1 | package org.greenstone.atea;
|
---|
2 |
|
---|
3 |
|
---|
4 | import java.io.*;
|
---|
5 | import java.nio.charset.StandardCharsets;
|
---|
6 | import java.util.Properties;
|
---|
7 | import java.util.zip.GZIPInputStream;
|
---|
8 | import java.util.Iterator;
|
---|
9 | import java.util.HashMap;
|
---|
10 | import java.util.Map;
|
---|
11 | import java.util.Set;
|
---|
12 | import java.util.TreeMap;
|
---|
13 | import java.util.TreeSet;
|
---|
14 |
|
---|
15 | import org.apache.commons.csv.*; // https://commons.apache.org/proper/commons-csv/download_csv.cgi
|
---|
16 | import org.apache.log4j.Logger;
|
---|
17 |
|
---|
18 |
|
---|
19 | /**
|
---|
20 | * The main() method of this class takes a folder of warc.wet(.gz) files and goes through
|
---|
21 | * the WET records in each, putting each WET record into a file. Each file is put into a
|
---|
22 | * keep or discard or greyListed folder, and its url listed written into a keep, discard
|
---|
23 | * or greylisted text file, based on based on
|
---|
24 | *
|
---|
25 | * 1. whether it's whitelisted, else greylisted else blacklisted
|
---|
26 | * 2. and if explicitly whitelisted or else not greylisted or blacklisted and there's
|
---|
27 | * enough content. Formerly, content-length and number of lines were used to determine if
|
---|
28 | * the content was sufficient. Now it's just word count and number of MAX characters
|
---|
29 | * (not MINIMUM characters) that determine a string is a word. These settings can be adjusted
|
---|
30 | * in conf/config.properties.
|
---|
31 | *
|
---|
32 | * Put a url-blacklist-filter.txt and/or url-greylist-filter.txt and/or url-whitelist-filter.txt
|
---|
33 | * into the conf folder to control any url patterns that are explicitly included or excluded or
|
---|
34 | * set aside for inspecting later. These filter text files don't use regexes, instead their
|
---|
35 | * format is:
|
---|
36 | * - precede URL by ^ to blacklist urls that match the given prefix
|
---|
37 | * - succeed URL by $ to blacklist urls that match the given suffix
|
---|
38 | * - ^url$ will blacklist urls that match the given url completely
|
---|
39 | * - Without either ^ or $ symbol, urls containing the given url will get blacklisted
|
---|
40 | *
|
---|
41 | * WETProcessor.java's current implementation is that explicit whitelisting has precedence
|
---|
42 | * over greylisting and which takes precedence over blacklisting in turn. However, even
|
---|
43 | * explicitly whitelisted urls still need to have sufficient content to end up in keepURLs.txt
|
---|
44 | * and in the seedURLs.txt file used for nutch, along with its domain in regex-urlfilter.txt
|
---|
45 | * also for nutch.
|
---|
46 | *
|
---|
47 | * A CCWETProcessor instance can be configured to process all the .warc.wet(.gz) files
|
---|
48 | * in the given input folder. Then use a single instance of the WETProcessor class to process
|
---|
49 | * each single unzipped warc.wet file.
|
---|
50 | *
|
---|
51 | * To compile, including the jars in lib/ for compiling.
|
---|
52 | * maori-lang-detection/src$ javac -cp ".:../lib/*" org/greenstone/atea/CCWETProcessor.java
|
---|
53 | *
|
---|
54 | * To run, passing the log4j and other properties files in conf/ folder:
|
---|
55 | * maori-lang-detection/src$ java -cp ".:../conf:../lib/*" org.greenstone.atea.CCWETProcessor <folder containing commoncrawls subfolders containing warc.wet(.gz) files> <outputFolder>
|
---|
56 | *
|
---|
57 | * e.g. (from maori-lang-detection/src)
|
---|
58 | *
|
---|
59 | * - java -cp ".:../conf:../lib/*" org.greenstone.atea.CCWETProcessor ../ccrawl-data /Scratch/ak19/gs3-extensions/maori-lang-detection/to_crawl
|
---|
60 | * - java -cp ".:../conf:../lib/*" org.greenstone.atea.CCWETProcessor ../ccrawl-data /Scratch/ak19/gs3-extensions/maori-lang-detection/to_crawl 2>&1 | less
|
---|
61 | *
|
---|
62 | */
|
---|
63 |
|
---|
64 | public class CCWETProcessor {
|
---|
65 | private static Logger logger = Logger.getLogger(org.greenstone.atea.CCWETProcessor.class.getName());
|
---|
66 |
|
---|
67 | // Properties shared across WETProcessor instances
|
---|
68 | public final int MAX_WORD_LENGTH;
|
---|
69 | public final int MIN_NUM_WORDS;
|
---|
70 | public final int MAX_WORDS_CAMELCASE;
|
---|
71 |
|
---|
72 | // constants for the possible fixed values in sites-too-big-to-exhaustively-crawl.txt file
|
---|
73 | public final String SUBDOMAIN_COPY = "SUBDOMAIN-COPY";
|
---|
74 | public final String SINGLEPAGE = "SINGLEPAGE";
|
---|
75 | public final String FOLLOW_LINKS_WITHIN_TOPSITE = "FOLLOW-LINKS-WITHIN-TOPSITE";
|
---|
76 |
|
---|
77 | /**
|
---|
78 | * Characters that need escaping if used as a string literal in a regex
|
---|
79 | * https://stackoverflow.com/questions/399078/what-special-characters-must-be-escaped-in-regular-expressions
|
---|
80 | * https://www.regular-expressions.info/refcharacters.html
|
---|
81 | * Put the \\ (escape char) at start so we don't double-escape chars already escaped,
|
---|
82 | * as would happen for any chars appearing earlier in this list than \\
|
---|
83 | */
|
---|
84 | public final String ESCAPE_CHARS_FOR_RE = "\\.^$*+?()[{|";
|
---|
85 | //public final String[] ESCAPE_CHARS_FOR_RE = ["\\", ".", "^", "$", "*", "+", "?", "(", ")", "[", "{", "|"];
|
---|
86 |
|
---|
87 | private Properties configProperties = new Properties();
|
---|
88 |
|
---|
89 | // File paths shared across WETProcessor instances
|
---|
90 | public final File commoncrawlDir;
|
---|
91 | public final File outputFolder;
|
---|
92 | public final File discardFolder;
|
---|
93 | public final File keepFolder;
|
---|
94 | public final File greyListedFolder;
|
---|
95 | public final File keepURLsFile;
|
---|
96 | public final File discardURLsFile;
|
---|
97 | public final File greyListedFile;
|
---|
98 |
|
---|
99 | /** Possible values stored in the blackList/whiteList/greyList Maps */
|
---|
100 | private final Integer LIST_ENTRY_CONTAINS = new Integer(0);
|
---|
101 | private final Integer LIST_ENTRY_STARTSWITH = new Integer(1);
|
---|
102 | private final Integer LIST_ENTRY_ENDSWITH = new Integer(2);
|
---|
103 | private final Integer LIST_ENTRY_MATCHES = new Integer(3);
|
---|
104 |
|
---|
105 | /**
|
---|
106 | * Store url patterns as keys and values indicated whether a url should
|
---|
107 | * match it exactly, start/end with it, or contain it
|
---|
108 | */
|
---|
109 | private HashMap<String, Integer> blackList;
|
---|
110 | private HashMap<String, Integer> greyList;
|
---|
111 | private HashMap<String, Integer> whiteList;
|
---|
112 |
|
---|
113 | /** map of topsites with allowable regexes: sites too big to exhaustively crawl
|
---|
114 | * with optional regex defining allowed exceptions, like subdomains or url suffixes
|
---|
115 | * off that top site. For example, wikipedia.org is a topsite, but mi.wikipedia.org
|
---|
116 | * is relevant. Or blogspot.com is a top site, but someone's pages in Maori off blogspot
|
---|
117 | * would be relevant.
