X-Git-Url: http://git.kpe.io/?p=umlisp.git;a=blobdiff_plain;f=parse-common.lisp;h=b1395665b68eb612ef4de7fcfccaa16436ce2981;hp=198041445fcf9baf870f8ac45a28af19436dc64c;hb=HEAD;hpb=dff3199405205cf99782dd3abf9d9dde187f5494 diff --git a/parse-common.lisp b/parse-common.lisp index 1980414..f7fab1e 100644 --- a/parse-common.lisp +++ b/parse-common.lisp @@ -2,15 +2,13 @@ ;;;; ************************************************************************* ;;;; FILE IDENTIFICATION ;;;; -;;;; Name: parse-common.lisp -;;;; Purpose: Common, stable parsing routines for UMLisp -;;;; Author: Kevin M. Rosenberg -;;;; Date Started: Apr 2000 -;;;; -;;;; $Id: parse-common.lisp,v 1.10 2003/05/07 22:53:36 kevin Exp $ +;;;; Name: parse-common.lisp +;;;; Purpose: Common, stable parsing routines for UMLisp +;;;; Author: Kevin M. Rosenberg +;;;; Created: Apr 2000 ;;;; ;;;; This file, part of UMLisp, is -;;;; Copyright (c) 2000-2002 by Kevin M. Rosenberg, M.D. +;;;; Copyright (c) 2000-2010 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. @@ -18,15 +16,21 @@ (in-package #:umlisp) -(eval-when (:compile-toplevel) - (declaim (optimize (speed 3) (safety 1) (compilation-speed 0) (debug 3)))) +(defun ensure-ucols+ufiles (&optional (alwaysclear nil)) + "Initialize all UMLS file and column structures if not already initialized" + (handler-case + (when (or alwaysclear (null *umls-files*)) + (setf *umls-cols* nil *umls-files* nil) + (gen-ufiles) + (gen-ucols) + (set-ucols-for-ufiles *umls-files*) + (ensure-field-lengths)) + (error (e) + (warn "Error reading ucols+ufiles: ~A." e) + (setf *umls-cols* nil *umls-files* nil) + nil)) + t) -(defun ensure-init-umls (&optional (alwaysclear nil)) -"Initialize all UMLS file and column structures if not already initialized" - (when (or alwaysclear (null *umls-files*)) - (gen-ucols) - (gen-ufiles) - (ensure-field-lengths))) (defun add-ucols (ucols) "Adds a ucol or list of ucols to *umls-cols*. Returns input value." @@ -38,30 +42,46 @@ (setq *umls-files* (append (mklist ufiles) *umls-files*)) ufiles) +(defun ufile-pathname (ufile &optional (extension "")) + "Return pathname for a umls filename with an optional extension" + (assert (typep ufile 'ufile)) + (let* ((dirs (nconc (list (dir ufile)) + (awhen (subdir ufile) (list it)))) + (name-list (delimited-string-to-list (fil ufile) #\.)) + (name (if (second name-list) + (first name-list) + (concatenate 'string (first name-list) (or extension "")))) + (type (when (second name-list) + (concatenate 'string (second name-list) (or extension ""))))) + (merge-pathnames + (make-pathname :name name :type type + :directory (cons :relative dirs)) + *umls-path*))) + (defun umls-pathname (filename &optional (extension "")) "Return pathname for a umls filename with an optional extension" (etypecase filename (string - (merge-pathnames - (make-pathname :name (concatenate 'string filename extension)) - (case (char filename 0) - ((#\M #\m) - *meta-path*) - ((#\L #\l) - *lex-path*) - ((#\S #\s) - *net-path*) - (t - *umls-path*)))) + (let* ((name-list (delimited-string-to-list filename #\.)) + (name (if (second name-list) + (first name-list) + (concatenate 'string (first name-list) (or extension "")))) + (type (when (second name-list) + (concatenate 'string (second name-list) (or extension ""))))) + (merge-pathnames + (make-pathname :name name :type type) + (case (schar filename 0) + ((#\M #\m) + *meta-path*) + ((#\L #\l) + *lex-path*) + ((#\S #\s) + *net-path*) + (t + *umls-path*))))) (pathname - filename))) + filename))) -(defun read-umls-line (strm &optional (eof 'eof)) - "Read a line from a UMLS stream, split into fields" - (let ((line (read-line strm nil eof))) - (if (eq line eof) - eof - (delimited-string-to-list line #\| t)))) ;;; Find field lengths for LEX and NET files @@ -69,56 +89,71 @@ "Initial colstruct field lengths for files that don't have a measurement. Currently, these are the LEX and NET files." (dolist (length-list (ufiles-field-lengths (ufiles-to-measure))) - (destructuring-bind (filename fields-max fields-av) length-list + (destructuring-bind (filename fields-max fields-av count-lines) length-list (let ((file (find-ufile filename))) - (unless file - (error "Can't find ~A filename in ufiles")) - (unless (= (length fields-max) (length (fields file))) - (error - "Number of file fields ~A not equal to field count in ufile ~S" - fields-max file)) - (dotimes (i (length (fields file))) - (declare (fixnum i)) - (let* ((field (nth i (fields file))) - (col (find-ucol field filename))) - (unless col - (error "can't find column ~A" field)) - (setf (cmax col) (aref fields-max i)) - (setf (av col) (aref fields-av i)) - (ensure-ucol-datatype col (datatype-for-colname (col col))))))))) - + (unless file + (error "Can't find ~A filename in ufiles" filename)) + (if (zerop count-lines) + (warn "File ~A is empty." filename) + (progn + (unless (= (length fields-max) (length (fields file))) + (error + "Number of file fields ~A not equal to field count in ufile ~S" + fields-max file)) + (dotimes (i (length (fields file))) + (declare (fixnum i)) + (let* ((field (nth i (fields file))) + (col (find-ucol field filename))) + (unless col + (error "can't find column ~A" field)) + (setf (cmax col) (aref fields-max i)) + (setf (av col) (aref fields-av i)) + (ensure-ucol-datatype col (datatype-for-colname (col col))))))))))) + (defun ufiles-to-measure () "Returns a list of ufiles to measure" (loop for ufile in *umls-files* - unless (or (char= #\M (char (fil ufile) 0)) - (char= #\m (char (fil ufile) 0))) - collect ufile)) - - + unless (or (char= #\M (schar (fil ufile) 0)) + (char= #\m (schar (fil ufile) 0))) + collect ufile)) + + (defun ufiles-field-lengths (ufiles) "Returns a list of lists of containing (FILE MAX AV)" - (loop for ufile in ufiles collect (file-field-lengths (fil ufile)))) + (loop for ufile in ufiles collect (file-field-lengths ufile))) -(defun file-field-lengths (filename) +(defun file-field-lengths (ufile) "Returns a list of FILENAME MAX AV" + (declare (optimize (speed 3) (safety 0))) (let (fields-max fields-av num-fields (count-lines 0)) - (with-umls-file (line filename) + (declare (fixnum count-lines)) + (with-umls-ufile (line ufile) (unless num-fields - (setq num-fields (length line)) - (setq fields-max (make-array num-fields :element-type 'fixnum - :initial-element 0)) - (setq fields-av (make-array num-fields :element-type 'number - :initial-element 0))) - (dotimes (i num-fields) - (declare (fixnum i)) - (let ((len (length (nth i line)))) - (incf (aref fields-av i) len) - (when (> len (aref fields-max i)) - (setf (aref fields-max i) len)))) + (setq num-fields (length line)) + (setq fields-max (make-array num-fields :element-type 'fixnum + :initial-element 0)) + (setq fields-av (make-array num-fields :element-type '(or integer float) + :initial-element 0))) + (dotimes (i (or num-fields 0)) + (declare (fixnum i)) + (let* ((str (nth i line)) + (len (length #-(and clisp unicode) str + #+(and clisp unicode) + (if *octet-sql-storage* + (ext:convert-string-to-bytes str charset:utf-8) + str)))) + #-(and clisp unicode) (declare (string str)) + (declare (type (integer 0 100000000) len)) + (incf (aref fields-av i) len) + (when (> len (aref fields-max i)) + (setf (aref fields-max i) len)))) (incf count-lines)) - (dotimes (i num-fields) - (setf (aref fields-av i) (float (/ (aref fields-av i) count-lines)))) - (list filename fields-max fields-av))) + (dotimes (i (or num-fields 0)) + (setf (aref fields-av i) + (if (plusp count-lines) + (float (/ (aref fields-av i) count-lines)) + 0))) + (list (fil ufile) fields-max fields-av count-lines))) ;;; UMLS column/file functions @@ -126,7 +161,7 @@ Currently, these are the LEX and NET files." "Returns list of umls-col structure for a column name and a filename" (dolist (ucol ucols nil) (when (and (string-equal filename (fil ucol)) - (string-equal colname (col ucol))) + (string-equal colname (col ucol))) (return-from find-ucol-of-colname ucol)))) (defun ensure-col-in-columns (colname filename ucols) @@ -137,15 +172,15 @@ Currently, these are the LEX and NET files." (defun make-ucol-for-column (colname filename ucols) ;; try to find column name without a terminal digit (let* ((len (length colname)) - (last-digit? (digit-char-p (char colname (1- len)))) - (base-colname (if last-digit? - (subseq colname 0 (1- len)) - colname)) - (ucol (when last-digit? - (find-ucol-of-colname base-colname filename ucols)))) + (last-digit? (digit-char-p (schar colname (1- len)))) + (base-colname (if last-digit? + (subseq colname 0 (1- len)) + colname)) + (ucol (when last-digit? + (find-ucol-of-colname base-colname filename ucols)))) (when (and last-digit? (null ucol)) (error "Couldn't find a base column for col ~A in file ~A" - colname filename)) + colname filename)) (copy-or-new-ucol colname filename ucol))) (defun copy-or-new-ucol (colname filename ucol) @@ -162,27 +197,33 @@ Currently, these are the LEX and NET files." (defun ensure-compiled-fun (fun) "Ensure that a function is compiled" (etypecase fun + (null + nil) (function (if (compiled-function-p fun) - fun - (compile nil fun))) + fun + (compile nil fun))) (list (compile nil fun)))) (defun make-ucol (col des ref min av max fil dty - &key (sqltype "VARCHAR") (parse-fun #'add-sql-quotes) - (quote-str "'") (custom-value-fun)) + &key (sqltype "VARCHAR") (parse-fun #'add-sql-quotes) + (quote-str "'") (custom-value-fun)) (let ((ucol (make-instance - 'ucol - :col col :des des :ref ref :min min :av av :max max :fil fil - :dty dty :sqltype sqltype :quote-str quote-str - :parse-fun (ensure-compiled-fun parse-fun) - :custom-value-fun (ensure-compiled-fun custom-value-fun)))) + 'ucol + :col col :des des :ref ref :min min :av av + :max (if (eql max 0) 1 max) ;; ensure at least one char wide + :fil fil + :dty dty + :sqltype sqltype + :quote-str quote-str + :parse-fun (ensure-compiled-fun parse-fun) + :custom-value-fun (ensure-compiled-fun custom-value-fun)))) (ensure-ucol-datatype ucol (datatype-for-colname col)) ucol)) (defun make-empty-ucol (colname filename) - (warn "call in make-empty-ucol") + ;;(format "call in make-empty-ucol: ~A/~A" colname filename) (make-ucol (copy-seq colname) "Unknown" "" nil nil nil filename nil)) (defun find-ucol (colname filename) @@ -190,67 +231,164 @@ Currently, these are the LEX and NET files." (ensure-col-in-columns colname filename *umls-cols*)) (defun