|
1
|
/**
|
|
2
|
* Copyright (C) 2021 EDIT
|
|
3
|
* European Distributed Institute of Taxonomy
|
|
4
|
* http://www.e-taxonomy.eu
|
|
5
|
*
|
|
6
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
|
7
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
|
8
|
*/
|
|
9
|
package eu.etaxonomy.cdm.api.service.media;
|
|
10
|
|
|
11
|
import java.io.IOException;
|
|
12
|
import java.io.InputStream;
|
|
13
|
|
|
14
|
import org.apache.commons.imaging.ImageInfo;
|
|
15
|
import org.apache.commons.imaging.ImageReadException;
|
|
16
|
import org.apache.commons.imaging.Imaging;
|
|
17
|
import org.apache.commons.imaging.common.GenericImageMetadata.GenericImageMetadataItem;
|
|
18
|
import org.apache.commons.imaging.common.ImageMetadata;
|
|
19
|
import org.apache.commons.imaging.common.ImageMetadata.ImageMetadataItem;
|
|
20
|
import org.apache.commons.lang3.StringUtils;
|
|
21
|
import org.apache.http.HttpException;
|
|
22
|
import org.apache.http.client.ClientProtocolException;
|
|
23
|
import org.apache.log4j.Logger;
|
|
24
|
|
|
25
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
|
26
|
import eu.etaxonomy.cdm.common.URI;
|
|
27
|
import eu.etaxonomy.cdm.common.UriUtils;
|
|
28
|
import eu.etaxonomy.cdm.common.media.CdmImageInfo;
|
|
29
|
|
|
30
|
/**
|
|
31
|
* TODO make use of timeOut ?
|
|
32
|
*
|
|
33
|
* @author a.kohlbecker
|
|
34
|
* @since May 6, 2021
|
|
35
|
*/
|
|
36
|
public class MediaMedadataFileReader {
|
|
37
|
|
|
38
|
|
|
39
|
private static Logger logger = Logger.getLogger(MediaMedadataFileReader.class);
|
|
40
|
|
|
41
|
private CdmImageInfo cdmImageInfo;
|
|
42
|
|
|
43
|
public static final Integer IMAGE_READ_TIMEOUT = 3000; // ms
|
|
44
|
|
|
45
|
private Integer timeOut = IMAGE_READ_TIMEOUT; // setting default
|
|
46
|
|
|
47
|
public MediaMedadataFileReader(CdmImageInfo cdmImageInfo) {
|
|
48
|
this.cdmImageInfo = cdmImageInfo;
|
|
49
|
}
|
|
50
|
|
|
51
|
public MediaMedadataFileReader(URI uri) {
|
|
52
|
this.cdmImageInfo = new CdmImageInfo(uri);
|
|
53
|
}
|
|
54
|
|
|
55
|
/**
|
|
56
|
* Reads the image info (width, height, bitPerPixel, metadata, format, mime type)
|
|
57
|
*/
|
|
58
|
public MediaMedadataFileReader readImageInfo() throws IOException, HttpException{
|
|
59
|
|
|
60
|
InputStream inputStream;
|
|
61
|
try {
|
|
62
|
inputStream = UriUtils.getInputStream(cdmImageInfo.getUri());
|
|
63
|
ImageInfo imageInfo = Imaging.getImageInfo(inputStream, null);
|
|
64
|
|
|
65
|
cdmImageInfo.setFormatName(imageInfo.getFormatName());
|
|
66
|
cdmImageInfo.setMimeType(imageInfo.getMimeType());
|
|
67
|
cdmImageInfo.setWidth(imageInfo.getWidth());
|
|
68
|
cdmImageInfo.setHeight(imageInfo.getHeight());
|
|
69
|
cdmImageInfo.setBitPerPixel(imageInfo.getBitsPerPixel());
|
|
70
|
inputStream.close();
|
|
71
|
|
|
72
|
} catch (ImageReadException e) {
|
|
73
|
logger.error("Could not read: " + cdmImageInfo.getUri() + ". " + e.getMessage());
|
|
74
|
throw new IOException(e);
|
|
75
|
}
|
|
76
|
|
|
77
|
return this;
|
|
78
|
}
|
|
79
|
|
|
80
|
public MediaMedadataFileReader readMetaData() throws IOException, HttpException {
|
|
81
|
|
|
82
|
ImageMetadata mediaData = null;
|
|
83
|
try {
|
|
84
|
InputStream inputStream = UriUtils.getInputStream(cdmImageInfo.getUri());
|
|
85
|
mediaData = Imaging.getMetadata(inputStream, null);
|
|
86
|
}catch (ImageReadException e) {
