Project

General

Profile

1 2276 jones
/**
2
 *  '$RCSfile$'
3
 *  Copyright: 2004 Regents of the University of California and the
4
 *             National Center for Ecological Analysis and Synthesis
5
 *
6
 *   '$Author$'
7
 *     '$Date$'
8
 * '$Revision$'
9
 *
10 2599 jones
 * This program is free software, you can redistribute it and/or modify
11 2276 jones
 * it under the terms of the GNU General Public License as published by
12 2599 jones
 * the Free Software Foundation, either version 2 of the License, or
13 2276 jones
 * (at your option) any later version.
14
 *
15
 * This program is distributed in the hope that it will be useful,
16 2599 jones
 * but WITHOUT ANY WARRANTY, without even the implied warranty of
17 2276 jones
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18
 * GNU General Public License for more details.
19
 *
20
 * You should have received a copy of the GNU General Public License
21 2599 jones
 * along with this program, if not, write to the Free Software
22 2276 jones
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
23
 */
24 2422 sgarg
25 2276 jones
/*
26 2599 jones
 * this is sql script does the same as the sql script named
27
 * xmltables.sql except that this script is to be use to
28
 * create the database tables on a Postgresql backend rather
29
 * than an Oracle Backend
30 2276 jones
 */
31
32
/*
33 2599 jones
 * Replication -- table to store servers that metacat is replicated to
34 2276 jones
 */
35
CREATE SEQUENCE xml_replication_id_seq;
36
CREATE TABLE xml_replication (
37 2422 sgarg
  serverid INT8 default nextval('xml_replication_id_seq'),
38 2276 jones
  server VARCHAR(512),
39
  last_checked DATE,
40
  replicate INT8,
41
  datareplicate INT8,
42
  hub INT8,
43
  CONSTRAINT xml_replication_pk PRIMARY KEY (serverid)
44 2422 sgarg
);
45 2276 jones
46 6814 leinfelder
INSERT INTO xml_replication (server, replicate, datareplicate, hub) VALUES ('localhost', '0', '0', '0');
47 2276 jones
48 6814 leinfelder
49 2422 sgarg
/*
50 2276 jones
 * Nodes -- table to store XML Nodes (both elements and attributes)
51
 */
52
CREATE SEQUENCE xml_nodes_id_seq;
53
CREATE TABLE xml_nodes (
54
	nodeid INT8 default nextval('xml_nodes_id_seq'),
55
					-- the unique node id (pk)
56
	nodeindex INT8,		-- order of nodes within parent
57
	nodetype VARCHAR(20),	-- type (DOCUMENT, COMMENT, PI,
58 2359 sgarg
				-- ELEMENT, ATTRIBUTE, TEXT)
59 2276 jones
	nodename VARCHAR(250),	-- the name of an element or attribute
60
        nodeprefix VARCHAR(50), -- the namespace prefix of the node
61 3336 tao
	nodedata TEXT, -- the data for this node (e.g.,
62 2359 sgarg
				-- for TEXT it is the content)
63
	parentnodeid INT8,	-- index of the parent of this node
64
	rootnodeid INT8,	-- index of the root node of this tree
65 2276 jones
	docid VARCHAR(250),	-- index to the document id
66
	date_created DATE,
67
	date_updated DATE,
68 6012 leinfelder
    nodedatanumerical FLOAT8, -- the data for this node if it is a number
69
    nodedatadate TIMESTAMP, -- the data for this node if it is a date
70 2276 jones
   CONSTRAINT xml_nodes_pk PRIMARY KEY (nodeid),
71 2422 sgarg
   CONSTRAINT xml_nodes_root_fk
72 2276 jones
		FOREIGN KEY (rootnodeid) REFERENCES xml_nodes,
73 2422 sgarg
   CONSTRAINT xml_nodes_parent_fk
74 2276 jones
