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
INSERT INTO xml_replication (server, replicate, datareplicate, hub) VALUES ('localhost', '0', '0', '0');
47
48 2422 sgarg
/*
49 2276 jones
 * Nodes -- table to store XML Nodes (both elements and attributes)
50
 */
51
CREATE SEQUENCE xml_nodes_id_seq;
52
CREATE TABLE xml_nodes (
53
	nodeid INT8 default nextval('xml_nodes_id_seq'),
54
					-- the unique node id (pk)
55
	nodeindex INT8,		-- order of nodes within parent
56
	nodetype VARCHAR(20),	-- type (DOCUMENT, COMMENT, PI,
57 2359 sgarg
				-- ELEMENT, ATTRIBUTE, TEXT)
58 2276 jones
	nodename VARCHAR(250),	-- the name of an element or attribute
59
        nodeprefix VARCHAR(50), -- the namespace prefix of the node
60 2422 sgarg
	nodedata VARCHAR(4000), -- the data for this node (e.g.,
61 2359 sgarg
				-- for TEXT it is the content)
62
	parentnodeid INT8,	-- index of the parent of this node
63
	rootnodeid INT8,	-- index of the root node of this tree
64 2276 jones
	docid VARCHAR(250),	-- index to the document id
65
	date_created DATE,
66
	date_updated DATE,
67 2359 sgarg
        nodedatanumerical FLOAT8, -- the data for this node if
68
				  -- if it is a number
69 2276 jones
   CONSTRAINT xml_nodes_pk PRIMARY KEY (nodeid),
70 2422 sgarg
   CONSTRAINT xml_nodes_root_fk
71 2276 jones
		FOREIGN KEY (rootnodeid) REFERENCES xml_nodes,
72 2422 sgarg
   CONSTRAINT xml_nodes_parent_fk
73 2276 jones
		FOREIGN KEY (parentnodeid) REFERENCES xml_nodes
74
);
75 2422 sgarg
/*
76 2276 jones
 * Indexes of rootnodeid, parentnodeid, and nodename in xml_nodes
77
 */
78
CREATE INDEX xml_nodes_idx1 ON xml_nodes (rootnodeid);
79
CREATE INDEX xml_nodes_idx2 ON xml_nodes (parentnodeid);
80
CREATE INDEX xml_nodes_idx3 ON xml_nodes (nodename);
81
82 2519 sgarg
83 2422 sgarg
/*
84 2519 sgarg
 * Table for storing the nodes for the old revisions of the document and the deleted documents
85
 */
86
CREATE TABLE xml_nodes_revisions (
87
        nodeid INT8,            -- the unique node id (pk)
88
        nodeindex INT8,         -- order of nodes within parent
89
        nodetype VARCHAR(20),   -- type (DOCUMENT, COMMENT, PI,
90
                                -- ELEMENT, ATTRIBUTE, TEXT)
91
        nodename VARCHAR(250),  -- the name of an element or attribute
92
        nodeprefix VARCHAR(50), -- the namespace prefix of the node
93
        nodedata VARCHAR(4000), -- the data for this node (e.g.,
94
                                -- for TEXT it is the content)
95
        parentnodeid INT8,      -- index of the parent of this node
96
        rootnodeid INT8,        -- index of the root node of this tree
97
        docid VARCHAR(250),     -- index to the document id
98
        date_created DATE,
99
        date_updated DATE,
100
        nodedatanumerical FLOAT8, -- the data for this node if
101
                                  -- if it is a number
102
   CONSTRAINT xml_nodes_revisions_pk PRIMARY KEY (nodeid),
103
   CONSTRAINT xml_nodes_revisions_root_fk
104
                FOREIGN KEY (rootnodeid) REFERENCES xml_nodes_revisions,
105
   CONSTRAINT xml_nodes_revisions_parent_fk
106
                FOREIGN KEY (parentnodeid) REFERENCES xml_nodes_revisions
107
);
108
109
/*
110
 * Indexes of rootnodeid, parentnodeid, and nodename in xml_nodes_revisions
111
 */
112
CREATE INDEX xml_nodes_revisions_idx1 ON xml_nodes_revisions (rootnodeid);
113
CREATE INDEX xml_nodes_revisions_idx2 ON xml_nodes_revisions (parentnodeid);
114
CREATE INDEX xml_nodes_revisions_idx3 ON xml_nodes_revisions (nodename);
115
116
117
118
/*
119 2276 jones
 * XML Catalog -- table to store all external sources for XML documents
120
 */
121
CREATE SEQUENCE xml_catalog_id_seq;
122
CREATE TABLE xml_catalog (
123
	catalog_id INT8 default nextval('xml_catalog_id_seq'),
124
                                        -- the id for this catalog entry
