我跟着https://gist.github.com/xrstf/b48a970098a8e76943b9整合了nutch和弹性搜索。一切正常，数据存储在Hbase＆＃39;网页＆＃39;表但我无法在弹性搜索中获取数据。我想知道如何在弹性搜索中获取数据。
答案 0 :(得分：0)
下面是我的代码
package com.process;
/*
import package will be here
*/
public class HbaseToElastic extends Configured implements
org.apache.hadoop.util.Tool {
static class Mapper extends TableMapper<Text, IndexWritable> {
public static String CLUSTER;
public static String SEARCH_HOST;
public static String SEARCH_PORT;
public static String SEARCH_INDEX_NAME;
public static String SEARCHtYPE;
public static int BULKSIZE;
public static String TABLENAME;
public static String FAMILY;
private static List<String> SPORTS_KEYWORDS;
private static List<String> BUSINESS_KEYWORDS;
private static List<String> GOSSIP_KEYWORDS;
private static List<String> CRIME_KEYWORDS;
private static Map<String, Map<String, String>> STATE_MAP = new HashMap<String, Map<String, String>>();
private static Map<String, String> CITY_MAP = new HashMap<String, String>();
private static Mapper mapper = new Mapper();
static {
try {
System.out.println("done1");
DetectorFactory.loadProfile("./profiles");
System.out.println("done2");
} catch (final LangDetectException e) {
System.out.println("done3");
e.printStackTrace();
}
}
Configuration hbaseConf = null;
HTable table = null;
List<Put> hbasePutErrorList = new ArrayList<Put>();
/**
* Clean up the hbase table object
*/
@Override
protected void cleanup(final Context context) throws IOException,
InterruptedException {
super.cleanup(context);
table.put(hbasePutErrorList);
table.close();
hbasePutErrorList.clear();
}
/**
* Initialize various variables
*/
@Override
protected void setup(
final org.apache.hadoop.mapreduce.Mapper<ImmutableBytesWritable, Result, Text, IndexWritable>.Context context)
throws IOException, InterruptedException {
final Configuration conf = context.getConfiguration();
CLUSTER = conf.get("cluster");
SEARCH_HOST = conf.get("search_host");
SEARCH_PORT = conf.get("search_port");
SEARCH_INDEX_NAME = conf.get("search_index_name");
SEARCHtYPE = conf.get("search_type");
BULKSIZE = conf.getInt("search_bulk_size", 500);
TABLENAME = conf.get("table_name");
FAMILY = conf.get("family");
hbaseConf = HBaseConfiguration.create();
hbaseConf.set("hbase.zookeeper.quorum",
conf.get("hbase.zookeeper.quorum"));
hbaseConf.set("hbase.zookeeper.property.clientPort",
conf.get("hbase.zookeeper.property.clientPort"));
hbaseConf.set("hbase.rpc.timeout", conf.get("hbase.rpc.timeout"));
hbaseConf.set("hbase.regionserver.lease.period",
conf.get("hbase.regionserver.lease.period"));
hbaseConf.set("hbase.master", conf.get("hbase.master"));
table = new HTable(hbaseConf, conf.get("table_name"));
SPORTS_KEYWORDS = new ArrayList<String>();
BUSINESS_KEYWORDS = new ArrayList<String>();
GOSSIP_KEYWORDS = new ArrayList<String>();
CRIME_KEYWORDS = new ArrayList<String>();
String keywrods = conf.get("sportskeywords");
String[] keyarr = keywrods.split(",");
for (final String key : keyarr) {
SPORTS_KEYWORDS.add(key.trim());
}
keywrods = conf.get("businesskeywords");
keyarr = keywrods.split(",");
for (final String key : keyarr) {
BUSINESS_KEYWORDS.add(key.trim());
}
keywrods = conf.get("gossipkeywords");
keyarr = keywrods.split(",");
for (final String key : keyarr) {
GOSSIP_KEYWORDS.add(key.trim());
}
keywrods = conf.get("crimekeywords");
keyarr = keywrods.split(",");
for (final String key : keyarr) {
CRIME_KEYWORDS.add(key.trim());
}
final String stateMap = conf.get("statemap");
final Gson g = new Gson();
STATE_MAP = g.fromJson(stateMap, Map.class);
}
/**