		FOREIGN KEY (parentnodeid) REFERENCES xml_nodes
75
);
76 2422 sgarg
/*
77 2276 jones
 * Indexes of rootnodeid, parentnodeid, and nodename in xml_nodes
78
 */
79
CREATE INDEX xml_nodes_idx1 ON xml_nodes (rootnodeid);
80
CREATE INDEX xml_nodes_idx2 ON xml_nodes (parentnodeid);
81
CREATE INDEX xml_nodes_idx3 ON xml_nodes (nodename);
82 3412 tao
CREATE INDEX xml_nodes_idx4 ON xml_nodes (docid);
83 2276 jones
84 2519 sgarg
85 2422 sgarg
/*
86 2519 sgarg
 * Table for storing the nodes for the old revisions of the document and the deleted documents
87
 */
88
CREATE TABLE xml_nodes_revisions (
89
        nodeid INT8,            -- the unique node id (pk)
90
        nodeindex INT8,         -- order of nodes within parent
91
        nodetype VARCHAR(20),   -- type (DOCUMENT, COMMENT, PI,
92
                                -- ELEMENT, ATTRIBUTE, TEXT)
93
        nodename VARCHAR(250),  -- the name of an element or attribute
94
        nodeprefix VARCHAR(50), -- the namespace prefix of the node
95 3336 tao
        nodedata TEXT, -- the data for this node (e.g.,
96 2519 sgarg
                                -- for TEXT it is the content)
97
        parentnodeid INT8,      -- index of the parent of this node
98
        rootnodeid INT8,        -- index of the root node of this tree
99
        docid VARCHAR(250),     -- index to the document id
100
        date_created DATE,
101
        date_updated DATE,
102 6012 leinfelder
        nodedatanumerical FLOAT8, -- the data for this node if it is a number
103
        nodedatadate TIMESTAMP, -- the data for this node if it is a date
104 2519 sgarg
   CONSTRAINT xml_nodes_revisions_pk PRIMARY KEY (nodeid),
105
   CONSTRAINT xml_nodes_revisions_root_fk
106
                FOREIGN KEY (rootnodeid) REFERENCES xml_nodes_revisions,
107
   CONSTRAINT xml_nodes_revisions_parent_fk
108
                FOREIGN KEY (parentnodeid) REFERENCES xml_nodes_revisions
109
);
110
111
/*
112
 * Indexes of rootnodeid, parentnodeid, and nodename in xml_nodes_revisions
113
 */
114
CREATE INDEX xml_nodes_revisions_idx1 ON xml_nodes_revisions (rootnodeid);
115
CREATE INDEX xml_nodes_revisions_idx2 ON xml_nodes_revisions (parentnodeid);
116
CREATE INDEX xml_nodes_revisions_idx3 ON xml_nodes_revisions (nodename);
117
118
119
120
/*
121 2276 jones
 * XML Catalog -- table to store all external sources for XML documents
122
 */
123
CREATE SEQUENCE xml_catalog_id_seq;
124
CREATE TABLE xml_catalog (
125
	catalog_id INT8 default nextval('xml_catalog_id_seq'),
126
                                        -- the id for this catalog entry
127
	entry_type VARCHAR(500),	-- the type of this catalog entry
128
					-- (e.g., DTD, XSD, XSL)
129
	source_doctype VARCHAR(500),	-- the source public_id for transforms
130
	target_doctype VARCHAR(500),	-- the target public_id for transforms
131
	public_id VARCHAR(500),	-- the unique id for this type
132
	system_id VARCHAR(1000),	-- the local location of the object
133
   CONSTRAINT xml_catalog_pk PRIMARY KEY (catalog_id),
134 2422 sgarg
   CONSTRAINT xml_catalog_uk UNIQUE
135 2276 jones
              (entry_type, source_doctype, target_doctype, public_id)
136
);
137
138 2422 sgarg
/*
139 2276 jones
 * Sequence to get uniqueID for Accession #
140
 */
141
CREATE SEQUENCE xml_documents_id_seq;
142 2422 sgarg
/*
143 2276 jones