125
	entry_type VARCHAR(500),	-- the type of this catalog entry
126
					-- (e.g., DTD, XSD, XSL)
127
	source_doctype VARCHAR(500),	-- the source public_id for transforms
128
	target_doctype VARCHAR(500),	-- the target public_id for transforms
129
	public_id VARCHAR(500),	-- the unique id for this type
130
	system_id VARCHAR(1000),	-- the local location of the object
131
   CONSTRAINT xml_catalog_pk PRIMARY KEY (catalog_id),
132 2422 sgarg
   CONSTRAINT xml_catalog_uk UNIQUE
133 2276 jones
              (entry_type, source_doctype, target_doctype, public_id)
134
);
135
136 2422 sgarg
/*
137 2276 jones
 * Sequence to get uniqueID for Accession #
138
 */
139
CREATE SEQUENCE xml_documents_id_seq;
140 2422 sgarg
/*
141 2276 jones
 * Documents -- table to store XML documents
142
 */
143
CREATE TABLE xml_documents (
144
	docid VARCHAR(250),	-- the document id #
145
	rootnodeid INT8,		-- reference to root node of the DOM
146
	docname VARCHAR(100),	-- usually the root element name
147
	doctype VARCHAR(100),	-- public id indicating document type
148
	user_owner VARCHAR(100),	-- the user owned the document
149
	user_updated VARCHAR(100),	-- the user updated the document
150
	server_location INT8,	-- the server on which this document resides
151
	rev INT8 default 1,   -- the revision number of the document
152
	date_created DATE,
153
	date_updated DATE,
154
	public_access INT8,	-- flag for public access
155 2422 sgarg
        catalog_id INT8,	-- reference to xml_catalog
156 2276 jones
     CONSTRAINT xml_documents_pk PRIMARY KEY (docid),
157
     CONSTRAINT xml_documents_rep_fk
158 2422 sgarg
     		FOREIGN KEY (server_location) REFERENCES xml_replication,
159
    CONSTRAINT xml_documents_root_fk
160 2276 jones
		FOREIGN KEY (rootnodeid) REFERENCES xml_nodes,
161 2422 sgarg
   CONSTRAINT xml_documents_catalog_fk
162 2276 jones
		FOREIGN KEY (catalog_id) REFERENCES xml_catalog
163
);
164
165 2422 sgarg
/*
166 2276 jones
 * Index of <docid,doctype> in xml_document
167
 */
168
CREATE INDEX xml_documents_idx1 ON xml_documents (docid, doctype);
169 3052 jones
CREATE INDEX xml_documents_idx2 ON xml_documents (lower(user_owner));
170 2276 jones
171 2422 sgarg
/*
172 2276 jones
 * Revised Documents -- table to store XML documents saved after an UPDATE
173
 *                    or DELETE
174
 */
175
CREATE SEQUENCE xml_revisions_id_seq;
176
CREATE TABLE xml_revisions (
177
	revisionid INT8  default nextval('xml_revisions_id_seq'),
178
                                        -- the revision number we are saving
179
	docid VARCHAR(250),	-- the document id #
180
	rootnodeid INT8,		-- reference to root node of the DOM
181
	docname VARCHAR(100),	-- usually the root element name
182
	doctype VARCHAR(100),	-- public id indicating document type
183
	user_owner VARCHAR(100),
184
	user_updated VARCHAR(100),
185
	server_location INT8,
186
	rev INT8,
187
	date_created DATE,
188
	date_updated DATE,
189
	public_access INT8,	-- flag for public access
190 2422 sgarg
        catalog_id INT8,	-- reference to xml_catalog
191 2276 jones
   CONSTRAINT xml_revisions_pk PRIMARY KEY (revisionid),
192
   CONSTRAINT xml_revisions_rep_fk
193
		FOREIGN KEY (server_location) REFERENCES xml_replication,
194 2422 sgarg
   CONSTRAINT xml_revisions_root_fk
195 2514 sgarg
		FOREIGN KEY (rootnodeid) REFERENCES xml_nodes_revisions,
196 2422 sgarg
   CONSTRAINT xml_revisions_catalog_fk
197 2276 jones
		FOREIGN KEY (catalog_id) REFERENCES xml_catalog
198
);
199
200
201 2422 sgarg
/*
202 2276 jones
 * ACL -- table to store ACL for XML documents by principals
203
 */
204
CREATE TABLE xml_access (
205
	docid VARCHAR(250),	-- the document id #
206
	accessfileid VARCHAR(250),	-- the document id # for the access file
207
	principal_name VARCHAR(100),	-- name of user, group, etc.
