X-Git-Url: http://git.kpe.io/?a=blobdiff_plain;f=create-sql.lisp;h=15506cca76bdb539e6cf812f0452186d8a853823;hb=eadf93ecf9164f5e0a1ec0be102624d1a383f677;hp=e212536b79bd5a1ca58445c7bf66937a6b5e7e44;hpb=aeade16272b79115d3f307906c7a3e9597137e97;p=umlisp.git diff --git a/create-sql.lisp b/create-sql.lisp index e212536..15506cc 100644 --- a/create-sql.lisp +++ b/create-sql.lisp @@ -23,15 +23,28 @@ (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" + (table file) + (mapcar col-func (ucols file)) + (if (and (eq *umls-sql-type* :mysql) + (string-equal (table file) "MRCXT")) + " MAX_ROWS=200000000" + "")))) (defun create-custom-table-cmd (tablename sql-cmd) "Return SQL command to create a custom table" @@ -93,9 +106,9 @@ (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 @@ -112,8 +125,9 @@ (format nil " (~d)" length) ""))) ((:postgresql :postgresql-socket) + ;; FIXME: incorrect syntax (if (integerp length) - (format nil "substr(~A,1,~D)" colname length) + (format nil "substr((~A)::text,1,~D)" colname length) colname)) (t colname)))) @@ -135,8 +149,36 @@ (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) + (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))) @@ -150,9 +192,21 @@ (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) + (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) @@ -161,7 +215,7 @@ (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")) + "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))) @@ -183,24 +237,19 @@ (sql-create-indexes conn +custom-index-cols+) (sql-create-special-tables conn))) -(defun create-umls-db (&optional (extension ".trans")) +(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." - (ignore-errors - (clsql:destroy-database (list *umls-sql-host* (lookup-db-name *umls-sql-db*) - *umls-sql-user* *umls-sql-passwd*) - :database-type *umls-sql-type*)) - (clsql:create-database (list *umls-sql-host* (lookup-db-name *umls-sql-db*) - *umls-sql-user* *umls-sql-passwd*) - :database-type *umls-sql-type*) (ensure-ucols+ufiles) (ensure-preparse) - (translate-all-files extension) + (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*) @@ -213,7 +262,7 @@ 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) - (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) @@ -222,17 +271,17 @@ 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 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))) + (let ((output-path (ufile-pathname 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)) + (with-umls-ufile (line input-ufile) (translate-line out-ufile line ostream) (princ #\newline ostream))))))) @@ -254,14 +303,15 @@ This is much faster that using create-umls-db-insert." "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) +(defun mysql-copy-cmd (file extension &key local-file) "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 " "") + (ufile-pathname file extension) (table file))) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;