X-Git-Url: http://git.kpe.io/?a=blobdiff_plain;f=sql-create.lisp;h=c9821fd742e7f77a108b89fcdcd5b4d9bc4cf813;hb=cc146d0b1c7d6d7a050713754f1ea9a8d71a46b5;hp=4d10d3da4c7216fa5d7b3678369684805fe5bec5;hpb=783a79a99f38169db68f1ded95aa507b0c964daa;p=umlisp.git diff --git a/sql-create.lisp b/sql-create.lisp index 4d10d3d..c9821fd 100644 --- a/sql-create.lisp +++ b/sql-create.lisp @@ -7,7 +7,7 @@ ;;;; Programmer: Kevin M. Rosenberg ;;;; Date Started: Apr 2000 ;;;; -;;;; $Id: sql-create.lisp,v 1.5 2003/05/04 08:52:13 kevin Exp $ +;;;; $Id: sql-create.lisp,v 1.9 2003/05/06 02:14:59 kevin Exp $ ;;;; ;;;; This file, part of UMLisp, is ;;;; Copyright (c) 2000-2002 by Kevin M. Rosenberg, M.D. @@ -17,25 +17,9 @@ ;;;; ************************************************************************* (in-package :umlisp) -(declaim (optimize (speed 3) (safety 1) (compilation-speed 0) (debug 3))) - - -(defun create-table-cmd-old (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)))))) +(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" @@ -50,38 +34,12 @@ (format nil "~a (~a)" sqltype (umls-col-max c)) sqltype)))))) (format nil "CREATE TABLE ~a (~{~a~^,~})" (umls-file-table file) - (mapcar col-func (umls-cols-for-umls-file file)))))) + (mapcar 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-old (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 insert-col-value (col value) (if (null (umls-col-parsefunc col)) value @@ -94,27 +52,16 @@ (let ((q (umls-col-quotechar col))) (concatenate 'string q (insert-col-value col value) q))))) (format - nil "INSERT INTO ~a (~{~a~^,~}) VALUES (~{~a~^,~})" + nil "INSERT INTO ~a (~{~a~^,~}) VALUES (~A)" (umls-file-table file) (umls-file-fields file) - (append + (concat-separated-strings + "," (mapcar insert-func (remove-custom-cols (umls-file-colstructs file)) values) (custom-col-values (custom-colstructs-for-file file) values t))))) -(defun custom-col-values-old (colstructs values delim doquote) - "Returns a 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 col-value (col doquote values) + +(defun custom-col-value (col doquote values) (let ((custom-value (funcall (umls-col-custom-value-func col) values))) (if doquote (let ((q (umls-col-quotechar col))) @@ -123,7 +70,7 @@ (defun custom-col-values (colstructs values doquote) "Returns a list of string column values for SQL inserts for custom columns" - (loop for col in colstructs collect (col-value col doquote values))) + (loop for col in colstructs collect (custom-col-value col doquote values))) (defun remove-custom-cols (cols) @@ -145,46 +92,43 @@ (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*)) + (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) - ""))) + (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*)) + (dolist (file *umls-files*) + (ignore-errors + (sql-execute (format nil "DROP TABLE ~a" (umls-file-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*)) + (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+)) + (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" @@ -193,19 +137,19 @@ (defun sql-insert-all-values (conn) "SQL Databases: inserts all values for all files" - (mapcar (lambda (file) (sql-insert-values conn file)) *umls-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)) + (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))) @@ -236,7 +180,7 @@ This is much faster that using create-umls-db-insert." (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) @@ -247,78 +191,53 @@ This is much faster that using create-umls-db-insert." (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*)) + (dolist (f *umls-files*) (translate-umls-file f extension))) -(defun translate-file (file extension) +(defun translate-file (filename extension files) "Translate a umls file into a format suitable for sql copy cmd" - (let ((path (umls-pathname (umls-file-fil file) extension))) + (let ((path (umls-pathname filename extension)) + (input-files (mklist files))) (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)))) + (format t "File ~A already exists: skipping~%" path) + (dolist (input-file input-files) + (with-open-file (ostream :direction :output) + (with-umls-file (line (umls-file-fil input-file)) + (umls-translate file line ostream) + (princ #\newline ostream))))))) + +(defun translate-umls-file (file extension) + "Translate a umls file into a format suitable for sql copy cmd" + (translate-file (umls-file-fil file) extension (umls-file-fil 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-file (find-umls-file "MRXW.NONENG" extension + (noneng-lang-index-files)))) (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))) + (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))) + (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-old (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)))) - -(defun umls-translate (file line) +(defun col-value (col value) + (if (eq (umls-col-datatype col) 'sql-u) + (write-to-string (parse-ui value "")) + (escape-backslashes value))) + +(defun umls-translate (file line strm) "Translate a single line for sql output" - (format nil "~{~A~^|~}" - (append - (mapcar - (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)))) + (print-separated-strings + strm "|" + (mapcar #'col-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 @@ -335,16 +254,21 @@ This is much faster that using create-umls-db-insert." (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) + (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))))) + (push (list (umls-file-fil file) (umls-col-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 (umls-file-fil file) (umls-col-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* @@ -374,16 +298,13 @@ This is much faster that using create-umls-db-insert." (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 (umls-file-colstructs file))) + (dolist (field fields) + (incf row (1+ (umls-col-max field)))) + (push row rowsizes))) + (car (sort rowsizes #'>))))