1 | # This class based on arcinfo.pm
|
---|
2 | package oaiinfo;
|
---|
3 |
|
---|
4 | use constant INFO_STATUS_INDEX => 0;
|
---|
5 | use constant INFO_TIMESTAMP_INDEX => 1;
|
---|
6 |
|
---|
7 | use constant NO_PD_STEP => 0; # no needing to mark db with "provisionally deleted", no PD pass
|
---|
8 | use constant DO_PD_STEP => 1; # PD pass required (pass to mark db "PD" and then later undo it).
|
---|
9 | # or SPECIAL_PROCESSING_REQUIRED => 1???
|
---|
10 |
|
---|
11 | use strict;
|
---|
12 |
|
---|
13 | use arcinfo;
|
---|
14 | use dbutil;
|
---|
15 |
|
---|
16 | # QUESTIONS:
|
---|
17 | # Should we use time or localtime(time) for timestamp? Just timestamp.
|
---|
18 | # What format should the timestamp be in, or is the basic format used by perl sufficient? Basic.
|
---|
19 |
|
---|
20 | # File format read in: OID <tab> Date-timestamp <tab> Deletion-Status
|
---|
21 |
|
---|
22 | # Deletion status can be:
|
---|
23 | # E = Doc with OID exists (has not been deleted from collection). Timestamp indicates last time of build
|
---|
24 | # D = Doc with OID has been deleted. Timestamp indicates time of deletion
|
---|
25 | # PD = Provisionally Deleted. Timestamp momentarily unaltered.
|
---|
26 |
|
---|
27 | # oaidb is "always incremental": always reflects the I/B/R/D status of archive info db,
|
---|
28 | # before the indexing step of the build phase that alters the I/B/R/D contents of archive info db.
|
---|
29 | # (I=index, B=been indexed, R=reindex; D=delete)
|
---|
30 |
|
---|
31 | sub new {
|
---|
32 | my $class = shift(@_);
|
---|
33 | my ($config_filename, $infodbtype) = @_;
|
---|
34 |
|
---|
35 | my $self = {
|
---|
36 | 'info'=>{} # map of {OID, array[deletion-status,timestamp]} pairs
|
---|
37 | };
|
---|
38 |
|
---|
39 | if(!defined $infodbtype) {
|
---|
40 | $infodbtype = &dbutil::get_default_infodb_type();
|
---|
41 | }
|
---|
42 | $infodbtype = "gdbm" if ($infodbtype eq "gdbm-txtgz");
|
---|
43 | $self->{'infodbtype'} = $infodbtype;
|
---|
44 |
|
---|
45 | # Create and store the db filenames we'll be working with (tmp and livedb)
|
---|
46 | my $etc_dir = &util::get_parent_folder($config_filename);
|
---|
47 |
|
---|
48 | my $perform_firsttime_init = 0;
|
---|
49 | $self->{'oaidb_live_filepath'} = &dbutil::get_infodb_file_path($infodbtype, "oai-inf", $etc_dir, $perform_firsttime_init);
|
---|
50 | $self->{'oaidb_tmp_filepath'} = &dbutil::get_infodb_file_path($infodbtype, "oai-inf-tmp", $etc_dir, $perform_firsttime_init);
|
---|
51 | $self->{'etc_dir'} = $etc_dir;
|
---|
52 | # print STDERR "############ LIVE DB: $self->{'oaidb_live_filepath'}\n";
|
---|
53 | # print STDERR "############ TMP DB: $self->{'oaidb_tmp_filepath'}\n";
|
---|
54 |
|
---|
55 | $self->{'oaidb_file_path'} = $self->{'oaidb_tmp_filepath'}; # db file we're working with
|
---|
56 |
|
---|
57 | return bless $self, $class;
|
---|
58 | }
|
---|
59 |
|
---|
60 | # this subroutine will work out the starting contents of the tmp-db (temporary oai db):
|
---|
61 | # whether it should start off empty, or with the contents of any existing live-db,
|
---|
62 | # or with the contents of any existing tmp-db.
|
---|
63 | sub init_tmpdb {
|
---|
64 | my $self = shift(@_);
|
---|
65 | my ($removeold, $have_manifest) = @_;
|
---|
66 |
|
---|
67 | # if we have a manifest file, then we pretend we are fully incremental for oaiinfo db.
