チェンジセット 54

AutoKeyword/trunk/README

r6	r54
63	63	Mecab 0.93
64	64	java 1.5.03（おそらく、これ以前のバージョンでもコンパイル/実行可能です）
	65	Jakarta Logging Log4j 1.2.12+ [http://logging.apache.org/log4j/]
65	66
66	67	■ 環境設定

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/analyze/Analyzer.java

r47	r54
2	2
3	3	import java.util.*;
4		~~import java.io.*;~~
5	4
6	5	/**
…	…
20	19	* �ݒ菉�� * @param properties �ݒ��e
21	20	* @param out �f�o�b�O�o��
22		* @return
	21	* @return ��
23	22	*/
24		public boolean init(Properties properties~~, PrintWriter out~~);
	23	public boolean init(Properties properties);
25	24
26	25	/**
…	…
30	29	* @param out ��s��f�o�b�O�o��
31	30	* @return ��s��ʃf�[�^ Collection<Keyword>
32		*
33		* @see #exec(String, String, PrintWriter)
34		* @see #parse(Collection)
35	31	*/
36		public Collection parse(String filename~~, PrintWriter out~~ );
	32	public Collection parse(String filename);
37	33	}
38	34	/*

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/analyze/Keyword.java

r47	r54
2	2
3	3	/**
4		* �`�ԑf��̌��ʂ��
	4	* �`�ԑf��̌��ʂ��an�N��X.
5	5	*
6	6	* <BR><BR>Copyright (C) 2006 Nihon Unisys, Ltd. All Rights Reserved.
…	…
17	17
18	18	/**
19		* �L�[��[�h
20		* @param word �L�[��[�h�� * @param part �i��
21		* @param sub �ו��i��
	19	* �L�[��[�h��̍쐬.
	20	*
	21	* @param word �L�[��[�h��ϒl]
	22	* @param part �i��[�s�ϒl]
	23	* @param sub �ו��i��[�s�ϒl]
22	24	* @param count �o��
23	25	*/
…	…
31	33
32	34	/**
33		* �L�[��[�h
34		* @param word �L�[��[�h�� * @param part �i��
35		* @param sub �ו��i��
	35	* �L�[��[�h��̍쐬.
	36	*
	37	* �o��񐔂��Ƃ��č쐬��
	38	*
	39	* @param word �L�[��[�h��ϒl]
	40	* @param part �i��[�s�ϒl]
	41	* @param sub �ו��i��[�s�ϒl]
	42	*
	43	* @see Keyword#Keyword(String, String, String, int)
36	44	*/
37	45	public Keyword(String word, String part, String sub)
…	…
40	48	}
41	49
	50	/**
	51	* �L�[��[�h�̕��擾�� * @return �L�[��[�h�� */
42	52	public String getWord()
43	53	{
…	…
45	55	}
46	56
	57	/**
	58	* ��̃L�[��[�h�̕i�� * @return �i��
	59	*/
47	60	public String getPart()
48	61	{
…	…
50	63	}
51	64
	65	/**
	66	* ��̃L�[��[�h�̕i��ו�� * @return �i��ו��
	67	*/
52	68	public String getSub()
53	69	{
…	…
55	71	}
56	72
	73	/**
	74	* ��̃L�[��[�h�̏o��񐔂��
	75	* @return �o��
	76	*/
57	77	public int getCount()
58	78	{
59	79	return count;
60	80	}
61		//
	81
	82	/**
	83	* ��̃L�[��[�h�̏o��񐔂�肷��
	84	* @param c �o��
	85	*/
62	86	public void setCount(int c) {
63	87	count = c;
64	88	}
65		public String info()
	89
	90	/**
	91	* ��̃C��X�^��X�̕��L��
	92	* @return ��̃C��X�^��X�̕��L
	93	* @see Object#toString()
	94	*/
	95	public String toString()
66	96	{
67		return getCount() + "\t" + getWord() + "\t" + getPart() + "\t" + getSub();
	97	return new StringBuffer().//
	98	append(getCount()).append("\t").//
	99	append(getWord()).append("\t").//
	100	append(getPart()).append("\t").//
	101	append(getSub()).toString();
68	102	}
69	103	}

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/analyze/KeywordFilter.java

r47	r54
22	22	/** ��߂̃o�b�t�@ */
23	23	private final ArrayList prevKeyword = new ArrayList(32);
24		/** �L�[��[�h�� Map<�L�[��[�h��L�[��[�h> */
	24	/** �W�v��ꂽ�L�[��[�h�� Map<�L�[��[�h��L�[��[�h> */
25	25	private HashMap /* <String, Keyword> */keywordMap = new HashMap();
26	26

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/analyze/Mecab.java

r47	r54
2	2
3	3	import java.util.*;
4		import java.io.*;
	4
5	5	import jp.tyzoh.rinza.folksonomy.common.*;
6	6
…	…
27	27	* ��̏�� * @param properties �ݒ��e
28	28	* @param out �f�o�b�O�o��
	29	* @return ��
	30	* @see Analyzer#init(Properties)
29	31	*/
30		public boolean init(Properties properties~~, PrintWriter out~~)
	32	public boolean init(Properties properties)
31	33	{
32	34	command = properties.getProperty(ANALYZER_COMMAND, command);
…	…
37	39
38	40	/**
39		* �R�}��h��s��ʂ�[�X��Ď擾�� * @param filename ��Ώۃt�@�C��
40		* @param out �f�o�b�O�o��
	41	* �R�}��h��s��ʂ�[�X��Ď擾��
	42	*
	43	*
	44	* <p>
	45	* <ol>�� * <li> Mecab��ăt�@�C��̓��ԑf�� * <li> {@link KeywordFilter}�ŉ��ꂽ�L�[��[�h��v��
	46	* </ol>
	47	*
	48	* </p>
	49	*
	50	* @param filename ��Ώۃt�@�C��
41	51	* @return �L�[��[�h�̃R��N�V��Collection<Keyword>
	52	* @see Analyzer#parse(String)
	53	* @see AutokeywordUtil#exec(String, String)
42	54	*/
43		public Collection parse(String filename~~, PrintWriter out~~)
	55	public Collection parse(String filename)
44	56	{
45
46	57	String command = this.command + " " + filename;
47		Collection execResults = ~~Util.exec(command, encodeType, out~~);
	58	Collection execResults = AutokeywordUtil.exec(command, encodeType);
48	59	Collection parseResults = parse(execResults);
49
50	60	return parseResults;
51	61	}
52	62
53	63	/**
54		* ��s��ʂ�[�X��āA�L�[��[�h��Ɋi�[�� * @param collection ��s��ʍs�̕��lection<String>
55		* @return
	64	* ��s��ʂ�[�X��āA�L�[��[�h��Ɋi�[��
	65	*
	66	* @param collection ��s��ʍs�̕��lection<String>
	67	* @return �p�[�X��L�[��[�h�̃R��N�V�� (Collection<Keyword>)
56	68	*/
57	69	private Collection parse(Collection collection)