* map function
*/
@Override
public void map(final ImmutableBytesWritable row, final Result result,
final Context context) throws IOException, InterruptedException {
try {
final byte b = 0;
int deleteFlag = 0;
final String keyString = Bytes.toString(row.get());
final Map<String, Object> mapobject = new HashMap<String, Object>();
for (final KeyValue kv : result.raw()) {
final String key = (new String(kv.getQualifier()));
final String value = (new String(kv.getValue()));
mapobject.put(key, value);
}
final Gson g = new Gson();
if (checkValidType(mapobject)) {
refineMetaTags(mapobject);
if (refineDescription(mapobject)) {
assignCity(mapobject);
if (checkTitleImage(mapobject)) {
if (setLang(mapobject)) {
setCorrectCategory(mapobject);
correctDuplicateTitle(mapobject);
final String json = g.toJson(mapobject);
context.write(new Text(keyString),
new IndexWritable(json, b));
deleteFlag = 1;
}
}
}
}
if (deleteFlag == 0) {
final Put put = new Put(Bytes.toBytes(keyString));
put.add(Bytes.toBytes("cf"), Bytes.toBytes("ErrorFlag"),
Bytes.toBytes("1"));
hbasePutErrorList.add(put);
}
} catch (final Exception e) {
e.printStackTrace();
}
}
/**
* Remove duplicate statement in the title
*
* @param mapobject
*/
private void correctDuplicateTitle(final Map<String, Object> mapobject) {
final String duplicateTitle = mapobject.get("title").toString();
final String stripedTitleArr[] = duplicateTitle.split(" ", 4);
if (stripedTitleArr.length == 4) {
final String subString = stripedTitleArr[0] + " "
+ stripedTitleArr[1] + " " + stripedTitleArr[2];
if (stripedTitleArr[3].contains(subString)) {
mapobject.put("title", duplicateTitle
.substring(duplicateTitle.indexOf(subString,
subString.length() - 1)));
mapobject.put("title", stripedTitleArr[3]
.substring(stripedTitleArr[3].indexOf(subString)));
}
}
}
/**
* Set category based on the various category specific keyword
*
* @param mapobject
*/
private void setCorrectCategory(final Map<String, Object> mapobject) {
final String url = mapobject.get("url") + "";
final String cat = mapobject.get("tags") + "";
if ("sports".equalsIgnoreCase(cat)
|| "cricket".equalsIgnoreCase(cat)) {
if (!(url.toLowerCase().contains("sport")
|| url.toLowerCase().contains("खेल")
|| url.toLowerCase().contains("cric") || url
.toLowerCase().contains("क्रिकेट"))) {
final String desc = mapobject.get("description").toString();
boolean isSports = false;
int count = 0;
for (final String keyword : SPORTS_KEYWORDS) {
if (desc.contains(keyword)) {
count++;
}
}
if (count > 1) {
isSports = true;
}
if (!isSports) {
mapobject.put("tags", "national");
}
if (isSports
&& (desc.contains("क्रिकेट")
|| url.toLowerCase().contains("cric")
|| desc.contains("टॉस")
|| desc.contains("वनडे") || desc
.contains("बल्लेबाज"))) {
mapobject.put("tags", "cricket");
}
}
} else if ("business".equalsIgnoreCase(cat)) {
if ((url.toLowerCase().contains("sport") || url.toLowerCase()
.contains("खेल"))) {
mapobject.put("tags", "sports");
} else if (url.toLowerCase().contains("cric")
|| url.toLowerCase().contains("क्रिकेट")) {
mapobject.put("tags", "cricket");
} else if (!(url.toLowerCase().contains("busines")
|| url.toLowerCase().contains("व्यापार")
|| url.toLowerCase().contains("economy")
|| url.toLowerCase().contains("finance")
|| url.toLowerCase().contains("बिजनेस")
|| url.toLowerCase().contains("market")
|| url.toLowerCase().contains("karobar") || url
.contains("कारोबार"))) {
final String desc = mapobject.get("description").toString();
int count = 0;
for (final String keyword : BUSINESS_KEYWORDS) {
if (desc.contains(keyword)) {