208
	permission INT8,		-- "read", "write", "all"
209
	perm_type VARCHAR(32),	-- "allowed" or "denied"
210
	perm_order VARCHAR(32),	-- "allow first" or "deny first"
211
	begin_time DATE,		-- the time that permission begins
212
	end_time DATE,		-- the time that permission ends
213
	ticket_count INT8,		-- ticket counter for that permission
214
  subtreeid VARCHAR(32),
215
  startnodeid INT8,
216
  endnodeid INT8,
217
   CONSTRAINT xml_access_ck CHECK (begin_time < end_time),
218 2422 sgarg
   CONSTRAINT xml_access_accessfileid_fk
219 2276 jones
		FOREIGN KEY (accessfileid) REFERENCES xml_documents
220
);
221 3052 jones
CREATE INDEX xml_access_idx1 ON xml_access (lower(principal_name));
222
CREATE INDEX xml_access_idx2 ON xml_access (permission);
223
CREATE INDEX xml_access_idx3 ON xml_access (perm_type);
224
CREATE INDEX xml_access_idx4 ON xml_access (perm_order);
225
CREATE INDEX xml_access_idx5 ON xml_access (subtreeid);
226 2276 jones
227 2422 sgarg
/*
228
 * Index of Nodes -- table to store precomputed paths through tree for
229 2276 jones
 * quick searching in structured searches
230
 */
231
CREATE TABLE xml_index (
232
	nodeid INT8,		-- the unique node id
233
	path VARCHAR(1000),	-- precomputed path through tree
234
	docid VARCHAR(250),	-- index to the document id
235
	doctype VARCHAR(100),	-- public id indicating document type
236 2377 sgarg
        parentnodeid INT8,      -- id of the parent of the node represented
237
				-- by this row
238 2276 jones
   CONSTRAINT xml_index_pk PRIMARY KEY (nodeid,path),
239
   CONSTRAINT xml_index_nodeid_fk FOREIGN KEY (nodeid) REFERENCES xml_nodes,
240 2422 sgarg
   CONSTRAINT xml_index_docid_fk
241 2276 jones
		FOREIGN KEY (docid) REFERENCES xml_documents
242
);
243
244 2422 sgarg
/*
245
 * Index of the paths in xml_index
246 2276 jones
 */
247
CREATE INDEX xml_index_idx1 ON xml_index (path);
248
249
CREATE SEQUENCE xml_relation_id_seq;
250
CREATE TABLE xml_relation (
251
	relationid INT8 default nextval('xml_relation_id_seq') PRIMARY KEY,
252
					     -- unique id
253
	docid VARCHAR(250) ,         -- the docid of the package file
254
	                                     -- that this relation came from
255
        packagetype VARCHAR(250),          -- the type of the package
256
	subject VARCHAR(512) NOT NULL, -- the subject of the relation
257
	subdoctype VARCHAR(128),         	-- the doctype of the subject
258
	relationship VARCHAR(128)  NOT NULL,-- the relationship type
259
	object VARCHAR(512) NOT NULL, -- the object of the relation
260
	objdoctype VARCHAR(128),          -- the doctype of the object
261
	CONSTRAINT xml_relation_uk UNIQUE (docid, subject, relationship, object),
262 2422 sgarg
	CONSTRAINT xml_relation_docid_fk
263 2276 jones
		FOREIGN KEY (docid) REFERENCES xml_documents
264
);
265
266 2422 sgarg
/*
267 2769 jones
 * Table used to store all document identifiers in metacat.  Each identifier
268
 * consists of 4 subparts, an authority, namespace, object, and revision as
269
 * defined in the LSID specification.
