1 | package org.greenstone.atea;
|
---|
2 |
|
---|
3 |
|
---|
4 | import java.io.*;
|
---|
5 | import java.util.Properties;
|
---|
6 | import java.util.zip.GZIPInputStream;
|
---|
7 | import java.util.Iterator;
|
---|
8 | import java.util.Set;
|
---|
9 | import java.util.TreeSet;
|
---|
10 |
|
---|
11 | import org.apache.log4j.Logger;
|
---|
12 |
|
---|
13 | /**
|
---|
14 | * The main() method of this class takes a folder of warc.wet(.gz) files and goes through
|
---|
15 | * the WET records in each, putting each WET record into a file. Each file is put into a
|
---|
16 | * keep or discard folder, based on content-length and number of lines.
|
---|
17 | * A single instance of the WETProcessor class processes a single unzipped warc.wet file.
|
---|
18 | *
|
---|
19 | * To compile, including the jars in lib/ for compiling.
|
---|
20 | * maori-lang-detection/src$ javac -cp ".:../lib/*" org/greenstone/atea/WETProcessor.java
|
---|
21 | *
|
---|
22 | * To run, passing the log4j and other properties files in conf/ folder:
|
---|
23 | * maori-lang-detection/src$ java -cp ".:../conf:../lib/*" org.greenstone.atea.WETProcessor <folder containing warc.wet(.gz) files> <outputFolder>
|
---|
24 | *
|
---|
25 | * e.g.
|
---|
26 | * - java -cp ".:../conf:../lib/*" org.greenstone.atea.WETProcessor ../tmp/processWET /Scratch/ak19/gs3-extensions/maori-lang-detection/tmp/processedWET
|
---|
27 | * - java -cp ".:../conf:../lib/*" org.greenstone.atea.WETProcessor ../tmp/processWET /Scratch/ak19/gs3-extensions/maori-lang-detection/tmp/processedWET 2>&1 | less
|
---|
28 | *
|
---|
29 | */
|
---|
30 | public class WETProcessor {
|
---|
31 | private static Logger logger = Logger.getLogger(org.greenstone.atea.WETProcessor.class.getName());
|
---|
32 | private static Properties configProperties = new Properties();
|
---|
33 |
|
---|
34 | // In Java, can initialize static final variables inside a static block
|
---|
35 | // But the unavoidable try/catch in this static block prevents initialization of
|
---|
36 | // the static final int variables (seen further below) inside the block itself,
|
---|
37 | // that therefore need to be declared and initialized thereafter.
|
---|
38 | static {
|
---|
39 | // load up the properties from the config file
|
---|
40 | try (InputStream infile = org.greenstone.atea.WETProcessor.class.getClassLoader().getResourceAsStream("config.properties")) {
|
---|
41 | configProperties = new Properties();
|
---|
42 | configProperties.load(infile);
|
---|
43 | //infile.close();
|
---|
44 |
|
---|
45 | } catch(Exception e) {
|
---|
46 | System.err.println("Exception attempting to read properties from config.properties.");
|
---|
47 | logger.error("Exception attempting to read properties from config.properties.");
|
---|
48 | e.printStackTrace();
|
---|
49 | }
|
---|
50 | }
|
---|
51 |
|
---|
52 | // Providing fall-back cuttoff values if config.properties doesn't load
|
---|
53 | // or doesn't have the named props. But what happens when Integer.parseInt throws an exception?
