;;;; -*- Mode: Lisp; Syntax: ANSI-Common-Lisp; Base: 10; Package: umlisp -*- ;;;; ************************************************************************* ;;;; FILE IDENTIFICATION ;;;; ;;;; Name: sql-create ;;;; Purpose: Create SQL database for UMLisp ;;;; Programmer: Kevin M. Rosenberg ;;;; Date Started: Apr 2000 ;;;; ;;;; $Id: sql-create.lisp,v 1.4 2002/11/07 23:19:02 kevin Exp $ ;;;; ;;;; This file, part of UMLisp, is ;;;; Copyright (c) 2000-2002 by Kevin M. Rosenberg, M.D. ;;;; ;;;; UMLisp users are granted the rights to distribute and use this software ;;;; as governed by the terms of the GNU General Public License. ;;;; ************************************************************************* (in-package :umlisp) (declaim (optimize (speed 3) (safety 1) (compilation-speed 0) (debug 3))) (defun create-table-cmd (file) "Return sql command to create a table" (let ((col-func (lambda (c) (let ((sqltype (umls-col-sqltype c))) (concatenate 'string (umls-col-col c) " " (if (or (string-equal sqltype "VARCHAR") (string-equal sqltype "CHAR")) (format nil "~a (~a)" sqltype (umls-col-max c)) sqltype) ","))))) (format nil "CREATE TABLE ~a (~a)" (umls-file-table file) (string-trim-last-character (mapcar-append-string col-func (umls-cols-for-umls-file file)))))) (defun create-custom-table-cmd (tablename sql-cmd) "Return SQL command to create a custom table" (format nil "CREATE TABLE ~a AS ~a;" tablename sql-cmd)) (defun insert-values-cmd (file values) "Return sql insert command for a row of values" (let ((insert-func (lambda (col value) (concatenate 'string (umls-col-quotechar col) (if (null (umls-col-parsefunc col)) value (format nil "~A" (funcall (umls-col-parsefunc col) value))) (umls-col-quotechar col) ",")))) (format nil "INSERT INTO ~a (~a) VALUES (~a)" (umls-file-table file) (string-trim-last-character (mapcar-append-string (lambda (c) (concatenate 'string c ",")) (umls-file-fields file))) (string-trim-last-character (concatenate 'string (mapcar2-append-string insert-func (remove-custom-cols (umls-file-colstructs file)) values) (custom-col-values (custom-colstructs-for-file file) values "," t))) ))) (defun custom-col-values (colstructs values delim doquote) "Returns string of column values for SQL inserts for custom columns" (let ((result "")) (dolist (col colstructs) (let* ((func (umls-col-custom-value-func col)) (custom-value (funcall func values))) (string-append result (if doquote (umls-col-quotechar col)) (escape-backslashes custom-value) (if doquote (umls-col-quotechar col)) delim))) result)) (defun remove-custom-cols (cols) "Remove custom cols from a list col umls-cols" (remove-if #'umls-col-custom-value-func cols)) (defun find-custom-cols-for-filename (filename) (remove-if-not (lambda (x) (string-equal filename (car x))) +custom-cols+)) (defun find-custom-col (filename col) (find-if (lambda (x) (and (string-equal filename (car x)) (string-equal col (cadr x)))) +custom-cols+)) (defun custom-colnames-for-filename (filename) (mapcar #'cadr (find-custom-cols-for-filename filename))) (defun custom-colstructs-for-file (file) (remove-if-not #'umls-col-custom-value-func (umls-file-colstructs file))) (defun noneng-lang-index-files () (remove-if-not (lambda (f) (and (> (length (umls-file-fil f)) 4) (string-equal (umls-file-fil f) "MRXW." :end1 5) (not (string-equal (umls-file-fil f) "MRXW.ENG")) (not (string-equal (umls-file-fil f) "MRXW.NONENG")))) *umls-files*)) ;;; SQL Command Functions (defun create-index-cmd (colname tablename length) "Return sql create index command" (format nil "CREATE INDEX ~a ON ~a (~a ~a)" (concatenate 'string tablename "_" colname "_X") tablename colname (if (integerp length) (format nil "(~d)" length) ""))) (defun create-all-tables-cmdfile () "Return sql commands to create all tables. Not need for automated SQL import" (mapcar (lambda (f) (format nil "~a~%~%" (create-table-cmd f))) *umls-files*)) ;; SQL Execution functions (defun sql-drop-tables (conn) "SQL Databases: drop all tables" (mapcar (lambda (file) (ignore-errors (sql-execute (format nil "DROP TABLE ~a" (umls-file-table file)) conn))) *umls-files*)) (defun sql-create-tables (conn) "SQL Databases: create all tables" (mapcar (lambda (file) (sql-execute (create-table-cmd file) conn)) *umls-files*)) (defun sql-create-custom-tables (conn) "SQL Databases: create all custom tables" (mapcar (lambda (ct) (sql-execute (create-custom-table-cmd (car ct) (cadr ct)) conn)) +custom-tables+)) (defun sql-insert-values (conn file) "SQL Databases: inserts all values for a file" (with-umls-file (line (umls-file-fil file)) (sql-execute (insert-values-cmd file line) conn))) (defun sql-insert-all-values (conn) "SQL Databases: inserts all values for all files" (mapcar (lambda (file) (sql-insert-values conn file)) *umls-files*)) (defun sql-create-indexes (conn &optional (indexes +index-cols+)) "SQL Databases: create all indexes" (mapcar (lambda (idx) (sql-execute (create-index-cmd (car idx) (cadr idx) (caddr idx)) conn)) indexes)) (defun make-usrl (conn) (sql-execute "drop table if exists USRL" conn) (sql-execute "create table USRL (sab varchar(80), srl integer)" conn) (dolist (tuple (mutex-sql-query "select distinct SAB,SRL from MRSO order by SAB asc")) (sql-execute (format nil "insert into USRL (sab,srl) values ('~a',~d)" (car tuple) (ensure-integer (cadr tuple))) conn))) (defun sql-create-special-tables (conn) (make-usrl conn)) (defun create-umls-db-by-insert () "SQL Databases: initializes entire database via SQL insert commands" (init-umls) (init-hash-table) (with-sql-connection (conn) (sql-drop-tables conn) (sql-create-tables conn) (sql-insert-all-values conn) (sql-create-indexes conn) (sql-create-custom-tables conn) (sql-create-indexes conn +custom-index-cols+) (sql-create-special-tables conn))) (defun create-umls-db (&optional (extension ".trans") (copy-cmd #'mysql-copy-cmd)) "SQL Databases: initializes entire database via SQL copy commands. This is much faster that using create-umls-db-insert." (init-umls) (init-hash-table) (translate-all-files extension) (with-sql-connection (conn) (sql-drop-tables conn) (sql-create-tables conn) (mapcar #'(lambda (file) (sql-execute (funcall copy-cmd file extension) conn)) *umls-files*) (sql-create-indexes conn) (sql-create-custom-tables conn) (sql-create-indexes conn +custom-index-cols+) (sql-create-special-tables conn))) (defun translate-all-files (&optional (extension ".trans")) "Copy translated files and return postgresql copy commands to import" (make-noneng-index-file extension) (mapcar (lambda (f) (translate-file f extension)) *umls-files*)) (defun translate-file (file extension) "Translate a umls file into a format suitable for sql copy cmd" (let ((path (umls-pathname (umls-file-fil file) extension))) (if (probe-file path) (progn (format t "File ~A already exists: skipping~%" path) nil) (with-open-file (ostream path :direction :output) (with-umls-file (line (umls-file-fil file)) (princ (umls-translate file line) ostream) (princ #\newline ostream)) t)))) (defun make-noneng-index-file (extension) "Make non-english index file" (let* ((outfile (find-umls-file "MRXW.NONENG")) (path (umls-pathname (umls-file-fil outfile) extension))) (if (probe-file path) (progn (format t "File ~A already exists: skipping~%" path) nil) (progn (with-open-file (ostream path :direction :output) (dolist (inputfile (noneng-lang-index-files)) (with-umls-file (line (umls-file-fil inputfile)) (princ (umls-translate outfile line) ostream) ;; use outfile for custom cols (princ #\newline ostream)))) t)))) (defun pg-copy-cmd (file extension) "Return postgresql copy statement for a file" (format nil "COPY ~a FROM '~a' using delimiters '|' with null as ''" (umls-file-table file) (umls-pathname (umls-file-fil file) extension))) (defun mysql-copy-cmd (file extension) "Return mysql copy statement for a file" (format nil "LOAD DATA LOCAL INFILE \"~a\" INTO TABLE ~a FIELDS TERMINATED BY \"|\"" (umls-pathname (umls-file-fil file) extension) (umls-file-table file))) (defun umls-translate (file line) "Translate a single line for sql output" (string-trim-last-character (concatenate 'string (mapcar2-append-string (lambda (col value) (concatenate 'string (if (eq (umls-col-datatype col) 'sql-u) (format nil "~d" (parse-ui value "")) (escape-backslashes value)) "|")) (remove-custom-cols (umls-file-colstructs file)) line) (custom-col-values (custom-colstructs-for-file file) line "|" nil)))) ;;; Routines for analyzing cost of fixed size storage (defun umls-fixed-size-waste () "Display storage waste if using all fixed size storage" (let ((totalwaste 0) (totalunavoidable 0) (totalavoidable 0) (unavoidable '()) (avoidable '())) (dolist (file *umls-files*) (dolist (col (umls-file-colstructs file)) (let* ((avwaste (- (umls-col-max col) (umls-col-av col))) (cwaste (* avwaste (umls-file-rws file)))) (unless (zerop cwaste) (if (<= avwaste 6) (progn (incf totalunavoidable cwaste) (setq unavoidable (append unavoidable (list (list (umls-file-fil file) (umls-col-col col) avwaste cwaste))))) (progn (incf totalavoidable cwaste) (setq avoidable (append avoidable (list (list (umls-file-fil file) (umls-col-col col) avwaste cwaste)))))) (incf totalwaste cwaste))))) (values totalwaste totalavoidable totalunavoidable avoidable unavoidable))) (defun display-waste () (unless *umls-files* (init-umls)) (multiple-value-bind (tw ta tu al ul) (umls-fixed-size-waste) (format t "Total waste: ~d~%" tw) (format t "Total avoidable: ~d~%" ta) (format t "Total unavoidable: ~d~%" tu) (format t "Avoidable:~%") (dolist (w al) (format t " (~a,~a): ~a,~a~%" (car w) (cadr w) (caddr w) (cadddr w))) (format t "Unavoidable:~%") (dolist (w ul) (format t " (~a,~a): ~a,~a~%" (car w) (cadr w) (caddr w) (cadddr w))) )) (defun max-umls-field () "Return length of longest field" (unless *umls-files* (init-umls)) (let ((max 0)) (declare (fixnum max)) (dolist (col *umls-cols*) (when (> (umls-col-max col) max) (setq max (umls-col-max col)))) max)) (defun max-umls-row () "Return length of longest row" (if t 6000 ;;; hack to use on systems without MRCOLS/MRFILES -- ok for UMLS2001 (progn (unless *umls-files* (init-umls)) (let ((rowsizes '())) (dolist (file *umls-files*) (let ((row 0) (fields (umls-file-colstructs file))) (dolist (field fields) (incf row (1+ (umls-col-max field)))) (push row rowsizes))) (car (sort rowsizes #'>))))))