270 2276 jones
 */
271 2769 jones
CREATE SEQUENCE identifier_id_seq;
272
CREATE TABLE identifier (
273
   id INT8 default nextval('identifier_id_seq') PRIMARY KEY, -- primary id
274
   authority VARCHAR(255),  -- the authority issuing the identifier
275
   namespace VARCHAR(255),  -- the namespace qualifying the identifier
276
   object    VARCHAR(255),  -- the local part of the identifier for a particular object
277
   revision  VARCHAR(255)   -- the revision part of the identifier
278 2276 jones
);
279
280 2422 sgarg
/*
281
 * accesssubtree -- table to store access subtree info
282 2276 jones
 */
283
CREATE TABLE xml_accesssubtree (
284
	docid		VARCHAR(250),	-- the document id #
285
  rev 		INT8 default 1, --the revision number of the docume
286
  controllevel VARCHAR(50), -- the level it control -- document or subtree
287 2422 sgarg
  subtreeid VARCHAR(250), -- the subtree id
288 2276 jones
	startnodeid	INT8,	-- the start node id of access subtree
289
  endnodeid INT8, -- the end node if of access subtree
290 2422 sgarg
  CONSTRAINT xml_accesssubtree_docid_fk
291 2276 jones
		FOREIGN KEY (docid) REFERENCES xml_documents
292
);
293
294
/*
295 2422 sgarg
 * Returnfields -- table to store combinations of returnfields requested
296
 *		    and the number of times this table is accessed
297
 */
298 2949 berkley
CREATE SEQUENCE xml_returnfield_id_seq;
299 2422 sgarg
CREATE TABLE xml_returnfield (
300
        returnfield_id     INT8 default nextval('xml_returnfield_id_seq'),   -- the id for this returnfield entry
301
        returnfield_string VARCHAR(2000),                                    -- the returnfield string
302
        usage_count        INT8,                                             -- the number of times this string has been requested
303
        CONSTRAINT xml_returnfield_pk PRIMARY KEY (returnfield_id)
304
);
305
CREATE INDEX xml_returnfield_idx1 ON xml_returnfield(returnfield_string);
306
307
/*
308
 * Queryresults -- table to store queryresults for a given docid
309
 * and returnfield_id
310
 */
311 2949 berkley
CREATE SEQUENCE xml_queryresult_id_seq;
312 2422 sgarg
CREATE TABLE xml_queryresult(
313
  queryresult_id INT8 default nextval('xml_queryresult_id_seq'), -- id for this entry
314
  returnfield_id       INT8,          -- id for the returnfield corresponding to this entry
315
  docid                VARCHAR(250),  -- docid of the document
316
  queryresult_string   VARCHAR(4000), -- resultant text generated for this docid and given
317
  				       -- returnfield
318
  CONSTRAINT xml_queryresult_pk PRIMARY KEY (queryresult_id),
319
  CONSTRAINT xml_queryresult_searchid_fk
320
               FOREIGN KEY (returnfield_id) REFERENCES xml_returnfield
321
);
322
323
CREATE INDEX xml_queryresult_idx1 ON xml_queryresult (returnfield_id, docid);
324
325
/*
326 2276 jones
 * Logging -- table to store metadata and data access log
327
 */
328
CREATE SEQUENCE access_log_id_seq;
329
CREATE TABLE access_log (
330
  entryid       INT8 default nextval ('access_log_id_seq'), -- the identifier for the log event
331
  ip_address    VARCHAR(512),   -- the ip address inititiating the event
332
  principal     VARCHAR(512),   -- the user initiiating the event
333
  docid         VARCHAR(250),	-- the document id #
334
  event         VARCHAR(512),   -- the code symbolizing the event type
335
  date_logged   TIMESTAMP,      -- the datetime on which the event occurred
336
  CONSTRAINT access_log_pk PRIMARY KEY (entryid)
337
);
338
339 2514 sgarg
340 2422 sgarg
/*
341 2514 sgarg
 * Table for indexing the paths specified the administrator in metacat.properties
342
 */
343
344
CREATE SEQUENCE xml_path_index_id_seq;