|
|
87
|
logger.error("Could not read: " + cdmImageInfo.getUri() + ". " + e.getMessage());
|
|
88
|
//throw new IOException(e);
|
|
89
|
}
|
|
90
|
|
|
91
|
if(mediaData != null) {
|
|
92
|
for (ImageMetadataItem imItem : mediaData.getItems()){
|
|
93
|
if (imItem instanceof GenericImageMetadataItem){
|
|
94
|
GenericImageMetadataItem item = (GenericImageMetadataItem)imItem;
|
|
95
|
if ("Keywords".equals(item.getKeyword())){
|
|
96
|
String value = text(item);
|
|
97
|
String[] splits = value.split(":");
|
|
98
|
if (splits.length == 2){
|
|
99
|
//convention used e.g. for Flora of cyprus (#9137)
|
|
100
|
cdmImageInfo.getMetaData().put(splits[0].trim(), splits[1].trim());
|
|
101
|
}else{
|
|
102
|
cdmImageInfo.getMetaData().put(
|
|
103
|
item.getKeyword(),
|
|
104
|
CdmUtils.concat("; ", cdmImageInfo.getMetaData().get(item.getKeyword()), value)
|
|
105
|
);
|
|
106
|
}
|
|
107
|
}else if (item.getKeyword().contains("/")){
|
|
108
|
//TODO: not sure where this syntax is used originally
|
|
109
|
String key = item.getKeyword();
|
|
110
|
//key.replace("/", "");
|
|
111
|
int index = key.indexOf("/");
|
|
112
|
key = key.substring(0, index);
|
|
113
|
cdmImageInfo.getMetaData().put(key, text(item));
|
|
114
|
}else{
|
|
115
|
cdmImageInfo.getMetaData().put(item.getKeyword(), text(item));
|
|
116
|
}
|
|
117
|
}
|
|
118
|
}
|
|
119
|
}
|
|
120
|
|
|
121
|
return this;
|
|
122
|
}
|
|
123
|
|
|
124
|
/**
|
|
125
|
* Reads the size of the image defined by the {@link #imageUri} in bytes
|
|
126
|
*/
|
|
127
|
public MediaMedadataFileReader readImageLength() throws ClientProtocolException, IOException, HttpException{
|
|
128
|
try {
|
|
129
|
long length = UriUtils.getResourceLength(cdmImageInfo.getUri(), null);
|
|
130
|
cdmImageInfo.setLength(length);
|
|
131
|
} catch (HttpException e) {
|
|
132
|
if (e.getMessage().equals("Could not retrieve Content-Length")){
|
|
133
|
InputStream inputStream = UriUtils.getInputStream(cdmImageInfo.getUri());
|
|
134
|
int n = 0;
|
|
135
|
while(inputStream.read() != -1){
|
|
136
|
n++;
|
|
137
|
}
|
|
138
|
inputStream.close();
|
|
139
|
logger.info("Content-Length not available in http header. Image size computed via input stream size: " + cdmImageInfo.getUri());
|
|
140
|
cdmImageInfo.setLength(n);
|
|
141
|
}else{
|
|
142
|
throw e;
|
|
143
|
}
|
|
144
|
}
|
|
145
|
return this;
|
|
146
|
}
|
|
147
|
|
|
148
|
public MediaMedadataFileReader readSuffix(){
|
|
149
|
String path = cdmImageInfo.getUri().getPath();
|
|
150
|
String suffix = path.substring(StringUtils.lastIndexOf(path, '.') + 1);
|
|
151
|
cdmImageInfo.setSuffix(suffix);
|
|
152
|
return this;
|
|
153
|
}
|
|
154
|
|
|
155
|
public CdmImageInfo getCdmImageInfo() {
|
|
156
|
return cdmImageInfo;
|
|
157
|
}
|
|
158
|
|
|
159
|
/**
|
|
160
|
* Wrapper for the Item.getText() method which applies cleaning of the text representation.
|
|
161
|
* <ol>
|
|
162
|
* <li>Strings are surrounded by single quotes, these must be removed</li>
|
|
163
|
* </ol>
|
|
164
|
* @param item
|
|
165
|
*/
|
|
166
|
private String text(GenericImageMetadataItem item) {
|
|
167
|
String text = item.getText();
|
|
168
|
if(text.startsWith("'") && text.endsWith("'")) {
|
|
169
|
text = text.substring(1 , text.length() - 1);
|
|
170
|
}
|
|
171
|
return text;
|
|
172
|
}
|
|
173
|
|
|
174
|
|
|
175
|
|
|
176
|
}
|