From 2c7083aa43b7e6636614460eb09c1be63bf3bd0e Mon Sep 17 00:00:00 2001
From: jendib <benjamin.gam@gmail.com>
Date: Sun, 18 Aug 2013 14:11:08 +0200
Subject: [PATCH] Force loading of corrupted PDF

---
 .../core/listener/async/FileCreatedAsyncListener.java     | 2 +-
 .../main/java/com/sismics/docs/core/util/FileUtil.java    | 4 ++--
 .../java/com/sismics/docs/rest/resource/AppResource.java  | 8 ++++----
 docs-web/src/main/webapp/partial/docs/document.edit.html  | 2 +-
 .../test/java/com/sismics/docs/rest/TestAppResource.java  | 2 +-
 5 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/docs-core/src/main/java/com/sismics/docs/core/listener/async/FileCreatedAsyncListener.java b/docs-core/src/main/java/com/sismics/docs/core/listener/async/FileCreatedAsyncListener.java
index 6ea1a214..ffb2af22 100644
--- a/docs-core/src/main/java/com/sismics/docs/core/listener/async/FileCreatedAsyncListener.java
+++ b/docs-core/src/main/java/com/sismics/docs/core/listener/async/FileCreatedAsyncListener.java
@@ -40,7 +40,7 @@ public class FileCreatedAsyncListener {
         final File file = fileCreatedAsyncEvent.getFile();
         long startTime = System.currentTimeMillis();
         final String content = FileUtil.extractContent(fileCreatedAsyncEvent.getDocument(), file);
-        log.info(MessageFormat.format("File OCR-ized in {0}ms", System.currentTimeMillis() - startTime));
+        log.info(MessageFormat.format("File content extracted in {0}ms", System.currentTimeMillis() - startTime));
         
         // Store the OCR-ization result in the database
         TransactionUtil.handle(new Runnable() {
diff --git a/docs-core/src/main/java/com/sismics/docs/core/util/FileUtil.java b/docs-core/src/main/java/com/sismics/docs/core/util/FileUtil.java
index 5ab00ae4..869b2f04 100644
--- a/docs-core/src/main/java/com/sismics/docs/core/util/FileUtil.java
+++ b/docs-core/src/main/java/com/sismics/docs/core/util/FileUtil.java
@@ -103,7 +103,7 @@ public class FileUtil {
         java.io.File storedfile = Paths.get(DirectoryUtil.getStorageDirectory().getPath(), file.getId()).toFile();
         try {
             PDFTextStripper stripper = new PDFTextStripper();
-            pdfDocument = PDDocument.load(storedfile);
+            pdfDocument = PDDocument.load(storedfile.getAbsolutePath(), true);
             content = stripper.getText(pdfDocument);
         } catch (IOException e) {
             log.error("Error while extracting text from the PDF " + storedfile, e);
@@ -153,7 +153,7 @@ public class FileUtil {
             image = ImageIO.read(originalFile);
         } else if(file.getMimeType().equals(MimeType.APPLICATION_PDF)) {
             // Generate preview from the first page of the PDF
-            PDDocument pdfDocument = PDDocument.load(originalFile);
+            PDDocument pdfDocument = PDDocument.load(originalFile.getAbsolutePath(), true);
             @SuppressWarnings("unchecked")
             List<PDPage> pageList = pdfDocument.getDocumentCatalog().getAllPages();
             if (pageList.size() > 0) {
diff --git a/docs-web/src/main/java/com/sismics/docs/rest/resource/AppResource.java b/docs-web/src/main/java/com/sismics/docs/rest/resource/AppResource.java
index 5e90d110..06402533 100644
--- a/docs-web/src/main/java/com/sismics/docs/rest/resource/AppResource.java
+++ b/docs-web/src/main/java/com/sismics/docs/rest/resource/AppResource.java
@@ -148,21 +148,21 @@ public class AppResource extends BaseResource {
     }
     
     /**
-     * OCR-ize all files again.
+     * Extract content from all files again.
      * 
      * @return Response
      * @throws JSONException
      */
     @POST
-    @Path("batch/ocr")
+    @Path("batch/extract")
     @Produces(MediaType.APPLICATION_JSON)
-    public Response batchOcr() throws JSONException {
+    public Response batchExtract() throws JSONException {
         if (!authenticate()) {
             throw new ForbiddenClientException();
         }
         checkBaseFunction(BaseFunction.ADMIN);
         
-        // Raise a OCR file event
+        // Raise an extract file content event
         AppContext.getInstance().getAsyncEventBus().post(new ExtractFileAsyncEvent());
         
         JSONObject response = new JSONObject();
diff --git a/docs-web/src/main/webapp/partial/docs/document.edit.html b/docs-web/src/main/webapp/partial/docs/document.edit.html
index c2fb318e..e3b7eb9d 100644
--- a/docs-web/src/main/webapp/partial/docs/document.edit.html
+++ b/docs-web/src/main/webapp/partial/docs/document.edit.html
@@ -33,7 +33,7 @@
   <div class="control-group">
     <label class="control-label" for="inputFiles">New files</label>
     <div class="controls">
-      <file class="input-block-level" id="inputFiles" multiple="multiple" ng-model="newFiles" accept="image/png,image/jpg,image/jpeg,image/gif" />
+      <file class="input-block-level" id="inputFiles" multiple="multiple" ng-model="newFiles" accept="image/png,image/jpg,image/jpeg,image/gif,application/pdf" />
     </div>
   </div>
   <div class="control-group">
diff --git a/docs-web/src/test/java/com/sismics/docs/rest/TestAppResource.java b/docs-web/src/test/java/com/sismics/docs/rest/TestAppResource.java
index 92600e29..3cfc0040 100644
--- a/docs-web/src/test/java/com/sismics/docs/rest/TestAppResource.java
+++ b/docs-web/src/test/java/com/sismics/docs/rest/TestAppResource.java
@@ -44,7 +44,7 @@ public class TestAppResource extends BaseJerseyTest {
         Assert.assertEquals(0, json.getInt("document_count"));
         
         // OCR-ize all files
-        appResource = resource().path("/app/batch/ocr");
+        appResource = resource().path("/app/batch/extract");
         appResource.addFilter(new CookieAuthenticationFilter(adminAuthenticationToken));
         response = appResource.post(ClientResponse.class);
         Assert.assertEquals(Status.OK, Status.fromStatusCode(response.getStatus()));