345
CREATE TABLE xml_path_index (
346
    nodeid INT8  default nextval('xml_path_index_id_seq'),
347
        docid VARCHAR(250),     -- the document id
348
        path VARCHAR(1000),     -- precomputed path through tree
349
        nodedata VARCHAR(4000), -- the data for this node (e.g.,
350
                                -- for TEXT it is the content)
351
        nodedatanumerical FLOAT8, -- the data for this node if
352
                                  -- if it is a number
353
        parentnodeid INT8,      -- id of the parent of the node represented
354
                                -- by this row
355
   CONSTRAINT xml_path_index_pk PRIMARY KEY (nodeid),
356 2515 sgarg
   CONSTRAINT xml_path_index_docid_fk
357 2514 sgarg
                FOREIGN KEY (docid) REFERENCES xml_documents
358
);
359
360
/*
361
 * Indexes of path, nodedata and nodedatanumerical in xml_path_index
362
 */
363
CREATE INDEX xml_path_index_idx1 ON xml_path_index (path);
364
CREATE INDEX xml_path_index_idx2 ON xml_path_index (nodedata);
365
CREATE INDEX xml_path_index_idx3 ON xml_path_index (nodedatanumerical);
366 3052 jones
CREATE INDEX xml_path_index_idx4 ON xml_path_index (upper(nodedata));
367 2514 sgarg
368
/*
369 2276 jones
 * harvest_site_schedule -- table to store harvest sites and schedule info
370
 */
371
CREATE TABLE harvest_site_schedule (
372
  site_schedule_id INT8,         -- unique id
373
  documentlisturl  VARCHAR(255), -- URL of the site harvest document list
374
  ldapdn           VARCHAR(255), -- LDAP distinguished name for site account
375
  datenextharvest  DATE,         -- scheduled date of next harvest
376
  datelastharvest  DATE,         -- recorded date of last harvest
377
  updatefrequency  INT8,         -- the harvest update frequency
378
  unit             VARCHAR(50),  -- update unit -- days weeks or months
379
  contact_email    VARCHAR(50),  -- email address of the site contact person
380
  ldappwd          VARCHAR(20),  -- LDAP password for site account
381
  CONSTRAINT harvest_site_schedule_pk PRIMARY KEY (site_schedule_id)
382
);
383
384 2422 sgarg
/*
385 2276 jones
 * harvest_log -- table to log entries for harvest operations
386
 */
387
CREATE TABLE harvest_log (
388
  harvest_log_id         INT8,          -- unique id
389
  harvest_date           DATE,          -- date of the current harvest
390
  status                 INT8,          -- non-zero indicates an error status
391
  message                VARCHAR(1000), -- text message for this log entry
392
  harvest_operation_code VARCHAR(30),   -- the type of harvest operation
393
  site_schedule_id       INT8,          -- site schedule id, or 0 if no site
394
  CONSTRAINT harvest_log_pk PRIMARY KEY (harvest_log_id)
395
);
396
397 2422 sgarg
/*
398 2276 jones
 * harvest_detail_log -- table to log detailed info about documents that
399
 *                       generated errors during the harvest
400
 */
401
CREATE TABLE harvest_detail_log (
402
  detail_log_id          INT8,          -- unique id
403
  harvest_log_id         INT8,          -- ponter to the related log entry
404
  scope                  VARCHAR(50),   -- document scope
405
  identifier             INT8,          -- document identifier
406
  revision               INT8,          -- document revision
407
  document_url           VARCHAR(255),  -- document URL
408
  error_message          VARCHAR(1000), -- text error message
409
  document_type          VARCHAR(100),  -- document type
410
  CONSTRAINT harvest_detail_log_pk PRIMARY KEY (detail_log_id),
411 2422 sgarg
  CONSTRAINT harvest_detail_log_fk
412 2276 jones
        FOREIGN KEY (harvest_log_id) REFERENCES harvest_log
413
);
414