|
---|
118 | * The map would store top site domain suffix and an optional regex string for allowable
|
---|
119 | * url patterns.
|
---|
120 | */
|
---|
121 | private HashMap<String, String> topSitesMap;
|
---|
122 |
|
---|
123 | /** Map of domains we keep and the full urls we're keeping that are of that domain.
|
---|
124 | * No need to use a TreeMap which preserves natural (alphabetical) ordering of keys,
|
---|
125 | * while a HashMap has no notion of ordering, because we just need to store urls with
|
---|
126 | * their domains. Whether the domains are sorted or the urls per domain are sorted becomes
|
---|
127 | * irrelevant. (Does it really? What if we have urls followed vs preceded by urls with the
|
---|
128 | * same prefix, e.g. pinky.com/toto/index.html and pinky.com/toto/nono/file.html
|
---|
129 | * Is there any benefit to nutch when crawling if these seedURLs are ordered or not?)
|
---|
130 | */
|
---|
131 | private Map<String, Set<String>> domainsToURLsMap;
|
---|
132 |
|
---|
133 | // Keep a count of all the records that all WETProcessors instantiated
|
---|
134 | // by our main method combined have processed
|
---|
135 | private int totalRecordCount = 0;
|
---|
136 |
|
---|
137 | private int wetFileCount = 0;
|
---|
138 |
|
---|
139 | private static ClassLoader MY_CLASSLOADER = org.greenstone.atea.CCWETProcessor.class.getClassLoader();
|
---|
140 |
|
---|
141 | public CCWETProcessor(File inFolder, File outFolder) throws Exception {
|
---|
142 | this.commoncrawlDir = inFolder;
|
---|
143 | this.outputFolder = outFolder;
|
---|
144 |
|
---|
145 | // load up the properties from the config file
|
---|
146 | try (InputStream infile = MY_CLASSLOADER.getResourceAsStream("config.properties")) {
|
---|
147 | configProperties = new Properties();
|
---|
148 | configProperties.load(infile);
|
---|
149 | //infile.close(); // not explicitly called in examples of try-with-resources
|
---|
150 |
|
---|
151 | } catch(Exception e) {
|
---|
152 | logger.error("Exception attempting to read properties from config.properties.", e);
|
---|
153 | }
|
---|
154 |
|
---|
155 | if(configProperties.size() == 0) {
|
---|
156 | logger.warn("*** Warning: no values read into config properties. Using defaults.");
|
---|
157 | }
|
---|
158 |
|
---|
159 | MAX_WORD_LENGTH = Integer.parseInt(configProperties.getProperty("WETprocessor.max.word.length", "15"));
|
---|
160 | MIN_NUM_WORDS = Integer.parseInt(configProperties.getProperty("WETprocessor.min.num.words", "20"));
|
---|
161 | MAX_WORDS_CAMELCASE = Integer.parseInt(configProperties.getProperty("WETprocessor.max.words.camelcase", "10"));
|
---|
162 |
|
---|
163 |
|
---|
164 | this.discardFolder = new File(outFolder, "discard");
|
---|
165 | if(!discardFolder.exists()) {
|
---|
166 | discardFolder.mkdir();
|
---|
167 | }
|
---|
168 | this.keepFolder = new File(outFolder, "keep");
|
---|
169 | if(!keepFolder.exists()) {
|
---|
170 | keepFolder.mkdir();
|
---|
171 | }
|
---|
172 |
|
---|
173 | this.greyListedFolder = new File(outFolder, "greylisted");
|
---|
174 | if(!greyListedFolder.exists()) {
|
---|
175 | greyListedFolder.mkdir();
|
---|
176 | }
|
---|
177 |
|
---|
178 | this.keepURLsFile = new File(outFolder, "keepURLs.txt");
|
---|
179 | if(keepURLsFile.exists() && !keepURLsFile.delete()) {
|
---|
180 | throw new Exception("Warning: Unable to delete " + this.keepURLsFile + ". Unable to proceed.");
|
---|
181 | }
|
---|
182 | this.discardURLsFile = new File(outFolder, "discardURLs.txt");
|
---|
183 | if(discardURLsFile.exists() && !discardURLsFile.delete()) {
|
---|
184 | throw new Exception ("Warning Unable to delete " + discardURLsFile + ". Unable to proceed.");
|
---|
185 | }
|
---|
186 | this.greyListedFile = new File(outFolder, "greyListed.txt");
|
---|
187 | if(greyListedFile.exists() && !greyListedFile.delete()) {
|
---|
188 | throw new Exception ("Warning Unable to delete " + greyListedFile + ". Unable to proceed.");
|
---|
189 | }
|
---|
190 |
|
---|
191 | // prepare our blacklist, greylist (for inspection) and whitelist
|
---|
192 | logger.info("Loading blacklist.");
|
---|
193 | blackList = new HashMap<String, Integer>();
|
---|
194 | initURLFilterList(blackList, "url-blacklist-filter.txt");
|
---|
195 |
|
---|
196 | logger.info("Loading greylist.");
|
---|
197 | greyList = new HashMap<String, Integer>();
|
---|
198 | initURLFilterList(greyList, "url-greylist-filter.txt");
|
---|
199 |
|
---|
200 | logger.info("Loading whitelist.");
|
---|
201 | whiteList = new HashMap<String, Integer>();
|
---|
202 | initURLFilterList(whiteList, "url-whitelist-filter.txt");
|
---|
203 |
|
---|
204 | // Create the map of topSites
|
---|
205 | logger.info("Loading map of topsites with regex of allowable url patterns for each topsite.");
|
---|
206 | topSitesMap = new HashMap<String, String>();
|
---|
207 |
|
---|
208 | // Read in our csv file of topsites and what to do when one hits a match with a seedURL