|
---|
68 | # removeold implies proper full-rebuild, whereas keepold or incremental means incremental
|
---|
69 | if($have_manifest) { # if we have a manifest file, we're not doing removeold/full-rebuild either
|
---|
70 | $removeold = 0;
|
---|
71 | }
|
---|
72 |
|
---|
73 | if($removeold) {
|
---|
74 | $self->{'removeold_no_manifest'} = 1;
|
---|
75 | $self->{'do_pd_step'} = DO_PD_STEP; # step where all E will be marked as PD
|
---|
76 | } else {
|
---|
77 | $self->{'removeold_no_manifest'} = 0;
|
---|
78 | $self->{'do_pd_step'} = NO_PD_STEP;
|
---|
79 | }
|
---|
80 |
|
---|
81 | my $oaidb_live_filepath = $self->{'oaidb_live_filepath'};
|
---|
82 | my $oaidb_tmp_filepath = $self->{'oaidb_tmp_filepath'};
|
---|
83 | my $infodbtype = $self->{'infodbtype'};
|
---|
84 | # Note: the live db can only exist if the collection has been activated at least once before
|
---|
85 | my $livedb_exists = &FileUtils::fileExists($oaidb_live_filepath);
|
---|
86 | my $tmpdb_exists = &FileUtils::fileExists($oaidb_tmp_filepath);
|
---|
87 |
|
---|
88 | # print STDERR "############ LIVE DB: $self->{'oaidb_live_filepath'}\n";
|
---|
89 | # print STDERR "############ TMP DB: $self->{'oaidb_tmp_filepath'}\n";
|
---|
90 |
|
---|
91 | my $initdb = 0;
|
---|
92 |
|
---|
93 | # work out what operation we need to do
|
---|
94 | # work with empty tmpdb
|
---|
95 | # copy_livedb_to_tmpdb
|
---|
96 | # work with existing tmpdb (so existing tmpdb will be topped up)
|
---|
97 |
|
---|
98 | # make_contents_of_tmpdb_empty
|
---|
99 | # make_contents_of_tmpdb_that_of_livedb
|
---|
100 | # continue_working_with_tmpdb ("contents_of_tmpdb_is_tmpdb")
|
---|
101 |
|
---|
102 | # We're going to prepare the starting state of tmpdb next.
|
---|
103 | # It can start off empty, start off with the contents of livedb, or it can start off with the contents
|
---|
104 | # of the existing tmp db. Which of these three it is depends on the 3 factors: whether livedb exists,
|
---|
105 | # whether tmpdb exists and whether or not removeold is true.
|
---|
106 | # i.o.w. which of the 3 outcomes it is depends on the truth table built on the following 3 variables:
|
---|
107 | # LDB = LiveDB exists
|
---|
108 | # TDB = TmpDB exists
|
---|
109 | # RO = Removeold
|
---|
110 | # OUTCOMES:
|
---|
111 | # clean slate (create an empty tmpdb/make tmpdb empty)
|
---|
112 | # top up tmpDB (work with existing tmpdb)
|
---|
113 | # copy LiveDB to TmpDB (liveDB's contents become the contents of TmpDB, and we'll work with that)
|
---|
114 | #
|
---|
115 | # TRUTH TABLE:
|
---|
116 | # ---------------------------------------
|
---|
117 | # LDB TDB RO | Outcome
|
---|
118 | # ---------------------------------------
|
---|
119 | # 0 0 0 | clean-slate
|
---|
120 | # 0 0 1 | clean-slate
|
---|
121 | # 0 1 0 | top-up-tmpdb
|
---|
122 | # 0 1 1 | erase tmpdb, clean-slate
|
---|
123 | # 1 0 0 | copy livedb to tmpdb
|
---|
124 | # 1 0 1 | copy livedb to tmpdb
|
---|
125 | # 1 1 0 | top-up-tmpdb
|
---|
126 | # 1 1 1 | copy livedb to tmpd
|
---|
127 | # ---------------------------------------
|
---|
128 | #
|
---|
129 | # Dr Bainbridge worked out using Karnaugh maps that, from the above truth table:
|
---|
130 | # => clean-slate/empty-tmpdb = !LDB && (RO || !TDB)
|
---|
131 | # => top-up-tmpdb/work-with-existing-tmpdb = !RO && TDB
|
---|
132 | # => copy-livedb-to-tmpdb = LDB && (!TDB || RO)
|
---|
133 | # I had most of these tests, except that I hadn't (yet) merged the two clean slate instances
|
---|
134 | # of first-build-ever and make-contents-of-tmpdb-empty
|
---|
135 |
|
---|
136 | #my $first_build_ever = (!$livedb_exists && !$tmpdb_exists);
|
---|
137 | #my $make_contents_of_tmpdb_empty = (!$livedb_exists && $tmpdb_exists && $removeold);
|
---|
138 | # Karnaugh map allows merging $first_build_ever and $make_contents_of_tmpdb_empty above
|
---|
139 | # into: my $work_with_empty_tmpdb = (!$livedb_exists && (!$tmpdb_exists || $removeold));
|
---|
140 | my $work_with_empty_tmpdb = (!$livedb_exists && (!$tmpdb_exists || $removeold));
|
---|
141 | my $make_contents_of_tmpdb_that_of_livedb = ($livedb_exists && (!$tmpdb_exists || $removeold));
|
---|
142 | my $work_with_existing_tmpdb = ($tmpdb_exists && !$removeold);
|
---|
143 |
|
---|
144 | if($work_with_empty_tmpdb) { # we'll use an empty tmpdb
|
---|
145 |
|
---|
146 | # If importing the collection for the very first time, neither db exists,
|
---|
147 | # so create an empty tmpdb.
|
---|
148 | #
|
---|
149 | # We also create an empty tmpdb when livedb doesn't exist and $removeold is true.
|
---|
150 | # This can happen if we've never run activate (so no livedb),
|
---|
151 | # yet had done some import (and perhaps building) followed by a full re-import now.
|
---|
152 | # Since there was no activate and we're doing a removeold/full-rebuild now, can just
|
---|
153 | # work with a new tmpdb, even though one already existed, its contents can be wiped out.
|
---|
154 | # In such a scenario, we'll be deleting tmpdb. Then there will be no livedb nor any tmpdb
|
---|
155 | # any more, so same situation as if importing the very first time when no oaidb exists either.
