;;;; Programmer: Kevin M. Rosenberg
;;;; Date Started: Apr 2000
;;;;
-;;;; $Id: sql-create.lisp,v 1.1 2002/10/18 03:57:39 kevin Exp $
+;;;; $Id: sql-create.lisp,v 1.19 2003/05/06 07:17:35 kevin Exp $
;;;;
;;;; This file, part of UMLisp, is
;;;; Copyright (c) 2000-2002 by Kevin M. Rosenberg, M.D.
;;;; *************************************************************************
(in-package :umlisp)
-(declaim (optimize (speed 3) (safety 1) (compilation-speed 0) (debug 3)))
+(eval-when (:compile-toplevel)
+ (declaim (optimize (speed 3) (safety 1) (compilation-speed 0) (debug 3))))
(defun create-table-cmd (file)
-"Return sql command to create a table"
+ "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))))))
+ (let ((sqltype (sqltype c)))
+ (concatenate 'string
+ (col c)
+ " "
+ (if (or (string-equal sqltype "VARCHAR")
+ (string-equal sqltype "CHAR"))
+ (format nil "~a (~a)" sqltype (cmax c))
+ sqltype))))))
+ (format nil "CREATE TABLE ~a (~{~a~^,~})" (table file)
+ (mapcar col-func (ucols-for-umls-file file)))))
(defun create-custom-table-cmd (tablename sql-cmd)
-"Return SQL command to create a custom table"
+ "Return SQL command to create a custom table"
(format nil "CREATE TABLE ~a AS ~a;" tablename sql-cmd))
+(defun insert-col-value (col value)
+ (if (null (parse-fun col))
+ value
+ (format nil "~A" (funcall (parse-fun col) value))))
+
(defun insert-values-cmd (file values)
-"Return sql insert command for a row of 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)
- ","))))
+ (let ((q (quotechar col)))
+ (concatenate 'string q (insert-col-value col value) q)))))
(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))
+ nil "INSERT INTO ~a (~{~a~^,~}) VALUES (~A)"
+ (table file)
+ (fields file)
+ (concat-separated-strings
+ ","
+ (mapcar insert-func (remove-custom-cols (ucols file)) values)
+ (custom-col-values (custom-ucols-for-file file) values t)))))
+
+
+(defun custom-col-value (col values doquote)
+ (let ((custom-value (funcall (custom-value-fun col) values)))
+ (if custom-value
+ (if doquote
+ (let ((q (quotechar col)))
+ (concatenate 'string q (escape-backslashes custom-value) q))
+ (escape-backslashes custom-value))
+ "")))
+
+(defun custom-col-values (ucols values doquote)
+ "Returns a list of string column values for SQL inserts for custom columns"
+ (loop for col in ucols collect (custom-col-value col values doquote)))
(defun remove-custom-cols (cols)
"Remove custom cols from a list col umls-cols"
- (remove-if #'umls-col-custom-value-func cols))
+ (remove-if #'custom-value-fun cols))
(defun find-custom-cols-for-filename (filename)
(remove-if-not (lambda (x) (string-equal filename (car 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 custom-ucols-for-file (file)
+ (remove-if-not #'custom-value-fun (ucols 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*))
+ (remove-if-not
+ (lambda (f) (and (> (length (fil f)) 4)
+ (string-equal (fil f) "MRXW." :end1 5)
+ (not (string-equal (fil f) "MRXW.ENG"))
+ (not (string-equal (fil f) "MRXW.NONENG"))))
+ *umls-files*))
;;; SQL Command Functions
(defun create-index-cmd (colname tablename length)
-"Return sql create index command"
+ "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)
- "")))
+ (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"
+ "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*))
+ "SQL Databases: drop all tables"
+ (dolist (file *umls-files*)
+ (ignore-errors
+ (sql-execute (format nil "DROP TABLE ~a" (table file)) conn))))
(defun sql-create-tables (conn)
-"SQL Databases: create all tables"
- (mapcar (lambda (file) (sql-execute (create-table-cmd file) conn)) *umls-files*))
+ "SQL Databases: create all tables"
+ (dolist (file *umls-files*)
+ (sql-execute (create-table-cmd file) conn)))
(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+))
+ "SQL Databases: create all custom tables"
+ (dolist (ct +custom-tables+)
+ (sql-execute (create-custom-table-cmd (car ct) (cadr ct)) conn)))
(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)))
+ "SQL Databases: inserts all values for a file"
+ (with-umls-file (line (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*))
+ "SQL Databases: inserts all values for all files"
+ (dolist (file *umls-files*)
+ (sql-insert-values conn file)))