 * Documents -- table to store XML documents
144
 */
145
CREATE TABLE xml_documents (
146
	docid VARCHAR(250),	-- the document id #
147
	rootnodeid INT8,		-- reference to root node of the DOM
148
	docname VARCHAR(100),	-- usually the root element name
149
	doctype VARCHAR(100),	-- public id indicating document type
150
	user_owner VARCHAR(100),	-- the user owned the document
151
	user_updated VARCHAR(100),	-- the user updated the document
152
	server_location INT8,	-- the server on which this document resides
153
	rev INT8 default 1,   -- the revision number of the document
154
	date_created DATE,
155
	date_updated DATE,
156
	public_access INT8,	-- flag for public access
157 2422 sgarg
        catalog_id INT8,	-- reference to xml_catalog
158 2276 jones
     CONSTRAINT xml_documents_pk PRIMARY KEY (docid),
159
     CONSTRAINT xml_documents_rep_fk
160 2422 sgarg
     		FOREIGN KEY (server_location) REFERENCES xml_replication,
161
    CONSTRAINT xml_documents_root_fk
162 2276 jones
		FOREIGN KEY (rootnodeid) REFERENCES xml_nodes,
163 2422 sgarg
   CONSTRAINT xml_documents_catalog_fk
164 2276 jones
		FOREIGN KEY (catalog_id) REFERENCES xml_catalog
165
);
166
167 2422 sgarg
/*
168 2276 jones
 * Index of <docid,doctype> in xml_document
169
 */
170
CREATE INDEX xml_documents_idx1 ON xml_documents (docid, doctype);
171 3052 jones
CREATE INDEX xml_documents_idx2 ON xml_documents (lower(user_owner));
172 3504 tao
CREATE INDEX xml_documents_idx3 ON xml_documents (rootnodeid);
173 7455 leinfelder
CREATE INDEX xml_documents_idx5 ON xml_documents (docid, rev);
174 2276 jones
175 2422 sgarg
/*
176 2276 jones
 * Revised Documents -- table to store XML documents saved after an UPDATE
177
 *                    or DELETE
178
 */
179
CREATE SEQUENCE xml_revisions_id_seq;
180
CREATE TABLE xml_revisions (
181
	revisionid INT8  default nextval('xml_revisions_id_seq'),
182
                                        -- the revision number we are saving
183
	docid VARCHAR(250),	-- the document id #
184
	rootnodeid INT8,		-- reference to root node of the DOM
185
	docname VARCHAR(100),	-- usually the root element name
186
	doctype VARCHAR(100),	-- public id indicating document type
187
	user_owner VARCHAR(100),
188
	user_updated VARCHAR(100),
189
	server_location INT8,
190
	rev INT8,
191
	date_created DATE,
192
	date_updated DATE,
193
	public_access INT8,	-- flag for public access
194 2422 sgarg
        catalog_id INT8,	-- reference to xml_catalog
195 2276 jones
   CONSTRAINT xml_revisions_pk PRIMARY KEY (revisionid),
196
   CONSTRAINT xml_revisions_rep_fk
197
		FOREIGN KEY (server_location) REFERENCES xml_replication,
198 2422 sgarg
   CONSTRAINT xml_revisions_root_fk
199 2514 sgarg
		FOREIGN KEY (rootnodeid) REFERENCES xml_nodes_revisions,
200 2422 sgarg
   CONSTRAINT xml_revisions_catalog_fk
201 2276 jones
		FOREIGN KEY (catalog_id) REFERENCES xml_catalog
202
);
203
204 5171 daigle
CREATE INDEX xml_revisions_idx1 ON xml_revisions (docid);
205 2276 jones
206 2422 sgarg
/*
207 2276 jones
 * ACL -- table to store ACL for XML documents by principals
208
 */
209
CREATE TABLE xml_access (
210 6122 leinfelder
	guid text,	-- foreign key to system metadata
211 6923 leinfelder
	accessfileid text,	-- the id for the access file
212 2276 jones
	principal_name VARCHAR(100),	-- name of user, group, etc.
