFFILIATIONS) String includeRawAffiliations) {
+        return processHeaderDocumentReturnXml_post(inputStream, consolidate, includeRawAffiliations);
+    }
+
+    @Path(PATH_HEADER)
+    @Consumes(MediaType.MULTIPART_FORM_DATA)
+    @Produces(BibTexMediaType.MEDIA_TYPE)
+    @POST
+    public Response processHeaderDocumentReturnBibTeX_post(
+        @FormDataParam(INPUT) InputStream inputStream,
+        @DefaultValue("0") @FormDataParam(CONSOLIDATE_HEADER) String consolidate,
+        @DefaultValue("0") @FormDataParam(INCLUDE_RAW_AFFILIATIONS) String includeRawAffiliations) {
+        int consol = validateConsolidationParam(consolidate);
+        return restProcessFiles.processStatelessHeaderDocument(
+            inputStream, consol,
+            validateIncludeRawParam(includeRawAffiliations),
+            ExpectedResponseType.BIBTEX
+        );
+    }
+
+    @Path(PATH_HEADER)
+    @Consumes(MediaType.MULTIPART_FORM_DATA)
+    @Produces(BibTexMediaType.MEDIA_TYPE)
+    @PUT
+    public Response processStatelessHeaderDocumentReturnBibTeX(
         @FormDataParam(INPUT) InputStream inputStream,
         @DefaultValue("0") @FormDataParam(CONSOLIDATE_HEADER) String consolidate,
         @DefaultValue("0") @FormDataParam(INCLUDE_RAW_AFFILIATIONS) String includeRawAffiliations) {
-        return processHeaderDocument_post(inputStream, consolidate, includeRawAffiliations);
+        return processHeaderDocumentReturnBibTeX_post(inputStream, consolidate, includeRawAffiliations);
     }
 
     @Path(PATH_FULL_TEXT)

diff --git a/grobid-service/src/main/java/org/grobid/service/process/GrobidRestProcessFiles.java b/grobid-service/src/main/java/org/grobid/service/process/GrobidRestProcessFiles.java
@@ -4,6 +4,7 @@
 import com.google.inject.Singleton;
 import org.apache.pdfbox.pdmodel.PDDocument;
 import org.grobid.core.data.BibDataSet;
+import org.grobid.core.data.BiblioItem;
 import org.grobid.core.data.PatentItem;
 import org.grobid.core.document.Document;
 import org.grobid.core.document.DocumentSource;
@@ -62,7 +63,8 @@ public GrobidRestProcessFiles() {
     public Response processStatelessHeaderDocument(
         final InputStream inputStream,
         final int consolidate,
-        final boolean includeRawAffiliations
+        final boolean includeRawAffiliations,
+        ExpectedResponseType expectedResponseType
     ) {
         LOGGER.debug(methodLogIn());
         String retVal = null;
@@ -91,17 +93,24 @@ public Response processStatelessHeaderDocument(
 
             String md5Str = DatatypeConverter.printHexBinary(digest).toUpperCase();
 
+            BiblioItem result = new BiblioItem();
+
             // starts conversion process
             retVal = engine.processHeader(
                 originFile.getAbsolutePath(),
                 md5Str,
                 consolidate,
                 includeRawAffiliations,
-                null
+                result
             );
 
             if (GrobidRestUtils.isResultNullOrEmpty(retVal)) {
                 response = Response.status(Response.Status.NO_CONTENT).build();
+            } else if (expectedResponseType == ExpectedResponseType.BIBTEX) {
+                response = Response.status(Response.Status.OK)
+                    .entity(result.toBibTeX("-1"))
+                    .header(HttpHeaders.CONTENT_TYPE, BibTexMediaType.MEDIA_TYPE + "; charset=UTF-8")
+                    .build();
             } else {
                 response = Response.status(Response.Status.OK)
                     .entity(retVal)