|
---|
156 |
|
---|
157 | &dbutil::remove_db_file($self->{'infodbtype'}, $oaidb_tmp_filepath) if $tmpdb_exists; # remove the db file and any assoc files
|
---|
158 | $initdb = 1; # new tmpdb
|
---|
159 |
|
---|
160 | # if the oai db is created the first time, it's like incremental and
|
---|
161 | # "keepold" (keepold means "only add, don't reprocess existing"). So
|
---|
162 | # no need to do the special passes dealing with "provisional deletes".
|
---|
163 | $self->{'do_pd_step'} = NO_PD_STEP;
|
---|
164 |
|
---|
165 | } elsif ($make_contents_of_tmpdb_that_of_livedb) {
|
---|
166 |
|
---|
167 | # If the livedb exists and we're doing a full rebuild ($removeold is true),
|
---|
168 | # copy livedb to tmp regardless of if tmpdb already exists.
|
---|
169 | # Or if the livedb exists and tmpdb doesn't exist, it doesn't matter
|
---|
170 | # if we're incremental or not: also copy live to tmp and work with tmp.
|
---|
171 |
|
---|
172 | # copy livedb to tmpdb
|
---|
173 | &dbutil::remove_db_file($self->{'infodbtype'}, $oaidb_tmp_filepath) if $tmpdb_exists; # remove the db file and any assoc files
|
---|
174 | &FileUtils::copyFiles($oaidb_live_filepath, $oaidb_tmp_filepath);
|
---|
175 |
|
---|
176 | $initdb = 0; # tmpdb exists, since we just copied livedb to tmpdb, so need to init existing tmpdb
|
---|
177 |
|
---|
178 | } else { # $work_with_existing_tmpdb, so we'll build on top of what's presently already in tmpdb
|
---|
179 | # (we'll be topping up the current tmpdb)
|
---|
180 |
|
---|
181 | # !$removeold, meaning incremental
|
---|
182 | # If incremental and have a tmpdb already, regardless of whether livedb exists,
|
---|
183 | # then work with the existing tmpdb file, as this means we've been
|
---|
184 | # importing (perhaps followed by building) repeatedly without activating the
|
---|
185 | # last time but want to maintain the (incremental) changes in tmpdb.
|
---|
186 |
|
---|
187 | $initdb = 0;
|
---|
188 |
|
---|
189 | } # Dr Bainbridge drew up Karnaugh maps on the truth table, which proved that all cases
|
---|
190 | # are indeed covered above, so don't need any other catch-all else here
|
---|
191 |
|
---|
192 | $self->{'oaidb_file_path'} = &dbutil::get_infodb_file_path($infodbtype, "oai-inf-tmp", $self->{'etc_dir'}, $initdb);
|
---|
193 | # final param follows jmt's $perform_firsttime_init in inexport.pm
|
---|
194 |
|
---|
195 | # print STDERR "@@@@@ oaidb: $self->{'oaidb_file_path'}\n";
|
---|
196 |
|
---|
197 | }
|
---|
198 |
|
---|
199 | sub set_proc_mode {
|
---|
200 | my $self = shift (@_);
|
---|
201 | my ($mode) = @_;
|
---|
202 |
|
---|
203 | $self->{'do_pd_step'} = $mode;
|
---|
204 | }
|
---|
205 |
|
---|
206 | # returns 1 if doing PD step (marking entries as provisionally deleted)
|
---|
207 | # returns 0 if not doing PD step, which happens when we're purely incremental or building first time.
|
---|
208 | sub get_proc_mode {
|
---|
209 | my $self = shift (@_);
|
---|
210 | return $self->{'do_pd_step'};
|
---|
211 | }
|
---|
212 |
|
---|
213 | sub get_filepath {
|
---|
214 | my $self = shift (@_);
|
---|
215 | return $self->{'oaidb_file_path'};
|
---|
216 | }
|
---|
217 |
|
---|
218 | sub import_stage {
|
---|
219 | my $self = shift (@_);
|
---|
220 | my ($removeold, $have_manifest) = @_;
|
---|
221 |
|
---|
222 | $self->init_tmpdb($removeold, $have_manifest);
|
---|
223 |
|
---|
224 | my $do_pd_step = $self->{'do_pd_step'}; # 1 if the step to mark oaidb entries as PD is required
|
---|
225 | # if we're doing full rebuilding and it's NOT the first time creating the oai_inf db,
|
---|
226 | # then the tasks to do with PD (provisionally deleted) OAI OIDs should be carried out
|
---|
227 |
|
---|
228 |
|
---|
229 | $self->load_info();
|
---|
230 | $self->print_info(); # DEBUGGING
|
---|
231 |
|
---|
232 | if ($do_pd_step) {
|
---|
233 | $self->mark_all_existing_as_provisionallydeleted();
|
---|
234 | $self->print_info(); # DEBUGGING
|
---|
235 |
|
---|
236 | # save to db file now that we're done
|
---|
237 | $self->save_info();
|
---|
238 | }
|
---|
239 |
|
---|
240 | }
|
---|
241 |
|
---|
242 | sub building_stage_before_indexing() {
|
---|
243 | my $self = shift (@_);
|
---|
244 | my ($archivedir) = @_;
|
---|
245 |
|
---|
246 | # load archive info db into memory
|
---|
247 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($self->{'infodbtype'}, "archiveinf-doc", $archivedir);
|
---|
248 | my $arcinfo_src_filename = &dbutil::get_infodb_file_path($self->{'infodbtype'}, "archiveinf-src", $archivedir);
|
---|
249 | my $archive_info = new arcinfo ($self->{'infodbtype'});
|
---|
250 | $archive_info->load_info ($arcinfo_doc_filename);
|
---|
251 |
|
---|
252 | #my $started_from_scratch = &FileUtils::fileTest($self->{'oaidb_tmp_filepath'}, '-z'); # 1 if tmpdb is empty
|
---|
253 | # -z test for file is empty http://www.perlmonks.org/?node_id=927447
|
---|
254 |
|
---|
255 | # load the oaidb file's contents into memory.
