X-Git-Url: http://git.kpe.io/?a=blobdiff_plain;f=create-sql.lisp;h=9311a820bc3980804a7e94ca9faf11677c23f1e0;hb=9e472725307c8f98e1590814b0cb77c62cf5dc14;hp=83c1a22e65bc28aaca13f13967682dc21b6a0829;hpb=a418a83108a16eb76ce432a2f781a572406ec0c3;p=umlisp.git

diff --git a/create-sql.lisp b/create-sql.lisp
index 83c1a22..9311a82 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))
@@ -208,9 +208,10 @@
      (format nil "DROP INDEX ~a"
 	     (concatenate 'string tablename "_" colname "_X")))))
 
-(defun sql-create-indexes (conn &optional (indexes +index-cols+))
+(defun sql-create-indexes (conn &key (indexes +index-cols+) verbose)
   "SQL Databases: create all indexes"
   (dolist (idx indexes)
+    (when verbose (format t "UMLS Import: Creating index ~S.~%" idx))
     (ignore-errors (sql-execute (drop-index-cmd (car idx) (cadr idx)) conn))
     (sql-execute (create-index-cmd (car idx) (cadr idx) (caddr idx)) conn)))
 
@@ -229,7 +230,7 @@
   (make-usrl conn)
   (make-ustats))
 
-(defun create-umls-db-by-insert ()
+(defun create-umls-db-by-insert (&key verbose)
   "SQL Databases: initializes entire database via SQL insert commands"
   (ensure-ucols+ufiles)
   (ensure-preparse)
@@ -239,16 +240,19 @@
     (sql-insert-all-values conn)
     (sql-create-indexes conn)
     (sql-create-custom-tables conn)
-    (sql-create-indexes conn +custom-index-cols+)
+    (sql-create-indexes conn :indexes +custom-index-cols+ :verbose verbose)
     (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
-    (translate-all-files extension))
+    (when verbose (format t "UMLS Import: Converting text UMLS files to optimized format.~%"))
+    (translate-all-files :extension extension :verbose verbose))
   (let ((copy-cmd
 	 (ecase (umls-sql-type)
 	   (:mysql #'mysql-copy-cmd)
@@ -258,17 +262,24 @@ 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))
-      (sql-create-indexes conn)
+      (When verbose (format t "UMLS Import: Creating SQL indices.~%"))
+      (sql-create-indexes conn :verbose verbose)
+      (When verbose (format t "UMLS Import: Creating custom tables.~%"))
       (sql-create-custom-tables conn)
-      (sql-create-indexes conn +custom-index-cols+)
+      (When verbose (format t "UMLS Import: Creating custom indices.~%"))
+      (sql-create-indexes conn :indexes +custom-index-cols+ :verbose verbose)
+      (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"))
-  "Copy translated files and return postgresql copy commands to import"
+(defun translate-all-files (&key (extension "-trans") verbose)
+  "Translate all *umls-files* to optimized import format."
   (make-noneng-index-file extension)
   (dolist (f (remove "MRXW_NONENG.RRF" *umls-files* :test #'string= :key #'fil))
+    (when verbose (format t "UMLS Import: translating file ~S.~%" f))
     (translate-umls-file f extension)))
 
 (defun translate-umls-file (file extension)
@@ -295,13 +306,10 @@ 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)
@@ -309,6 +317,9 @@ 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)