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/analyze/Sen.java

r47	r54
4	4	import java.io.File;
5	5	import java.io.IOException;
6		~~import java.io.PrintWriter;~~
7	6	import java.util.Collection;
8	7	import java.util.Properties;
9	8
	9	import jp.tyzoh.rinza.folksonomy.common.AutokeywordUtil;
10	10	import jp.tyzoh.rinza.folksonomy.extractor.ContentsExtractorLookuper;
	11	import jp.tyzoh.rinza.folksonomy.extractor.Extractor;
11	12	import net.java.sen.StreamTagger;
12	13	import net.java.sen.StringTagger;
…	…
44	45	* ��
45	46	* @param properties �ݒ��e
46		* @param out �f�o�b�O�o��
47	47	* @param ��
	48	* @see Analyzer#init(Properties)
48	49	*/
49		public boolean init(Properties properties~~, PrintWriter out~~) {
	50	public boolean init(Properties properties) {
50	51	command = properties.getProperty(ANALYZER_COMMAND, command);
51	52
…	…
60	61
61	62	/**
62		* �t�@�C��~~[�X~~��
	63	* �t�@�C��ԑf��
63	64	*
	65	* <p>
	66	* <ol>�� * <li> {@link Extractor}�Ńe�L�X�g��o��
	67	* <li> {@link StreamTagger}��ăt�@�C��̓��ԑf�� * <li> {@link KeywordFilter}�ŉ��ꂽ�L�[��[�h��v��
	68	* </ol>
	69	*
	70	* </p>
64	71	* @param filename
65	72	* �t�@�C��
…	…
67	74	* ��s�G��[�̏o��
68	75	* @return �L�[��[�h�̃R��N�V��(Collection<Keyword>)
	76	* @see Analyzer#parse(String)
69	77	*/
70		public Collection parse(String filename~~, PrintWriter out~~) {
	78	public Collection parse(String filename) {
71	79	final KeywordFilter kf = new KeywordFilter();
72	80	BufferedReader br = null;
…	…
79	87	}
80	88	} catch (IOException e) {
81		~~e.printStackTrace(out~~);
	89	AutokeywordUtil.getLogger().warn("Sen��͒��ɃG��[��܂��", e);
82	90	} finally {
83	91	if (br != null) {

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/common/AutokeywordUtil.java

r43	r54
1	1	package jp.tyzoh.rinza.folksonomy.common;
2	2
3		import java.io.*;
	3	import java.io.BufferedReader;
	4	import java.io.File;
	5	import java.io.IOException;
	6	import java.io.InputStream;
	7	import java.io.InputStreamReader;
4	8	import java.lang.reflect.Constructor;
5	9	import java.lang.reflect.InvocationTargetException;
6		import java.util.*;
	10	import java.util.Collection;
	11	import java.util.Vector;
7	12
8	13	import javax.activation.MimetypesFileTypeMap;
9	14
	15	import org.apache.log4j.Logger;
	16
10	17	/**
11	18	* folksonomy �p�b�P�[�W��ʂŎg��鏈��
12		*
13		* <BR><BR>Copyright (C) 2006 Nihon Unisys, Ltd. All Rights Reserved.
	19	*
	20	* <BR>
	21	* <BR>
	22	* Copyright (C) 2006 Nihon Unisys, Ltd. All Rights Reserved.
14	23	*/
15		public class Util {
	24	public class AutokeywordUtil {
16	25	/** MIME�̃t�@�C��^�C�v�}�b�v */
17	26	private static final MimetypesFileTypeMap mimeMap = new MimetypesFileTypeMap();
18	27
19
20		public static Collection exec(String command, String encodeType, PrintWriter out)
21		{
22		return exec(new String[]{command}, encodeType, out);
	28	/** Autokeyword�p�̃��O�o��*/
	29	private static final Logger log = Logger.getLogger("Autokeyword");
	30
	31	/**
	32	* �R�}��h�̎�s
	33	*
	34	* @param command
	35	* �R�}��h
	36	* @param encodeType
	37	* �R�}��h��ʏo�̓G��R�[�h
	38	* @return ��s��ʂ̏o�͓�(Collection<String)
	39	* @see #exec(String[], String)
	40	*/
	41	public static Collection exec(String command, String encodeType) {
	42	return exec(new String[] { command }, encodeType);
23	43	}
24	44
25		public static Collection exec(String[] commandArray, String encodeType, PrintWriter out)
26		{
27		Collection collection = new Vector();
	45	/**
	46	* �R�}��h�̎�s
	47	*
	48	* @param commandArray
	49	* �R�}��h�� * @param encodeType
	50	* �R�}��h��ʏo�̓G��R�[�h
	51	* @return ��s��ʂ̏o�͓�(Collection<String)
	52	*/
	53	public static Collection exec(String[] commandArray, String encodeType) {
	54	Collection collection = new Vector();
28	55
29	56	try {
30		if (out != null) {
31		for(int i = 0; i < commandArray.length; i++) {
32		out.print(commandArray[i] + " ");
	57
	58	if (log.isDebugEnabled()) {
	59	final StringBuffer buf = new StringBuffer();
	60	buf.append("Util.exec ��s: ��");
	61	for (int i = 0; i < commandArray.length; i++) {
	62	buf.append(commandArray[i]).append(" ");
33	63	}
34		~~out.println("<br>"~~);
	64	log.debug(buf.toString());
35	65	}
36		/*
37		for(int i = 0; i < commandArray.length; i++) {
38		System.out.print(commandArray[i] + " ");
39		}
40		System.out.println("\n");
41		*/
42		Process process = null;
	66
	67	Process process = null;
43	68	if (commandArray.length == 1) {
44	69	process = Runtime.getRuntime().exec(commandArray[0]);
…	…
47	72	}
48	73	InputStream is = process.getInputStream();
49		InputStreamReader isr = null;
50		BufferedReader br = null;
	74	InputStreamReader isr = null;
	75	BufferedReader br = null;
51	76	try {
52		//~~int~~ count = 0;
	77	// int count = 0;
53	78	if (encodeType.equals("non")) {
54	79	isr = new InputStreamReader(is);
55	80	} else {
56		isr = new InputStreamReader(is, encodeType);
	81	isr = new InputStreamReader(is, encodeType);
57	82	}
58	83	br = new BufferedReader(isr);
59		while(true) {
60		String line = br.readLine();
61		if (line == null) break;
	84	while (true) {
	85	String line = br.readLine();
	86	if (line == null)
	87	break;
62	88	collection.add(line);
63		if (out != null) out.println(line+"<br>");
64		// System.out.println(line);
	89	if (log.isDebugEnabled()) {
	90	log.debug("Util.exec():[��:" + line);
	91	}
65	92	}
66		} catch(IOException ioe) {
	93	} catch (IOException ioe) {
67	94	ioe.printStackTrace();
68	95	try {
69		if (br != null) br.close();
70		if (isr != null) isr.close();
71		if (is != null) is.close();
72		} catch(IOException ioe2) {
	96	if (br != null)
	97	br.close();
	98	if (isr != null)
	99	isr.close();
	100	if (is != null)
	101	is.close();
	102	} catch (IOException ioe2) {
73	103	ioe2.printStackTrace();
74	104	}
75	105	}
76		} catch(Exception ex) {
77		ex.printStackTrace();
78		if (out != null) out.println("error " + ex.getMessage());
	106	} catch (Exception ex) {
	107	log.error("Util.exec()�̎�s��ɃG��[��܂��:" + ex.getMessage(), ex);
79	108	}
80	109
…	…
85	114	* �C��X�^��X�쐬��.
86	115	*
87		* @param name �N��X��
88		* @param argumentClasses ��z�� * @param argumentObjects ��z�� * @return �쐬��C��X�^��X
	116	* @param name
	117	* �N��X��
	118	* @param argumentClasses
	119	* ��z�� * @param argumentObjects
	120	* ��z�� * @return �쐬��C��X�^��X
89	121	*
90		* @throws ClassNotFoundException
	122	* @throws ClassNotFoundException
91	123	* @throws InstantiationException
92		* @throws IllegalAccessException
93		* @throws SecurityException
94		* @throws NoSuchMethodException
95		* @throws IllegalArgumentException
	124	* @throws IllegalAccessException
	125	* @throws SecurityException
	126	* @throws NoSuchMethodException
	127	* @throws IllegalArgumentException
96	128	* @throws InvocationTargetException
97	129	*/
98		public static Object createInstance(String name, Class[] argumentClasses, Object[] argumentObjects) throws ClassNotFoundException, InstantiationException, IllegalAccessException, SecurityException, NoSuchMethodException, IllegalArgumentException, InvocationTargetException {
	130	public static Object createInstance(String name, Class[] argumentClasses,
	131	Object[] argumentObjects) throws ClassNotFoundException,
	132	InstantiationException, IllegalAccessException, SecurityException,
	133	NoSuchMethodException, IllegalArgumentException,
	134	InvocationTargetException {
99	135	final Class clazz = Class.forName(name);
100		if~~(argumentClasses==~~null) {
	136	if (argumentClasses == null) {
101	137	return clazz.newInstance();
102	138	} else {
…	…
106	142	}
107	143
	144	//
108	145	public static String findType(File file) {
109	146	return mimeMap.getContentType(file);
…	…
113	150	return mimeMap.getContentType(fileName);
114	151	}
	152
	153	//
	154	public static Logger getLogger() {
	155	return log;
	156	}
115	157	}
116		/*
117		* This Program is distributed under version 1.0 of the Rinza Public
118		* ~~License Agreement, that is bundled with this package in the file~~
119		* ~~LICENSE, and is~~ available through the website at the following URL:
	158	/*
	159	* This Program is distributed under version 1.0 of the Rinza Public License
	160	* Agreement, that is bundled with this package in the file LICENSE, and is
	161	* available through the website at the following URL:
120	162	* http://www.tyzoh.jp/rinza/licenses/LICENSE-1.0.txt.
121	163	*
122		* This is the Original Program.
123		* The Initial Developer of the Original Program is Nihon Unisys, Ltd.
124		* The Original Program is copyrighted (C) 2006 by Nihon Unisys, Ltd. with
125		* all rights reserved.
126		* There is NO WARRANTY OF ANY KIND by the Initial Developer of the
127		* Original Program.
	164	* This is the Original Program. The Initial Developer of the Original Program
	165	* is Nihon Unisys, Ltd. The Original Program is copyrighted (C) 2006 by Nihon
	166	* Unisys, Ltd. with all rights reserved. There is NO WARRANTY OF ANY KIND by
	167	* the Initial Developer of the Original Program.
128	168	*/