|
---|
256 | $self->load_info();
|
---|
257 | $self->print_info(); # DEBUGGING
|
---|
258 |
|
---|
259 | # process all the index, reindex and delete operations as indicated in arcinfo,
|
---|
260 | # all the while ensuring all PDs are changed back to E for OIDs that exist in both arcinfo and oaiinfo db.
|
---|
261 |
|
---|
262 | foreach my $OID (keys $archive_info->{'info'}) {
|
---|
263 | my $arcinf_tuple = $archive_info->{'info'}->{$OID};
|
---|
264 | my $indexing_status = $arcinf_tuple->[arcinfo::INFO_STATUS_INDEX];
|
---|
265 | # use packageName::constant to refer to constants declared in another package,
|
---|
266 | # see http://perldoc.perl.org/constant.html
|
---|
267 |
|
---|
268 | print STDERR "######## OID: $OID - status: $indexing_status\n";
|
---|
269 |
|
---|
270 | if($indexing_status eq "I") {
|
---|
271 | $self->index($OID); # add new as E with current timestamp/or set existing as E with orig timestamp
|
---|
272 | } elsif($indexing_status eq "R") {
|
---|
273 | $self->reindex($OID); # update timestamp and ensure marked as E (if oid doesn't exist, add new)
|
---|
274 | } elsif($indexing_status eq "D") {
|
---|
275 | $self->delete($OID); # set as D with current timestamp
|
---|
276 | } elsif($indexing_status eq "B") { # B for "been indexed"
|
---|
277 | $self->been_indexed($OID); # will flip any PD to E if oid exists, else will add new entry for oid
|
---|
278 | # A new entry may be required if the collection had been built prior to turning this into
|
---|
279 | # an oaicollection. But what if we always maintain an oaidb? Still call $self->index() here.
|
---|
280 | } else {
|
---|
281 | print STDERR "### oaiinfo::building_stage_before_indexing(): Unrecognised indexing status $indexing_status\n";
|
---|
282 | }
|
---|
283 | }
|
---|
284 |
|
---|
285 | # once all docs processed, go through oaiiinfo db changing any PDs to D along with current timestamp
|
---|
286 | # to indicate that they're deleted
|
---|
287 | $self->mark_all_provisionallydeleted_as_deleted();
|
---|
288 | $self->print_info();
|
---|
289 |
|
---|
290 | # let's save to db file now that we're done
|
---|
291 | $self->save_info();
|
---|
292 |
|
---|
293 | }
|
---|
294 |
|
---|
295 | sub activate_collection { # move tmp db to live db
|
---|
296 | my $self = shift (@_);
|
---|
297 |
|
---|
298 | my $oaidb_live_filepath = $self->{'oaidb_live_filepath'};
|
---|
299 | my $oaidb_tmp_filepath = $self->{'oaidb_tmp_filepath'};
|
---|
300 |
|
---|
301 | my $livedb_exists = &FileUtils::fileExists($oaidb_live_filepath);
|
---|
302 | my $tmpdb_exists = &FileUtils::fileExists($oaidb_tmp_filepath);
|
---|
303 |
|
---|
304 | if($tmpdb_exists) {
|
---|
305 | if($livedb_exists) {
|
---|
306 | #&dbutil::remove_db_file($self->{'infodbtype'}, s$oaidb_live_filepath); # remove the db file and any assoc files
|
---|
307 | &dbutil::rename_db_file_to($self->{'infodbtype'}, $oaidb_live_filepath, $oaidb_live_filepath.".bak"); # rename the db file and any assoc files
|
---|
308 | }
|
---|
309 | #&FileUtils::moveFiles($oaidb_tmp_filepath, $oaidb_live_filepath);
|
---|
310 | &dbutil::rename_db_file_to($self->{'infodbtype'}, $oaidb_tmp_filepath, $oaidb_live_filepath); # rename the db file and any assoc files
|
---|
311 |
|
---|
312 | print STDERR "#### Should now have MOVED $self->{'oaidb_tmp_filepath'} to $self->{'oaidb_live_filepath'}\n";
|
---|
313 |
|
---|
314 | } else {
|
---|
315 | print STDERR "@@@@@ In oaiinfo::activate_collection():\n";
|
---|
316 | print STDERR "@@@@@ No tmpdb at $self->{'oaidb_tmp_filepath'}\n";
|
---|
317 | print STDERR "@@@@@ to make 'live' by moving to $self->{'oaidb_live_filepath'}.\n";
|
---|
318 | }
|
---|
319 | }
|
---|
320 |
|
---|
321 | ############################### FOR FULL AND INCR BUILDING ##########################
|
---|
322 | # add or reindex if incremental (updating timestamp to current),
|
---|
323 | # add or (re-)mark PDs as status=E for existing if full build, keeping timestamp.