|
---|
209 | // and put these in our topSitesMap
|
---|
210 | // https://commons.apache.org/proper/commons-csv/apidocs/index.html
|
---|
211 | // https://commons.apache.org/proper/commons-csv/apidocs/org/apache/commons/csv/CSVParser.html
|
---|
212 | //https://commons.apache.org/proper/commons-csv/apidocs/org/apache/commons/csv/CSVFormat.html
|
---|
213 | CSVFormat customisedCSVFormat = CSVFormat.DEFAULT
|
---|
214 | .withCommentMarker('#')
|
---|
215 | .withSkipHeaderRecord()
|
---|
216 | .withIgnoreSurroundingSpaces();
|
---|
217 |
|
---|
218 | File topSitesCSVData = new File(MY_CLASSLOADER.getResource("sites-too-big-to-exhaustively-crawl.txt").getFile());
|
---|
219 | // CSVParser is AutoCloseable and throws exceptions, so putting it in a try-with-resources
|
---|
220 | try (
|
---|
221 | CSVParser parser = CSVParser.parse(topSitesCSVData, StandardCharsets.UTF_8, customisedCSVFormat);
|
---|
222 | ) {
|
---|
223 | for (CSVRecord csvRecord : parser) {
|
---|
224 | String topsite = csvRecord.get(0);
|
---|
225 | String allowed_url_pattern = (csvRecord.size() >= 2) ? csvRecord.get(1) : "";
|
---|
226 | topSitesMap.put(topsite, allowed_url_pattern);
|
---|
227 |
|
---|
228 | //logger.debug("@@@@ topsite: " + topsite + " - " + allowed_url_pattern);
|
---|
229 |
|
---|
230 | }
|
---|
231 | } catch(Exception e) {
|
---|
232 | logger.error("@@@@@@@@@ Error attempting to parse CSV format of text file " + topSitesCSVData, e);
|
---|
233 | }
|
---|
234 |
|
---|
235 | //logger.debug("Prematurely terminating for testing purposes.");
|
---|
236 | //System.exit(-1);
|
---|
237 | }
|
---|
238 |
|
---|
239 |
|
---|
240 | /** Utility function to help escape regex characters in URL to go into regex-urlfilter.txt */
|
---|
241 | private String escapeStringForRegex(String str) {
|
---|
242 | for(int i = 0; i < ESCAPE_CHARS_FOR_RE.length(); i++) {
|
---|
243 | char c = ESCAPE_CHARS_FOR_RE.charAt(i);
|
---|
244 | str = str.replace(Character.toString(c), "\\"+c);
|
---|
245 | }
|
---|
246 | return str;
|
---|
247 | }
|
---|
248 |
|
---|
249 | /**
|
---|
250 | * Using the keepURLs.txt file generated by running WETProcessor instances, this produces
|
---|
251 | * as output the URL seed list and regex-urlfilter text files required by nutch, see
|
---|
252 | * https://cwiki.apache.org/confluence/display/nutch/NutchTutorial
|
---|
253 | */
|
---|
254 | public void createSeedURLsFiles(File seedURLsFile, File urlFilterFile,
|
---|
255 | File domainURLsFile, File topSiteMatchesFile,
|
---|
256 | File possibleProductSitesFile) {
|
---|
257 | // Maintain a Map of unique domains mapped to seed urls at that domain
|
---|
258 | // TreeSet: by default, "the elements are ordered using their natural ordering"
|
---|
259 | // (or by a Comparator provided at set creation time).
|
---|
260 | // Whereas HashSet doesn't guarantee ordering.
|
---|
261 | // So we get alphabetic sorting for free. And guaranteed log(n) for basic operations.
|
---|
262 | // Would be a similar distinction for Maps.
|
---|
263 | domainsToURLsMap = new TreeMap<String, Set<String>>();
|
---|
264 |
|
---|
265 | final String PROTOCOL_REGEX_PREFIX = "+^https?://";
|
---|
266 | final String FILTER_REGEX_PREFIX = PROTOCOL_REGEX_PREFIX + "([a-z0-9-]+\\.)*"; // https?://([a-z0-9-]+\.)* for nutch's regex-urlfilter.txt
|
---|
267 |
|
---|
268 | // keep an eye out on URLs we need to inspect later
|
---|
269 | Set<String> possibleProductDomains = new TreeSet<String>();
|
---|
270 | File geoLiteCityDatFile = new File(MY_CLASSLOADER.getResource("GeoLiteCity.dat").getFile());
|
---|
271 |
|
---|
272 | try (
|
---|
273 | BufferedReader reader = new BufferedReader(new FileReader(this.keepURLsFile));
|
---|
274 | BufferedWriter possibleProductSitesWriter = new BufferedWriter(new FileWriter(possibleProductSitesFile));
|
---|
275 | ) {
|
---|
276 |
|
---|
277 | // read a URL at a time from urlsFile
|
---|
278 | String url = null;
|
---|
279 | String domainWithProtocol = null;
|
---|
280 | while((url = reader.readLine()) != null) { // readLine removes newline separator
|
---|
281 |
|
---|
282 | // work out domain. This retains any www. or subdomain prefix
|
---|
283 | // passing true to further also retain the http(s) protocol
|
---|
284 | domainWithProtocol = Utility.getDomainForURL(url, true);
|
---|
285 |
|
---|
286 | Set<String> urlsSet;
|
---|
287 | if(!domainsToURLsMap.containsKey(domainWithProtocol)) {
|
---|
288 | urlsSet = new TreeSet<String>();
|
---|
289 | urlsSet.add(url);
|
---|
290 | domainsToURLsMap.put(domainWithProtocol, urlsSet);
|
---|
291 | } else {
|
---|
292 | urlsSet = domainsToURLsMap.get(domainWithProtocol);
|
---|
293 | urlsSet.add(url);
|
---|
294 | }
|
---|
295 |
|
---|
296 | /*
|
---|
297 | // Dr Nichols said that a url that was located outside the country and
|
---|
298 | // which had /mi/ URLs was more likely to be an autotranslated (product) site.