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/ContentsExtractorLookuper.java

r43	r54
7	7	import java.util.HashMap;
8	8
9		import jp.tyzoh.rinza.folksonomy.common.Util;
	9	import jp.tyzoh.rinza.folksonomy.common.AutokeywordUtil;
10	10
11	11	/**
…	…
32	32
33	33	/**
34		* �e�L�X�g��o�N��X�̕ێ��N��X�̏�� *
	34	* �e�L�X�g��o�N��X�̕ێ��N��X�̏��
	35	*
35	36	*/
36	37	public ContentsExtractorLookuper() {
…	…
40	41
41	42	/**
42		* ��o��N��X
	43	* ��o��N��X�̉�
	44	*
43	45	* @param mimeType MIME�^�C�v
44	46	* @return ��o��N��X
…	…
50	52
51	53	/**
52		* �t�@�C��ader��
	54	* �t�@�C��L�X�g�ǂݍ��ݗpReader��
53	55	* @param file ��ΏۃN��X.
54	56	* @return Reader
55	57	* @throws IOException �t�@�C��ǂݍ��݂Ȃǂł�O�G��[
	58	* @see AutokeywordUtil#findType(File) MIME�^�C�v�̉�
56	59	*/
57	60	public BufferedReader getExtractorReader(File file) throws IOException {
58		final String mimeType = Util.findType(file);
	61	final String mimeType = AutokeywordUtil.findType(file);
59	62	return new BufferedReader(getExtractor(mimeType).extract(new FileInputStream(file)));
60	63	}
61	64
62	65	/**
63		* �e�L�X�g��o�N��X�ƃR��e��c�^�C�v�̑Ή��\�̏�� *
	66	* �e�L�X�g��o�N��X�ƃR��e��c�^�C�v�̑Ή��\�̏��
	67	*
	68	* <p>
	69	* �W�J�N��X( {@link Extractor}�̎��N��X )�̃C��X�^��X��A
	70	* {@link #map}�ɃR��e��c�^�C�v�̕��L�[�Ɋi�[�� * </p>
64	71	*/
65	72	private void initMap() {
…	…
68	75	Object obj = null;
69	76	try {
70		obj = Util.createInstance(name, //
	77	obj = AutokeywordUtil.createInstance(name, //
71	78	new Class[] {,}, new Object[] {,});
72	79	} catch (Throwable e) {

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/DefaultExtractor.java

r43	r54
14	14	*/
15	15	public class DefaultExtractor implements Extractor {
	16	/** �t�@�C��̃G��R�[�h */
16	17	private final String encode;
17
	18	/** ��̓W�J��̃R��e��c�^�C�v�萔 */
18	19	private static String CONTENT_TYPE = "text/plain";
19
	20	/** �t�@�C��G��R�[�h�̃f�B�t�H��g�w��/
20	21	private static String DEFAULT_ENCODING = "jisautodetect";
21	22
	23	/**
	24	* �t�@�C��̓W�J��̏��
	25	*
	26	* �G��R�[�h��{@value #DEFAULT_ENCODING}�ŏ��
	27	* @see DefaultExtractor#DefaultExtractor(String)
	28	*/
22	29	public DefaultExtractor() {
23	30	this(null);
24	31	}
	32
	33	/**
	34	* �t�@�C��̓W�J��̏��
	35	*
	36	* @param encode �e�L�X�g�̃G��R�[�h
	37	*/
25	38	public DefaultExtractor(String encode) {
26	39	this.encode = encode;
27	40	}
	41
	42	/**
	43	* �e�L�X�g�ǂݍ��eader�̎擾.
	44	*
	45	* @param content ��g��[��
	46	* @return �e�L�X�g�̓Ǎ��p {@link Reader}
	47	* @throws IOException �ǂݍ��ݏ��O�G��[��
	48	*/
28	49	public Reader extract(InputStream content) throws IOException {
29	50	return new InputStreamReader(content, encode==null?DEFAULT_ENCODING:encode);
30	51	}
31	52
	53	/**
	54	* ��̃e�L�X�g��o�N��X�̃R��e��c�^�C�v.
	55	*
	56	* @return �R��e��c�^�C�v
	57	*/
32	58	public String getContentType() {
33	59	return CONTENT_TYPE;

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/Extractor.java

r43	r54
12	12	public interface Extractor {
13	13	/**
14		* �f�[�^�t�@�C��̃X�g��[��[�_�[�֕ϊ��
	14	* �f�[�^�t�@�C��̃X�g��[��L�X�g��o��[�_�[�֕ϊ��
15	15	* @param content �R��e��c�X�g��[��
16	16	* @return �e�L�X�g��̂ݒ��o��Reader

