public/docs
1
0
Fork 0
mirror of https://github.com/sismics/docs.git synced 2025-12-13 09:46:17 +00:00
Code Issues Releases Wiki Activity

Closes #53: Build thumbnails for DOCX and ODT files

Browse Source
This commit is contained in:
jendib
2015-12-11 22:00:44 +01:00
parent 1a37d97a61
commit 7708f61343
10 changed files with 228 additions and 118 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
docs-web/src/main/java/com/sismics/docs/rest/resource/FileResource.java
View File

@@ -146,8 +146,14 @@ public class FileResource extends BaseResource {
file.setUserId(principal.getId());
String fileId = fileDao.create(file);
// Guess the mime type a second time, for open document format (first detected as simple ZIP file)
file.setMimeType(MimeTypeUtil.guessOpenDocumentFormat(file, fileInputStream));
// Convert to PDF if necessary (for thumbnail and text extraction)
InputStream pdfIntputStream = FileUtil.convertToPdf(fileInputStream, file);
// Save the file
FileUtil.save(fileInputStream, file, user.getPrivateKey());
FileUtil.save(fileInputStream, pdfIntputStream, file, user.getPrivateKey());
// Update the user quota
user.setStorageCurrent(user.getStorageCurrent() + fileData.length);
@@ -159,6 +165,7 @@ public class FileResource extends BaseResource {
fileCreatedAsyncEvent.setDocument(document);
fileCreatedAsyncEvent.setFile(file);
fileCreatedAsyncEvent.setInputStream(fileInputStream);
fileCreatedAsyncEvent.setPdfInputStream(pdfIntputStream);
AppContext.getInstance().getAsyncEventBus().post(fileCreatedAsyncEvent);
}

3
docs-web/src/main/webapp/src/partial/docs/document.edit.html
View File

@@ -39,7 +39,8 @@
<label class="col-sm-2 control-label" for="inputFiles">New files</label>
<div class="col-sm-6">
<file class="form-control" id="inputFiles" multiple="multiple" ng-model="newFiles"
accept="image/png,image/jpg,image/jpeg,image/gif,application/pdf" ng-disabled="fileIsUploading"></file>
accept="image/png,image/jpg,image/jpeg,image/gif,application/pdf,application/vnd.oasis.opendocument.text,application/vnd.openxmlformats-officedocument.wordprocessingml.document"
ng-disabled="fileIsUploading"></file>
</div>
<div class="col-sm-4" ng-if="orphanFiles.length > 0">
+ {{ orphanFiles.length }} file{{ orphanFiles.length > 1 ? 's' : '' }}

132
docs-web/src/test/java/com/sismics/docs/rest/TestDocumentResource.java
View File