|
---|
299 | // Following Dr Nichols' idea, let's keep a look out for more product sites:
|
---|
300 | // if any URL contains /mi AND the tld of its domain is outside of New Zealand
|
---|
301 | // then add that domain (if not already added) and that url into a file
|
---|
302 | // for later manual inspection
|
---|
303 | if(!domainWithProtocol.endsWith(".nz")
|
---|
304 | && (url.contains("/mi/") || url.endsWith("/mi"))) {
|
---|
305 |
|
---|
306 | if(!possibleProductDomains.contains(domainWithProtocol)) {
|
---|
307 |
|
---|
308 | String countryCode = "";
|
---|
309 | try {
|
---|
310 | // more expensive test, so do this only if above conditions are true:
|
---|
311 | countryCode = Utility.getCountryCodeOfDomain(domainWithProtocol, geoLiteCityDatFile);
|
---|
312 | System.err.println("@@@@ Got country code: " + countryCode);
|
---|
313 | } catch(Exception exceptObj) {
|
---|
314 | countryCode = ""; // forces domain to be included for inspection
|
---|
315 |
|
---|
316 | logger.error("Could not check if domain " + domainWithProtocol
|
---|
317 | + " was in country: " + countryCode,
|
---|
318 | exceptObj);
|
---|
319 | }
|
---|
320 |
|
---|
321 | boolean isInNZ = countryCode.toLowerCase().equals("nz");
|
---|
322 |
|
---|
323 |
|
---|
324 | //if(!Utility.isDomainInCountry(domainWithProtocol, "nz", geoLiteCityDatFile)) {
|
---|
325 | if(!isInNZ) {
|
---|
326 | possibleProductDomains.add(domainWithProtocol);
|
---|
327 | // write both domain and a sample seedURL on that site out to file
|
---|
328 | possibleProductSitesWriter.write(countryCode + " : " + domainWithProtocol + "\n");
|
---|
329 | possibleProductSitesWriter.write("\t" + url + "\n");
|
---|
330 | }
|
---|
331 | }
|
---|
332 | //else {
|
---|
333 | // already wrote out domain to file at some point, write just the URL out to file
|
---|
334 | //possibleProductSitesWriter.write("\t" + url + "\n");
|
---|
335 | //}
|
---|
336 | }
|
---|
337 | */
|
---|
338 | }
|
---|
339 | } catch (IOException ioe) {
|
---|
340 | logger.error("@@@@@@@@@ Error reading in urls from file " + this.keepURLsFile, ioe);
|
---|
341 | }
|
---|
342 |
|
---|
343 | // We'd have pruned out duplicates by now and have a sorted list of domains,
|
---|
344 | // each of which maps to seed URLs in the commoncrawl for that domain
|
---|
345 |
|
---|
346 | int domainCount = 0;
|
---|
347 | File sitesFolder = new File(outputFolder, "sites");
|
---|
348 | if(!sitesFolder.exists()) {
|
---|
349 | sitesFolder.mkdir();
|
---|
350 | }
|
---|
351 | final String FORMATSTR = "%05d";
|
---|
352 |
|
---|
353 | // write out each domain followed in sequence by all urls we found in that domain
|
---|
354 | // (urls with tab up front)
|
---|
355 | try (
|
---|
356 | // global lists of all domains, seedURLs and regex-urlfilters across all wet files of all commoncrawls
|
---|
357 | // Also a global file listing any urls that matched top sites that didn't specify
|
---|
358 | // allowed regex patterns
|
---|
359 | BufferedWriter domainURLsWriter = new BufferedWriter(new FileWriter(domainURLsFile));
|
---|
360 | BufferedWriter seedURLsWriter = new BufferedWriter(new FileWriter(seedURLsFile));
|
---|
361 | BufferedWriter urlFilterWriter = new BufferedWriter(new FileWriter(urlFilterFile));
|
---|
362 | BufferedWriter topSiteMatchesWriter = new BufferedWriter(new FileWriter(topSiteMatchesFile));
|
---|
363 | ) {
|
---|
364 |
|
---|
365 | // initialise topSiteMatchesFile with some instructional text.
|
---|
366 | topSiteMatchesWriter.write("The following domain with seedURLs are on a major/top 500 site\n");
|
---|
367 | topSiteMatchesWriter.write("for which no allowed URL pattern regex has been specified.\n");
|
---|
368 | topSiteMatchesWriter.write("Specify one for this domain in the tab-spaced sites-too-big-to-exhaustively-crawl.txt file\n");
|
---|
369 |
|
---|
370 | //Set<Map.Entry<String, Set<String>>> domainsSet = domainsToURLsMap.keySet();
|
---|
371 | Set<String> domainsSet = domainsToURLsMap.keySet();
|
---|
372 | Iterator<String> domainIterator = domainsSet.iterator();
|
---|
373 |
|
---|
374 | /*
|
---|
375 | // DEBUG
|
---|
376 | String value = topSitesMap.get("wikipedia.org");
|
---|
377 | if(value == null) {
|
---|
378 | logger.debug("### wikipedia.org had null value");
|
---|
379 | } else {
|
---|
380 | logger.debug("### wikipedia.org had value: " + value);
|
---|
381 | } // DEBUG
|
---|
382 | */
|
---|
383 |
|
---|
384 | while(domainIterator.hasNext()) {
|
---|
385 | String domainWithProtocol = domainIterator.next();
|
---|
386 | // Also get domain without protocol prefix
|
---|
387 | int startIndex = domainWithProtocol.indexOf("//"); // http:// or https:// prefix
|
---|
388 | startIndex = (startIndex == -1) ? 0 : (startIndex+2); // skip past the protocol's // portion
|
---|
389 | String domain = domainWithProtocol.substring(startIndex);
|
---|
390 |
|
---|
391 | /*if(domain.contains("docs.google.com")) {
|
---|
392 | logger.debug("domain with protocol: " + domainWithProtocol);
|
---|
393 | logger.debug("domain: " + domain);
|
---|
394 | }*/
|
---|
395 |
|
---|
396 | String allowedURLPatternRegex = isURLinTopSitesMap(domain);
|
---|
397 | // If the domain is of a topsite for which no allowed URL pattern has been provided
|
---|
398 | // in sites-too-big-to-exhaustively-crawl.txt,
|
---|
399 | // then we don't know how to crawl the site. Warn the user by writing the affected
|
---|
400 | // domain and seedURLs to the topSiteMatchesFile.
|
---|
401 | if(allowedURLPatternRegex != null && allowedURLPatternRegex.equals("")) {
|
---|
402 |
|
---|
403 | // topsite, but we don't (yet) know what portion can be crawled
|
---|
404 | // Append the top site and url to a global/toplevel file that
|
---|
405 | // the user needs to check later and we're done with this domain as it
|
---|
406 | // won't go into any other file hereafter
|
---|
407 |
|
---|
408 | Set<String> urlsForDomainSet = domainsToURLsMap.get(domainWithProtocol);
|
---|
409 | for(String url : urlsForDomainSet) {
|
---|
410 | topSiteMatchesWriter.write("\t" + url + "\n");
|
---|
411 | }
|
---|
412 |
|
---|
413 | continue; // done with this domain
|
---|
414 | }
|
---|
415 |
|
---|
416 | // start counting the domains we're actually going to process
|
---|
417 | domainCount++;
|
---|
418 |
|
---|
419 | String siteID = String.format(FORMATSTR, domainCount);
|
---|
420 | File domainFolder = new File(sitesFolder, siteID);
|
---|
421 | domainFolder.mkdir();
|
---|
422 |
|
---|
423 | // write out the domain
|
---|
424 | //seedURLsWriter.write(domainWithProtocol + "\n");
|
---|
425 |
|
---|
426 |
|
---|
427 | // for every domain, we need a sites/0000x/ folder, where x is domain#, containing
|
---|
428 | // its own INDIVIDUAL seedURLs.txt and regex-urlfilter.txt
|
---|
429 | // We still have a global seedURLs.txt and regex-urlfilter.txt too.