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/pdfbox/PDFExtractor.java

r43	r54
29	29	*/
30	30	public class PDFExtractor implements Extractor {
31
	31	/** PDF�̃R��e��c�^�C�v�� */
32	32	private static final String CONTENT_TYPE_PDF_1 = "application/pdf";
	33	/** PDF�̃R��e��c�^�C�v�� */
33	34	private static final String CONTENT_TYPE_PDF_2 = "application/x-pdf";
	35	/** CSV�ŕ��L�q��PDF�t�@�C��R��e��c�^�C�v */
34	36	private static final String CONTENT_TYPE_PDF_ALL_CSV = CONTENT_TYPE_PDF_1+","+CONTENT_TYPE_PDF_2;
35	37
36
	38	/**
	39	* PDF�t�@�C��L�X�g�f�[�^�𒊏o��ader��
	40	*
	41	* <p>
	42	* {@link PDFTextStripper}��ăe�L�X�g��o��B
	43	* ��PDFBox�œ��Map��܂��Ȃ��̂�
	44	* ��PDF��ǂ߂܂��
	45	* </p>
	46	*
	47	* @param content PDF�t�@�C��̓��̓�g��[��
	48	* @return �e�L�X�g�f�[�^�̒��o��ꂽReader
	49	* @throws IOException �X�g��[��ǂݍ��ݎ��A��PDF��
	50	* @see PDFTextStripper PDFBox�̃e�L�X�g��o��N��X
	51	*/
37	52	public Reader extract(InputStream content) throws IOException {
38	53	try {
39	54	PDFParser parser = new PDFParser(content);
40	55	parser.parse();
41
42
	56
43	57	PDDocument document = parser.getPDDocument();
44	58	CharArrayWriter writer = new CharArrayWriter();
…	…
57	71	}
58	72
59
	73	/**
	74	* PDF�t�@�C��̃R��e��c�^�C�v�� * @return PDF�t�@�C��̃R��e��c�^�C�v
	75	*/
60	76	public String getContentType() {
61	77	return CONTENT_TYPE_PDF_ALL_CSV;

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/poi/MSExcelExtractor.java

r43	r54
31	31	*/
32	32	public class MSExcelExtractor implements Extractor {
33
	33	/** Excel�̃R��e��c�^�C�v�� */
34	34	static final String CONTENT_TYPE_EXCEL_1 = "application/msexcel";
	35	/** Excel�̃R��e��c�^�C�v�� */
35	36	static final String CONTENT_TYPE_EXCEL_2 = "application/vnd.ms-excel";
	37	/** CSV�ŕ��L�q��Excel�t�@�C��R��e��c�^�C�v */
36	38	static final String CONTENT_TYPE_EXCEL_ALL_CSV = CONTENT_TYPE_EXCEL_1+","+CONTENT_TYPE_EXCEL_2;
37	39
	40	/**
	41	* Excel�t�@�C��L�X�g�f�[�^�𒊏o��ader��
	42	*
	43	* <p>
	44	* POI��SSF�̎��𗘗p��xcel�̃Z��當��o�� * </p>
	45	*
	46	* @param content Excel�t�@�C��̓��̓�g��[��
	47	* @return �e�L�X�g�f�[�^�̒��o��ꂽReader
	48	* @throws IOException �X�g��[��ǂݍ��ݎ��A��Excel�t�@�C��
	49	* @see <a href="http://poi.apache.org/">Apache POI</a>
	50	*/
38	51	public Reader extract(InputStream content) throws IOException {
39	52	try {
…	…
76	89	}
77	90
	91	/**
	92	* Excel�t�@�C��̃R��e��c�^�C�v�� * @return Excel�t�@�C��̃R��e��c�^�C�v
	93	*/
78	94	public String getContentType() {
79	95	return CONTENT_TYPE_EXCEL_ALL_CSV;

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/poi/MSPowerPointExtractor.java

r43	r54
27	27	*/
28	28	public class MSPowerPointExtractor implements Extractor {
	29	/** PowerPoint�̃R��e��c�^�C�v�� */
29	30	static final String CONTENT_TYPE_POWERPOINT_1 = "application/mspowerpoint";
30
	31	/** PowerPoint�̃R��e��c�^�C�v�� */
31	32	static final String CONTENT_TYPE_POWERPOINT_2 = "application/vnd.ms-powerpoint";
32
	33	/** CSV�ŕ��L�q��PowerPoint�t�@�C��R��e��c�^�C�v */
33	34	static final String CONTENT_TYPE_POWERPOINT_ALL_CSV = CONTENT_TYPE_POWERPOINT_1
34	35	+ "," + CONTENT_TYPE_POWERPOINT_2;
35	36
	37	/**
	38	* PowerPoint�t�@�C��L�X�g�f�[�^�𒊏o��ader��
	39	*
	40	* <p>
	41	* POI��SLF�̎��𗘗p��owerPoint��當��o�� * </p>
	42	*
	43	* @param content PowerPoint�t�@�C��̓��̓�g��[��
	44	* @return �e�L�X�g�f�[�^�̒��o��ꂽReader
	45	* @throws IOException �X�g��[��ǂݍ��ݎ��A��PowerPoint�t�@�C��
	46	* @see <a href="http://poi.apache.org/">Apache POI</a>
	47	*/
36	48	public Reader extract(InputStream content) throws IOException {
37	49	final CharArrayWriter writer = new CharArrayWriter();
…	…
69	81	}
70	82
	83	/**
	84	* PowerPoint�t�@�C��̃R��e��c�^�C�v�� * @return PowerPoint�t�@�C��̃R��e��c�^�C�v
	85	*/
71	86	public String getContentType() {
72	87	return CONTENT_TYPE_POWERPOINT_ALL_CSV;

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/poi/MSWordExtractor.java

r43	r54
22	22	*/
23	23	public class MSWordExtractor implements Extractor {
	24	/** MS-Word�̃R��e��c�^�C�v�� */
24	25	static final String CONTENT_TYPE_WORD_1 = "application/msword";
	26	/** MS-Word�̃R��e��c�^�C�v�� */
25	27	static final String CONTENT_TYPE_WORD_2 = "application/vnd.ms-word";
	28	/** CSV�ŕ��L�q��MS-Word�t�@�C��R��e��c�^�C�v */
26	29	static final String CONTENT_TYPE_WORD_ALL_CSV = CONTENT_TYPE_WORD_1+","+CONTENT_TYPE_WORD_2;
27	30
	31	/**
	32	* MS-Word�t�@�C��L�X�g�f�[�^�𒊏o��ader��
	33	*
	34	* <p>
	35	* POI��WPF�̎��𗘗p��ĕ��o�� * </p>
	36	*
	37	* @param content MS-Word�t�@�C��̓��̓�g��[��
	38	* @return �e�L�X�g�f�[�^�̒��o��ꂽReader
	39	* @throws IOException �X�g��[��ǂݍ��ݎ��A��MS-Word�t�@�C��
	40	* @see <a href="http://poi.apache.org/">Apache POI</a>
	41	*/
28	42	public Reader extract(InputStream content) throws IOException {
29	43	try {
…	…
36	50	}
37	51
	52	/**
	53	* MS-Word�t�@�C��̃R��e��c�^�C�v�� * @return MS-Word�t�@�C��̃R��e��c�^�C�v
	54	*/
38	55	public String getContentType() {
39	56	return CONTENT_TYPE_WORD_ALL_CSV;