213
	permission INT8,		-- "read", "write", "all"
214
	perm_type VARCHAR(32),	-- "allowed" or "denied"
215
	perm_order VARCHAR(32),	-- "allow first" or "deny first"
216
	begin_time DATE,		-- the time that permission begins
217
	end_time DATE,		-- the time that permission ends
218
	ticket_count INT8,		-- ticket counter for that permission
219
  subtreeid VARCHAR(32),
220
  startnodeid INT8,
221
  endnodeid INT8,
222 6209 leinfelder
   CONSTRAINT xml_access_ck CHECK (begin_time < end_time)
223 2276 jones
);
224 3052 jones
CREATE INDEX xml_access_idx1 ON xml_access (lower(principal_name));
225
CREATE INDEX xml_access_idx2 ON xml_access (permission);
226
CREATE INDEX xml_access_idx3 ON xml_access (perm_type);
227
CREATE INDEX xml_access_idx4 ON xml_access (perm_order);
228
CREATE INDEX xml_access_idx5 ON xml_access (subtreeid);
229 7455 leinfelder
CREATE INDEX xml_access_idx6 on xml_access(guid);
230 6122 leinfelder
/*
231
 * ALTER TABLE xml_access ADD COLUMN guid text;
232
*/
233 2276 jones
234 2422 sgarg
/*
235
 * Index of Nodes -- table to store precomputed paths through tree for
236 2276 jones
 * quick searching in structured searches
237
 */
238
CREATE TABLE xml_index (
239
	nodeid INT8,		-- the unique node id
240 3352 tao
	path TEXT,	-- precomputed path through tree
241 2276 jones
	docid VARCHAR(250),	-- index to the document id
242
	doctype VARCHAR(100),	-- public id indicating document type
243 2377 sgarg
        parentnodeid INT8,      -- id of the parent of the node represented
244
				-- by this row
245 2276 jones
   CONSTRAINT xml_index_pk PRIMARY KEY (nodeid,path),
246
   CONSTRAINT xml_index_nodeid_fk FOREIGN KEY (nodeid) REFERENCES xml_nodes,
247 2422 sgarg
   CONSTRAINT xml_index_docid_fk
248 2276 jones
		FOREIGN KEY (docid) REFERENCES xml_documents
249
);
250
251 2422 sgarg
/*
252
 * Index of the paths in xml_index
253 2276 jones
 */
254
CREATE INDEX xml_index_idx1 ON xml_index (path);
255 3347 tao
CREATE INDEX xml_index_idx2 ON xml_index (docid);
256 3504 tao
CREATE INDEX xml_index_idx3 ON xml_index (nodeid);
257 2276 jones
258
CREATE SEQUENCE xml_relation_id_seq;
259
CREATE TABLE xml_relation (
260
	relationid INT8 default nextval('xml_relation_id_seq') PRIMARY KEY,
261
					     -- unique id
262
	docid VARCHAR(250) ,         -- the docid of the package file
263
	                                     -- that this relation came from
264
        packagetype VARCHAR(250),          -- the type of the package
265
	subject VARCHAR(512) NOT NULL, -- the subject of the relation
266
	subdoctype VARCHAR(128),         	-- the doctype of the subject
267
	relationship VARCHAR(128)  NOT NULL,-- the relationship type
268
	object VARCHAR(512) NOT NULL, -- the object of the relation
269
	objdoctype VARCHAR(128),          -- the doctype of the object
270
	CONSTRAINT xml_relation_uk UNIQUE (docid, subject, relationship, object),
271 2422 sgarg
	CONSTRAINT xml_relation_docid_fk
272 2276 jones
		FOREIGN KEY (docid) REFERENCES xml_documents
273
);
274
275 2422 sgarg
/*
276 2769 jones
 * Table used to store all document identifiers in metacat.  Each identifier
277 5276 jones
 * has a globally unique, unconstrained string, which we will refer to as a
278
 * GUID, and a local metacat identifier, which consists of the docid
279
 * and revision fields. Each row maps one global identifier to the local
280
 * identifier (docid) used within metacat.