|
---|
324 | # This subroutine will end up adding if firstbuild (since then the OID won't already exist)
|
---|
325 | sub Xprocess_new_or_existing {
|
---|
326 | my $self = shift (@_);
|
---|
327 | my ($OID) = @_;
|
---|
328 |
|
---|
329 | my $OID_info = $self->{'info'}->{$OID};
|
---|
330 | if (defined $OID_info) {
|
---|
331 |
|
---|
332 | if ($self->{'do_pd_step'} == NO_PD_STEP) { # update modification time for reindexed doc
|
---|
333 | $OID_info->[INFO_TIMESTAMP_INDEX] = $self->get_current_time();
|
---|
334 | }
|
---|
335 | else { # mode is DO_PD_STEP
|
---|
336 | $OID_info->[INFO_STATUS_INDEX] = "E"; # will flip any PD status of existing documents back to E,
|
---|
337 | # so need to keep existing timestamps.
|
---|
338 | # Can't be called on deleted docs, so status was never D
|
---|
339 | # $self->set_status_info($OID, "E"); # will only set it if OID exists
|
---|
340 |
|
---|
341 |
|
---|
342 | }
|
---|
343 | } else { # if OID is not present, then it's new and now added as existing from current time on
|
---|
344 | $self->set_info($OID, "E", $self->get_current_time());
|
---|
345 | }
|
---|
346 |
|
---|
347 | }
|
---|
348 |
|
---|
349 | ##################### SPECIFIC TO PD-STEP ####################
|
---|
350 |
|
---|
351 |
|
---|
352 | # mark all existing, E (non-deleted) OIDs as Provisionally Deleted (PD)
|
---|
353 | # this subroutine doesn't save to oai-inf.DB
|
---|
354 | # the caller should call save_info when they want to save to the db
|
---|
355 | sub mark_all_existing_as_provisionallydeleted {
|
---|
356 | my $self = shift (@_);
|
---|
357 |
|
---|
358 | print STDERR "@@@@@ oaiinfo::mark_all_E_as_PD(): Marking the E entries as PD\n";
|
---|
359 |
|
---|
360 | foreach my $OID (keys $self->{'info'}) {
|
---|
361 | my $curr_status = $self->get_status_info($OID);
|
---|
362 | if(defined $curr_status && $curr_status eq "E") {
|
---|
363 | $self->set_status_info($OID, "PD");
|
---|
364 | }
|
---|
365 | }
|
---|
366 | }
|
---|
367 |
|
---|
368 | # mark all OIDs that are Provisionally Deleted (PD) as deleted, and set to current timestamp
|
---|
369 | # To be called at end of build. Again, the caller should save to DB by calling save_info.
|
---|
370 | sub mark_all_provisionallydeleted_as_deleted {
|
---|
371 | my $self = shift (@_);
|
---|
372 |
|
---|
373 | print STDERR "@@@@@ oaiinfo::mark_all_PD_as_D(): Marking the PD entries as D\n";
|
---|
374 |
|
---|
375 | foreach my $OID (keys $self->{'info'}) {
|
---|
376 | my $curr_status = $self->get_status_info($OID);
|
---|
377 | if(defined $curr_status && $curr_status eq "PD") {
|
---|
378 | $self->set_info($OID, "D", $self->get_current_time());
|
---|
379 | }
|
---|
380 | }
|
---|
381 | }
|
---|
382 |
|
---|
383 | # find the OID, if it exists, make its status=E for existing.
|
---|
384 | sub Xset_status_to_existing_if_OID_present {
|
---|
385 | my $self = shift (@_);
|
---|
386 | my ($OID) = @_;
|
---|
387 |
|
---|
388 | $self->set_status_info($OID, "E"); # will only set it if OID exists
|
---|
389 | }
|
---|
390 |
|
---|
391 | ##################### GENERAL, NOT SPECIFIC TO PD-STEP ####################
|
---|
392 |
|
---|
393 | sub print_info {
|
---|
394 | my $self = shift (@_);
|
---|
395 |
|
---|
396 | print STDERR "###########################################################\n";
|
---|
397 | print STDERR "@@@@@ oaiinfo::print_info(): oaidb in memory contains: \n";
|
---|
398 |
|
---|
399 | foreach my $OID (keys $self->{'info'}) {
|
---|
400 | print STDERR "OID: $OID";
|
---|
401 | print STDERR " status: " . $self->{'info'}->{$OID}->[INFO_STATUS_INDEX];
|
---|
402 | print STDERR " time: " . $self->{'info'}->{$OID}->[INFO_TIMESTAMP_INDEX];
|
---|
403 | print STDERR "\n";
|
---|
404 | }
|
---|
405 |
|
---|
406 | print STDERR "###########################################################\n";
|
---|
407 | }
|
---|
408 |
|
---|
409 |
|
---|
410 | # Find the OID, if it exists, make its status=E for existing. Leave its timestamp alone.
|
---|
411 | # If the OID doesn't yet exist, add it as a new entry with status=E and with current timestamp.