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/keyword/Auto.java

r15	r54
4	4	import java.io.*;
5	5
	6	import jp.tyzoh.rinza.folksonomy.common.AutokeywordUtil;
	7
	8	import org.apache.log4j.BasicConfigurator;
	9	import org.apache.log4j.ConsoleAppender;
	10	import org.apache.log4j.Level;
	11	import org.apache.log4j.Logger;
	12
6	13	/**
7		* �~~L�[��[�h��o~~
	14	* �Ώۃt�@�C��̃L�[��[�h��o�ƑΉ��F-IDF�l��
8	15	*
9	16	* <BR><BR>Copyright (C) 2006 Nihon Unisys, Ltd. All Rights Reserved.
10	17	*/
11	18	public class Auto {
12		Engine engine = new Engine();
13		PrintWriter out = new PrintWriter(System.out);
	19	/** �L�[��[�h��o�G��W�� */
	20	private final Engine engine = new Engine();
	21	/** �f�o�b�O�o��*/
	22	private final PrintWriter out = new PrintWriter(System.out);
14	23
	24	/**
	25	* �R�}��h��s
	26	* @param args ��s��߂̈��t�@�C��肷��
	27	*/
15	28	public static void main(String args[])
16	29	{
…	…
24	37	}
25	38
	39	BasicConfigurator.resetConfiguration();
	40	final Logger log = AutokeywordUtil.getLogger();
	41	log.addAppender(new ConsoleAppender());
	42	log.setLevel(Level.DEBUG);
	43
26	44	Auto auto = new Auto();
27	45	auto.fire(file.getAbsolutePath());
28	46	}
29	47
	48	/**
	49	* ��̃R�}��h�̎g�p��\��ƃR�}��h��s�̏I��.
	50	*
	51	*/
30	52	private static void usageAndExit()
31	53	{
…	…
34	56	}
35	57
	58	/**
	59	* ��̏��̏��
	60	*
	61	* <p>
	62	* ��̃p�b�P�[�W�Ɠ��f�B��N�g��ɂ��@�C��
	63	* (<code>autokeyword.properties</code>)��ݍ��݁A
	64	* {@link Engine}�� * </p>
	65	*/
36	66	public void init()
37	67	{
…	…
39	69	try {
40	70	properties.load(Auto.class.getClassLoader().getResourceAsStream("autokeyword.properties"));
41		engine.init(properties~~, out~~);
	71	engine.init(properties);
42	72	} catch (IOException ex) {
43	73	ex.printStackTrace(out);
…	…
45	75	}
46	76
47
	77	/**
	78	* �t�@�C��̃L�[��[�h�Ƃ��F-IDF�l�� * @param filename �L�[��[�h��o�Ώۃt�@�C��
	79	* @return �L�[��[�h��F-IDF�� {@link Result}�̃R��N�V��
	80	* Collection<Result>
	81	*/
48	82	public Collection getKeywords(String filename)
49	83	{
…	…
52	86	}
53	87
	88	/**
	89	* ��s�Ǝ�s��ʕ\��
	90	* @param filename �L�[��[�h��o�Ώۃt�@�C��
	91	*/
54	92	public void fire(String filename)
55	93	{
56	94	init();
57		Collection collection = engine.getKeywords(filename~~, new PrintWriter(System.out)~~);
	95	Collection collection = engine.getKeywords(filename);
58	96
59	97	out.println("TFIDF �J�n");
…	…
61	99	while(iterator.hasNext()) {
62	100	Result result = (Result)iterator.next();
63		out.println(result.~~info~~());
	101	out.println(result.toString());
64	102	}
65	103	out.println("TFIDF �I��");

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/keyword/Engine.java

r47	r54
4	4	import java.io.*;
5	5
	6	import org.apache.log4j.Logger;
	7
6	8	import jp.tyzoh.rinza.folksonomy.analyze.*;
	9	import jp.tyzoh.rinza.folksonomy.common.AutokeywordUtil;
7	10	import jp.tyzoh.rinza.folksonomy.search.*;
8	11
…	…
33	36	* @return ��
34	37	*/
35		public boolean init(Properties properties~~, PrintWriter out~~)
	38	public boolean init(Properties properties)
36	39	{
37	40	String analyzerName = properties.getProperty(ANALYZER_CLASS_NAME, ANALYZER_CLASS_NAME_DEFAULT);
38	41	try {
39	42	analyzer = (Analyzer)Class.forName(analyzerName).newInstance();
40		if (!analyzer.init(properties~~, out~~)) {
41		~~if (out != null) out.println(analyzerName + " is not initialized~~");
	43	if (!analyzer.init(properties)) {
	44	AutokeywordUtil.getLogger().fatal("��̓N��X"+analyzerName+"�̏��s��܂��");
42	45	return false;
43	46	}
44	47	} catch(Exception ex) {
45		~~if (out != null) out.println(analyzerName + " is not supported."~~);
	48	AutokeywordUtil.getLogger().fatal("��̓N��X"+analyzerName+"�̏��s��܂��", ex);
46	49	return false;
47	50	}
…	…
50	53	try {
51	54	searcher = (Searcher)Class.forName(searcherName).newInstance();
52		if (!searcher.init(properties, out)) {
53		String message = searcherName + " is not initialized";
54		if (out != null) out.println(message);
55		System.err.println(message);
	55	if (!searcher.init(properties)) {
	56	AutokeywordUtil.getLogger().fatal("�S��X"+searcherName+"�̏��s��܂��");
56	57	return false;
57	58	}
58	59	} catch(Exception ex) {
59		String message = searcherName + " is not initialized";
60		if (out != null) out.println(message);
61		System.err.println(message);
	60	AutokeywordUtil.getLogger().fatal("�S��X"+searcherName+"�̏��s��܂��", ex);
62	61	return false;
63	62	}
…	…
68	67	/**
69	68	* �L�[��[�h�R��N�V��̈ꗗ�\��
70		* @param out �f�o�b�O�o��
71	69	* @param keywords �L�[��[�h�R��N�V�� (Collection<Keyword>)
72	70	*/
73		void debug(PrintWriter out, Collection keywords)
74		{
75		if (out != null) {
76		out.println("�L�[��[�h�ꗗ �J�n");
	71	void debug(Collection keywords)
	72	{
	73	final Logger log = AutokeywordUtil.getLogger();
	74	if(log.isDebugEnabled()) {
	75	log.debug("�L�[��[�h�ꗗ �J�n");
77	76	for(Iterator iterator = keywords.iterator(); iterator.hasNext(); ) {
78	77	final Keyword keyword = (Keyword)iterator.next();
79		out.println(keyword.info());
80		}
81		out.println("�L�[��[�h�ꗗ �I��");
82		}
83		}
84
85		/**
86		* �L�[��[�h��o��(�f�o�b�O�o�͂Ȃ�)
87		* @param filename �L�[��[�h��o�̑Ώۃt�@�C��
88		* @return �L�[��[�h�̃R��N�V�� (Collection<Keyword>)
89		* @see #getKeywords(String, PrintWriter) �f�o�b�O�o�͂Ȃ��Ōďo�B
90		*/
91		public Collection getKeywords(String filename)
92		{
93		return getKeywords(filename, null);
94		}
95
	78	log.debug(keyword.toString());
	79	}
	80	log.debug("�L�[��[�h�ꗗ �I��");
	81	}
	82	}
	83
96	84	/**
97	85	* �L�[��[�h��o��
…	…
104	92	*
105	93	* @param filename �L�[��[�h��o�̑Ώۃt�@�C��
106		* @param out �f�o�b�O�o��
107		* @return �L�[��[�h�̃R��N�V�� (Collection<Keyword>)
108		*/
109		public Collection getKeywords(String filename, PrintWriter out)
	94	* @return �L�[��[�h�̃R��N�V�� (Collection<Result>)
	95	*/
	96	public Collection getKeywords(String filename)
110	97	{
111	98	ArrayList results = new ArrayList();
…	…
115	102	File[] fileArray = file.listFiles();
116	103	int fileCount = fileArray.length;
117		if (out != null) out.println("N " + fileCount);
118
119		final Collection localKeywords = getLocalKeywords(filename, out);
	104	final Logger log = AutokeywordUtil.getLogger();
	105	if(log.isDebugEnabled()) {
	106	log.debug("N " + fileCount);
	107	}
	108
	109	final Collection localKeywords = getLocalKeywords(filename);
120	110	for(Iterator iterator = localKeywords.iterator(); iterator.hasNext();) {
121	111	final Keyword keyword = (Keyword)iterator.next();
…	…
134	124
135	125	/**
136		* TI-IDF�l�̌v�Z
	126	* TF-IDF�l�̌v�Z
137	127	* @param keyword �L�[��[�h��
138	128	* @param fileCount �W�{��c�̐�
139		* @return
	129	* @return TF-IDF�l
140	130	*/
141	131	private double calclateTfidf(Keyword keyword, int fileCount) {
…	…
152	142	return keyword.getCount()*Math.log((double)fileCount/df);
153	143	}
	144
154	145	/**
155	146	* �L�[��[�h�̒��o
…	…
159	150	* </p>
160	151	* @param filename �Ώۃt�@�C��
161		* @param out �f�o�b�O�o��
162	152	* @return Collection<Keyword>
163	153	*/
164		public Collection getLocalKeywords(String filename~~, PrintWriter out~~)
	154	public Collection getLocalKeywords(String filename)
165	155	{
166	156	//Analyzer�ɃL�[��[�h�̐��K��߂�
167		return analyzer.parse(filename~~, null~~);
	157	return analyzer.parse(filename);
168	158	}
169	159