(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))
+ "SQL Databases: create all indexes"
+ (dolist (idx indexes)
+ (sql-execute (create-index-cmd (car idx) (cadr idx) (caddr idx)) conn)))
(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"))
+ (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)))
(make-usrl conn))
(defun create-umls-db-by-insert ()
-"SQL Databases: initializes entire database via SQL insert commands"
+ "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)))
+ (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))
+ (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)
(with-sql-connection (conn)
(sql-drop-tables conn)
(sql-create-tables conn)
- (mapcar
+ (map 'nil
#'(lambda (file) (sql-execute (funcall copy-cmd file extension) conn))
*umls-files*)
(sql-create-indexes conn)
(sql-create-special-tables conn)))
(defun translate-all-files (&optional (extension ".trans"))
-"Copy translated files and return postgresql copy commands to import"
+ "Copy translated files and return postgresql copy commands to import"
(make-noneng-index-file extension)
- (mapcar (lambda (f) (translate-file f extension)) *umls-files*))
+ (dolist (f *umls-files*) (translate-umls-file f extension)))
-(defun translate-file (file extension)
+(defun translate-umls-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))))
+ (translate-files (fil file) extension (list file)))
(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))))
+ (translate-files "MRXW.NONENG" extension (noneng-lang-index-files)))
+
+(defun translate-files (output-basename extension input-files)
+ "Translate a umls file into a format suitable for sql copy cmd"
+ (let ((output-path (umls-pathname output-basename extension)))
+ (if (probe-file output-path)
+ (format t "File ~A already exists: skipping~%" output-path)
+ (with-open-file (ostream output-path :direction :output)
+ (dolist (input-file input-files)
+ (with-umls-file (line (fil input-file))
+ (umls-translate input-file line ostream)
+ (princ #\newline ostream)))))))
(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)))
+ "Return postgresql copy statement for a file"
+ (format
+ nil "COPY ~a FROM '~a' using delimiters '|' with null as ''"
+ (table file) (umls-pathname (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))))
+ "Return mysql copy statement for a file"
+ (format
+ nil
+ "LOAD DATA LOCAL INFILE \"~a\" INTO TABLE ~a FIELDS TERMINATED BY \"|\""
+ (umls-pathname (fil file) extension) (table file)))
+
+(defun col-value (col value)
+ (if (eq (datatype col) 'sql-u)
+ (let ((ui (parse-ui value "")))
+ (if (stringp ui)
+ ui
+ (write-to-string ui)))
+ (escape-backslashes value)))
+
+(defun umls-translate (file line strm)
+ "Translate a single line for sql output"
+ (print-separated-strings
+ strm "|"
+ (mapcar #'col-value (remove-custom-cols (ucols file)) line)
+ (custom-col-values (custom-ucols-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)
(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)
+ (dolist (col (ucols file))
+ (let* ((avwaste (- (cmax col) (av col)))
+ (cwaste (* avwaste (rws file))))
+ (when (plusp 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
+ (push (list (fil file) (col col)
+ avwaste cwaste)
+ unavoidable))
+ (progn
(incf totalavoidable cwaste)
- (setq avoidable (append avoidable (list (list (umls-file-fil file) (umls-col-col col) avwaste cwaste))))))
+ (push (list (fil file) (col col)
+ avwaste cwaste)
+ avoidable)))
(incf totalwaste cwaste)))))
- (values totalwaste totalavoidable totalunavoidable avoidable unavoidable)))
+ (values totalwaste totalavoidable totalunavoidable
+ (nreverse avoidable) (nreverse unavoidable))))
(defun display-waste ()
(unless *umls-files*
(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"
(let ((max 0))
(declare (fixnum max))
(dolist (col *umls-cols*)
- (when (> (umls-col-max col) max)
- (setq max (umls-col-max col))))
+ (when (> (cmax col) max)
+ (setq max (cmax 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 #'>))))))
+ (unless *umls-files*
+ (init-umls))
+ (let ((rowsizes '()))
+ (dolist (file *umls-files*)
+ (let ((row 0)
+ (fields (ucols file)))
+ (dolist (field fields)
+ (incf row (1+ (cmax field))))
+ (push row rowsizes)))
+ (car (sort rowsizes #'>))))
projects / umlisp.git / blobdiff