|
---|
54 | /*
|
---|
55 | private static final int MIN_CONTENT_LENGTH = Integer.parseInt(configProperties.getProperty("WETprocessor.min.content.length", "100"));
|
---|
56 | private static final int MIN_LINE_COUNT= Integer.parseInt(configProperties.getProperty("WETprocessor.min.line.count", "2"));
|
---|
57 | private static final int MIN_CONTENT_LENGTH_WRAPPED_LINE = Integer.parseInt(configProperties.getProperty("WETprocessor.min.content.length.wrapped.line", "500"));
|
---|
58 | private static final int MIN_SPACES_IN_A_WRAPPED_LINE = Integer.parseInt(configProperties.getProperty("WETprocessor.min.spaces.per.wrapped.line", "10"));
|
---|
59 | */
|
---|
60 | private static final int MAX_WORD_LENGTH = Integer.parseInt(configProperties.getProperty("WETprocessor.max.word.length", "15")); // to identify and skip web pages where content consists of words glued together (with no spaces)
|
---|
61 | private static final int MIN_NUM_WORDS = Integer.parseInt(configProperties.getProperty("WETprocessor.min.num.words", "20"));
|
---|
62 | private static final int MAX_WORDS_CAMELCASE = Integer.parseInt(configProperties.getProperty("WETprocessor.max.words.camelcase", "10"));
|
---|
63 |
|
---|
64 | // File paths shared across WETProcessor instances
|
---|
65 | private static File discardFolder;
|
---|
66 | private static File keepFolder;
|
---|
67 | private static File keepURLsFile;
|
---|
68 | private static File discardURLsFile;
|
---|
69 |
|
---|
70 | // WARC WET header lines and header line prefixes of interest
|
---|
71 | static final String WARC_RECORD_START = "WARC/1.0";
|
---|
72 | static final String WARC_INFO_HEADER = "WARC-Type: warcinfo";
|
---|
73 | static final String WARC_TARGET_URI_HEADER_PREFIX = "WARC-Target-URI:";
|
---|
74 | static final String WARC_CONTENT_LENGTH_HEADER_PREFIX = "Content-Length:";
|
---|
75 |
|
---|
76 | // Keep a count of all the records that all WETProcessors instantiated
|
---|
77 | // by our main method combined have processed
|
---|
78 | //private static int recordCount = 0;
|
---|
79 |
|
---|
80 | private final File outputFolder;
|
---|
81 | private final String WETFileID;
|
---|
82 |
|
---|
83 |
|
---|
84 | /**
|
---|
85 | * WET processor processes a single warc.wet file containing multiple WET records
|
---|
86 | * containing text identified as primary langcode=mri. Each individual WET record is written
|
---|
87 | * out to a uniquely named file in either the keep or discard folder depending on the WET
|
---|
88 | * record's content length and number of lines of actual content (excluding WARC headers).
|
---|
89 | */
|
---|
90 | public WETProcessor(File inFile, File outFolder) {
|
---|
91 | this.outputFolder = outFolder;
|
---|
92 |
|
---|
93 | StringBuilder record = null;
|
---|
94 | String line = null;
|
---|
95 | boolean readingRecord = false;
|
---|
96 |
|
---|
97 | String WARCtargetURI = "";
|
---|
98 |
|
---|
99 | int recordCount = 0;
|
---|
100 |
|
---|
101 | int contentLength = -1; // of record
|
---|
102 | int lineCount = -1; // actual number of non-empty lines in record body (i.e. excludes WET/WARC headers)
|
---|
103 |
|
---|
104 | // We just want a unique recordID prefix, which we get from the wet file name suffix:
|
---|
105 | // inFile name looks something like MAORI-CC-2019-30-20190902100139-000000.warc.wet
|
---|
106 | // the prefix will be everything after the last hyphen and without file extension,
|
---|
107 | // so "000000" in our example. Then suffix the recordCount (keeping track of the current
|
---|
108 | // WET record) to get a unique filename to store each WET record into.