count++;
}
}
if (count < 2) {
mapobject.put("tags", "national");
}
}
} else if ("gossip".equalsIgnoreCase(cat)) {
if ((url.toLowerCase().contains("sport") || url.toLowerCase()
.contains("खेल"))) {
mapobject.put("tags", "sports");
} else if (url.toLowerCase().contains("cric")
|| url.toLowerCase().contains("क्रिकेट")) {
mapobject.put("tags", "cricket");
} else if (url.toLowerCase().contains("busines")) {
mapobject.put("tags", "business");
} else if (!(url.toLowerCase().contains("masala")
|| url.toLowerCase().contains("gossip")
|| url.toLowerCase().contains("gupshup") || url
.toLowerCase().contains("garam"))) {
final String desc = mapobject.get("description").toString();
int count = 0;
for (final String keyword : GOSSIP_KEYWORDS) {
if (desc.contains(keyword)) {
count++;
}
}
if (count < 2) {
mapobject.put("tags", "national");
}
}
} else if ("crime".equalsIgnoreCase(cat)) {
if ((url.toLowerCase().contains("sport") || url.toLowerCase()
.contains("खेल"))) {
mapobject.put("tags", "sports");
} else if (url.toLowerCase().contains("cric")
|| url.toLowerCase().contains("क्रिकेट")) {
mapobject.put("tags", "cricket");
} else if (url.toLowerCase().contains("busines")) {
mapobject.put("tags", "business");
} else if (!(url.toLowerCase().contains("crime")
|| url.toLowerCase().contains("terrorist")
|| url.toLowerCase().contains("abuse")
|| url.toLowerCase().contains("forgery")
|| url.toLowerCase().contains("assault")
|| url.toLowerCase().contains("violence")
|| url.toLowerCase().contains("rape")
|| url.toLowerCase().contains("teasing")
|| url.toLowerCase().contains("molestation")
|| url.toLowerCase().contains("scandal") || url
.toLowerCase().contains("murder"))) {
final String desc = mapobject.get("description").toString();
int count = 0;
for (final String keyword : CRIME_KEYWORDS) {
if (desc.contains(keyword)) {
count++;
}
}
if (count < 2) {
mapobject.put("tags", "national");
}
}
} else if (cat != null && cat.startsWith("local")) {
}
}
/**
* Check valid type of the HTML pages
*
* @param mapobject
* @return
*/
private boolean checkValidType(final Map<String, Object> mapobject) {
if (mapobject.containsKey("type")
&& !(mapobject.get("type").toString().contains("image") || mapobject
.get("type").toString().contains("rss"))) {
return true;
}
return false;
}
/**
* refine the description according to its length and must starting with
* english and it the description is not present get the description
* from the metatags description
*
* @param mapobject
* @return {@link Boolean}
*/
private boolean refineDescription(final Map<String, Object> mapobject) {
if (mapobject.containsKey("description")
&& mapobject.get("description").toString().length() > 75
&& !mapobject.get("description").toString().contains(";}")
&& !mapobject.get("description").toString()
.contains("<cite>")
&& !mapobject.get("description").toString()
.contains("href=")
&& !mapobject.get("description").toString()
.contains("All rights reserved")) {
return true;
} else if (mapobject.containsKey("metatag.description")
&& mapobject.get("metatag.description").toString().length() > 75
&& !mapobject.get("metatag.description").toString()
.contains(";}")
&& !mapobject.get("metatag.description").toString()
.contains("<cite>")) {
mapobject.put("description",
mapobject.get("metatag.description"));
return true;
}
return false;
}
/**
* refine metatags by refining meta keyword to only include the English
* keyword only that has at most three keyword and if not present then
* create the keyword with title field of the html and if none of the
