r11412: warn on empty files -- occurs when subsetting UMLS

[umlisp.git] / parse-rrf.lisp

diff --git a/parse-rrf.lisp b/parse-rrf.lisp
index 4c617a02655c539da1e370a52de14c7faec1dcdc..4076dc7c47635615d12b3813d2beb2030922f814 100644 (file)
--- a/parse-rrf.lisp
+++ b/parse-rrf.lisp
@@ -2,7 +2,7 @@
 ;;;; *************************************************************************
 ;;;; FILE IDENTIFICATION
 ;;;;
 ;;;; *************************************************************************
 ;;;; FILE IDENTIFICATION
 ;;;;

-;;;; Name:     parse-2002.lisp
+;;;; Name:     parse-rrf.lisp

 ;;;; Purpose:  Parsing and SQL insertion routines for UMLisp which may
 ;;;;           change from year to year
 ;;;; Author:   Kevin M. Rosenberg
 ;;;; Purpose:  Parsing and SQL insertion routines for UMLisp which may
 ;;;;           change from year to year
 ;;;; Author:   Kevin M. Rosenberg


@@ -40,40 +40,47 @@
           (clrhash cuisui-lrl-hash)
           (clrhash sab-srl-hash))
       (setf
           (clrhash cuisui-lrl-hash)
           (clrhash sab-srl-hash))
       (setf

-         pfstr-hash (make-hash-table :size 800000)
-         cui-lrl-hash (make-hash-table :size 800000)
-         lui-lrl-hash (make-hash-table :size 1500000)
-         sui-lrl-hash (make-hash-table :size 1500000)
-         cuisui-lrl-hash (make-hash-table :size 1800000)
+         pfstr-hash (make-hash-table :size 1300000)
+         cui-lrl-hash (make-hash-table :size 1300000)
+         lui-lrl-hash (make-hash-table :size 4600000)
+         sui-lrl-hash (make-hash-table :size 5100000)
+         cuisui-lrl-hash (make-hash-table :size 2000000)

          sab-srl-hash (make-hash-table :size 100 :test 'equal))))
 
   (defun ensure-preparse (&optional (force-read nil))
     (when (and *preparse-hash-init?* (not force-read))
       (return-from ensure-preparse 'already-done))
     (make-preparse-hash-table)
          sab-srl-hash (make-hash-table :size 100 :test 'equal))))
 
   (defun ensure-preparse (&optional (force-read nil))
     (when (and *preparse-hash-init?* (not force-read))
       (return-from ensure-preparse 'already-done))
     (make-preparse-hash-table)

-    (with-umls-file (line "MRCONSO.RRF")
-      (let ((cui (parse-ui (nth 0 line)))
-            (lui (parse-ui (nth 3 line)))
-            (sui (parse-ui (nth 5 line)))
-            (sab (nth 11 line))
-            (srl (parse-integer (nth 15 line))))
-        (unless (gethash cui pfstr-hash)  ;; if haven't stored pfstr for cui
-            (if (and (string-equal (nth 1 line) "ENG") ; LAT
-                     (string-equal (nth 2 line) "P") ; ts
-                     (string-equal (nth 4 line) "PF")) ; stt
+    (let ((counter 0))
+      (declare (fixnum counter))
+      (with-umls-file (line "MRCONSO.RRF")
+        (let ((cui (parse-ui (nth 0 line)))
+              (lui (parse-ui (nth 3 line)))
+              (sui (parse-ui (nth 5 line)))
+              (sab (nth 11 line))
+              (srl (parse-integer (nth 15 line))))
+          #+sbcl
+          (when (= 0 (mod (incf counter) 100000)) (sb-ext:gc :full t))
+
+          ;; pfstr deprecated by KPKENG field in MRCONSO
+          #+nil
+          (unless (gethash cui pfstr-hash)  ;; if haven't stored pfstr for cui
+            (when (and (string-equal (nth 1 line) "ENG") ; LAT
+                       (string-equal (nth 2 line) "P") ; ts
+                       (string-equal (nth 4 line) "PF")) ; stt