|
---|
109 |
|
---|
110 | String fileID = inFile.getName();
|
---|
111 | fileID = fileID.substring(fileID.lastIndexOf("-")+1);
|
---|
112 | fileID = fileID.substring(0, fileID.indexOf("."));
|
---|
113 | this.WETFileID = fileID;
|
---|
114 |
|
---|
115 |
|
---|
116 | // read from WETfile
|
---|
117 | try (
|
---|
118 | BufferedReader reader = new BufferedReader(new FileReader(inFile));
|
---|
119 | BufferedWriter keepURLsWriter = new BufferedWriter(new FileWriter(keepURLsFile, true));
|
---|
120 | BufferedWriter discardURLsWriter = new BufferedWriter(new FileWriter(discardURLsFile, true)); // true to append
|
---|
121 | ) {
|
---|
122 |
|
---|
123 | while((line = reader.readLine()) != null) { // readLine removes newline separator
|
---|
124 |
|
---|
125 | if(line.startsWith(WARC_INFO_HEADER)) { // "WARC-Type: warcinfo"
|
---|
126 | readingRecord = false;
|
---|
127 | record = null; // drop this record, which is just an info record not actual web page's text
|
---|
128 | recordCount--;
|
---|
129 | continue;
|
---|
130 | }
|
---|
131 |
|
---|
132 | if(line.startsWith(WARC_RECORD_START)) { // "WARC/1.0" means finished previous WET record
|
---|
133 | // process any previous record
|
---|
134 | if(record != null) {
|
---|
135 | processWETrecord(keepURLsWriter, discardURLsWriter,
|
---|
136 | recordCount, contentLength, lineCount,
|
---|
137 | WARCtargetURI, record.toString());
|
---|
138 | record = null;
|
---|
139 | contentLength = -1;
|
---|
140 | lineCount = -1;
|
---|
141 | }
|
---|
142 |
|
---|
143 | recordCount++;
|
---|
144 | // get ready to start a new record
|
---|
145 | readingRecord = true;
|
---|
146 | record = new StringBuilder();
|
---|
147 | }
|
---|
148 |
|
---|
149 | if(readingRecord) { // append current line to current record
|
---|
150 |
|
---|
151 | if(line.startsWith(WARC_TARGET_URI_HEADER_PREFIX)) { // "WARC-Target-URI:"
|
---|
152 | // get and store the value
|
---|
153 | WARCtargetURI = line.substring(WARC_TARGET_URI_HEADER_PREFIX.length()).trim();
|
---|
154 | }
|
---|
155 |
|
---|
156 | record.append(line + "\n"); // add back (unix style) line ending
|
---|
157 |
|
---|
158 | // if the line is non-empty
|
---|
159 | // AND if we've started counting lines, which happens only when the current
|
---|
160 | // line is past WARC/WET headers and we're into the actual body portion
|
---|
161 | // of the WET record,
|
---|
162 | // start incrementing the line counter.
|
---|
163 | if(lineCount >= 0 && !line.trim().equals("")) {
|
---|
164 | lineCount++;
|
---|
165 | }
|
---|
166 | else if(line.startsWith(WARC_CONTENT_LENGTH_HEADER_PREFIX)) { // "Content-Length:"
|
---|
167 | String contentLengthStr = line.substring(WARC_CONTENT_LENGTH_HEADER_PREFIX.length()).trim();
|
---|
168 | contentLength = Integer.parseInt(contentLengthStr);
|
---|
169 | lineCount = 0;
|
---|
170 | }
|
---|
171 |
|
---|
172 | }
|
---|
173 |
|
---|
174 | }
|
---|
175 |
|
---|
176 | // flush the last record. If it was a warcinfo record, record would be null here
|
---|
177 | if(record != null) {
|
---|
178 | processWETrecord(keepURLsWriter, discardURLsWriter,
|
---|
179 | recordCount, contentLength, lineCount,
|
---|
180 | WARCtargetURI, record.toString());
|
---|
181 | record = null;
|
---|
182 | }
|
---|
183 |
|
---|
184 | } catch(IOException ioe) {
|
---|
185 | ioe.printStackTrace();
|
---|
186 | }
|
---|
187 | }
|
---|
188 |
|
---|
189 | /**
|
---|
190 | * Determines if a WET record belongs in the keep or discard pile depending on if it
|
---|
191 | * contains enough text, based on contentLength and line count of the record body.
|
---|
192 | * Then writes out the WET record to a uniquely named file in the keep or discard folder,
|
---|
193 | * and writes out the WET record's URL to the keepURLs.txt file or discardURLs.txt file.