|
---|
430 | File siteSeedsFile = new File(domainFolder, "seedURLs.txt"); // e.g. sites/00001/seedURLs.txt
|
---|
431 | File siteRegexFile = new File(domainFolder, "regex-urlfilter.txt"); // e.g. sites/00001/regex-urlfilter.txt
|
---|
432 | try (
|
---|
433 | BufferedWriter siteURLsWriter = new BufferedWriter(new FileWriter(siteSeedsFile));
|
---|
434 | BufferedWriter siteRegexWriter = new BufferedWriter(new FileWriter(siteRegexFile));
|
---|
435 | ) {
|
---|
436 |
|
---|
437 | // write all sorted unique domains into global domains file
|
---|
438 | // Using the domain withuot protocol since the global domains file is for
|
---|
439 | // informational purposes
|
---|
440 | domainURLsWriter.write(domain + "\n");
|
---|
441 |
|
---|
442 | // Only write urls and no domain into single global seedurls file
|
---|
443 | // But write domain and tabbed urls into individual sites/0000#/seedURLs.txt
|
---|
444 | // files (and write regexed domain into each sites/0000#/regex-urlfilter.txt)
|
---|
445 | // If we ever run nutch on a single seedURLs listing containing
|
---|
446 | // all seed pages to crawl sites from, the above two files will work for that.
|
---|
447 |
|
---|
448 | // first write out the urls for the domain into the sites/0000x/seedURLs.txt file
|
---|
449 | // also write into the global seeds file (with a tab prefixed to each?)
|
---|
450 | Set<String> urlsForDomainSet = domainsToURLsMap.get(domainWithProtocol);
|
---|
451 | for(String url : urlsForDomainSet) {
|
---|
452 | seedURLsWriter.write(url + "\n"); // global seedURLs file
|
---|
453 | siteURLsWriter.write(url + "\n");
|
---|
454 | }
|
---|
455 |
|
---|
456 | if(allowedURLPatternRegex == null) { // entire site can be crawled
|
---|
457 | siteURLsWriter.write(domainWithProtocol + "\n");
|
---|
458 |
|
---|
459 | // Write out filter in the following form for a site, e.g. for nutch.apache.org:
|
---|
460 | // nutch.apache.org => +^https?://([a-z0-9-]+\.)*nutch\.apache\.org/
|
---|
461 | String regexed_domain = FILTER_REGEX_PREFIX + escapeStringForRegex(domain) + "/";
|
---|
462 | //String regexed_domain = FILTER_REGEX_PREFIX + domain.replace(".", "\\.") + "/";
|
---|
463 | urlFilterWriter.write(regexed_domain + "\n"); //global file
|
---|
464 | siteRegexWriter.write(regexed_domain + "\n"); // site file
|
---|
465 | }
|
---|
466 | else { // domain belongs to a top site where only portion of site can be crawled
|
---|
467 |
|
---|
468 | if(allowedURLPatternRegex.equals(SUBDOMAIN_COPY)) { // COPY existing domain as url-filter
|
---|
469 | siteURLsWriter.write(domainWithProtocol + "\n");
|
---|
470 | // e.g. pinky.blogspot.com will add a filter for pinky.blogspot.com
|
---|
471 | // and not for all of blogspot.com
|
---|
472 |
|
---|
473 | String regexed_domain = PROTOCOL_REGEX_PREFIX+escapeStringForRegex(domain) + "/";
|
---|
474 | //String regexed_domain = PROTOCOL_REGEX_PREFIX+domain.replace(".", "\\.") + "/";
|
---|
475 | urlFilterWriter.write(regexed_domain + "\n");
|
---|
476 | siteRegexWriter.write(regexed_domain + "\n");
|
---|
477 |
|
---|
478 | } else if(allowedURLPatternRegex.equals(SINGLEPAGE)) {
|
---|
479 | // don't write out domain. We want individual pages
|
---|
480 | //DON'T DO THIS HERE: siteURLsWriter.write(domainWithProtocol + "\n");
|
---|
481 |
|
---|
482 | // don't write out domain as a regex expression url filter either,
|
---|
483 | // write out the individual seed urls for the domain instead
|
---|
484 | // since we will only be downloading the single page
|
---|
485 |
|
---|
486 | urlsForDomainSet = domainsToURLsMap.get(domainWithProtocol);
|
---|
487 | for(String urlInDomain : urlsForDomainSet) {
|
---|
488 | // don't append slash to end this time
|
---|
489 | String regexed_url = "+^"+escapeStringForRegex(urlInDomain);
|
---|
490 | //String regexed_url = "+^"+urlInDomain.replace(".", "\\.");
|
---|
491 | urlFilterWriter.write(regexed_url + "\n");
|
---|
492 | siteRegexWriter.write(regexed_url + "\n");
|
---|
493 | }
|
---|
494 | } else if(allowedURLPatternRegex.equals(FOLLOW_LINKS_WITHIN_TOPSITE)) {
|
---|
495 |
|
---|
496 | // DON'T write out domain into siteURLs file,
|
---|
497 | // BUT DO write it into urlFilter file
|
---|
498 | String regexed_domain = PROTOCOL_REGEX_PREFIX + escapeStringForRegex(domain) + "/";
|
---|
499 |
|
---|
500 | urlFilterWriter.write(regexed_domain + "\n");
|
---|
501 | siteRegexWriter.write(regexed_domain + "\n");
|
---|
502 | } else { // allowedURLPatternRegex is a url-form - convert to regex
|
---|
503 | if(!allowedURLPatternRegex.endsWith("/")) {
|
---|
504 | allowedURLPatternRegex += "/";
|
---|
505 | }
|
---|
506 | String regexed_pattern = PROTOCOL_REGEX_PREFIX+escapeStringForRegex(allowedURLPatternRegex);
|
---|
507 | //String regexed_pattern = PROTOCOL_REGEX_PREFIX+allowedURLPatternRegex.replace(".", "\\.");
|
---|
508 | siteURLsWriter.write(domainWithProtocol + "\n");
|
---|
509 | urlFilterWriter.write(regexed_pattern + "\n");
|
---|
510 | siteRegexWriter.write(regexed_pattern + "\n");
|
---|
511 |
|
---|
512 | }
|
---|
513 | }
|
---|
514 |
|
---|
515 | } catch (IOException ioe) {
|
---|
516 | logger.error("@@@@@@@@@ Error writing to one of:" + siteSeedsFile + " or " + siteRegexFile, ioe);
|
---|
517 | }
|
---|
518 |
|
---|
519 | }
|
---|
520 |
|
---|
521 | } catch (IOException ioe) {
|
---|
522 | logger.error("\n@@@@@@@@@ Error writing to one of:\n\t" + seedURLsFile
|
---|
523 | + "\n\t" + urlFilterFile
|
---|
524 | + "\n\t" + domainURLsFile
|
---|
525 | + "\n\t" + topSiteMatchesFile, ioe);
|
---|
526 | }
|
---|
527 |
|
---|
528 | /*
|
---|
529 | // BEGIN DEBUG
|
---|
530 | logger.debug("@@@@ TopSitesMap contains: ");
|
---|
531 | for(Map.Entry<String, String> entry : topSitesMap.entrySet()) {
|
---|
532 | String topSite = entry.getKey();
|
---|
533 | String urlPattern = entry.getValue();
|
---|
534 | logger.debug(topSite + " - " + urlPattern);
|
---|
535 | } // END DEBUG
|
---|
536 | */
|
---|
537 | }
|
---|
538 |
|
---|
539 | private String stripSubDomain(String url) {
|
---|
540 | int index = url.indexOf(".");
|
---|
541 | if(index != -1) {
|
---|
542 | url = url.substring(index+1);
|
---|
543 | }
|
---|
544 | return url;
|
---|
545 | }
|
---|
546 |
|
---|
547 |
|
---|
548 | /**
|
---|
549 | * @return true when a seedURL's domain exactly matches a topsite such as blogspot.com,
|
---|
550 | * with or without www. prefix. This method tests for such as case as it would be dangerous
|
---|
551 | * to do a SUBDOMAIN-COPY on such a site and thereby crawl that entire domain.