              (setf (gethash cui pfstr-hash) (nth 14 line))))
              (setf (gethash cui pfstr-hash) (nth 14 line))))

-        (set-lrl-hash cui srl cui-lrl-hash)
-        (set-lrl-hash lui srl lui-lrl-hash)
-        (set-lrl-hash sui srl sui-lrl-hash)
-        (set-lrl-hash (make-cuisui cui sui) srl cuisui-lrl-hash)
-        (multiple-value-bind (val found) (gethash sab sab-srl-hash)
-          (declare (ignore val))
-          (unless found
-            (setf (gethash sab sab-srl-hash) srl)))))
+          (set-lrl-hash cui srl cui-lrl-hash)
+          (set-lrl-hash lui srl lui-lrl-hash)
+          (set-lrl-hash sui srl sui-lrl-hash)
+          (set-lrl-hash (make-cuisui cui sui) srl cuisui-lrl-hash)
+          (multiple-value-bind (val found) (gethash sab sab-srl-hash)
+            (declare (ignore val))
+            (unless found
+              (setf (gethash sab sab-srl-hash) srl))))))

     (setq *preparse-hash-init?* t)
     t)
 
     (setq *preparse-hash-init?* t)
     t)
 

-  (defun pfstr-hash (cui) (gethash cui pfstr-hash))
+  #+nil (defun pfstr-hash (cui) (gethash cui pfstr-hash))

   (defun cui-lrl (cui)    (gethash cui cui-lrl-hash))
   (defun lui-lrl (lui)    (gethash lui lui-lrl-hash))
   (defun sui-lrl (sui)    (gethash sui sui-lrl-hash))
   (defun cui-lrl (cui)    (gethash cui cui-lrl-hash))
   (defun lui-lrl (lui)    (gethash lui lui-lrl-hash))
   (defun sui-lrl (sui)    (gethash sui sui-lrl-hash))


@@ -100,9 +107,10 @@
 (defparameter +col-datatypes+
     '(("AV" sql-f) ("BTS" sql-i) ("CLS" sql-i) ("COF" sql-i) ("CUI1" sql-u)
       ("AUI" sql-u) ("AUI1" sql-u) ("AUI2" sql-u) ("PCUI" sql-u)
 (defparameter +col-datatypes+
     '(("AV" sql-f) ("BTS" sql-i) ("CLS" sql-i) ("COF" sql-i) ("CUI1" sql-u)
       ("AUI" sql-u) ("AUI1" sql-u) ("AUI2" sql-u) ("PCUI" sql-u)

-      ("PLUI" sql-u) ("PAUI" sql-u)
+      ("PLUI" sql-u) ("PAUI" sql-u) ("RUI" sql-u)

       ("CUI2" sql-u) ("CUI" sql-u) ("CXN" sql-s) ("FR" sql-i) ("LRL" sql-s)
       ("LUI" sql-u) ("MAX" sql-s) ("MIN" sql-s) ("RANK" sql-s) ("REF" sql-c)
       ("CUI2" sql-u) ("CUI" sql-u) ("CXN" sql-s) ("FR" sql-i) ("LRL" sql-s)
       ("LUI" sql-u) ("MAX" sql-s) ("MIN" sql-s) ("RANK" sql-s) ("REF" sql-c)

+      ("PTR" sql-c)

       ("RNK" sql-s) ("RWS" sql-i) ("SRL" sql-s) ("SUI" sql-u) ("TUI" sql-u)
       ("MAPRANK" sql-s)
       ;;; Custom columns
       ("RNK" sql-s) ("RWS" sql-i) ("SRL" sql-s) ("SUI" sql-u) ("TUI" sql-u)
       ("MAPRANK" sql-s)
       ;;; Custom columns


@@ -127,9 +135,16 @@
   "Custom tables to create")
 
 (defparameter +custom-cols+
   "Custom tables to create")
 