|
---|
194 | */
|
---|
195 | private void processWETrecord(BufferedWriter keepURLsWriter, BufferedWriter discardURLsWriter,
|
---|
196 | int recordID, int contentLength, int lineCount,
|
---|
197 | String recordURI, String record)
|
---|
198 | {
|
---|
199 | System.err.println("WET #" + this.WETFileID + " record #" + recordID
|
---|
200 | + " - contentLength: " + contentLength
|
---|
201 | + " - lineCount: " + lineCount);
|
---|
202 | System.err.println("URI: " + recordURI);
|
---|
203 | //System.err.println(record);
|
---|
204 | //System.err.println("--------------------------");
|
---|
205 |
|
---|
206 | String paddedFileName = String.format("%04d.txt", recordID);
|
---|
207 |
|
---|
208 | File parentFolder = null;
|
---|
209 |
|
---|
210 | // want to match "product(s)" but not "production"
|
---|
211 | //if(recordURI.matches(".*/?product[^a-rt-z].*")) {//if(recordURI.matches(".*/?products?/?.*")) {
|
---|
212 |
|
---|
213 |
|
---|
214 | /*
|
---|
215 | if(recordURI.contains("product") && !recordURI.contains("production")) {
|
---|
216 |
|
---|
217 | // don't want a "translated" product site/online store
|
---|
218 | // These curiously often tend to have "product(s)" in the URL
|
---|
219 | parentFolder = WETProcessor.discardFolder;
|
---|
220 | }
|
---|
221 |
|
---|
222 | else if(lineCount >= MIN_LINE_COUNT && contentLength >= MIN_CONTENT_LENGTH) {
|
---|
223 | parentFolder = WETProcessor.keepFolder;
|
---|
224 | System.err.println("@@@KEEPING");
|
---|
225 | } else if(contentLength >= MIN_CONTENT_LENGTH_WRAPPED_LINE) {
|
---|
226 | int countSpaces = 0;
|
---|
227 | for(int i = 0; i < record.length(); i++) {
|
---|
228 | if(record.charAt(i) == ' ') countSpaces++;
|
---|
229 | }
|
---|
230 | if(countSpaces >= MIN_SPACES_IN_A_WRAPPED_LINE) {
|
---|
231 | // So we have at least 500 chars (possibly on a single wrapped line)
|
---|
232 | // containing at least 10 spaces. Such a record is also worth keeping.
|
---|
233 | parentFolder = WETProcessor.keepFolder;
|
---|
234 | }
|
---|
235 | }
|
---|
236 | */
|
---|
237 |
|
---|
238 | if(isInDiscardFilter(recordURI)) {
|
---|
239 | parentFolder = WETProcessor.discardFolder;
|
---|
240 | }
|
---|
241 | else if(isInCheckFilter(recordURI)) { // products sites
|
---|
242 | parentFolder = WETProcessor.discardFolder; // TODO: checkfolder
|
---|
243 | } else {
|
---|
244 | // If a web page's WET record contains a certain minimum number of words,
|
---|
245 | // we will think it's a meaningful web page and has sufficient content for text analysis
|
---|
246 | // to have been successful. Cut off values at present are:
|
---|
247 | // - a minimum of 20 words
|
---|
248 | // - a word consists of 1 to 15 chars demarcated by spaces. Any more chars may point
|
---|
249 | // to words having been glued together. This is used by irrelevant sites and moreover
|
---|
250 | // can't be analysed for language, so may not be actually MRI.
|
---|
251 |
|
---|
252 | // Though StringTokenizer still in use, as seen in discussion at
|
---|
253 | // https://stackoverflow.com/questions/6983856/why-is-stringtokenizer-deprecated
|
---|
254 | // String.split(regex) seems better for splitting on general whitespace
|
---|
255 | String[] allWords = record.split("\\s");
|
---|
256 | int validWordCount = 0;
|
---|
257 | int numCamelCaseWords = 0;
|
---|
258 | for(int i = 0; i < allWords.length; i++) {
|
---|
259 | String word = allWords[i];
|
---|
260 |
|
---|
261 | // throw away if n words contain camelcase, which is another case of words glued together
|
---|
262 | if(word.matches(".*[a-z][A-Z].*") && word.length() >= 5) {
|
---|
263 | numCamelCaseWords++;
|
---|
264 | }
|
---|
265 |
|
---|
266 | // In Maori, word length of 1 is not uncommon
|
---|
267 | // but let's skip camelcased words when counting valid words
|
---|
268 | else if(word.length() >= 1 && word.length() <= MAX_WORD_LENGTH) validWordCount++;
|
---|
269 | }
|
---|
270 |
|
---|
271 | // dump if too many camelcase words (ideally keep none of that kind?)