281 2276 jones
 */
282 2769 jones
CREATE TABLE identifier (
283 5276 jones
   guid   text,          -- the globally unique string identifier
284
   docid  VARCHAR(250),	 -- the local document id #
285
   rev    INT8,          -- the revision part of the local identifier
286
   CONSTRAINT identifier_pk PRIMARY KEY (guid)
287 2276 jones
);
288 7455 leinfelder
CREATE INDEX identifier_guid on identifier(guid);
289
CREATE INDEX identifier_docid on identifier(docid);
290
CREATE INDEX identifier_rev on identifier(rev);
291
CREATE INDEX identifier_docid_rev on identifier(docid, rev);
292 8335 leinfelder
CREATE INDEX identifier_docid_rev_log ON identifier((docid||'.'||rev));
293 2276 jones
294 2422 sgarg
/*
295 5333 berkley
 * Table used to store all document identifiers for system metadata objects
296
 * similar restraints to identifier.  Cannot use identifier table for this
297
 * purpose because then you have to worry about whether you insert the
298
 * data first or the systemMetadata first.
299
 */
300
CREATE TABLE systemMetadata (
301 6107 leinfelder
	guid   text,          -- the globally unique string identifier of the object that the system metadata describes
302 8810 leinfelder
	series_id text, -- the series identifier
303 6561 leinfelder
	serial_version VARCHAR(256), --the serial version of the object
304 6107 leinfelder
	date_uploaded TIMESTAMP, -- the date/time the document was first submitted
305
	rights_holder VARCHAR(250), --the user who has rights to the document, usually the first persons to upload it
306
	checksum VARCHAR(512), --the checksum of the doc using the given algorithm (see below)
307
	checksum_algorithm VARCHAR(250), --the algorithm used to calculate the checksum
308
	origin_member_node VARCHAR(250), --the member node where the document was first uploaded
309
	authoritive_member_node VARCHAR(250), --the member node that currently controls the document
310
	date_modified TIMESTAMP, -- the last date/time that the file was changed
311
	submitter VARCHAR(256), -- the user who originally submitted the doc
312
	object_format VARCHAR(256), --the format of the object
313
	size VARCHAR(256), --the size of the object
314 6766 leinfelder
	archived boolean,	 -- specifies whether this an archived object
315 6107 leinfelder
	replication_allowed boolean,	 -- replication allowed
316
	number_replicas INT8, 	-- the number of replicas allowed
317 6375 leinfelder
	obsoletes   text,       -- the identifier that this record obsoletes
318
	obsoleted_by   text,     -- the identifier of the record that replaces this record
319 9216 tao
  media_type   text,      -- the media type of this object
320
  file_name    text,      -- the suggested file name for this object
321 6107 leinfelder
	CONSTRAINT systemMetadata_pk PRIMARY KEY (guid)
322 5333 berkley
);
323 9224 tao
324 6100 leinfelder
/*
325 9224 tao
 * Table used to store the properties for media types. They are part of the system metadata. But a media type
326
 * can have multiple properties, we have to store them in a separate table. The guids in this table refer
327 9225 tao
 * the guids in the systemMetadata.