 (defparameter +custom-cols+

-    '(("MRCONSO.RRF" "KPFSTR" "TEXT"
-                    (slot-value (find-ucol "STR" "MRCONSO.RRF") 'max)
-                    (lambda (x) (pfstr-hash (parse-ui (nth 0 x)))))
+    '(#+nil ("MRCONSO.RRF" "KPFSTR" "TEXT"
+             (slot-value (find-ucol "STR" "MRCONSO.RRF") 'max)
+             (lambda (x) (pfstr-hash (parse-ui (nth 0 x)))))
+      ;; Set to 1 if term is prefered term for english
+      ("MRCONSO.RRF" "KPFENG" "TINYINT" 0
+       (lambda (x)  (if (and (string-equal (nth 1 x) "ENG") ; LAT
+                             (string-equal (nth 2 x) "P") ; ts
+                             (string-equal (nth 4 x) "PF")) ; stt
+                      "1"
+                      "0")))

       ("MRCONSO.RRF" "KCUISUI" "BIGINT" 0
        (lambda (x) (write-to-string (make-cuisui (parse-ui (nth 0 x)) (parse-ui (nth 5 x))))))
       ("MRCONSO.RRF" "KCUILUI" "BIGINT" 0
       ("MRCONSO.RRF" "KCUISUI" "BIGINT" 0
        (lambda (x) (write-to-string (make-cuisui (parse-ui (nth 0 x)) (parse-ui (nth 5 x))))))
       ("MRCONSO.RRF" "KCUILUI" "BIGINT" 0


@@ -140,13 +155,6 @@
        (lambda (x) (write-to-string (lui-lrl (parse-ui (nth 3 x))))))
       ("MRCONSO.RRF" "KSUILRL" "SMALLINT" 0
        (lambda (x) (write-to-string (sui-lrl (parse-ui (nth 5 x))))))
        (lambda (x) (write-to-string (lui-lrl (parse-ui (nth 3 x))))))
       ("MRCONSO.RRF" "KSUILRL" "SMALLINT" 0
        (lambda (x) (write-to-string (sui-lrl (parse-ui (nth 5 x))))))

-      ;; Deprecated, last in 2004AA -- skip index
-      #+ignore
-      ("MRLO.RRF" "KLRL" "SMALLINT" 0
-       (lambda (x) (write-to-string
-                   (if (zerop (length (nth 4 x)))
-                       (cui-lrl (parse-ui (nth 0 x)))
-                     (cuisui-lrl (make-cuisui (parse-ui (nth 0 x)) (parse-ui (nth 4 x))))))))

       ("MRSTY.RRF" "KLRL" "SMALLINT" 0
        (lambda (x) (write-to-string (cui-lrl (parse-ui (nth 0 x))))))
       ("MRCOC.RRF" "KLRL" "SMALLINT" 0
       ("MRSTY.RRF" "KLRL" "SMALLINT" 0
        (lambda (x) (write-to-string (cui-lrl (parse-ui (nth 0 x))))))
       ("MRCOC.RRF" "KLRL" "SMALLINT" 0


@@ -159,10 +167,15 @@
        (lambda (x) (write-to-string (sab-srl (nth 10 x)))))
       ("MRRANK.RRF" "KSRL" "SMALLINT" 0
        (lambda (x) (write-to-string (sab-srl (nth 1 x)))))
        (lambda (x) (write-to-string (sab-srl (nth 10 x)))))
       ("MRRANK.RRF" "KSRL" "SMALLINT" 0
        (lambda (x) (write-to-string (sab-srl (nth 1 x)))))

+      ("MRHIER.RRF" "KSRL" "SMALLINT" 0
+       (lambda (x) (write-to-string (sab-srl (nth 4 x)))))
+      ("MRMAP.RRF" "KSRL" "SMALLINT" 0
+       (lambda (x) (write-to-string (sab-srl (nth 1 x)))))
+      ("MRSMAP.RRF" "KSRL" "SMALLINT" 0
+       (lambda (x) (write-to-string (sab-srl (nth 1 x)))))