|
---|
552 | */
|
---|
553 | private boolean isExactDomainMatch(String seedURLDomain, String domain) {
|
---|
554 | // check for an exact match as-is
|
---|
555 | if(seedURLDomain.equals(domain)) {
|
---|
556 | return true;
|
---|
557 | }
|
---|
558 |
|
---|
559 | // else check if with or without a www. prefix we have an exact match with domain
|
---|
560 | if(seedURLDomain.startsWith("www.")) {
|
---|
561 | if(seedURLDomain.substring(4).equals(domain)) {
|
---|
562 | return true;
|
---|
563 | }
|
---|
564 | } else {
|
---|
565 | if(domain.equals("www."+seedURLDomain)) {
|
---|
566 | return true;
|
---|
567 | }
|
---|
568 | }
|
---|
569 |
|
---|
570 | return false;
|
---|
571 | }
|
---|
572 |
|
---|
573 |
|
---|
574 | /**
|
---|
575 | * Check if the domain of the seedurl, either in its entirety or when stripped of
|
---|
576 | * www/subdomains, is in the list of top sites.
|
---|
577 | * If it is, and the given url matches the regex for that topsite, then add the url to the
|
---|
578 | * whitelist and a regex disallowing the rest of the topsite to the url regex filter file.
|
---|
579 | * @param fullSeedDomain: domain of seedURL without the protocol. May include www. prefix.
|
---|
580 | * @return one of the following values:
|
---|
581 | * - This function returns null if the seedURL's domain does not match any of the topsites.
|
---|
582 | * - The empty String is returned if the seedURL's domain matched a topsite but no (allowed-
|
---|
583 | * url-pattern) value was defined for it. The empty String is also returned if the seedURL's
|
---|
584 | * domain exactly matched a topsite and had a value of SUBDOMAIN-COPY, because we still don't
|
---|
585 | * want to blindly crawl a topsite (as would happen with SUBDOMAIN-COPY).
|
---|
586 | * - A non-emptry String is returned if the seedURL's domain matched a topsite and a value
|
---|
587 | * was defined for it. (The value will be one of "SUBDOMAIN-COPY", "SINGLEPAGE" or an allowed
|
---|
588 | * URL pattern.
|
---|
589 | */
|
---|
590 | private String isURLinTopSitesMap(String fullSeedDomain) {
|
---|
591 | boolean keepLooping = true;
|
---|
592 |
|
---|
593 | String domain = fullSeedDomain;
|
---|
594 |
|
---|
595 | // domain aprameter will have retained www or subdomains, but is stripped of protocol
|
---|
596 |
|
---|
597 | // keep looping, stripping subdomains from url and checking if it matches a topsite domain
|
---|
598 | // if it does, return the value for that topsite domain in the topSitesMap
|
---|
599 | // If no match at all, return null.
|
---|
600 | do {
|
---|
601 |
|
---|
602 | String allowed_url_pattern = topSitesMap.get(domain);
|
---|
603 | if(allowed_url_pattern != null) { // if topSitesMap.containsKey(domain);
|
---|
604 | // there's an entry for the URL in the topSitesMap
|
---|
605 | logger.debug("##### A top site matches URL domain " + domain);
|
---|
606 |
|
---|
607 | // if we're dealing with SUBDOMAIN-COPY, then the fullSeedDomain, with or without
|
---|
608 | // www prefix, should not exactly match the topSitesMap domain
|
---|
609 | // e.g. we don't want to crawl a seed URL with domain www.blogspot.com
|
---|
610 | // despite it matching topsite blogspot.com with a value of SUBDOMAIN-COPY.
|
---|
611 |
|
---|
612 | if(allowed_url_pattern.equals(SUBDOMAIN_COPY) && isExactDomainMatch(fullSeedDomain, domain)) {
|
---|
613 | return ""; // means don't crawl site, write url into unprocessed-topsite-matches file
|
---|
614 | }
|
---|
615 | return allowed_url_pattern;
|
---|
616 | }
|
---|
617 | // else, no entry for the URL in the topSitesMap
|
---|
618 | // We're not done yet: strip subDomain from URL and check it against topSitesMap again
|
---|
619 |
|
---|
620 | String newDomain = stripSubDomain(domain);
|
---|
621 | if(domain.equals(newDomain)) {
|
---|
622 | keepLooping = false;
|
---|
623 | } else {
|
---|
624 | domain = newDomain;
|
---|
625 | }
|
---|
626 | } while(keepLooping);
|
---|
627 |
|
---|
628 | // url in entirety or stripped of subdomains did not match any of the topsites
|
---|
629 | return null;
|
---|
630 | }
|
---|
631 |
|
---|
632 | private boolean isListedInFilterList(Map<String, Integer> filterListMap, String url) {
|
---|
633 | //Set<Map.Entry<String,Integer>> entries = filterListMap.entrySet();
|
---|
634 | //Iterator<Map.Entry<String, Integer>> i = entries.iterator();
|
---|
635 | //while(i.hasNext()) {
|
---|
636 | // Map.Entry<String, Integer> entry = i.next();
|
---|
637 | for(Map.Entry<String,Integer> entry : filterListMap.entrySet()) {
|
---|
638 | String urlPattern = entry.getKey();
|
---|
639 | Integer matchRule = entry.getValue();
|
---|
640 |
|
---|
641 | if(matchRule == LIST_ENTRY_CONTAINS && url.contains(urlPattern)) {
|
---|
642 | return true;
|
---|
643 | }
|
---|
644 | else if(matchRule == LIST_ENTRY_STARTSWITH && url.startsWith(urlPattern)) {
|
---|
645 | return true;
|
---|
646 | }
|
---|
647 | else if(matchRule == LIST_ENTRY_ENDSWITH && url.endsWith(urlPattern)) {
|
---|
648 | return true;
|
---|
649 | }
|
---|
650 | else if(matchRule == LIST_ENTRY_MATCHES && url.equals(urlPattern)) {
|
---|
651 | return true;
|
---|
652 | }
|
---|
653 | // else check the rest of the filter list against this url
|
---|
654 | // before returning false to be certain it's not been listed in the filter list
|
---|
655 | }
|
---|
656 |
|
---|
657 | return false;
|
---|
658 | }
|
---|
659 |
|
---|
660 | /**
|
---|
661 | * Returns true if the url or pattern is found in the blacklist file.
|
---|
662 | * Note that if eventually the same url pattern is found in the greylist or whitelist too,
|
---|
663 | * it won't get blacklisted after all. But that's not implemented here.