@@ -267,6 +267,124 @@ public class TestDocumentResource extends BaseJerseyTest {
return json.getJsonArray("documents").size();
}
/**
* Test ODT extraction.
*
* @throws Exception
*/
@Test
public void testOdtExtraction() throws Exception {
// Login document_odt
clientUtil.createUser("document_odt");
String documentOdtToken = clientUtil.login("document_odt");
// Create a document
long create1Date = new Date().getTime();
JsonObject json = target().path("/document").request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentOdtToken)
.put(Entity.form(new Form()
.param("title", "My super title document 1")
.param("description", "My super description for document 1")
.param("language", "eng")
.param("create_date", Long.toString(create1Date))), JsonObject.class);
String document1Id = json.getString("id");
Assert.assertNotNull(document1Id);
// Add a PDF file
String file1Id = null;
try (InputStream is = Resources.getResource("file/document.odt").openStream()) {
StreamDataBodyPart streamDataBodyPart = new StreamDataBodyPart("file", is, "document.odt");
try (FormDataMultiPart multiPart = new FormDataMultiPart()) {
json = target()
.register(MultiPartFeature.class)
.path("/file").request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentOdtToken)
.put(Entity.entity(multiPart.field("id", document1Id).bodyPart(streamDataBodyPart),
MediaType.MULTIPART_FORM_DATA_TYPE), JsonObject.class);
file1Id = json.getString("id");
Assert.assertNotNull(file1Id);
}
}
// Search documents by query in full content
json = target().path("/document/list")
.queryParam("search", "full:ipsum")
.request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentOdtToken)
.get(JsonObject.class);
Assert.assertTrue(json.getJsonArray("documents").size() == 1);
// Get the file thumbnail data
Response response = target().path("/file/" + file1Id + "/data")
.queryParam("size", "thumb")
.request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentOdtToken)
.get();
InputStream is = (InputStream) response.getEntity();
byte[] fileBytes = ByteStreams.toByteArray(is);
Assert.assertTrue(fileBytes.length > 0); // Images rendered from PDF differ in size from OS to OS due to font issues
Assert.assertEquals(MimeType.IMAGE_JPEG, MimeTypeUtil.guessMimeType(fileBytes));
}
/**
* Test DOCX extraction.
*
* @throws Exception
*/
@Test
public void testDocxExtraction() throws Exception {
// Login document_docx
clientUtil.createUser("document_docx");
String documentDocxToken = clientUtil.login("document_docx");
// Create a document
long create1Date = new Date().getTime();
JsonObject json = target().path("/document").request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentDocxToken)
.put(Entity.form(new Form()
.param("title", "My super title document 1")
.param("description", "My super description for document 1")
.param("language", "eng")
.param("create_date", Long.toString(create1Date))), JsonObject.class);
String document1Id = json.getString("id");
Assert.assertNotNull(document1Id);
// Add a PDF file
String file1Id = null;
try (InputStream is = Resources.getResource("file/document.docx").openStream()) {
StreamDataBodyPart streamDataBodyPart = new StreamDataBodyPart("file", is, "document.docx");
try (FormDataMultiPart multiPart = new FormDataMultiPart()) {
json = target()
.register(MultiPartFeature.class)
.path("/file").request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentDocxToken)
.put(Entity.entity(multiPart.field("id", document1Id).bodyPart(streamDataBodyPart),
MediaType.MULTIPART_FORM_DATA_TYPE), JsonObject.class);
file1Id = json.getString("id");
Assert.assertNotNull(file1Id);
}
}
// Search documents by query in full content
json = target().path("/document/list")
.queryParam("search", "full:dolor")
.request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentDocxToken)
.get(JsonObject.class);
Assert.assertTrue(json.getJsonArray("documents").size() == 1);
// Get the file thumbnail data
Response response = target().path("/file/" + file1Id + "/data")
.queryParam("size", "thumb")
.request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentDocxToken)
.get();
InputStream is = (InputStream) response.getEntity();
byte[] fileBytes = ByteStreams.toByteArray(is);
Assert.assertTrue(fileBytes.length > 0); // Images rendered from PDF differ in size from OS to OS due to font issues
Assert.assertEquals(MimeType.IMAGE_JPEG, MimeTypeUtil.guessMimeType(fileBytes));
}
/**
* Test PDF extraction.
*
@@ -274,14 +392,14 @@ public class TestDocumentResource extends BaseJerseyTest {
*/
@Test
public void testPdfExtraction() throws Exception {
// Login document2
clientUtil.createUser("document2");
String document2Token = clientUtil.login("document2");
// Login document_pdf
clientUtil.createUser("document_pdf");
String documentPdfToken = clientUtil.login("document_pdf");
// Create a document
long create1Date = new Date().getTime();
JsonObject json = target().path("/document").request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, document2Token)
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentPdfToken)
.put(Entity.form(new Form()
.param("title", "My super title document 1")
.param("description", "My super description for document 1")
@@ -298,7 +416,7 @@ public class TestDocumentResource extends BaseJerseyTest {
json = target()
.register(MultiPartFeature.class)
.path("/file").request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, document2Token)
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentPdfToken)
.put(Entity.entity(multiPart.field("id", document1Id).bodyPart(streamDataBodyPart),
MediaType.MULTIPART_FORM_DATA_TYPE), JsonObject.class);
file1Id = json.getString("id");
@@ -310,7 +428,7 @@ public class TestDocumentResource extends BaseJerseyTest {
json = target().path("/document/list")
.queryParam("search", "full:vrandecic")
.request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, document2Token)
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentPdfToken)
.get(JsonObject.class);
Assert.assertTrue(json.getJsonArray("documents").size() == 1);
@@ -318,7 +436,7 @@ public class TestDocumentResource extends BaseJerseyTest {
Response response = target().path("/file/" + file1Id + "/data")
.queryParam("size", "thumb")
.request()
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, document2Token)
.cookie(TokenBasedSecurityFilter.COOKIE_NAME, documentPdfToken)
.get();
InputStream is = (InputStream) response.getEntity();
byte[] fileBytes = ByteStreams.toByteArray(is);

BIN
docs-web/src/test/resources/file/document.docx Normal file
View File

Binary file not shown.

BIN
docs-web/src/test/resources/file/document.odt Normal file
View File

Binary file not shown.