       ("MRDEF.RRF" "KSRL" "SMALLINT" 0
        (lambda (x) (write-to-string (sab-srl (nth 4 x)))))
       ("MRDEF.RRF" "KSRL" "SMALLINT" 0
        (lambda (x) (write-to-string (sab-srl (nth 4 x)))))

-      ("MRCXT.RRF" "KSRL" "SMALLINT" 0
-       (lambda (x) (write-to-string (sab-srl (nth 2 x)))))
+      #+nil  ("MRCXT.RRF" "KSRL" "SMALLINT" 0 (lambda (x) (write-to-string (sab-srl (nth 2 x)))))

       ("MRXW_ENG.RRF" "KLRL" "SMALLINT" 0
        (lambda (x) (write-to-string (cuisui-lrl (make-cuisui
                                                 (parse-ui (nth 2 x))
       ("MRXW_ENG.RRF" "KLRL" "SMALLINT" 0
        (lambda (x) (write-to-string (cuisui-lrl (make-cuisui
                                                 (parse-ui (nth 2 x))


@@ -179,12 +192,9 @@
        (lambda (x) (write-to-string (cuisui-lrl (make-cuisui
                                                 (parse-ui (nth 2 x))
                                                 (parse-ui (nth 4 x)))))))
        (lambda (x) (write-to-string (cuisui-lrl (make-cuisui
                                                 (parse-ui (nth 2 x))
                                                 (parse-ui (nth 4 x)))))))

-      ("MRREL.RRF" "KPFSTR2" "TEXT" 1024
-       (lambda (x) (pfstr-hash (parse-ui (nth 4 x)))))
-      ("MRCOC.RRF" "KPFSTR2" "TEXT" 1024
-       (lambda (x) (pfstr-hash (parse-ui (nth 2 x)))))
-      ("MRCXT.RRF" "KCUISUI" "BIGINT" 0
-       (lambda (x) (write-to-string (make-cuisui (parse-ui (nth 0 x)) (parse-ui (nth 1 x))))))
+
+      #+nil  ("MRREL.RRF" "KPFSTR2" "TEXT" 1024 (lambda (x) (pfstr-hash (parse-ui (nth 4 x)))))
+      #+nil  ("MRCOC.RRF" "KPFSTR2" "TEXT" 1024 (lambda (x) (pfstr-hash (parse-ui (nth 2 x)))))

       ("MRSAT.RRF" "KCUILUI" "BIGINT" 0
        (lambda (x) (write-to-string (make-cuilui (parse-ui (nth 0 x)) (parse-ui (nth 1 x))))))
       ("MRSAT.RRF" "KCUISUI" "BIGINT" 0
       ("MRSAT.RRF" "KCUILUI" "BIGINT" 0
        (lambda (x) (write-to-string (make-cuilui (parse-ui (nth 0 x)) (parse-ui (nth 1 x))))))
       ("MRSAT.RRF" "KCUISUI" "BIGINT" 0


@@ -206,21 +216,26 @@
 
 (defparameter +index-cols+
     '(("CUI1" "MRCOC") ("CUI" "MRCONSO") ("LUI" "MRCONSO")
 
 (defparameter +index-cols+
     '(("CUI1" "MRCOC") ("CUI" "MRCONSO") ("LUI" "MRCONSO")

-      ("SRL" "MRCONSO") ("AUI" "MRCONSO")
-      ("SUI" "MRCONSO") ("CUI" "MRCXT") ("CUI" "MRDEF") ("CUI" "MRLO")
-      ("CUI1" "MRREL") ("CUI" "MRSAT") ("LUI" "MRSAT") ("SUI" "MRSAT")
-      ("CUI" "MRSTY")
-      ("TUI" "MRSTY") ("CUI" "MRXNS_ENG")
+      ("SRL" "MRCONSO") ("AUI" "MRCONSO") ("KPFENG" "MRCONSO")
+      ("SUI" "MRCONSO") ("SAUI" "MRCONSO") ("CODE" "MRCONSO")
+      ("SCUI" "MRCONSO")
+      ("CUI" "MRDEF")
+      ("CUI1" "MRREL") ("CUI2" "MRREL") ("SAB" "MRREL")
+      ("RUI" "MRREL") ("AUI1" "MRREL") ("AUI2" "MRREL")
+      ("CUI" "MRSAT") ("LUI" "MRSAT") ("SUI" "MRSAT")
+      ("METAUI" "MRSAT") ("ATN" "MRSAT")
+      ("CUI" "MRSTY")  ("TUI" "MRSTY") ("CUI" "MRXNS_ENG")
+      ("AUI" "MRHIER") ("CUI" "MRHIER") ("CXN" "MRHIER") ("RELA" "MRHIER") ("PAUI" "MRHIER")
+      ("SAB" "MRHIER")