328 9224 tao
 */
329
CREATE TABLE smMediaTypeProperties (
330
	guid    text,  -- id refer to guid in the system metadata table
331
  name    text, -- name of the property
332
  value    text, -- value of the property
333 9225 tao
  CONSTRAINT smMediaTypeProperties_fk
334 9224 tao
     FOREIGN KEY (guid) REFERENCES systemMetadata DEFERRABLE
335
);
336
/*
337 6100 leinfelder
 * For devs to remove docid, rev
338
 * ALTER TABLE systemMetadata DROP COLUMN docid;
339
 * ALTER TABLE systemMetadata DROP COLUMN rev;
340 6107 leinfelder
 * ALTER TABLE systemMetadata ADD COLUMN replication_allowed boolean;
341
 * ALTER TABLE systemMetadata ADD COLUMN number_replicas INT8;
342 6100 leinfelder
 */
343 5333 berkley
344 9483 tao
CREATE SEQUENCE policy_id_seq;
345 7392 leinfelder
CREATE TABLE smReplicationPolicy (
346 9483 tao
  policy_id INT8 default nextval('policy_id_seq'),
347 6107 leinfelder
	guid text,	-- the globally unique string identifier of the object that the system metadata describes
348
	member_node VARCHAR(250),	 -- replication member node
349
	policy text,	 -- the policy (preferred, blocked, etc...TBD)
350 7392 leinfelder
	CONSTRAINT smReplicationPolicy_fk
351 6933 jones
		FOREIGN KEY (guid) REFERENCES systemMetadata DEFERRABLE
352 6107 leinfelder
);
353
354 7392 leinfelder
CREATE TABLE smReplicationStatus (
355 6107 leinfelder
	guid text,	-- the globally unique string identifier of the object that the system metadata describes
356
	member_node VARCHAR(250),	 -- replication member node
357
	status VARCHAR(250),	 -- replication status
358
	date_verified TIMESTAMP, 	-- the date replication was verified
359 7392 leinfelder
	CONSTRAINT smReplicationStatus_fk
360 6933 jones
		FOREIGN KEY (guid) REFERENCES systemMetadata DEFERRABLE
361 6107 leinfelder
);
362
363 6098 leinfelder
/*
364 2422 sgarg
 * accesssubtree -- table to store access subtree info
365 2276 jones
 */
366
CREATE TABLE xml_accesssubtree (
367
	docid		VARCHAR(250),	-- the document id #
368
  rev 		INT8 default 1, --the revision number of the docume
369
  controllevel VARCHAR(50), -- the level it control -- document or subtree
370 2422 sgarg
  subtreeid VARCHAR(250), -- the subtree id
371 2276 jones
	startnodeid	INT8,	-- the start node id of access subtree
372
  endnodeid INT8, -- the end node if of access subtree
373 2422 sgarg
  CONSTRAINT xml_accesssubtree_docid_fk
374 2276 jones
		FOREIGN KEY (docid) REFERENCES xml_documents
375
);
376
377
/*
378 2422 sgarg
 * Returnfields -- table to store combinations of returnfields requested
379
 *		    and the number of times this table is accessed
380
 */
381 2949 berkley
CREATE SEQUENCE xml_returnfield_id_seq;
382 2422 sgarg
CREATE TABLE xml_returnfield (
383
        returnfield_id     INT8 default nextval('xml_returnfield_id_seq'),   -- the id for this returnfield entry
384
        returnfield_string VARCHAR(2000),                                    -- the returnfield string
385
        usage_count        INT8,                                             -- the number of times this string has been requested
386
        CONSTRAINT xml_returnfield_pk PRIMARY KEY (returnfield_id)
387
);
388
CREATE INDEX xml_returnfield_idx1 ON xml_returnfield(returnfield_string);
389
390
/*
391
 * Queryresults -- table to store queryresults for a given docid
392
 * and returnfield_id
393
 */
394 2949 berkley
CREATE SEQUENCE xml_queryresult_id_seq;
395 2422 sgarg
CREATE TABLE xml_queryresult(
396
  queryresult_id INT8 default nextval('xml_queryresult_id_seq'), -- id for this entry
397
  returnfield_id       INT8,          -- id for the returnfield corresponding to this entry
398
  docid                VARCHAR(250),  -- docid of the document
399 3336 tao
  queryresult_string   TEXT, -- resultant text generated for this docid and given