|
---|
272 | if(numCamelCaseWords >= MAX_WORDS_CAMELCASE) {
|
---|
273 | parentFolder = WETProcessor.discardFolder;
|
---|
274 | System.err.println("@@@DISCARDING - CAMELCASED CONTENTS");
|
---|
275 | }
|
---|
276 | else if(validWordCount >= MIN_NUM_WORDS) { // otherwise, keep anything with a sufficient number of valid words
|
---|
277 | parentFolder = WETProcessor.keepFolder;
|
---|
278 | System.err.println("@@@KEEPING");
|
---|
279 | }
|
---|
280 | }
|
---|
281 | // if parentFolder still not set, set to discard pile folder
|
---|
282 | if(parentFolder == null) {
|
---|
283 | parentFolder = WETProcessor.discardFolder;
|
---|
284 | System.err.println("@@@DISCARDING");
|
---|
285 | }
|
---|
286 |
|
---|
287 | try {
|
---|
288 | if (parentFolder == WETProcessor.keepFolder) {
|
---|
289 | keepURLsWriter.write(recordURI + "\n");
|
---|
290 | } else {
|
---|
291 | discardURLsWriter.write(recordURI + "\n");
|
---|
292 | }
|
---|
293 | } catch(Exception e) {
|
---|
294 | System.err.println("Unable to write URL");
|
---|
295 | e.printStackTrace();
|
---|
296 | }
|
---|
297 |
|
---|
298 | System.err.println("--------------------------");
|
---|
299 |
|
---|
300 | File outFile = new File(parentFolder, this.WETFileID + "-" + paddedFileName);
|
---|
301 |
|
---|
302 | try (BufferedWriter writer = new BufferedWriter(new FileWriter(outFile))) {
|
---|
303 | writer.write(record);
|
---|
304 | // Try-with-resources examples don't call close() explicitly:
|
---|
305 | // https://docs.oracle.com/javase/tutorial/essential/exceptions/tryResourceClose.html
|
---|
306 | //writer.close();
|
---|
307 | } catch(IOException ioe) {
|
---|
308 | ioe.printStackTrace();
|
---|
309 | System.err.println("\n@@@@@@@@@ Error writing to file " + outFile);
|
---|
310 | }
|
---|
311 | }
|
---|
312 |
|
---|
313 |
|
---|
314 | /**
|
---|
315 | * Takes as input the keepURLs.txt file generated by running WETProcessor instances.
|
---|
316 | * As output produces the URL seed list and regex-urlfilter text files required by nutch,
|
---|
317 | * https://cwiki.apache.org/confluence/display/nutch/NutchTutorial
|
---|
318 | */
|
---|
319 | public static void createSeedURLsFiles(File urlsFile, File seedURLsFile, File urlFilterFile) {
|
---|
320 | // Maintain Sets of unique domains and urls
|
---|
321 | // TreeSet: by default, "the elements are ordered using their natural ordering"
|
---|
322 | // (or by a Comparator provided at set creation time).
|
---|
323 | // Whereas HashSet doesn't guarantee ordering.
|
---|
324 | // So we get alphabetic sorting for free. And guaranteed log(n) for basic operations.
