X-Git-Url: http://git.kpe.io/?p=umlisp.git;a=blobdiff_plain;f=create-sql.lisp;h=4a4413a1aa2e93e7f1f4431c7fb6ed66fedbfcf2;hp=f842eb21af84ca166f342766430180db7e7c01fc;hb=57c4b059639968903aec88e65103a7263cb97535;hpb=3adbc09e353b3c4dc09c8c9da6cddee8075eaa14 diff --git a/create-sql.lisp b/create-sql.lisp index f842eb2..4a4413a 100644 --- a/create-sql.lisp +++ b/create-sql.lisp @@ -2,15 +2,15 @@ ;;;; ************************************************************************* ;;;; FILE IDENTIFICATION ;;;; -;;;; Name: sql-create -;;;; Purpose: Create SQL database for UMLisp -;;;; Author: Kevin M. Rosenberg -;;;; Date Started: Apr 2000 +;;;; Name: sql-create +;;;; Purpose: Create SQL database for UMLisp +;;;; Author: Kevin M. Rosenberg +;;;; Created: Apr 2000 ;;;; ;;;; $Id$ ;;;; ;;;; This file, part of UMLisp, is -;;;; Copyright (c) 2000-2003 by Kevin M. Rosenberg, M.D. +;;;; Copyright (c) 2000-2006 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. @@ -20,30 +20,46 @@ (defun create-table-cmd (file) "Return sql command to create a table" - (let ((col-func - (lambda (c) + (let ((col-func + (lambda (c) (let ((sqltype (sqltype c))) + (case *umls-sql-type* + (:oracle + (cond + ((string-equal sqltype "VARCHAR") + (setq sqltype "VARCHAR2")) + ((string-equal sqltype "BIGINT") + (setq sqltype "VARCHAR2(20)"))))) + (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 file))))) + (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~^,~})~A~A" + (table file) + (mapcar col-func (ucols file)) + (if (and (eq *umls-sql-type* :mysql) + (string-equal (table file) "MRCXT")) + " MAX_ROWS=200000000" + "") + (if (eq *umls-sql-type* :mysql) + " TYPE=MYISAM CHARACTER SET utf8" + "")))) (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-col-value (col value) - (if (null (parse-fun col)) + (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 (quote-str col) @@ -54,7 +70,7 @@ (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))))) @@ -93,19 +109,31 @@ (defun noneng-lang-index-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")))) + (string-equal (fil f) "MRXW_" :end1 5) + (not (string-equal (fil f) "MRXW_ENG.RRF")) + (not (string-equal (fil f) "MRXW_NONENG.RRF")))) *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)" + (format nil "CREATE INDEX ~a ON ~a (~a)" (concatenate 'string tablename "_" colname "_X") - tablename colname - (if (integerp length) (format nil "(~d)" length) ""))) + tablename + (case *umls-sql-type* + (:mysql + (concatenate 'string colname + (if (integerp length) + (format nil " (~d)" length) + ""))) + ((:postgresql :postgresql-socket) + ;; FIXME: incorrect syntax + (if (integerp length) + (format nil "substr((~A)::text,1,~D)" colname length) + colname)) + (t + colname)))) (defun create-all-tables-cmdfile () "Return sql commands to create all tables. Not need for automated SQL import" @@ -116,40 +144,84 @@ (defun sql-drop-tables (conn) "SQL Databases: drop all tables" (dolist (file *umls-files*) - (ignore-errors + (ignore-errors (sql-execute (format nil "DROP TABLE ~a" (table file)) conn)))) (defun sql-create-tables (conn) - "SQL Databases: create all tables" + "SQL Databases: create all tables" (dolist (file *umls-files*) (sql-execute (create-table-cmd file) conn))) +#+ignore +(defun sql-create-kcon-table (conn) + "Create concept table, one row per concept." + (ignore-errors (execute-command "DROP TABLE KCON" :database conn)) + (execute-command + (format nil "CREATE TABLE KCON (CUI INTEGER, STR ~A, LRL ~A)" + (case *umls-sql-type* + (:oracle + (format nil "VARCHAR2(~D)" + (slot-value (find-ucol "STR" "MRCONSO.RRF") 'max))) + (t "TEXT")) + (case *umls-sql-type* + (:mysql "TINYINT") + ((:postgresql :postgresql-socket) "INT2") + (:oracle "NUMBER(2,0)") + (t "INTEGER"))) + :database conn) + ;; KCON deprecated by KPFENG field in MRCONSO + #+nil + (dolist (tuple (query "select distinct cui from MRCONSO order by cui" + :database conn)) + (let ((cui (car tuple))) + (execute-command + (format nil "INSERT into KCON VALUES (~D,'~A',~D)" + cui + (add-sql-quotes (pfstr-hash cui) ) + (cui-lrl cui)) + :database conn)))) + (defun sql-create-custom-tables (conn) "SQL Databases: create all custom tables" + ;;(sql-create-kcon-table conn) (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" + "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" + "SQL Databases: inserts all values for all files" (dolist (file *umls-files*) (sql-insert-values conn file))) +(defun drop-index-cmd (colname tablename) + "Return sql create index command" + (case *umls-sql-type* + (:mysql + (format nil "DROP INDEX ~a ON ~a" + (concatenate 'string tablename "_" colname "_X") + tablename)) + (t + (format nil "DROP INDEX ~a" + (concatenate 'string tablename "_" colname "_X"))))) + (defun sql-create-indexes (conn &optional (indexes +index-cols+)) "SQL Databases: create all indexes" (dolist (idx indexes) - (sql-execute (create-index-cmd (car idx) (cadr idx) (caddr idx)) conn))) + (ignore-errors (sql-execute (drop-index-cmd (car idx) (cadr idx)) conn)) + (sql-execute (create-index-cmd (car idx) (cadr idx) (caddr idx)) conn))) (defun make-usrl (conn) - (sql-execute "drop table if exists USRL" conn) + (if (eql :mysql *umls-sql-type*) + (sql-execute "drop table if exists USRL" conn) + (ignore-errors (sql-execute "drop table 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)" + "select distinct SAB,SRL from MRCONSO order by SAB asc")) + (sql-execute (format nil "insert into USRL (sab,srl) values ('~a',~d)" (car tuple) (ensure-integer (cadr tuple))) conn))) @@ -170,27 +242,32 @@ (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. +(defun create-umls-db (&key (extension "-trans") (skip-translation nil)) + "SQL Databases: initializes entire database via SQL copy commands. This is much faster that using create-umls-db-insert." (ensure-ucols+ufiles) (ensure-preparse) - (translate-all-files extension) - (with-sql-connection (conn) - (sql-drop-tables conn) - (sql-create-tables conn) - (dolist (file *umls-files*) - (sql-execute (funcall copy-cmd file extension) conn)) - (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")) + (unless skip-translation + (translate-all-files extension)) + (let ((copy-cmd + (ecase (umls-sql-type) + (:mysql #'mysql-copy-cmd) + (:postgresql #'pg-copy-cmd)))) + (with-sql-connection (conn) + (clsql:truncate-database :database conn) + (sql-drop-tables conn) + (sql-create-tables conn) + (dolist (file *umls-files*) + (sql-execute (funcall copy-cmd file extension) conn)) + (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) - (dolist (f (remove "MRXW.NONENG" *umls-files* :test #'string= :key #'fil)) + (dolist (f (remove "MRXW_NONENG.RRF" *umls-files* :test #'string= :key #'fil)) (translate-umls-file f extension))) (defun translate-umls-file (file extension) @@ -199,19 +276,56 @@ This is much faster that using create-umls-db-insert." (defun make-noneng-index-file (extension) "Make non-english index file" - (translate-files (find-ufile "MRXW.NONENG") + (translate-files (find-ufile "MRXW_NONENG.RRF") extension (noneng-lang-index-files))) +(defun verify-translation-file (output-path input-ufiles) + "Returns t if translation file exists and is correct size. Warns and deletes incomplete translation file." + (when (probe-file output-path) + (let ((translated-lines 0) + (input-lines 0) + (eof (cons nil nil))) + (catch 'done-counting + (with-open-file (ts output-path :direction :input + #+(and clisp unicode) :external-format + #+(and clisp unicode) charset:utf-8) + (do () + ((eq (read-line ts nil eof) eof)) + (incf translated-lines))) + (dolist (input-ufile input-ufiles) + (with-umls-ufile (line input-ufile) + (incf input-lines) + (when (> input-lines translated-lines) + (throw 'done-counting 'incomplete))))) + (cond + ((eql input-lines 0) + (error "Input lines is 0") + nil) + ((< input-lines translated-lines) + (format t "Translated file ~A incomplete, deleting...~%" output-path) + (delete-file output-path) + nil) + ((eql input-lines translated-lines) + (format t "Translated file ~A already exists: skipping...~%" output-path) + t) + ((> translated-lines input-lines) + (error "Shouldn't happen. Translated lines of ~A is ~D, greater than input lines ~D" + output-path translated-lines input-lines) + (delete-file output-path) + nil))))) + (defun translate-files (out-ufile extension input-ufiles) "Translate a umls file into a format suitable for sql copy cmd" - (let ((output-path (umls-pathname (fil out-ufile) 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-ufile input-ufiles) - (with-umls-file (line (fil input-ufile)) - (translate-line out-ufile line ostream) - (princ #\newline ostream))))))) + (let ((output-path (ufile-pathname out-ufile extension))) + (when (verify-translation-file output-path input-ufiles) + (return-from translate-files output-path)) + (with-open-file (ostream output-path :direction :output + #+(and clisp unicode) :external-format + #+(and clisp unicode) charset:utf-8) + (dolist (input-ufile input-ufiles) + (with-umls-ufile (line input-ufile) + (translate-line out-ufile line ostream) + (princ #\newline ostream)))))) (defun translate-line (file line strm) "Translate a single line for sql output" @@ -223,24 +337,25 @@ This is much faster that using create-umls-db-insert." (write-to-string ui))) (escape-backslashes value)))) (print-separated-strings - strm "|" + strm "|" (mapcar #'col-value (remove-custom-cols (ucols file)) line) (custom-col-values (custom-ucols-for-file file) line nil)))) (defun pg-copy-cmd (file extension) - "Return postgresql copy statement for a file" + "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))) + (table file) (ufile-pathname file extension))) -(defun mysql-copy-cmd (file extension) - "Return mysql copy statement for a file" +(defun mysql-copy-cmd (file extension &key (local-file t)) + "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))) + "LOAD DATA ~AINFILE \"~a\" INTO TABLE ~a FIELDS TERMINATED BY \"|\"" + (if local-file "LOCAL " "") + (namestring (ufile-pathname file extension)) (table file))) + - ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;; ;;; Routines for analyzing cost of fixed size storage