|
---|
412 | sub index { # Add a new oid with current time and E. If the oid was already present, mark as E
|
---|
413 | my $self = shift (@_);
|
---|
414 | my ($OID) = @_;
|
---|
415 |
|
---|
416 | my $OID_info = $self->{'info'}->{$OID};
|
---|
417 |
|
---|
418 | if (defined $OID_info) { # if OID is present, this will change status back to E, timestamp unchanged
|
---|
419 | $OID_info->[INFO_STATUS_INDEX] = "E";
|
---|
420 |
|
---|
421 | } else { # if OID is not present, then it's now added as existing from current time on
|
---|
422 | $self->set_info($OID, "E", $self->get_current_time());
|
---|
423 | }
|
---|
424 | }
|
---|
425 |
|
---|
426 | # Upon reindexing a document with identifier OID, change its timestamp to current time
|
---|
427 | # if a new OID, then add as new entry with status=E and current timestamp
|
---|
428 | sub reindex { # update timestamp if oid is already present, if not (unlikely), add as new
|
---|
429 | my $self = shift (@_);
|
---|
430 | my ($OID) = @_;
|
---|
431 |
|
---|
432 | my $OID_info = $self->{'info'}->{$OID};
|
---|
433 | $self->set_info($OID, "E", $self->get_current_time()); # Takes care of 3 things:
|
---|
434 | # if OID exists, updates modified time to indicate the doc has been reindexed
|
---|
435 | # if OID exists, ensures any status=PD is flipped back to E for this OID doc (as we know it exists);
|
---|
436 | # if the OID doesn't yet exist, adds a new OID entry with status=E and current timestamp.
|
---|
437 |
|
---|
438 | }
|
---|
439 |
|
---|
440 |
|
---|
441 | # Does the same as index():
|
---|
442 | # OIDs that have been indexed upon rebuild may still be new to the oaidb: GS2 collections
|
---|
443 | # are not OAI collections by default, unlike GS3 collections. Imagine rebuilding a (GS2) collection
|
---|
444 | # 5 times and then setting them to be an OAI collection. In that case, the doc OIDs in the collection
|
---|
445 | # may not be in the oaidb yet. Unless, we decide (as is the present case) to always maintain an oaidb
|
---|
446 | # (always creating an oaidb regardless of whether the collection has OAI support turned on or not).
|
---|
447 | sub been_indexed {
|
---|
448 | my $self = shift (@_);
|
---|
449 | my ($OID) = @_;
|
---|
450 |
|
---|
451 | $self->index($OID);
|
---|
452 | }
|
---|
453 |
|
---|
454 | # Upon deleting a document with identifier OID,
|
---|
455 | # set status to deleted and change its timestamp to current time
|
---|
456 | sub delete {
|
---|
457 | my $self = shift (@_);
|
---|
458 | my ($OID) = @_;
|
---|
459 |
|
---|
460 | # the following method will set to current time if no timestamp provided,
|
---|
461 | # But by explicit here, the code is easier to follow
|
---|
462 | $self->set_info($OID, "D", $self->get_current_time());
|
---|
463 |
|
---|
464 | }
|
---|
465 |
|
---|
466 | #############################################################
|
---|
467 | sub get_current_time {
|
---|
468 | my $self = shift (@_);
|
---|
469 | return time; # current time
|
---|
470 |
|
---|
471 | # localtime(time) returns an array of values (day, month, year, hour, min, seconds) or singular string
|
---|
472 | # return localtime; # same as localtime(time); # http://perldoc.perl.org/functions/localtime.html
|
---|
473 |
|
---|
474 | }
|
---|
475 |
|
---|
476 | sub _load_info_txt
|
---|
477 | {
|
---|
478 | my $self = shift (@_);
|
---|
479 | my ($filename) = @_;
|
---|
480 |
|
---|
481 | if (defined $filename && &FileUtils::fileExists($filename)) {
|
---|
482 | open (INFILE, $filename) ||
|
---|
483 | die "oaiinfo::load_info couldn't read $filename\n";
|
---|
484 |
|
---|
485 | my ($line, @lineparts);
|
---|
486 | while (defined ($line = <INFILE>)) {
|
---|
487 | $line =~ s/\cM|\cJ//g; # remove end-of-line characters
|
---|
488 | @lineparts = split ("\t", $line);
|
---|
489 | if (scalar(@lineparts) >= 2) {
|
---|
490 | $self->add_info (@lineparts);
|
---|
491 | }
|
---|
492 | }
|
---|
493 | close (INFILE);
|
---|
494 | }
|
---|
495 |
|
---|
496 |
|
---|
497 | }
|
---|
498 |
|
---|
499 | sub _load_info_db
|
---|
500 | {
|
---|
501 | my $self = shift (@_);
|
---|
502 | my ($filename) = @_;
|
---|
503 |
|
---|
