1 ;;;; -*- Mode: Lisp; Syntax: ANSI-Common-Lisp; Base: 10; Package: umlisp -*-
2 ;;;; *************************************************************************
3 ;;;; FILE IDENTIFICATION
6 ;;;; Purpose: Create SQL database for UMLisp
7 ;;;; Programmer: Kevin M. Rosenberg
8 ;;;; Date Started: Apr 2000
10 ;;;; $Id: sql-create.lisp,v 1.15 2003/05/06 05:35:48 kevin Exp $
12 ;;;; This file, part of UMLisp, is
13 ;;;; Copyright (c) 2000-2002 by Kevin M. Rosenberg, M.D.
15 ;;;; UMLisp users are granted the rights to distribute and use this software
16 ;;;; as governed by the terms of the GNU General Public License.
17 ;;;; *************************************************************************
21 (eval-when (:compile-toplevel)
22 (declaim (optimize (speed 3) (safety 1) (compilation-speed 0) (debug 3))))
24 (defun create-table-cmd (file)
25 "Return sql command to create a table"
28 (let ((sqltype (umls-col-sqltype c)))
32 (if (or (string-equal sqltype "VARCHAR")
33 (string-equal sqltype "CHAR"))
34 (format nil "~a (~a)" sqltype (umls-col-max c))
36 (format nil "CREATE TABLE ~a (~{~a~^,~})" (umls-file-table file)
37 (mapcar col-func (umls-cols-for-umls-file file)))))
39 (defun create-custom-table-cmd (tablename sql-cmd)
40 "Return SQL command to create a custom table"
41 (format nil "CREATE TABLE ~a AS ~a;" tablename sql-cmd))
43 (defun insert-col-value (col value)
44 (if (null (umls-col-parsefunc col))
46 (format nil "~A" (funcall (umls-col-parsefunc col) value))))
48 (defun insert-values-cmd (file values)
49 "Return sql insert command for a row of values"
52 (let ((q (umls-col-quotechar col)))
53 (concatenate 'string q (insert-col-value col value) q)))))
55 nil "INSERT INTO ~a (~{~a~^,~}) VALUES (~A)"
56 (umls-file-table file)
57 (umls-file-fields file)
58 (concat-separated-strings
60 (mapcar insert-func (remove-custom-cols (umls-file-colstructs file)) values)
61 (custom-col-values (custom-colstructs-for-file file) values t)))))
64 (defun custom-col-value (col values doquote)
65 (let ((custom-value (funcall (umls-col-custom-value-func col) values)))
68 (let ((q (umls-col-quotechar col)))
69 (concatenate 'string q (escape-backslashes custom-value) q))
70 (escape-backslashes custom-value))
73 (defun custom-col-values (colstructs values doquote)
74 "Returns a list of string column values for SQL inserts for custom columns"
75 (loop for col in colstructs collect (custom-col-value col values doquote)))
77 (defun remove-custom-cols (cols)
78 "Remove custom cols from a list col umls-cols"
79 (remove-if #'umls-col-custom-value-func cols))
81 (defun find-custom-cols-for-filename (filename)
82 (remove-if-not (lambda (x) (string-equal filename (car x))) +custom-cols+))
84 (defun find-custom-col (filename col)
85 (find-if (lambda (x) (and (string-equal filename (car x))
86 (string-equal col (cadr x)))) +custom-cols+))
89 (defun custom-colnames-for-filename (filename)
90 (mapcar #'cadr (find-custom-cols-for-filename filename)))
92 (defun custom-colstructs-for-file (file)
93 (remove-if-not #'umls-col-custom-value-func (umls-file-colstructs file)))
95 (defun noneng-lang-index-files ()
97 (lambda (f) (and (> (length (umls-file-fil f)) 4)
98 (string-equal (umls-file-fil f) "MRXW." :end1 5)
99 (not (string-equal (umls-file-fil f) "MRXW.ENG"))
100 (not (string-equal (umls-file-fil f) "MRXW.NONENG"))))
103 ;;; SQL Command Functions
105 (defun create-index-cmd (colname tablename length)
106 "Return sql create index command"
107 (format nil "CREATE INDEX ~a ON ~a (~a ~a)"
108 (concatenate 'string tablename "_" colname "_X")
110 (if (integerp length) (format nil "(~d)" length) "")))
112 (defun create-all-tables-cmdfile ()
113 "Return sql commands to create all tables. Not need for automated SQL import"
114 (mapcar (lambda (f) (format nil "~a~%~%" (create-table-cmd f))) *umls-files*))
116 ;; SQL Execution functions
118 (defun sql-drop-tables (conn)
119 "SQL Databases: drop all tables"
120 (dolist (file *umls-files*)
122 (sql-execute (format nil "DROP TABLE ~a" (umls-file-table file)) conn))))
124 (defun sql-create-tables (conn)
125 "SQL Databases: create all tables"
126 (dolist (file *umls-files*)
127 (sql-execute (create-table-cmd file) conn)))
129 (defun sql-create-custom-tables (conn)
130 "SQL Databases: create all custom tables"
131 (dolist (ct +custom-tables+)
132 (sql-execute (create-custom-table-cmd (car ct) (cadr ct)) conn)))
134 (defun sql-insert-values (conn file)
135 "SQL Databases: inserts all values for a file"
136 (with-umls-file (line (umls-file-fil file))
137 (sql-execute (insert-values-cmd file line) conn)))
139 (defun sql-insert-all-values (conn)
140 "SQL Databases: inserts all values for all files"
141 (dolist (file *umls-files*)
142 (sql-insert-values conn file)))
144 (defun sql-create-indexes (conn &optional (indexes +index-cols+))
145 "SQL Databases: create all indexes"
146 (dolist (idx indexes)
147 (sql-execute (create-index-cmd (car idx) (cadr idx) (caddr idx)) conn)))
149 (defun make-usrl (conn)
150 (sql-execute "drop table if exists USRL" conn)
151 (sql-execute "create table USRL (sab varchar(80), srl integer)" conn)
152 (dolist (tuple (mutex-sql-query
153 "select distinct SAB,SRL from MRSO order by SAB asc"))
154 (sql-execute (format nil "insert into USRL (sab,srl) values ('~a',~d)"
155 (car tuple) (ensure-integer (cadr tuple)))
158 (defun sql-create-special-tables (conn)
161 (defun create-umls-db-by-insert ()
162 "SQL Databases: initializes entire database via SQL insert commands"
165 (with-sql-connection (conn)
166 (sql-drop-tables conn)
167 (sql-create-tables conn)
168 (sql-insert-all-values conn)
169 (sql-create-indexes conn)
170 (sql-create-custom-tables conn)
171 (sql-create-indexes conn +custom-index-cols+)
172 (sql-create-special-tables conn)))
174 (defun create-umls-db (&optional (extension ".trans")
175 (copy-cmd #'mysql-copy-cmd))
176 "SQL Databases: initializes entire database via SQL copy commands.