|
---|
664 | */
|
---|
665 | public boolean isBlacklisted(String url) {
|
---|
666 | boolean isBlackListed = isListedInFilterList(blackList, url);
|
---|
667 |
|
---|
668 | // if any portion of the URL contains the word "livejasmin", or even "jasmin" actually,
|
---|
669 | // then it's an adult site, so blacklist the entire domain if it wasn't already blacklisted
|
---|
670 | String domainWithoutProtocol = Utility.getDomainForURL(url, false); // remove protocol
|
---|
671 | if(!isBlackListed && url.contains("jasmin")) {
|
---|
672 | logger.warn("### Blacklisting additional domain (likely an adult site): " + domainWithoutProtocol);
|
---|
673 | blackList.put(domainWithoutProtocol, LIST_ENTRY_CONTAINS);
|
---|
674 | }
|
---|
675 | return isBlackListed;
|
---|
676 | }
|
---|
677 |
|
---|
678 | /**
|
---|
679 | * Returns true if the url or pattern is explicitly mentioned in the greylist file.
|
---|
680 | * Will eventually take precedence over if the same URL pattern was mentioned in the blacklist.
|
---|
681 | * Will eventually be pre-empted into the whitelist if mentioned in the whitelist.
|
---|
682 | */
|
---|
683 | public boolean isGreylisted(String url) {
|
---|
684 | // auto-translated product sites
|
---|
685 | return isListedInFilterList(greyList, url);
|
---|
686 | }
|
---|
687 |
|
---|
688 | /**
|
---|
689 | * Returns true if the url or pattern is explicitly mentioned in the whitelist file
|
---|
690 | * Its mention in a whitelist moreover overrides any mention in the blacklist and greylist.
|
---|
691 | */
|
---|
692 | public boolean isWhitelisted(String url) {
|
---|
693 | return isListedInFilterList(whiteList, url);
|
---|
694 | }
|
---|
695 |
|
---|
696 | /**
|
---|
697 | * Checks URL parameter against each line ("filter") of conf/url-black|grey|whitelist-filter.txt to decide
|
---|
698 | * whether it is in the mentioned black|grey|white list.
|
---|
699 | * Filters don't represent actual regex, just ^ and $ as start and end terminators.
|
---|
700 | * By not having this method deal with actual regex for filters, this has the advantage that
|
---|
701 | * we don't have to remember to escape or double escape each filter to turn it into a regex.
|
---|
702 | */
|
---|
703 | public void initURLFilterList(Map<String, Integer> list, String filterListFilename) {
|
---|
704 |
|
---|
705 | // if filterListFilename does not exist in the conf folder, just return
|
---|
706 | if(MY_CLASSLOADER.getResource(filterListFilename) == null) {
|
---|
707 | logger.warn("Filter list filename: " + filterListFilename + " does not exist");
|
---|
708 | return;
|
---|
709 | }
|
---|
710 |
|
---|
711 | try (
|
---|
712 | BufferedReader reader = new BufferedReader(new InputStreamReader(MY_CLASSLOADER.getResourceAsStream(filterListFilename), "UTF-8"));
|
---|
713 | ) {
|
---|
714 | String filter = null;
|
---|
715 | while((filter = reader.readLine()) != null) {
|
---|
716 | // skip comments and empty lines
|
---|
717 | filter = filter.trim();
|
---|
718 | if(filter.equals("") || filter.startsWith("#")) {
|
---|
719 | continue;
|
---|
720 | }
|
---|
721 |
|
---|
722 | if(filter.startsWith("^") && filter.endsWith("$")) {
|
---|
723 | filter = filter.substring(1, filter.length()-1);
|
---|
724 | list.put(filter, LIST_ENTRY_MATCHES);
|
---|
725 | }
|
---|
726 | else if(filter.startsWith("^")) {
|
---|
727 | filter = filter.substring(1);
|
---|
728 | list.put(filter, LIST_ENTRY_STARTSWITH);
|
---|
729 | //logger.debug("Match filter startswith: " + filter);
|
---|
730 | }
|
---|
731 | else if(filter.endsWith("$")) {
|
---|
732 | filter = filter.substring(0, filter.length()-1);
|
---|
733 | list.put(filter, LIST_ENTRY_ENDSWITH);
|
---|
734 | //logger.debug("@@@ Match filter endswith: " + filter);
|
---|
735 | }
|
---|
736 | else {
|
---|
737 | list.put(filter, LIST_ENTRY_CONTAINS);
|
---|
738 | }
|
---|
739 | //logger.debug("Got filter: " + filter);
|
---|
740 | }
|
---|
741 |
|
---|
742 | } catch (IOException ioe) {
|
---|
743 | logger.error("@@@@@@@@@ Error reading into map from file " + filterListFilename, ioe);
|
---|
744 | }
|
---|
745 |
|
---|
746 | }
|
---|
747 |
|
---|
748 | /** Maintain a count of all WET files processed. */
|
---|
749 | public void setWETFileCount(int count) { this.wetFileCount = count; }
|
---|
750 |
|
---|
751 | /** Maintain a count of all WET records processed. */
|
---|
752 | //public int getRecordCount() { return this.totalRecordCount; }
|
---|
753 | //public void addToRecordCount(int count) { this.totalRecordCount += count; }
|
---|
754 | public void setRecordCount(int count) { this.totalRecordCount = count; }
|
---|
755 |
|
---|
756 | public void processAllWETFilesOfCrawl(File ccrawlWETFileDir) {
|
---|
757 |
|
---|
758 | // Will list all the warc.wet files in the input directory or else their gzipped versions
|
---|
759 | File[] WETFiles = ccrawlWETFileDir.listFiles(new WETFilenameFilter());
|
---|
760 |
|
---|
761 | int wetRecordCount = 0;
|
---|
762 | int wetFileCount = 0;
|
---|
763 |
|
---|
764 | for(int i = 0; i < WETFiles.length; i++) {
|
---|
765 | File WETFile = WETFiles[i];
|
---|
766 | logger.debug("Processing WETfile: " + WETFile);
|
---|
767 |
|
---|
768 | // Any .gz files listed means they haven't been unzipped yet. So unzip.