504 | my $infodb_map = {};
|
---|
505 |
|
---|
506 | &dbutil::read_infodb_file($self->{'infodbtype'}, $filename, $infodb_map);
|
---|
507 |
|
---|
508 | foreach my $oid ( keys $infodb_map ) {
|
---|
509 | my $vals = $infodb_map->{$oid};
|
---|
510 | # interested in oid, timestamp, deletion status
|
---|
511 |
|
---|
512 | my ($deletion_status) = ($vals=~/^<status>(.*)$/m);
|
---|
513 | my ($timestamp) = ($vals=~/^<timestamp>(.*)$/m);
|
---|
514 |
|
---|
515 | $self->add_info ($oid, $deletion_status, $timestamp);
|
---|
516 | }
|
---|
517 | }
|
---|
518 |
|
---|
519 | # if no filename is passed in (and you don't generally want to), then
|
---|
520 | # it tries to load in <collection>/etc/oai-inf.<db> if it exists
|
---|
521 | sub load_info {
|
---|
522 | my $self = shift (@_);
|
---|
523 | my ($filename) = @_;
|
---|
524 |
|
---|
525 | $self->{'info'} = {};
|
---|
526 |
|
---|
527 | $filename = $self->{'oaidb_file_path'} unless defined $filename;
|
---|
528 |
|
---|
529 | if (&FileUtils::fileExists($filename)) {
|
---|
530 | if ($filename =~ m/\.inf$/) {
|
---|
531 | $self->_load_info_txt($filename);
|
---|
532 | }
|
---|
533 | else {
|
---|
534 | $self->_load_info_db($filename);
|
---|
535 | }
|
---|
536 | }
|
---|
537 |
|
---|
538 | }
|
---|
539 |
|
---|
540 | sub _save_info_txt {
|
---|
541 | my $self = shift (@_);
|
---|
542 | my ($filename) = @_;
|
---|
543 |
|
---|
544 | my ($OID, $info);
|
---|
545 |
|
---|
546 | open (OUTFILE, ">$filename") ||
|
---|
547 | die "oaiinfo::save_info couldn't write $filename\n";
|
---|
548 |
|
---|
549 | foreach $info (@{$self->get_OID_list()}) {
|
---|
550 | if (defined $info) {
|
---|
551 | print OUTFILE join("\t", @$info), "\n";
|
---|
552 | }
|
---|
553 | }
|
---|
554 | close (OUTFILE);
|
---|
555 | }
|
---|
556 |
|
---|
557 | # if no filename is passed in (and you don't generally want to), then
|
---|
558 | # this subroutine tries to write to <collection>/etc/oai-inf.<db>.
|
---|
559 | sub _save_info_db {
|
---|
560 | my $self = shift (@_);
|
---|
561 | my ($filename) = @_;
|
---|
562 |
|
---|
563 | $filename = $self->{'oaidb_file_path'} unless defined $filename;
|
---|
564 | my $infodbtype = $self->{'infodbtype'};
|
---|
565 |
|
---|
566 | # write out again. Open file for overwriting, not appending.
|
---|
567 | # Then write out data structure $self->{'info'} that's been maintaining the data in-memory.
|
---|
568 | my $infodb_handle = &dbutil::open_infodb_write_handle($infodbtype, $filename);
|
---|
569 | foreach my $oid ( keys $self->{'info'} ) {
|
---|
570 | my $OID_info = $self->{'info'}->{$oid};
|
---|
571 | #my $val_hash = {
|
---|
572 | # "status" => $OID_info->[INFO_STATUS_INDEX],
|
---|
573 | # "timestamp" => $OID_info->[INFO_TIMESTAMP_INDEX]
|
---|
574 | #};
|
---|
575 |
|
---|
576 | #&dbutil::write_infodb_rawentry($infodbtype,$infodb_handle,$oid,$val_hash);
|
---|
577 |
|
---|
578 | my $val = "<status>".$OID_info->[INFO_STATUS_INDEX]."\n<timestamp>".$OID_info->[INFO_TIMESTAMP_INDEX]."\n";
|
---|
579 | &dbutil::write_infodb_rawentry($infodbtype,$infodb_handle,$oid,$val);
|
---|
580 | }
|
---|
581 | &dbutil::close_infodb_write_handle($infodbtype, $infodb_handle);
|
---|
582 | }
|
---|
583 |
|
---|
584 | sub save_info {
|
---|
585 | my $self = shift (@_);
|
---|
586 | my ($filename) = @_;
|
---|
587 |
|
---|
588 | if(defined $filename) {
|
---|
589 | if ($filename =~ m/(contents)|(\.inf)$/) {
|
---|
590 | $self->_save_info_txt($filename);
|
---|
591 | }
|
---|
592 | else {
|
---|
593 | $self->_save_info_db($filename);
|
---|
594 | }
|
---|
595 | } else {
|
---|
596 | $self->_save_info_db();
|
---|
597 | }
|
---|
598 | }
|
---|
599 |
|
---|
600 | sub delete_info {
|
---|
601 | my $self = shift (@_);
|
---|
602 | my ($OID) = @_;
|
---|
603 |
|
---|
604 | print STDERR "@@@@ ERROR oaiinfo::delete_info: Not allowed to delete entries in oai DB.\n";
|
---|
605 |
|
---|
606 | #if (defined $self->{'info'}->{$OID}) {
|
---|
607 | # delete $self->{'info'}->{$OID};
|
---|
608 | #}
|
---|
609 | }
|
---|
610 |
|
---|
611 | sub set_info { # sets existing or appends
|
---|
612 | my $self = shift (@_);
|
---|
613 | my ($OID, $del_status, $timestamp) = @_;
