X-Git-Url: http://git.kpe.io/?p=umlisp.git;a=blobdiff_plain;f=create-sql.lisp;h=4a4413a1aa2e93e7f1f4431c7fb6ed66fedbfcf2;hp=fdc444c9cf2bd8fb1279674bff3f8e5c0e414cf0;hb=57c4b059639968903aec88e65103a7263cb97535;hpb=612a2df000b3ff47d2454dbad0b901c1aa5558e7

diff --git a/create-sql.lisp b/create-sql.lisp
index fdc444c..4a4413a 100644
--- a/create-sql.lisp
+++ b/create-sql.lisp
@@ -46,7 +46,7 @@
 		" 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)
@@ -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)))
@@ -240,7 +242,7 @@
     (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)
@@ -262,7 +264,7 @@ This is much faster that using create-umls-db-insert."
       (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))
@@ -277,18 +279,53 @@ This is much faster that using create-umls-db-insert."
   (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
-                               #+(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)))))))
+    (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"
@@ -310,13 +347,13 @@ 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)))
 
 
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;