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/keyword/Hidden.java

r43	r54
1	1	package jp.tyzoh.rinza.folksonomy.keyword;
2	2
3		import java.util.*;
4		import java.io.*;
	3	import java.io.File;
	4	import java.io.IOException;
	5	import java.io.PrintWriter;
	6	import java.util.Collection;
	7	import java.util.Iterator;
	8	import java.util.Properties;
	9
	10	import jp.tyzoh.rinza.folksonomy.common.AutokeywordUtil;
	11
	12	import org.apache.log4j.BasicConfigurator;
	13	import org.apache.log4j.ConsoleAppender;
	14	import org.apache.log4j.Level;
	15	import org.apache.log4j.Logger;
5	16
6	17	/**
7		* ~~��~~
	18	*
8	19	*
9	20	* <BR><BR>Copyright (C) 2007 Nihon Unisys, Ltd. All Rights Reserved.
…	…
23	34	}
24	35
	36	BasicConfigurator.resetConfiguration();
	37	final Logger log = AutokeywordUtil.getLogger();
	38	log.addAppender(new ConsoleAppender());
	39	log.setLevel(Level.DEBUG);
	40
25	41	Hidden hidden = new Hidden();
26	42	hidden.fire(file.getAbsolutePath());
…	…
38	54	try {
39	55	properties.load(Auto.class.getClassLoader().getResourceAsStream("autokeyword.properties"));
40		init(properties~~, out~~);
	56	init(properties);
41	57	} catch (IOException ex) {
42		ex.printStackTrace(~~out~~);
	58	ex.printStackTrace();
43	59	}
44	60	}
…	…
47	63	{
48	64	init();
49		Collection collection = getKeywords(filename~~, new PrintWriter(System.out)~~);
	65	Collection collection = getKeywords(filename);
50	66
51	67	out.println("TFIDF �J�n");
…	…
53	69	while(iterator.hasNext()) {
54	70	Result result = (Result)iterator.next();
55		out.println(result.~~info~~());
	71	out.println(result.toString());
56	72	}
57	73	out.println("TFIDF �I��");
…	…
65	81	}
66	82
67		public Collection getLocalKeywords(String filename~~, PrintWriter out~~)
	83	public Collection getLocalKeywords(String filename)
68	84	{
69		return super.getLocalKeywords(filename~~, out~~);
	85	return super.getLocalKeywords(filename);
70	86	}
71	87	}

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/keyword/Result.java

r47	r54
17	17	private final double tfidf;
18	18
19		public Result(Keyword keyword, double tidf)
	19	/**
	20	* �L�[��[�h�{TFIDF�l�̒lBean�̍쐬.
	21	* <p>
	22	* ��̃C��X�^��X��mmutable�ł��
	23	*
	24	* </p>
	25	* @param keyword �L�[��[�h
	26	* @param tfidf TFIDF�l
	27	*/
	28	public Result(Keyword keyword, double tfidf)
20	29	{
21	30	this.keyword = keyword;
22		this.tfidf = tidf;
	31	this.tfidf = tfidf;
23	32	}
24	33
	34	/**
	35	* TF-IDF�l�̎擾.
	36	*
	37	* @return TF-IDF�l
	38	*/
25	39	public double getTFIDF()
26	40	{
…	…
28	42	}
29	43
	44	/**
	45	* �L�[��[�h��擾
	46	* @return �L�[��[�h�� * @see Keyword#getWord() ��͂ł̃L�[��[�h��
	47	*/
30	48	public String getWord()
31	49	{
…	…
34	52
35	53	/**
	54	* ��̃C��X�^��X�̕��̕��擾��
	55	*
	56	* @return ��̃C��X�^��X�̕��
	57	* @see Object#toString()
	58	*/
	59	public String toString()
	60	{
	61	return new StringBuffer().append(getTFIDF()).append("\t").append(keyword.toString()).toString();
	62	}
	63
	64	/**
36	65	* �\�[�g�̂��߂̎��R��\�b�h
37	66	*
38	67	* <p>
39		* tfidf�l�̑傫��Ƀ\�[�g�� * </p>
	68	* ��̃C��X�^��X�̎��R��͈ȉ��̂Ƃ�� * <ol>
	69	* <li>tfidf(�~��)
	70	* <li>word(��)
	71	* </ol>
	72	* </p>
40	73	*
41		* TODO �\�[�g��ʂɓ��-IDF�l��̃L�[��[�h�ł̃\�[�g��
42	74	*
43	75	* @param obj ��Ώ�
…	…
53	85	return -1;
54	86	} else if (result.getTFIDF() == getTFIDF()) {
55		return 0;
	87	return getWord().compareTo(result.getWord());
	88	//return 0;
56	89	} else {
57	90	return 1;
58	91	}
59		}
60
61		~~public String info()~~
62		{
63		~~return getTFIDF() + "\t" + keyword.info();~~
64	92	}
65	93	}