* keyword found then form it using the help of the url and exclude the
* number from the keywords
*
* @param mapobject
*/
private void refineMetaTags(final Map<String, Object> mapobject) {
String metaTag = "";
int tagFlag = 0;
if (mapobject.containsKey("metatag.keywords")) {
final String metaTags[] = mapobject.get("metatag.keywords")
.toString().replaceAll("\\|", ",").split(",");
String domain = null;
StringBuilder temp = null;
for (final String metaTag2 : metaTags) {
if (mapobject.containsKey("host")) {
domain = mapobject.get("host") + "";
if (domain.split("\\.").length > 1
&& (metaTag2
.contains(domain.split("\\.")[domain
.split("\\.").length - 2]) || metaTag2
.contains(domain.split("\\.")[0])))
{
continue;
}
}
String[] arr = metaTag2.split(" ");
arr = removeUnicodeWords(arr);
if (arr.length > 0 && arr.length < 5) {
temp = new StringBuilder();
for (final String str : arr) {
temp.append(str);
temp.append(" ");
}
if (metaTag.length() + temp.length() < 70) {
metaTag = metaTag + "," + temp.toString();
}
}
}
if (metaTag.startsWith(",")) {
metaTag = metaTag.trim();
metaTag = metaTag.substring(1, metaTag.length());
}
}
if (metaTag.length() < 1 && mapobject.containsKey("title")) {
/**
* Extracting tags from the title tag if the length of the
* keyword is greater than 4
*/
final String title = (String) mapobject.get("title");
final String splitTitle[] = title.split(" ");
int count = 0;
for (int i = 0; i < splitTitle.length; i++) {
if (splitTitle[i].length() > 4
&& !splitTitle[i].matches("^[\\u0900-\\u097F].*")) {
metaTag = metaTag + splitTitle[i] + ",";
count++;
if (count == 5) {
break;
}
}
}
if (metaTag.split(",").length > 3) {
if (metaTag.endsWith(",")) {
metaTag = metaTag.trim();
metaTag = metaTag.substring(0, metaTag.length() - 1);
}
} else {
metaTag = "";
}
}
if (metaTag.length() < 1) {
/**
* Extracting the tags from the url if the length of the keyword
* is greater than 4
*/
final String splitUrl[] = mapobject.get("url").toString()
.split("/");
final String lastSplitValue = splitUrl[splitUrl.length - 1];
final String tagList[] = generateTokens(lastSplitValue);
if (tagList != null) {
int count = 0;
for (int i = 0; i < tagList.length; i++) {
if (tagList[i].length() > 4
&& !tagList[i].matches("^[\\u0900-\\u097F].*")) {
metaTag = metaTag + tagList[i] + ",";
count++;
if (count == 5) {
break;
}
}
}
}
if (metaTag.endsWith(",")) {
metaTag = metaTag.trim();
metaTag = metaTag.substring(0, metaTag.length() - 1);
}
}
if (metaTag.length() > 0) {
metaTag = metaTag.replaceAll("\\[", "");
metaTag = metaTag.replaceAll("\"", "");
metaTag = metaTag.replaceAll(";", "");
metaTag = metaTag.replaceAll(":", "");
metaTag = metaTag.replaceAll("\u0027", "");
metaTag = metaTag.replaceAll("\u003d", "");
metaTag = metaTag.replaceAll("\u0026", "");
tagFlag = 1;
}
mapobject.put("TagFlag", tagFlag);
mapobject.put("metatag.keywords", metaTag);
}
/**
* Remove unicode character
*
* @param arr
* @return
*/
private String[] removeUnicodeWords(final String[] arr) {
final List<String> returnArr = new ArrayList<String>();
for (final String str : arr) {
if (str != null && str.trim().length() > 3
&& !str.matches("^[\\u0900-\\u097F].*")
&& !(str.matches("^[0-9].*"))) {
returnArr.add(str.trim());
}
}
final String[] retrnArr = new String[returnArr.size()];
returnArr.toArray(retrnArr);
return retrnArr;
}
/**
* Generate Token list with the help of the lucene analyzer
*
* @param lastSplitValue
* @return {@link ArrayIndexOutOfBoundsException} of the list of the
* keywords
*/
private String[] generateTokens(String lastSplitValue) {