|
---|
614 | if(!defined $timestamp) { # get current date timestamp
|
---|
615 | $timestamp = $self->get_current_time();
|
---|
616 | }
|
---|
617 | $self->{'info'}->{$OID} = [$del_status, $timestamp];
|
---|
618 |
|
---|
619 | }
|
---|
620 |
|
---|
621 | sub add_info { # appends iff it doesn't exist already
|
---|
622 | my $self = shift (@_);
|
---|
623 | my ($OID, $del_status, $timestamp) = @_;
|
---|
624 |
|
---|
625 | if (defined($self->{'info'}->{$OID})) {
|
---|
626 | print STDERR "@@@@ ERROR oaiinfo::add_info: id $OID already exists. Not adding.\n";
|
---|
627 | } else {
|
---|
628 | $self->set_info($OID, $del_status, $timestamp);
|
---|
629 | }
|
---|
630 | }
|
---|
631 |
|
---|
632 | sub set_status_info {
|
---|
633 | my $self = shift (@_);
|
---|
634 | my ($OID, $del_status) = @_;
|
---|
635 |
|
---|
636 | my $OID_info = $self->{'info'}->{$OID};
|
---|
637 | if (defined $OID_info) {
|
---|
638 | $OID_info->[INFO_STATUS_INDEX] = $del_status;
|
---|
639 | }
|
---|
640 | else {
|
---|
641 | print STDERR "oaiinfo::set_status_info: Unable to find OAI document id $OID\n";
|
---|
642 | }
|
---|
643 | }
|
---|
644 |
|
---|
645 | sub get_status_info {
|
---|
646 | my $self = shift (@_);
|
---|
647 | my ($OID) = @_;
|
---|
648 |
|
---|
649 | my $del_status = undef;
|
---|
650 |
|
---|
651 | my $OID_info = $self->{'info'}->{$OID};
|
---|
652 | if (defined $OID_info) {
|
---|
653 | $del_status = $OID_info->[INFO_STATUS_INDEX];
|
---|
654 | }
|
---|
655 | else {
|
---|
656 | print STDERR "oaiinfo::get_status_info: Unable to find OAI document id $OID\n";
|
---|
657 | }
|
---|
658 |
|
---|
659 | return $del_status;
|
---|
660 | }
|
---|
661 |
|
---|
662 | # if no timestamp provided, uses current timestamp
|
---|
663 | sub set_timestamp_info {
|
---|
664 | my $self = shift (@_);
|
---|
665 | my ($OID, $timestamp) = @_;
|
---|
666 |
|
---|
667 | my $OID_info = $self->{'info'}->{$OID};
|
---|
668 | if (defined $OID_info) {
|
---|
669 | if(!defined $timestamp) {
|
---|
670 | $timestamp = $self->get_current_time();
|
---|
671 | }
|
---|
672 | $OID_info->[INFO_TIMESTAMP_INDEX] = $timestamp;
|
---|
673 | }
|
---|
674 | else {
|
---|
675 | print STDERR "oaiinfo::set_timestamp_info: Unable to find OAI document id $OID\n";
|
---|
676 | }
|
---|
677 | }
|
---|
678 |
|
---|
679 | sub get_timestamp_info {
|
---|
680 | my $self = shift (@_);
|
---|
681 | my ($OID) = @_;
|
---|
682 |
|
---|
683 | my $timestamp = undef;
|
---|
684 |
|
---|
685 | my $OID_info = $self->{'info'}->{$OID};
|
---|
686 | if (defined $OID_info) {
|
---|
687 | $timestamp = $OID_info->[INFO_TIMESTAMP_INDEX];
|
---|
688 | }
|
---|
689 | else {
|
---|
690 | print STDERR "oaiinfo::get_timestamp_info: Unable to find OAI document id $OID\n";
|
---|
691 | }
|
---|
692 |
|
---|
693 | return $timestamp;
|
---|
694 | }
|
---|
695 |
|
---|
696 |
|
---|
697 | # returns a list of the form [[OID, timestamp, deletion_status], ...]
|
---|
698 | sub get_OID_list
|
---|
699 | {
|
---|
700 | my $self = shift (@_);
|
---|
701 |
|
---|
702 | my @list = ();
|
---|
703 |
|
---|
704 | foreach my $OID (keys $self->{'info'}) {
|
---|
705 | my $OID_info = $self->{'info'}->{$OID};
|
---|
706 |
|
---|
707 | push (@list, [$OID, $OID_info->[INFO_STATUS_INDEX],
|
---|
708 | $OID_info->[INFO_TIMESTAMP_INDEX]]);
|
---|
709 | }
|
---|
710 |
|
---|
711 | return \@list;
|
---|
712 | }
|
---|
713 |
|
---|
714 |
|
---|
715 |
|
---|
716 | # returns an array/list of the form [deletion_status, timestamp]
|
---|
717 | sub get_info {
|
---|
718 | my $self = shift (@_);
|
---|
719 | my ($OID) = @_;
|
---|
720 |
|
---|
721 | if (defined $self->{'info'}->{$OID}) {
|
---|
722 | return $self->{'info'}->{$OID};
|
---|
723 | }
|
---|
724 |
|
---|
725 | return undef;
|
---|
726 | }
|
---|
727 |
|
---|
728 |
|
---|
729 |
|
---|
730 | # returns the number of entries so far, including deleted ones
|
---|
731 | # http://stackoverflow.com/questions/1109095/how-can-i-find-the-number-of-keys-in-a-hash-in-perl
|
---|
732 | sub size {
|
---|
733 | my $self = shift (@_);
|
---|
734 | return (scalar keys $self->{'info'});
|
---|
735 | }
|
---|
736 |
|
---|
737 | 1;
|
---|