|
---|
325 |
|
---|
326 | Set<String> domainsSet = new TreeSet<String>();
|
---|
327 | Set<String> urlsSet = new TreeSet<String>();
|
---|
328 |
|
---|
329 | final String FILTER_REGEX_PREFIX = "+https?://([a-z0-9-]+\\.)*"; // https?://([a-z0-9-]+\.)*
|
---|
330 |
|
---|
331 | try (
|
---|
332 | BufferedReader reader = new BufferedReader(new FileReader(urlsFile));
|
---|
333 | ) {
|
---|
334 |
|
---|
335 | // read a URL at a time from urlsFile
|
---|
336 | String url = null;
|
---|
337 | String domain = null;
|
---|
338 | while((url = reader.readLine()) != null) { // readLine removes newline separator
|
---|
339 |
|
---|
340 | // work out domain. This retains any www. or subdomain prefix:
|
---|
341 | int startIndex = url.indexOf("//"); // http:// or https:// prefix
|
---|
342 | startIndex = (startIndex == -1) ? 0 : (startIndex+2); // skip past the protocol's // portion
|
---|
343 | domain = url.substring(startIndex);
|
---|
344 | int endIndex = domain.indexOf("/");
|
---|
345 | if(endIndex == -1) endIndex = domain.length();
|
---|
346 | domain = domain.substring(0, endIndex);
|
---|
347 |
|
---|
348 | //if(!domainsMap.containsKey(domain)) {
|
---|
349 | urlsSet.add(url);
|
---|
350 | domainsSet.add(domain);
|
---|
351 | //}
|
---|
352 | }
|
---|
353 | } catch (IOException ioe) {
|
---|
354 | ioe.printStackTrace();
|
---|
355 | System.err.println("\n@@@@@@@@@ Error reading in urls from file " + urlsFile);
|
---|
356 | }
|
---|
357 |
|
---|
358 | try (BufferedWriter seedURLsWriter = new BufferedWriter(new FileWriter(seedURLsFile))) {
|
---|
359 | Iterator<String> i = urlsSet.iterator();
|
---|
360 | while(i.hasNext()) {
|
---|
361 | String url = i.next();
|
---|
362 | seedURLsWriter.write(url + "\n");
|
---|
363 | }
|
---|
364 |
|
---|
365 | } catch (IOException ioe) {
|
---|
366 | ioe.printStackTrace();
|
---|
367 | System.err.println("\n@@@@@@@@@ Error writing to " + seedURLsFile);
|
---|
368 | }
|
---|
369 |
|
---|
370 | try (BufferedWriter urlFilterWriter = new BufferedWriter(new FileWriter(urlFilterFile))) {
|
---|
371 | Iterator<String> i = domainsSet.iterator();
|
---|
372 | // nutch.apache.org => +^https?://([a-z0-9-]+\.)*nutch\.apache\.org/
|
---|
373 | while(i.hasNext()) {
|
---|
374 | String domain = i.next();
|
---|
375 | domain = FILTER_REGEX_PREFIX + domain.replace(".", "\\.") + "/";
|
---|
376 | urlFilterWriter.write(domain + "\n");
|
---|
377 | }
|
---|
378 |
|
---|
379 | } catch (IOException ioe) {
|
---|
380 | ioe.printStackTrace();
|
---|
381 | System.err.println("\n@@@@@@@@@ Error writing to " + urlFilterFile);
|
---|
382 | }
|
---|
383 | }
|
---|
384 |
|
---|
385 | /**
|
---|
386 | * Checks URL parameter against each line ("filter") of conf/url-discard-filter.txt to decide
|
---|
387 | * whether it is in the discard list.
|
---|
388 | * Filters don't represent actual regex, just ^ and $ as start and end terminators.
|
---|
389 | * By not having this method deal with actual regex for filters, this has the advantage that
|
---|
390 | * we don't have to remember to escape or double escape each filter to turn it into a regex.