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/search/LuceneJa.java

r47	r54
3	3	import java.io.File;
4	4	import java.io.IOException;
5		~~import java.io.PrintWriter;~~
6	5	import java.io.Reader;
7	6	import java.util.ArrayList;
…	…
13	12	import java.util.Properties;
14	13
	14	import jp.tyzoh.rinza.folksonomy.common.AutokeywordUtil;
15	15	import jp.tyzoh.rinza.folksonomy.extractor.ContentsExtractorLookuper;
16	16
…	…
38	38	*/
39	39	public class LuceneJa implements Searcher {
	40	/** Lucene�̑S��Ώۂ̃f�[�^�i�[�L�[ */
	41	private static final String LUCENE_SEARCH_CONTENTS = "contents";
	42	/** Lucene�̍ŏI�X�V��Ԃ̃f�[�^�i�[�L�[ */
	43	private static final String LUCENE_MODIFIED = "modified";
	44	/** Lucene�̉�Ώۃt�@�C��̃p�X��f�[�^�̊i�[�L�[ */
	45	private static final String LUCENE_PATH_INFO = "path";
	46
40	47	//
41		private static final String LUCENE_SEARCH_CONTENTS = "contents";
42
43		private static final String LUCENE_MODIFIED = "modified";
44
45		private static final String LUCENE_PATH_INFO = "path";
46
47		//
	48	/** Lucene�̃C��f�b�N�X��̊i�[�f�B��N�g�� */
48	49	private String indexDirectory = DEFAULT_LUCENE_INDEX_DIRECTORY;
49
	50	/** Lucene�̐ݒ��@�C��(Sen�ݒ�*/
50	51	private String command = DEFAULT_COMMAND;
51
	52	/** Lucene�̉�Ώۃt�@�C��f�B��N�g�� */
52	53	private String fileDirectory = DEFAULT_FILESTORE_DIRECTORY;
53
	54	/** ��͗p�̃C��^�[�t�F�[�X */
54	55	private Analyzer analyzer = null;
55
	56	/** Lucene�̌��}�l�[�W��N��X */
56	57	private IndexSearcherManager searcherManager = null;
57
	58	/** �e�L�X�g��o��p�̎��擾�p�N��X */
58	59	private ContentsExtractorLookuper lookuper = null;
59
	60	/** �t�@�C��i�[�f�B��N�g��̋K�� */
60	61	private static final String DEFAULT_FILESTORE_DIRECTORY = "/var/lib/tPod/files";
61
	62	/** Lucene�̐ݒ��@�C��̋K��(Sen�ݒ�*/
62	63	private static final String DEFAULT_COMMAND = "/var/lib/tPod/sen/conf/sen.xml";
63
	64	/** Lucene�̃C��f�b�N�X��̊i�[�f�B��N�g��̋K�� s*/
64	65	private static final String DEFAULT_LUCENE_INDEX_DIRECTORY = "/var/lib/tPod/index";
65	66
…	…
67	68	* ��
68	69	*
69		* @param properties �ݒ� * @param out �f�o�b�O�o��
70		* @return ��s�̐��
71		*/
72		public boolean init(Properties properties, PrintWriter out) {
	70	* @param properties �ݒ� * @return ��s�̐��
	71	*/
	72	public boolean init(Properties properties) {
73	73	fileDirectory = properties.getProperty(CONTENTS_DIRECTORY, //
74	74	properties.getProperty(CONTENTS_DIRECTORY2, fileDirectory));
…	…
79	79	searcherManager = new IndexSearcherManager(indexDirectory);
80	80	lookuper = new ContentsExtractorLookuper();
	81
	82	AutokeywordUtil.getLogger().info("Lucene(ja)��܂��");
81	83	return true;
82	84	}
83	85
	86	/**
	87	* Lucene�̏I��
	88	*/
84	89	public void destroy() {
85
86		}
87
	90	}
	91
	92	/**
	93	* Lucene�̉�Ώۃt�@�C��̔z�u�ʒu.
	94	* @return �Ώۃf�B��N�g��
	95	*/
88	96	public File getTargetDirectory() {
89	97	return new File(fileDirectory);
…	…
94	102	*
95	103	* @param keywordList �L�[��[�h��X�g(List<String>)
96		* @param out �f�o�b�O�o��
97	104	* @return ��ʂ̃t�@�C��̃R��N�V�� Collection<String>
98	105	*/
99		public Collection /* <String> */search(List keywordList, ~~PrintWriter out,~~ int offset, int limit) {
	106	public Collection /* <String> */search(List keywordList, int offset, int limit) {
100	107	final ArrayList al = new ArrayList();
101	108	IndexSearcherPool isp = null;
…	…
121	128	}
122	129	} catch (ParseException e) {
123		if (out != null) {
124		e.printStackTrace(out);
125		}
	130	AutokeywordUtil.getLogger().warn("Lucene��ɉ�G��[��܂��", e);
126	131	} catch (IOException e) {
127		if (out != null) {
128		e.printStackTrace(out);
129		}
	132	AutokeywordUtil.getLogger().warn("Lucene��O�G��[��܂��", e);
130	133	}
131	134	}
…	…
139	142	* �S��
140	143	*
141		* @see #search(List, PrintWriter, int, int) �J�n�ʒu/�\��0�Ƃ��đS��\��
142		*/
143		public Collection search(List keywordList, PrintWriter out) {
144		return search(keywordList, out, 0, 0);
	144	* @param keywordList �S��̌��X�g
	145	* @see #search(List, int, int) �J�n�ʒu/�\��0�Ƃ��đS��\��
	146	*/
	147	public Collection search(List keywordList) {
	148	return search(keywordList, 0, 0);
145	149	}
146	150
147	151	/**
148	152	* �S��s��ꍇ�̊Y��
	153	* @param keywordList �S��̌��X�g
	154	* @return �Ή��t�@�C��̑��
149	155	*/
150	156	public int searchCount(List keywordList) {
…	…
177	183	* �L�[��[�h��󔒋��Ȃ�
178	184	*
179		* @param keywordList
180		* @return
181		*/
182		private String parseKeyword(List keywordList) {
	185	* @param keywordList ��ΏۃL�[��[�h�� * @return��󔒂łȂ�� */
	186	private static String parseKeyword(List keywordList) {
183	187	final StringBuffer buf = new StringBuffer();
184	188	for (int i = 0; i < keywordList.size(); i++) {
…	…
194	198	* �S��C��f�b�N�X��.
195	199	*
196		* @param out
197		* �f�o�b�O�o�͗p
198		*/
199		public Collection indexing(PrintWriter out) {
	200	*/
	201	public void indexing() {
200	202	synchronized (this) {
201	203	final boolean isCreate = (!IndexReader.indexExists(indexDirectory));
…	…
230	232	reader.close();
231	233	} catch (IOException e) {
232		~~e.printStackTrace(~~);
	234	AutokeywordUtil.getLogger().warn("Lucene�C��f�b�N�X�Đ��ɃG��[��܂��", e);
233	235	}
234	236	}
…	…
253	255	searcherManager.updateSearcher();
254	256	}
255		~~return null;~~
256	257	}
257	258
258	259	/**
259		* �t�@�C��L�X�g��𒊏o��
	260	* �t�@�C��L�X�g��𒊏o��S��t�@�C��
260	261	* @param parent �e�f�B��N�g��
261	262	* @param name �t�@�C��
262		* @return
	263	* @return �e�L�X�g��o��f�[�^��b�g��ۃf�[�^
263	264	*/
264	265	private Document createDocument(File parent, String name) {
…	…
282	283	}
283	284
	285	/**
	286	* �ŏI�X�V��Ԃ̕��쐬��
	287	* @param file �Ώۃt�@�C��
	288	* @return Lucene�̓��`��p�̎�� * @see DateField#timeToString(long)
	289	*/
284	290	private static String findLastModified(File file) {
285	291	return DateField.timeToString(file.lastModified());
…	…
305	311	private IndexSearcherPool searcherPool = null;
306	312
	313	/**
	314	* �C��f�b�N�X�ł̌��}�l�[�W��.
	315	*
	316	* @param indexDirectory ��ۂ̃C��f�b�N�X�f�B��N�g��
	317	*/
307	318	public IndexSearcherManager(String indexDirectory) {
308	319	this.indexDirectory = indexDirectory;
…	…
310	321	}
311	322
312		//
	323	/**
	324	* ��̎؎� * @return ��N��X.
	325	*/
313	326	private IndexSearcherPool rentSearcher() {
314	327	synchronized (this) {
…	…
319	332	}
320	333
	334	/**
	335	* ��̕ԋp
	336	* @param isp �ԋp��@link IndexSearcherPool}
	337	*/
321	338	private void returnSearcher(IndexSearcherPool isp) {
322	339	IndexSearcher is = null;
…	…
330	347	}
331	348
	349	/**
	350	* �t�@�C��ǉ�Ȃǂł̌��̍X�V��.
	351	*
	352	*/
332	353	private void updateSearcher() {
333	354	IndexSearcher is = null;
…	…
344	365	}
345	366
	367	/**
	368	* ��̏��
	369	* @param indexDirectory ��f�b�N�X�f�B��N�g��
	370	* @return ��.
	371	*/
346	372	private static IndexSearcherPool openSearcher(String indexDirectory) {
347	373	IndexSearcher is = null;
…	…
353	379	}
354	380
	381	/**
	382	* ��̏I��
	383	* @param is ��
	384	*/
355	385	private static void closeSearcher(IndexSearcher is) {
356	386	if (is != null) {
…	…
363	393	}
364	394
365		//
	395	/**
	396	* �T�[�`�p�̃I�u�W�F�N�g�̃v�[��N��X.
	397	*
	398	*
	399	* <BR>
	400	* <BR>
	401	* Copyright (C) 2007 Nihon Unisys, Ltd. All Rights Reserved.
	402	*/
366	403	private static class IndexSearcherPool {
367	404	private final IndexSearcher searcher;
368
	405	/** ��̌��͍X�V�҂��H */
369	406	private boolean isUpdated;
370
	407	/** ��s��X��b�h�̏� ��s��̃X��b�h��o�^�� */
371	408	private final HashSet /* <Thread> */runningSet = new HashSet();
372
	409	/**
	410	* Lucene�̌��肵�ď��
	411	* @param searcher Lucene�C��f�b�N�X�T�[�`
	412	*/
373	413	public IndexSearcherPool(IndexSearcher searcher) {
374	414	this.searcher = searcher;
375	415	}
376	416
377		//
	417	/**
	418	* �i�[��Ă��錟��̎擾
	419	* @return �i�[��Ă��錟��
	420	*/
378	421	public IndexSearcher getSearcher() {
379	422	return searcher;
380	423	}
381	424
382		//
	425	/**
	426	* ��̌��͍X�V��ꂽ��H
	427	* @param isUpdated
	428	*/
383	429	public void setUpdated(boolean isUpdated) {
384	430	this.isUpdated = isUpdated;
385	431	}
386	432
	433	/**
	434	* ��̌��̗��p��n��
	435	*
	436	*/
387	437	public void lock() {
388	438	runningSet.add(Thread.currentThread());
389	439	}
390	440
	441	/**
	442	* ��̌��̗��p��߂�
	443	*
	444	*/
391	445	public void unlock() {
392	446	runningSet.remove(Thread.currentThread());
393	447	}
394	448
	449	/**
	450	* ��̃v�[��͕��v��āA��邩�H
	451	* @return ��v��g�p��̂ŕ��trye
	452	*/
395	453	public boolean needClose() {
396	454	return isUpdated && runningSet.isEmpty();