177 This is much faster that using create-umls-db-insert."
180 (translate-all-files extension)
181 (with-sql-connection (conn)
182 (sql-drop-tables conn)
183 (sql-create-tables conn)
185 #'(lambda (file) (sql-execute (funcall copy-cmd file extension) conn))
187 (sql-create-indexes conn)
188 (sql-create-custom-tables conn)
189 (sql-create-indexes conn +custom-index-cols+)
190 (sql-create-special-tables conn)))
192 (defun translate-all-files (&optional (extension ".trans"))
193 "Copy translated files and return postgresql copy commands to import"
194 (make-noneng-index-file extension)
195 (dolist (f *umls-files*) (translate-umls-file f extension)))
197 (defun translate-umls-file (file extension)
198 "Translate a umls file into a format suitable for sql copy cmd"
199 (translate-files (umls-file-fil file) extension (list file)))
201 (defun make-noneng-index-file (extension)
202 "Make non-english index file"
203 (translate-files "MRXW.NONEN" extension (noneng-lang-index-files)))
205 (defun translate-files (output-basename extension input-files)
206 "Translate a umls file into a format suitable for sql copy cmd"
207 (let ((output-path (umls-pathname output-basename extension)))
208 (if (probe-file output-path)
209 (format t "File ~A already exists: skipping~%" output-path)
210 (with-open-file (ostream output-path :direction :output)
211 (dolist (input-file input-files)
212 (with-umls-file (line (umls-file-fil input-file))
213 (umls-translate input-file line ostream)
214 (princ #\newline ostream)))))))
216 (defun pg-copy-cmd (file extension)
217 "Return postgresql copy statement for a file"
219 nil "COPY ~a FROM '~a' using delimiters '|' with null as ''"
220 (umls-file-table file) (umls-pathname (umls-file-fil file) extension)))
222 (defun mysql-copy-cmd (file extension)
223 "Return mysql copy statement for a file"
226 "LOAD DATA LOCAL INFILE \"~a\" INTO TABLE ~a FIELDS TERMINATED BY \"|\""
227 (umls-pathname (umls-file-fil file) extension) (umls-file-table file)))
229 (defun col-value (col value)
231 (if (eq (umls-col-datatype col) 'sql-u)
232 (write-to-string (parse-ui value ""))
233 (escape-backslashes value))
236 (defun umls-translate (file line strm)
237 "Translate a single line for sql output"
238 (print-separated-strings
240 (mapcar #'col-value (remove-custom-cols (umls-file-colstructs file)) line)
241 (custom-col-values (custom-colstructs-for-file file) line nil)))
244 ;;; Routines for analyzing cost of fixed size storage
247 (defun umls-fixed-size-waste ()
248 "Display storage waste if using all fixed size storage"
254 (dolist (file *umls-files*)
255 (dolist (col (umls-file-colstructs file))
256 (let* ((avwaste (- (umls-col-max col) (umls-col-av col)))
257 (cwaste (* avwaste (umls-file-rws file))))
261 (incf totalunavoidable cwaste)
262 (push (list (umls-file-fil file) (umls-col-col col)
266 (incf totalavoidable cwaste)
267 (push (list (umls-file-fil file) (umls-col-col col)
270 (incf totalwaste cwaste)))))
271 (values totalwaste totalavoidable totalunavoidable
272 (nreverse avoidable) (nreverse unavoidable))))
274 (defun display-waste ()
277 (multiple-value-bind (tw ta tu al ul) (umls-fixed-size-waste)
278 (format t "Total waste: ~d~%" tw)
279 (format t "Total avoidable: ~d~%" ta)
280 (format t "Total unavoidable: ~d~%" tu)
281 (format t "Avoidable:~%")
283 (format t " (~a,~a): ~a,~a~%" (car w) (cadr w) (caddr w) (cadddr w)))
284 (format t "Unavoidable:~%")
286 (format t " (~a,~a): ~a,~a~%" (car w) (cadr w) (caddr w) (cadddr w)))
289 (defun max-umls-field ()
290 "Return length of longest field"
294 (declare (fixnum max))
295 (dolist (col *umls-cols*)
296 (when (> (umls-col-max col) max)
297 (setq max (umls-col-max col))))
300 (defun max-umls-row ()
301 "Return length of longest row"
304 (let ((rowsizes '()))
305 (dolist (file *umls-files*)
307 (fields (umls-file-colstructs file)))
308 (dolist (field fields)
309 (incf row (1+ (umls-col-max field))))
310 (push row rowsizes)))
311 (car (sort rowsizes #'>))))