|
---|
391 | */
|
---|
392 | public boolean isInDiscardFilter(String url) {
|
---|
393 |
|
---|
394 | String discardFilterFile = "url-discard-filter.txt"; // in conf folder
|
---|
395 |
|
---|
396 | try (
|
---|
397 | BufferedReader reader = new BufferedReader(new InputStreamReader(org.greenstone.atea.WETProcessor.class.getClassLoader().getResourceAsStream(discardFilterFile), "UTF-8"));
|
---|
398 | ) {
|
---|
399 | String filter = null;
|
---|
400 | while((filter = reader.readLine()) != null) {
|
---|
401 | if(filter.trim().equals("")) {
|
---|
402 | continue;
|
---|
403 | }
|
---|
404 | //System.err.println("Got filter: " + filter);
|
---|
405 | if(filter.startsWith("^") && filter.endsWith("$") && url.equals(filter.substring(1, filter.length()-1))) {
|
---|
406 | System.err.println("*** Discarding url " + url + "\n\tas it MATCHES filter " + filter);
|
---|
407 | }
|
---|
408 | else if(filter.startsWith("^") && url.startsWith(filter.substring(1))) {
|
---|
409 | System.err.println("*** Discarding url " + url + "\n\tas it STARTS WITH filter " + filter);
|
---|
410 | return true;
|
---|
411 | }
|
---|
412 | else if(filter.endsWith("$") && url.endsWith(filter.substring(0, filter.length()-1))) {
|
---|
413 | System.err.println("*** Discarding url " + url + "\n\tas it ENDS WITH filter " + filter);
|
---|
414 | return true;
|
---|
415 | }
|
---|
416 | else if(url.contains(filter)) {
|
---|
417 | System.err.println("*** Discarding url " + url + "\n\tas it CONTAINS filter " + filter);
|
---|
418 | return true;
|
---|
419 | }
|
---|
420 |
|
---|
421 | }
|
---|
422 |
|
---|
423 | } catch (IOException ioe) {
|
---|
424 | ioe.printStackTrace();
|
---|
425 | System.err.println("\n@@@@@@@@@ Error reading from " + discardFilterFile);
|
---|
426 | }
|
---|
427 |
|
---|
428 | return false;
|
---|
429 | }
|
---|
430 |
|
---|
431 | // TODO
|
---|
432 | public boolean isInCheckFilter(String url) {
|
---|
433 | //System.err.println("isInCheckFilter(url) is not yet implemented");
|
---|
434 | return false;
|
---|
435 | }
|
---|
436 |
|
---|
437 | //public static int getRecordCount() { return recordCount; }
|
---|
438 |
|
---|
439 | public static void printUsage() {
|
---|
440 | System.err.println("Run this program as:");
|
---|
441 | System.err.println("\tWetProcessor <folder containing wet(.gz) files> <output folder path>");
|
---|
442 | }
|
---|
443 |
|
---|
444 | /** Filename filter to only list warc.wet files or else warc.wet.gz files
|
---|
445 | * for which unzipped warc.wet equivalents don't yet exist.
|
---|
446 | */
|
---|
447 | private static class WETFilenameFilter implements FilenameFilter {
|
---|
448 |
|
---|
449 | public boolean accept(File dir, String name) {
|
---|
450 | if(name.endsWith(".warc.wet")) {
|
---|
451 | logger.debug("Will include " + name + " for processing.");
|
---|
452 | return true;
|
---|
453 | }
|
---|
454 |
|
---|
455 | if(name.endsWith(".warc.wet.gz")) {
|
---|
456 | String nameWithoutGZext = name.substring(0, name.lastIndexOf(".gz"));
|
---|
457 | File unzippedVersion = new File(dir, nameWithoutGZext);
|
---|
458 | if(unzippedVersion.exists()) {
|
---|
459 | logger.debug("--- Unzipped version " + unzippedVersion + " exists.");
|
---|
460 | logger.debug("Skipping " + name);
|
---|
461 | return false; // don't count gzipped version if unzipped version exists.