400 2422 sgarg
  				       -- returnfield
401
  CONSTRAINT xml_queryresult_pk PRIMARY KEY (queryresult_id),
402
  CONSTRAINT xml_queryresult_searchid_fk
403
               FOREIGN KEY (returnfield_id) REFERENCES xml_returnfield
404
);
405
406
CREATE INDEX xml_queryresult_idx1 ON xml_queryresult (returnfield_id, docid);
407
408
/*
409 2276 jones
 * Logging -- table to store metadata and data access log
410
 */
411
CREATE SEQUENCE access_log_id_seq;
412
CREATE TABLE access_log (
413
  entryid       INT8 default nextval ('access_log_id_seq'), -- the identifier for the log event
414
  ip_address    VARCHAR(512),   -- the ip address inititiating the event
415 6542 leinfelder
  user_agent    VARCHAR(512),   -- the user agent for the request
416
  principal     VARCHAR(512),   -- the user initiating the event
417 2276 jones
  docid         VARCHAR(250),	-- the document id #
418
  event         VARCHAR(512),   -- the code symbolizing the event type
419
  date_logged   TIMESTAMP,      -- the datetime on which the event occurred
420
  CONSTRAINT access_log_pk PRIMARY KEY (entryid)
421
);
422 8335 leinfelder
CREATE INDEX access_log_docid ON access_log(docid);
423 2276 jones
424 8335 leinfelder
425 7830 leinfelder
/*
426
 * the index_event table for solr-based indexing
427
 */
428
CREATE TABLE index_event (
429
	guid text,
430
	event_action VARCHAR(250),
431
	description text,
432
	event_date TIMESTAMP
433
);
434 2514 sgarg
435 2422 sgarg
/*
436 2514 sgarg
 * Table for indexing the paths specified the administrator in metacat.properties
437
 */
438
439
CREATE SEQUENCE xml_path_index_id_seq;
440
CREATE TABLE xml_path_index (
441
    nodeid INT8  default nextval('xml_path_index_id_seq'),
442
        docid VARCHAR(250),     -- the document id
443 7053 leinfelder
        path TEXT,     -- precomputed path through tree
444 3336 tao
        nodedata TEXT, -- the data for this node (e.g.,
445 2514 sgarg
                                -- for TEXT it is the content)
446 6012 leinfelder
        nodedatanumerical FLOAT8, -- the data for this node if it is a number
447
        nodedatadate TIMESTAMP, -- the data for this node if it is a date
448 2514 sgarg
        parentnodeid INT8,      -- id of the parent of the node represented
449
                                -- by this row
450
   CONSTRAINT xml_path_index_pk PRIMARY KEY (nodeid),
451 2515 sgarg
   CONSTRAINT xml_path_index_docid_fk
452 2514 sgarg
                FOREIGN KEY (docid) REFERENCES xml_documents
453
);
454
455
/*
456
 * Indexes of path, nodedata and nodedatanumerical in xml_path_index
457
 */
458
CREATE INDEX xml_path_index_idx1 ON xml_path_index (path);
459
CREATE INDEX xml_path_index_idx2 ON xml_path_index (nodedata);
460
CREATE INDEX xml_path_index_idx3 ON xml_path_index (nodedatanumerical);
461 3052 jones
CREATE INDEX xml_path_index_idx4 ON xml_path_index (upper(nodedata));
462 6012 leinfelder
CREATE INDEX xml_path_index_idx5 ON xml_path_index (nodedatadate);
463 7455 leinfelder
CREATE INDEX xml_path_index_idx6 ON xml_path_index (docid);
464 2514 sgarg
465
/*
466 2276 jones
 * harvest_site_schedule -- table to store harvest sites and schedule info
467
 */
468
CREATE TABLE harvest_site_schedule (
469
  site_schedule_id INT8,         -- unique id
470
  documentlisturl  VARCHAR(255), -- URL of the site harvest document list
471
  ldapdn           VARCHAR(255), -- LDAP distinguished name for site account
472
  datenextharvest  DATE,         -- scheduled date of next harvest
473
  datelastharvest  DATE,         -- recorded date of last harvest
474
  updatefrequency  INT8,         -- the harvest update frequency
475
  unit             VARCHAR(50),  -- update unit -- days weeks or months
476
