X-Git-Url: http://git.kpe.io/?a=blobdiff_plain;f=create-sql.lisp;h=1440fd32911e755fc16da94c974021cf8e644c2c;hb=fe0f8af77c3bc5bb58c1cc0ab3701688fa8089a8;hp=f2762861637a5422dc055db70f43fecc5bbaf844;hpb=6ebaa3627ee89b8f7f429ebcd01c4dc8d9892e0b;p=umlisp.git

diff --git a/create-sql.lisp b/create-sql.lisp
index f276286..1440fd3 100644
--- a/create-sql.lisp
+++ b/create-sql.lisp
@@ -128,7 +128,7 @@
 			      (format nil " (~d)" length)
 			      "")))
 	    ((:postgresql :postgresql-socket)
-	     ;; FIXME: incorrect syntax
+	     ;; FIXME: incorrect syntax for postgresql?
 	     (if (integerp length)
 		 (format nil "substr((~A)::text,1,~D)" colname length)
 		 colname))
@@ -169,6 +169,8 @@
 	     (: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)))
@@ -240,12 +242,15 @@
     (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) (verbose nil))
   "SQL Databases: initializes entire database via SQL copy commands.
 This is much faster that using create-umls-db-insert."
+  (when verbose (format t "UMLS Import: Starting.~%"))
   (ensure-ucols+ufiles)
+  (when verbose (format t "UMLS Import: Preparsing files.~%"))
   (ensure-preparse)
   (unless skip-translation
+    (when verbose (format t "UMLS Import: Converting text UMLS files to optimized format.~%"))
     (translate-all-files extension))
   (let ((copy-cmd
 	 (ecase (umls-sql-type)
@@ -256,13 +261,20 @@ This is much faster that using create-umls-db-insert."
       (sql-drop-tables conn)
       (sql-create-tables conn)
       (dolist (file *umls-files*)
+        (when verbose (format t "UMLS Import: Importing file ~A.~%" file))
 	(sql-execute (funcall copy-cmd file extension) conn))
+      (When verbose (format t "UMLS Import: Creating SQL indices.~%"))
       (sql-create-indexes conn)
+      (When verbose (format t "UMLS Import: Creating custom tables.~%"))
       (sql-create-custom-tables conn)
+      (When verbose (format t "UMLS Import: Creating custom indices.~%"))
       (sql-create-indexes conn +custom-index-cols+)
-      (sql-create-special-tables conn))))
+      (When verbose (format t "UMLS Import: Creating special tables.~%"))
+      (sql-create-special-tables conn)))
+  (When verbose (format t "UMLS Import: Completed.~%"))
+  t)
 
-(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))
@@ -292,13 +304,10 @@ This is much faster that using create-umls-db-insert."
             (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)))))
+                           (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)
@@ -306,6 +315,9 @@ This is much faster that using create-umls-db-insert."
         ((eql input-lines translated-lines)
           (format t "Translated file ~A already exists: skipping...~%" output-path)
           t)
+        ((eql input-lines 0)
+          (warn "The number of input lines is 0 for output file ~A." output-path)
+          nil)
         ((> 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)
@@ -345,13 +357,13 @@ This is much faster that using create-umls-db-insert."
    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)))
 
 
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;