       #+ignore ("NSTR" "MRXNS_ENG" 10)
       ("CUI" "MRXNW_ENG") ("NWD" "MRXNW_ENG") ("WD" "MRXW_ENG")
       ("KCUISUI" "MRCONSO") ("KCUILUI" "MRCONSO") ("KCUILRL" "MRCONSO")
       #+ignore ("NSTR" "MRXNS_ENG" 10)
       ("CUI" "MRXNW_ENG") ("NWD" "MRXNW_ENG") ("WD" "MRXW_ENG")
       ("KCUISUI" "MRCONSO") ("KCUILUI" "MRCONSO") ("KCUILRL" "MRCONSO")

-      ("KLUILRL" "MRCONSO") ("KCUISUI" "MRCXT")
+      ("KLUILRL" "MRCONSO")

       ("KCUISUI" "MRSAT")  ("KCUILUI" "MRSAT")
       ("KCUISUI" "MRXW_ENG") ("KCUISUI" "MRXNW_ENG")
       ("KCUISUI" "MRXNS_ENG") ("KCUISUI" "MRXW_NONENG")
       ("KCUISUI" "MRSAT")  ("KCUILUI" "MRSAT")
       ("KCUISUI" "MRXW_ENG") ("KCUISUI" "MRXNW_ENG")
       ("KCUISUI" "MRXNS_ENG") ("KCUISUI" "MRXW_NONENG")

-      ("KSRL" "MRCXT") ("KSRL" "MRDEF") ("KSRL" "MRRANK")
+      ("KSRL" "MRDEF") ("KSRL" "MRRANK")

       ("KSRL" "MRREL") ("KSRL" "MRSAT") ("KLRL" "MRCOC")
       ("KSRL" "MRREL") ("KSRL" "MRSAT") ("KLRL" "MRCOC")

-      #+ignore ("KLRL" "MRLO")  ;; deprecated

       ("KLRL" "MRSTY") ("KLRL" "MRXW_ENG") ("KLRL" "MRXNW_ENG")
       ("KLRL" "MRXNS_ENG") ("KLRL" "MRXW_NONENG")
       ;; LEX indices
       ("KLRL" "MRSTY") ("KLRL" "MRXW_ENG") ("KLRL" "MRXNW_ENG")
       ("KLRL" "MRXNS_ENG") ("KLRL" "MRXW_NONENG")
       ;; LEX indices


@@ -235,7 +250,7 @@
 
       ("SRL" "MRSAB") ("RSAB" "MRSAB") ("VSAB" "MRSAB") ("RCUI" "MRSAB")
       ("VCUI" "MRSAB") ("LAT" "MRSAB") ("MAPSETCUI" "MRMAP")  ("MAPSETCUI" "MRSMAP")
 
       ("SRL" "MRSAB") ("RSAB" "MRSAB") ("VSAB" "MRSAB") ("RCUI" "MRSAB")
       ("VCUI" "MRSAB") ("LAT" "MRSAB") ("MAPSETCUI" "MRMAP")  ("MAPSETCUI" "MRSMAP")

-      ("CUI" "MRHIER") ("AUI" "MRHIER") ("PAUI" "MRHIER"))
+      ("CUI" "MRHIER"))

   "Columns in files to index")
 
 
   "Columns in files to index")