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/search/Namazu.java

r47	r54
34	34	return new File(fileDirectory);
35	35	}
36
37		public boolean init(Properties properties, PrintWriter out)
	36	/**
	37	* Namazu��
	38	*
	39	* @param properties �ݒ��e
	40	* @param out �f�o�b�O�o��
	41	*/
	42	public boolean init(Properties properties)
38	43	{
39	44	encodeType = properties.getProperty(ENCODE_TYPE, //
…	…
45	50	index = properties.getProperty(INDEX_COMMAND, "c:\\namazu\\bin\\mknmz.bat \"" + fileDirectory + "\" -O");
46	51	indexDirectory = properties.getProperty(INDEX_DIRECTORY, indexDirectory);
47
	52	AutokeywordUtil.getLogger().info("Namazu ��܂��");
48	53	return true;
49	54	}
50
51		public Collection search(List keywordList, PrintWriter out) {
52		return search(keywordList, out, 0,0);
	55	/**
	56	* Namazu�o�R�ł̌��.
	57	*
	58	* @param keywordList ��[��[�h�� */
	59	public Collection search(List keywordList) {
	60	return search(keywordList, 0,0);
53	61	}
54		public Collection search(List keywordList, PrintWriter out, int offset, int limit)
	62
	63	/**
	64	* Namazu�o�R�ł̌��.
	65	*
	66	* @param keywordList ��[��[�h�� * @param offset ��ʊJ�n�ʒu
	67	* @param limit ��ʏI��
	68	*/
	69	public Collection search(List keywordList, int offset, int limit)
55	70	{
56	71	StringBuffer execBuf = new StringBuffer();
…	…
69	84	execBuf.append(" ").append(indexDirectory);
70	85
71		return ~~Util.exec(execBuf.toString(), encodeType, out~~);
	86	return AutokeywordUtil.exec(execBuf.toString(), encodeType);
72	87	}
73	88
	89	/**
	90	* ��őΉ��錟��ʐ��̎擾.
	91	* @param �� * @return ��ʏW
	92	*/
74	93	public int searchCount(List keywordList) {
75	94	StringBuffer execBuf = new StringBuffer();
…	…
83	102	execBuf.append(" ").append(indexDirectory);
84	103
85		int count = -1;
86		final Collection c = ~~Util.exec(execBuf.toString(), encodeType, null~~);
	104	int count = 0;
	105	final Collection c = AutokeywordUtil.exec(execBuf.toString(), encodeType);
87	106	if(c.size()==1) {
88	107	final String s = (String) c.iterator().next();
…	…
95	114	}
96	115
97		public Collection indexing(PrintWriter out)
	116	/**
	117	* �C��f�b�N�X�쐬��
	118	* @param out �C��f�b�N�X�쐬��̃f�o�b�O�o��
	119	*/
	120	public void indexing()
98	121	{
99	122	String command = index + " " + indexDirectory;
100	123
101		~~return Util.exec(command, encodeType, out~~);
	124	AutokeywordUtil.exec(command, encodeType);
102	125	}
103	126

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/search/Searcher.java

r47	r54
29	29	/**
30	30	* �� * @param properties �ݒ� * @param out �f�o�b�O�o��
31		* @return
	31	* @return ��̐��
32	32	*/
33		public boolean init(Properties properties, PrintWriter out);
	33	public boolean init(Properties properties);
	34
34	35	/**
35		* �Ώۃ~~f�B��N�g��~~
36		* @return
	36	* �Ώۃt�@�C��̂��B��N�g��̎擾
	37	* @return �f�B��N�g��
37	38	*/
38	39	public File getTargetDirectory();
…	…
41	42	* ��
42	43	* @param keywordList �L�[��[�h��X�g
43		* @param out �f�o�b�O�o��
44	44	* @return Collection<String>
45	45	*/
46		public Collection search(List keywordList~~, PrintWriter out~~);
	46	public Collection search(List keywordList);
47	47
48	48	/**
49	49	* ��(��)
50	50	* @param keywordList �L�[��[�h��X�g
51		* @param out �f�o�b�O�o��
52	51	* @param offset ��ʊJ�n�ʒu
53	52	* @param limit ��ʍő吔
54	53	* @return Collection<String>
55	54	*/
56		public Collection search(List keywordList, ~~PrintWriter out,~~ int offset, int limit);
	55	public Collection search(List keywordList, int offset, int limit);
57	56
58	57	/**
…	…
70	69	* @param out �f�o�b�O�o��
71	70	*/
72		public ~~Collection indexing(PrintWriter out~~);
	71	public void indexing();
73	72
74	73	/**

Context Navigation

凡例:

AutoKeyword/trunk/README

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/analyze/Analyzer.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/analyze/Keyword.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/analyze/KeywordFilter.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/analyze/Mecab.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/analyze/Sen.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/common/AutokeywordUtil.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/ContentsExtractorLookuper.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/DefaultExtractor.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/Extractor.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/pdfbox/PDFExtractor.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/poi/MSExcelExtractor.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/poi/MSPowerPointExtractor.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/extractor/poi/MSWordExtractor.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/keyword/Auto.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/keyword/Engine.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/keyword/Hidden.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/keyword/Result.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/search/LuceneJa.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/search/Namazu.java

AutoKeyword/trunk/src/jp/tyzoh/rinza/folksonomy/search/Searcher.java

異なるフォーマットでダウンロード: