;;;; $Id$
;;;;
;;;; This file, part of UMLisp, is
-;;;; Copyright (c) 2000-2004 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.
" MAX_ROWS=200000000"
"")
(if (eq *umls-sql-type* :mysql)
- " TYPE=MYISAM DEFAULT CHARACTER latin1"
+ " TYPE=MYISAM CHARACTER SET utf8"
""))))
(defun create-custom-table-cmd (tablename sql-cmd)
(: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)))
(sql-create-indexes conn +custom-index-cols+)
(sql-create-special-tables conn)))
-(defun create-umls-db (&key (extension ".trans") (skip-translation nil))
+(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)
(sql-create-indexes conn +custom-index-cols+)
(sql-create-special-tables conn))))
-(defun translate-all-files (&optional (extension ".trans"))
+(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.RRF" *umls-files* :test #'string= :key #'fil))
(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 (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-ufile (line input-ufile)
- (translate-line out-ufile line ostream)
- (princ #\newline ostream)))))))
+ (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"
nil "COPY ~a FROM '~a' using delimiters '|' with null as ''"
(table file) (ufile-pathname file extension)))
-(defun mysql-copy-cmd (file extension &key local-file)
+(defun mysql-copy-cmd (file extension &key (local-file t))
"Return mysql copy statement for a file"
(format
nil
"LOAD DATA ~AINFILE \"~a\" INTO TABLE ~a FIELDS TERMINATED BY \"|\""
(if local-file "LOCAL " "")
- (ufile-pathname file extension) (table file)))
+ (namestring (ufile-pathname file extension)) (table file)))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;