final List<String> list = new ArrayList<String>();
lastSplitValue = lastSplitValue.replace("\\.", " ").replace("%20",
" ");
try {
final Version matchVersion = Version.LUCENE_45;
final Analyzer analyzer = new HindiAnalyzer(matchVersion);
final TokenStream ts = analyzer.tokenStream("field",
new StringReader(lastSplitValue));
ts.reset();
while (ts.incrementToken()) {
final CharTermAttribute cta = ts
.getAttribute(CharTermAttribute.class);
if (cta.toString().length() > 4
&& !cta.toString().matches("^[0-9].*")) {
list.add(cta.toString());
}
}
ts.end();
ts.close();
analyzer.close();
} catch (final Exception e) {
e.printStackTrace();
}
if (list.size() > 3) {
return list.toArray(new String[list.size()]);
} else {
return null;
}
}
/**
* Checks title and assign their language based on their first character
* of the title
*
* @param mapobject
* @return {@link Map}
*/
private boolean setLang(final Map<String, Object> mapobject) {
final String title = mapobject.get("title").toString();
final String description = mapobject.get("title").toString();
String language = "";
try {
language = mapper.detect(title);
mapper.detect(description);
} catch (final LangDetectException e) {
System.out.println("\n title with error is - " + title);
System.out.println("\n description with error is - "
+ description);
e.printStackTrace();
/*
* String title = mapobject.get("title").toString(); language =
* mapobject.get("lang") + ""; language = language.trim(); if
* (language.trim().equalsIgnoreCase("hi") ||
* language.trim().startsWith("en") ||
* language.trim().equalsIgnoreCase("lt")) { String[] titleArr =
* title.trim().split(" "); int i = 0; for (String titlePart :
* titleArr) { if
* (titlePart.trim().matches("^[\\u0900-\\u097F].*")) { i++; } }
* if (i >= titleArr.length * 0.5) { mapobject.put("lang",
* "hi"); } else { mapobject.put("lang", "lt"); } return true; }
*/
return false;
}
if (language.trim().equalsIgnoreCase("hi")
|| language.trim().startsWith("en")
|| language.trim().equalsIgnoreCase("lt")) {
mapobject.put("lang", language);
return true;
}
return false;
}
private String detect(final String text) throws LangDetectException {
final Detector detector = DetectorFactory.create();
detector.append(text);
return detector.detect();
}
/**
* Checks whether to include the doc based on their title and get the
* title from anchor tag title to choose the title that has largest
* number of the words and in hindi and it also gets the image from
* anchor tag href attribute
*
* @param mapobject
* of the key value pair
* @return {@link Boolean}
*/
private boolean checkTitleImage(final Map<String, Object> mapobject) {
final TreeSet<String> set = new TreeSet<String>(new SetSort());
final Gson gson = new Gson();
JsonArray array = null;
JsonObject object2 = null;
if (mapobject.containsKey("anchor")
&& mapobject.get("anchor") != null) {
final String arr = (String) mapobject.get("anchor");
try {
array = gson.fromJson(arr, JsonArray.class);
for (final JsonElement jsonElement : array) {
try {
object2 = gson.fromJson(jsonElement.getAsString(),
JsonObject.class);
} catch (final Exception e) {
if (object2 == null) {
object2 = new JsonObject();
object2.addProperty("title",
jsonElement.getAsString());
object2.addProperty("href", "");
object2.addProperty("alt", "");
}
}
if (object2 != null) {
assignTitleImage(mapobject, set, object2);
}
object2 = null;
}
} catch (final ClassCastException e) {
object2 = gson.fromJson(arr, JsonObject.class);
assignTitleImage(mapobject, set, object2);
} catch (final Exception e) {