  contact_email    VARCHAR(50),  -- email address of the site contact person
477
  ldappwd          VARCHAR(20),  -- LDAP password for site account
478
  CONSTRAINT harvest_site_schedule_pk PRIMARY KEY (site_schedule_id)
479
);
480
481 2422 sgarg
/*
482 2276 jones
 * harvest_log -- table to log entries for harvest operations
483
 */
484
CREATE TABLE harvest_log (
485
  harvest_log_id         INT8,          -- unique id
486
  harvest_date           DATE,          -- date of the current harvest
487
  status                 INT8,          -- non-zero indicates an error status
488
  message                VARCHAR(1000), -- text message for this log entry
489 8028 leinfelder
  harvest_operation_code VARCHAR(1000),   -- the type of harvest operation
490 2276 jones
  site_schedule_id       INT8,          -- site schedule id, or 0 if no site
491
  CONSTRAINT harvest_log_pk PRIMARY KEY (harvest_log_id)
492
);
493
494 2422 sgarg
/*
495 2276 jones
 * harvest_detail_log -- table to log detailed info about documents that
496
 *                       generated errors during the harvest
497
 */
498
CREATE TABLE harvest_detail_log (
499
  detail_log_id          INT8,          -- unique id
500
  harvest_log_id         INT8,          -- ponter to the related log entry
501
  scope                  VARCHAR(50),   -- document scope
502
  identifier             INT8,          -- document identifier
503
  revision               INT8,          -- document revision
504
  document_url           VARCHAR(255),  -- document URL
505
  error_message          VARCHAR(1000), -- text error message
506
  document_type          VARCHAR(100),  -- document type
507
  CONSTRAINT harvest_detail_log_pk PRIMARY KEY (detail_log_id),
508 2422 sgarg
  CONSTRAINT harvest_detail_log_fk
509 2276 jones
        FOREIGN KEY (harvest_log_id) REFERENCES harvest_log
510
);
511
512 4080 daigle
/*
513
 * db_version -- table to store the version history of this database
514
 */
515
CREATE SEQUENCE db_version_id_seq;
516
CREATE TABLE db_version (
517
  db_version_id   INT8 default nextval ('db_version_id_seq'), -- the identifier for the version
518
  version         VARCHAR(250),     -- the version number
519
  status          INT8,             -- status of the version
520
  date_created    TIMESTAMP,        -- the datetime on which the version was created
521
  CONSTRAINT db_version_pk PRIMARY KEY (db_version_id)
522
);
523 4975 daigle
524
/*
525
 * scheduled_job -- table to store scheduled jobs
526
 */
527
CREATE SEQUENCE scheduled_job_id_seq;
528
CREATE TABLE scheduled_job (
529
  id INT8 NOT NULL default nextval('scheduled_job_id_seq'),
530
  date_created TIMESTAMP NOT NULL,
531
  date_updated TIMESTAMP NOT NULL,
532
  status VARCHAR(64) NOT NULL,
533
  name VARCHAR(512) NOT NULL,
534
  trigger_name VARCHAR(512) NOT NULL,
535
  group_name VARCHAR(512) NOT NULL,
536
  class_name VARCHAR(1024) NOT NULL,
537
  start_time TIMESTAMP NOT NULL,
538 5027 daigle
  end_time TIMESTAMP,
539 4975 daigle
  interval_value INT NOT NULL,
540
  interval_unit VARCHAR(8) NOT NULL,
541
  CONSTRAINT scheduled_job_pk PRIMARY KEY (id),
542
  CONSTRAINT scheduled_job_uk UNIQUE (name)
543
);
544
545
/*
546
 * scheduled_job_params -- table to store scheduled jobs
547
 */
548
CREATE SEQUENCE scheduled_job_params_id_seq;
549
CREATE TABLE scheduled_job_params (
550
  id INT8  NOT NULL default nextval('scheduled_job_params_id_seq'),
551
  date_created TIMESTAMP NOT NULL,
552
  date_updated TIMESTAMP  NOT NULL,
553
  status VARCHAR(64)  NOT NULL,
554
  job_id INT8 NOT NULL,
555
  key VARCHAR(64) NOT NULL,
556
  value VARCHAR(1024) NOT NULL,
557
  CONSTRAINT scheduled_job_params_pk PRIMARY KEY (id),
558
  CONSTRAINT scheduled_job_params_fk
559
        FOREIGN KEY (job_id) REFERENCES scheduled_job(id)
560
);