|
---|
462 | }
|
---|
463 | else {
|
---|
464 | logger.debug("Only zipped version " + name + " exists.");
|
---|
465 | return true; // No unzipped version, so have to work with gzipped version
|
---|
466 | }
|
---|
467 | }
|
---|
468 |
|
---|
469 | // we're not even interested in any other file extensions
|
---|
470 | logger.debug("Not a WET file. Skipping " + name);
|
---|
471 | return false;
|
---|
472 | }
|
---|
473 | }
|
---|
474 |
|
---|
475 | public static void main(String[] args) {
|
---|
476 | if(args.length != 2) {
|
---|
477 | printUsage();
|
---|
478 | return;
|
---|
479 | }
|
---|
480 |
|
---|
481 |
|
---|
482 | File WETFileDir = new File(args[0]);
|
---|
483 | if(!WETFileDir.exists() || !WETFileDir.isDirectory()) {
|
---|
484 | System.out.println("Error: " + args[0] + " does not exist or is not a directory");
|
---|
485 | return;
|
---|
486 | }
|
---|
487 |
|
---|
488 | File outFolder = new File(args[1]);
|
---|
489 | if(!outFolder.exists() || !outFolder.isDirectory()) {
|
---|
490 | System.out.println("Error: " + args[1] + " does not exist or is not a directory.");
|
---|
491 | return;
|
---|
492 | }
|
---|
493 |
|
---|
494 | // static folders and files to be shared across all WETProcessor instances
|
---|
495 | WETProcessor.discardFolder = new File(outFolder, "discard");
|
---|
496 | if(!WETProcessor.discardFolder.exists()) {
|
---|
497 | WETProcessor.discardFolder.mkdir();
|
---|
498 | }
|
---|
499 | WETProcessor.keepFolder = new File(outFolder, "keep");
|
---|
500 | if(!WETProcessor.keepFolder.exists()) {
|
---|
501 | WETProcessor.keepFolder.mkdir();
|
---|
502 | }
|
---|
503 |
|
---|
504 | WETProcessor.keepURLsFile = new File(outFolder, "keepURLs.txt");
|
---|
505 | if(WETProcessor.keepURLsFile.exists() && !WETProcessor.keepURLsFile.delete()) {
|
---|
506 | System.err.println("Unable to delete " + WETProcessor.keepURLsFile + ". Unable to proceed.");
|
---|
507 | return;
|
---|
508 | }
|
---|
509 | WETProcessor.discardURLsFile = new File(outFolder, "discardURLs.txt");
|
---|
510 | if(WETProcessor.discardURLsFile.exists() && !WETProcessor.discardURLsFile.delete()) {
|
---|
511 | System.err.println("Unable to delete " + WETProcessor.discardURLsFile + ". Unable to proceed.");
|
---|
512 | return;
|
---|
513 | }
|
---|
514 |
|
---|
515 | // Will list all the warc.wet files in the input directory or else their gzipped versions
|
---|
516 | File[] WETFiles = WETFileDir.listFiles(new WETFilenameFilter());
|
---|
517 |
|
---|
518 | for(int i = 0; i < WETFiles.length; i++) {
|
---|
519 | File WETFile = WETFiles[i];
|
---|
520 | logger.debug("Processing WETfile: " + WETFile);
|
---|
521 |
|
---|
522 | // Any .gz files listed means they haven't been unzipped yet. So unzip.
|
---|
523 | String WETFilename = WETFile.toString();
|
---|
524 | if(WETFilename.endsWith(".gz")) {
|
---|
525 | File GZippedWETFile = WETFile;
|
---|
526 | String WETGZippedFilename = WETFilename;
|
---|
527 | WETFilename = WETFilename.substring(0, WETFilename.lastIndexOf(".gz"));
|
---|
528 |
|
---|
529 | WETFile = new File(WETFilename);
|
---|
530 | Utility.unzipFile(GZippedWETFile, WETFile);
|
---|
531 | }
|
---|
532 | // hereafter all WETFiles should refer to the unzipped version
|
---|
533 | // Check the unzipped WETFile exists
|
---|
534 |
|
---|
535 | if(!WETFile.exists() || !WETFile.isFile()) {
|
---|
536 | System.err.println("Error: " + WETFile + " does not exist (failure to unzip?)");
|
---|
537 | logger.error("Error: " + WETFile + " does not exist (failure to unzip?)");
|
---|
538 | return;
|
---|
539 | }
|
---|
540 |
|
---|
541 | // Finally, we can process this WETFile's records into the keep and discard pile
|
---|
542 | logger.debug("Off to process " + WETFile);
|
---|
543 | WETProcessor processor = new WETProcessor(WETFile, outFolder);
|
---|
544 |
|
---|
545 | }
|
---|
546 |
|
---|
547 | File seedURLsFile = new File(outFolder, "seedURLs.txt");
|
---|
548 | File urlFilterFile = new File(outFolder, "regex-urlfilter.txt");
|
---|
549 | WETProcessor.createSeedURLsFiles(WETProcessor.keepURLsFile, seedURLsFile, urlFilterFile);
|
---|
550 |
|
---|
551 | return;
|
---|
552 | }
|
---|
553 | }
|
---|