|
---|
769 | String WETFilename = WETFile.toString();
|
---|
770 | if(WETFilename.endsWith(".gz")) {
|
---|
771 | File GZippedWETFile = WETFile;
|
---|
772 | String WETGZippedFilename = WETFilename;
|
---|
773 | WETFilename = WETFilename.substring(0, WETFilename.lastIndexOf(".gz"));
|
---|
774 |
|
---|
775 | WETFile = new File(WETFilename);
|
---|
776 | Utility.unzipFile(GZippedWETFile, WETFile);
|
---|
777 | }
|
---|
778 | // hereafter all WETFiles should refer to the unzipped version
|
---|
779 | // Check the unzipped WETFile exists
|
---|
780 |
|
---|
781 | if(!WETFile.exists() || !WETFile.isFile()) {
|
---|
782 | logger.error("Error: " + WETFile + " does not exist (failure to unzip?)");
|
---|
783 | return;
|
---|
784 | }
|
---|
785 |
|
---|
786 | // Finally, we can process this WETFile's records into the keep and discard pile
|
---|
787 | wetFileCount++;
|
---|
788 | logger.debug("Off to process " + WETFile);
|
---|
789 | String crawlID = ccrawlWETFileDir.getName(); // something like CC-MAIN-YYYY-##-wet-files
|
---|
790 | crawlID = crawlID.substring("CC-MAIN-".length(), crawlID.indexOf("-wet-files")); // YYYY-##
|
---|
791 | WETProcessor wetFileProcessor = new WETProcessor(WETFile, crawlID, this);
|
---|
792 | wetFileProcessor.processWETFile();
|
---|
793 | wetRecordCount += wetFileProcessor.getRecordCount();
|
---|
794 | }
|
---|
795 |
|
---|
796 | // for information purposes
|
---|
797 | this.setWETFileCount(wetFileCount);
|
---|
798 | this.setRecordCount(wetRecordCount);
|
---|
799 | }
|
---|
800 |
|
---|
801 |
|
---|
802 | // --------------- STATIC METHODS AND INNER CLASSED USED BY MAIN -------------- //
|
---|
803 | public static void printUsage() {
|
---|
804 | System.err.println("Run this program as:");
|
---|
805 | System.err.println("\tCCWetProcessor <path to 'ccrawl-data' input folder> <output folder path>");
|
---|
806 | }
|
---|
807 |
|
---|
808 | /** Filename filter to only list warc.wet files or else warc.wet.gz files
|
---|
809 | * for which unzipped warc.wet equivalents don't yet exist.
|
---|
810 | */
|
---|
811 | private static class WETFilenameFilter implements FilenameFilter {
|
---|
812 |
|
---|
813 | public boolean accept(File dir, String name) {
|
---|
814 | if(name.endsWith(".warc.wet")) {
|
---|
815 | logger.debug("Will include " + name + " for processing.");
|
---|
816 | return true;
|
---|
817 | }
|
---|
818 |
|
---|
819 | if(name.endsWith(".warc.wet.gz")) {
|
---|
820 | String nameWithoutGZext = name.substring(0, name.lastIndexOf(".gz"));
|
---|
821 | File unzippedVersion = new File(dir, nameWithoutGZext);
|
---|
822 | if(unzippedVersion.exists()) {
|
---|
823 | logger.debug("--- Unzipped version " + unzippedVersion + " exists.");
|
---|
824 | logger.debug("Skipping " + name);
|
---|
825 | return false; // don't count gzipped version if unzipped version exists.
|
---|
826 | }
|
---|
827 | else {
|
---|
828 | logger.debug("Only zipped version " + name + " exists.");
|
---|
829 | return true; // No unzipped version, so have to work with gzipped version
|
---|
830 | }
|
---|
831 | }
|
---|
832 |
|
---|
833 | // we're not even interested in any other file extensions
|
---|
834 | logger.debug("Not a WET file. Skipping " + name);
|
---|
835 | return false;
|
---|
836 | }
|
---|
837 | }
|
---|
838 |
|
---|
839 |
|
---|
840 | private static class CCrawlWETFolderFilenameFilter implements FilenameFilter {
|
---|
841 |
|
---|
842 | public boolean accept(File dir, String name) {
|
---|
843 | File f = new File (dir, name);
|
---|
844 | if(f.isDirectory()) {
|
---|
845 | if(name.matches("CC-MAIN-\\d{4}-\\d{2}-wet-files")) {
|
---|
846 | return true;
|
---|
847 | }
|
---|
848 | }
|
---|
849 | else {
|
---|
850 | logger.info("File " + f + " is not a directory");
|
---|
851 | }
|
---|
852 | return false;
|
---|
853 | }
|
---|
854 | }
|
---|
855 |
|
---|
856 | public static void main(String[] args) {
|
---|
857 | if(args.length != 2) {
|
---|
858 | printUsage();
|
---|
859 | return;
|
---|
860 | }
|
---|
861 |
|
---|
862 | File commoncrawlDir = new File(args[0]);
|
---|
863 | if(!commoncrawlDir.exists() || !commoncrawlDir.isDirectory()) {
|
---|
864 | logger.error("Error: " + args[0] + " does not exist or is not a directory");
|
---|
865 | return;
|
---|
866 | }
|
---|
867 |
|
---|
868 | File outFolder = new File(args[1]);
|
---|
869 | if(!outFolder.exists() || !outFolder.isDirectory()) {
|
---|
870 | logger.error("Error: " + args[1] + " does not exist or is not a directory.");
|
---|
871 | return;
|
---|
872 | }
|
---|
873 |
|
---|
874 | try {
|
---|
875 | CCWETProcessor ccWETFilesProcessor = new CCWETProcessor(commoncrawlDir, outFolder);
|
---|
876 |
|
---|
877 | File[] ccrawlFolders = commoncrawlDir.listFiles(new CCrawlWETFolderFilenameFilter());
|
---|
878 |
|
---|
879 | for(int i = 0; i < ccrawlFolders.length; i++) {
|
---|
880 | File ccrawlFolder = ccrawlFolders[i];
|
---|
881 | logger.info("About to process commoncrawl WET files folder: " + ccrawlFolder);
|
---|
882 | ccWETFilesProcessor.processAllWETFilesOfCrawl(ccrawlFolder);
|
---|
883 | }
|
---|
884 |
|
---|
885 | // create the global files of all domains, seedURLs and regex-urlfilters across all wet files of all commoncrawls
|
---|
886 | // The former is the only unique one. seedURLs and regex-urlfilters are
|
---|
887 | // repeated on a per site/domain basis too, stored in the sites folder
|
---|
888 | File seedURLsFile = new File(outFolder, "seedURLs.txt");
|
---|
889 | File urlFilterFile = new File(outFolder, "regex-urlfilter.txt");
|
---|
890 | File domainURLsFile = new File(outFolder, "all-domain-urls.txt");
|
---|
891 | File topSitesMatchedFile = new File(outFolder, "unprocessed-topsite-matches.txt");
|
---|
892 | File possibleProductSitesFile = new File(outFolder, "possible-product-sites.txt");
|
---|
893 |
|
---|
894 | ccWETFilesProcessor.createSeedURLsFiles(seedURLsFile, urlFilterFile, domainURLsFile, topSitesMatchedFile, possibleProductSitesFile);
|
---|
895 |
|
---|
896 | logger.info("\n*** Inspect urls in greylist at " + ccWETFilesProcessor.greyListedFile + "\n");
|
---|
897 |
|
---|
898 | logger.info("\n*** Check " + topSitesMatchedFile + " for sites not prepared for crawling because they matched top sites for which no regex of allowed url patterns were specified in sites-too-big-to-exhaustively-crawl.txt.\n");
|
---|
899 |
|
---|
900 |
|
---|
901 | } catch(Exception e) {
|
---|
902 | // can get an exception when instantiating CCWETProcessor instance
|
---|
903 | logger.error(e.getMessage(), e);
|
---|
904 | }
|
---|
905 |
|
---|
906 | return;
|
---|
907 |
|
---|
908 | }
|
---|
909 | }
|
---|