e.printStackTrace();
}
if (!set.isEmpty()) {
int loop = 0;
final List<String> tempList = new LinkedList<String>();
for (final String string : set) {
final String title = string;
tempList.add(title.trim());
loop++;
if (loop == 2) {
break;
}
}
if (!tempList.isEmpty()) {
if (tempList.get(0).matches("^[\\u0900-\\u097F].*")) {
mapobject.put("title", tempList.get(0));
} else if (tempList.size() > 1
&& !(tempList.get(0)
.matches("^[\\u0900-\\u097F].*"))
&& tempList.get(1).matches(
"^[\\u0900-\\u097F].*")) {
mapobject.put("title", tempList.get(1));
} else {
mapobject.put("title", tempList.get(0));
}
}
}
}
if (mapobject.containsKey("title")
&& mapobject.get("title").toString().length() > 0
&& mapobject.get("title").toString().split(" ").length > 2
&& mapobject.get("title").toString().split(" ").length < 20
&& !mapobject.get("title").toString().contains("<")) {
if (set.isEmpty()) {
mapobject.put("title",
getTitleRefined(mapobject.get("title") + ""));
}
return true;
}
return false;
}
/**
* @param mapobject
* @param set
* @param object2
*/
private void assignTitleImage(final Map<String, Object> mapobject,
final TreeSet<String> set, final JsonObject object2) {
if (!mapobject.containsKey("ImgH1")
&& !mapobject.containsKey("ImgH2")) {
if (object2.get("href") != null
&& object2.get("href").getAsString().length() > 0
&& (object2.get("href").getAsString().toLowerCase()
.contains(".jpg")
|| object2.get("href").getAsString()
.toLowerCase().contains(".jpeg") || object2
.get("href").getAsString().toLowerCase()
.contains(".gif"))) {
putImages(mapobject, object2.get("href").getAsString()
.trim(), mapobject.get("tags").toString().trim()
.toLowerCase());
}
}
if (object2.get("title") != null
&& object2.get("title").getAsString().length() > 0
&& object2.get("title").getAsString().split(" ").length > 2
&& object2.get("title").getAsString().split(" ").length < 20
&& !object2.get("title").getAsString().contains("<")) {
final String newTitle = getTitleRefined(object2.get("title")
.getAsString());
set.add(newTitle.trim());
}
}
/**
* This function used to refine the title based on specific bad keyword
* during observation
*
* @param title
* @return refined title
*/
private String getTitleRefined(String title) {
title = title.replaceAll("\u0027", "");
title = title.replaceAll("\u0026", "");
title = title.replaceAll("\u003d", "");
if (title.contains("-")) {
if (title.trim().split("-").length > 1
&& !title.trim().split("-")[1].trim().matches(
"^[\\u0900-\\u097F].*")) {
return title.trim().split("-")[0].trim();
}
} else if (title.contains(":")) {
if (!title.trim().split(":")[0].trim().matches(
"^[\\u0900-\\u097F].*")
&& title.trim().split(":").length > 1) {
return title.trim().split(":")[1].trim();
}
}
return title;
}
/**
* Creates the path for the images
*
* @param map
* of the key value pair
* @param imageUrl
* @param category
*/
private void putImages(final Map<String, Object> map2,
final String imageUrl, final String category) {
try {
map2.put("ImgSrc", StringEscapeUtils.unescapeHtml(imageUrl)
.trim());
if (map2.containsKey("ImgSrc") && map2.get("ImgSrc") != null
&& map2.get("ImgSrc").toString().length() > 0) {
map2.put(
"ImgSrc",
StringEscapeUtils.unescapeHtml(map2.get("ImgSrc")
.toString())
+ "##RAFTAAR##"
+ imageUrl.trim());
} else {
return;
}
String imgNamearr[] = null;
try {
imgNamearr = imageUrl.split("/");
} catch (final Exception e) {
e.printStackTrace();
}
String imgName = null;
try {
imgName = imgNamearr[imgNamearr.length - 1];
} catch (final Exception e) {
e.printStackTrace();
}
final String imagePath = "/"
+ String.valueOf(imgName.charAt(0));
imgName = imgName.replaceAll(" ", "_").replaceAll("%20", "_");
if (imgName.split(".jpg").length > 0) {
imgName = imgName.split(".jpg")[0];
imgName = imgName + ".jpg";
}
map2.put("ImgH1", "h1/" + category + imagePath + "/" + imgName);
map2.put("ImgH2", "h2/" + category + imagePath + "/" + imgName);
} catch (final Exception e) {
e.printStackTrace();
}
}
/**
* Inserts the data to the elasticsearch
*
* @param mapobject
* @param key
* unique id generally it is the unique url
*/
public static void insertToElastic(final Map<String, Object> mapobject,
final String key) {
final Settings settings = ImmutableSettings.settingsBuilder()
.put("cluster.name", CLUSTER).build();/*
* change ccluster.name
* to cluster
*/
final Client client = new TransportClient(settings)
.addTransportAddress(new InetSocketTransportAddress(
SEARCH_HOST, Integer.parseInt(SEARCH_PORT)));
client.prepareIndex(SEARCH_INDEX_NAME, SEARCHtYPE, key)
.setSource(mapobject).execute().actionGet();
client.close();
}
/**
* Assign the city to the news without city
*
* @param category
* @param description
* @return update category with city
*/
private static void assignCity(final Map<String, Object> mapobject) {
String category = mapobject.get("tags").toString();
if (category.endsWith("/")) {
boolean flag = true;
final String catArr[] = category.split("/");
if (catArr.length == 2) {
final String state = catArr[1];
CITY_MAP = STATE_MAP.get(state);
for (final Entry<String, String> e : CITY_MAP.entrySet()) {
final String description = mapobject.get("description")
.toString();
if (description.contains(e.getValue())) {
category = category + e.getKey();
mapobject.put("tags", category);
flag = false;
break;
}
}
}
if (flag) {
mapobject.put("tags", "national");
}
}
}
}
/**
* Update the data to hbase
*
* @param tableName
* @param rowKey
* @param family
* @param qualifier
* @param value
* @param conf
*/
public static void updateIntoHbase(final String tableName,
final String rowKey, final String family, final String qualifier,
final String value, final Configuration conf) {
HTable table = null;
try {
table = new HTable(conf, tableName);
} catch (final IOException e) {
e.printStackTrace();
}
final Put put = new Put(Bytes.toBytes(rowKey));
put.add(Bytes.toBytes(family), Bytes.toBytes(qualifier),
Bytes.toBytes(value));
try {
table.put(put);
table.close();
} catch (final IOException e) {
e.printStackTrace();
}
}
/**
* Return the map of the all states and city
*
* @param fileName
* @return
*/
private static Map<String, Map<String, String>> returnMap(
final String fileName) {
final Map<String, Map<String, String>> map = new HashMap<String, Map<String, String>>();
BufferedReader br = null;
try {
br = new BufferedReader(new FileReader(fileName));
String line;
while ((line = br.readLine()) != null) {
final String arr[] = line.split("\t", 3);
if (arr.length == 3) {
if (map.containsKey(arr[0])) {
Map<String, String> m = new HashMap<String, String>();
m = map.get(arr[0]);
m.put(arr[1], arr[2]);
} else {
final Map<String, String> m = new HashMap<String, String>();
m.put(arr[1], arr[2]);
map.put(arr[0], m);
}
}
}
} catch (final FileNotFoundException e) {
e.printStackTrace();
} catch (final IOException e) {
e.printStackTrace();
} finally {
if (br != null) {
try {
br.close();
} catch (final Exception e) {
e.printStackTrace();
}
}
}
return map;
}
public static void main(final String[] args) throws Exception {
int c = 0;
c = ToolRunner.run(new Configuration(), new HbaseToElastic(), args);
System.exit(c);
}
}