find-ufile (filename) - "Returns umls-file structure for a filename" - (find-if #'(lambda (f) (string-equal filename (fil f))) *umls-files*)) + "Returns umls-file structure for a filename" + (find-if #'(lambda (f) (string= filename (fil f))) *umls-files*)) + +(defun position-field-file (filename fieldname) + "Returns the position of a field in a file" + (let ((ufile (find-ufile filename))) + (unless ufile + (warn "Unable to find ufile for filename ~A." filename) + (return-from position-field-file nil)) + (let ((pos (position fieldname (fields ufile) :test #'string=))) + (unless pos + (warn "Unable to find field ~A in ufile ~S." fieldname ufile) + (return-from position-field-file nil)) + pos))) -(defun find-ucols-for-filename (filename) +(defun find-ucols-for-ufile (ufile) "Returns list of umls-cols for a file structure" - (loop for colname in (fields (find-ufile filename)) - collect (find-ucol colname filename))) + (loop for colname in (fields ufile) + collect (find-ucol colname + (if (subdir ufile) + (concatenate 'string (subdir ufile) "/" (fil ufile)) + (fil ufile))))) (defun umls-field-string-to-list (fmt) "Converts a comma delimited list of fields into a list of field names. Will append a unique number (starting at 2) onto a column name that is repeated in the list" (let ((col-counts (make-hash-table :test 'equal))) (loop for colname in (delimited-string-to-list (escape-column-name fmt) #\,) - collect - (multiple-value-bind (value found) (gethash col col-counts) - (cond - (found - (incf (gethash colname col-counts)) - (concatenate 'string colname (write-to-string (1+ value)))) - (t - (setf (gethash colname col-counts) 1) - colname)))))) - -(defun make-ufile (fil des table cls rws bts fields) - (let ((ufile (make-instance 'ufile :fil fil :des des :table table :cls cls - :rws rws :bts bts :fields fields))) - (setf (ucols ufile) (find-ucols-for-filename fil)) - ufile)) + collect + (multiple-value-bind (value found) (gethash colname col-counts) + (cond + (found + (incf (gethash colname col-counts)) + (concatenate 'string colname (write-to-string (1+ value)))) + (t + (setf (gethash colname col-counts) 1) + colname)))))) + +(defun decompose-fil (fil) + (if fil + (let ((pos (position #\/ fil))) + (if pos + (values (subseq fil (1+ pos)) (subseq fil 0 pos)) + (values fil nil))) + (values nil nil))) + +(defun filename-to-tablename (file) + (let ((pos (search ".RRF" file))) + (when pos + (setf file (subseq file 0 pos)))) + (substitute #\_ #\. file)) + +(defun make-ufile (dir fil des cls rws bts fields) + (multiple-value-bind (file subdir) (decompose-fil fil) + (let ((ufile (make-instance 'ufile :dir dir :fil file :subdir subdir + :des des :cls cls + :rws rws :bts bts :fields fields + :table (filename-to-tablename file)))) + ufile))) + +(defun set-ucols-for-ufiles (ufiles) + (dolist (ufile ufiles) + (setf (ucols ufile) (find-ucols-for-ufile ufile)))) (defun datatype-for-colname (colname) -"Return datatype for column name" +"Return datatype for column name" (second (find colname +col-datatypes+ :key #'car :test #'string-equal))) +(defun canonicalize-column-type (type) + (cond + ((string-equal type "TINYINT") + (case *umls-sql-type* + (:mysql "TINYINT") + ((:postgresql :postgresql-socket) "INT1") + (:oracle "NUMBER(3,0)") + (t "INTEGER"))) + ((string-equal type "SMALLINT") + (case *umls-sql-type* + (:mysql "SMALLINT") + ((:postgresql :postgresql-socket) "INT2") + (:oracle "NUMBER(5,0)") + (t "INTEGER"))) + ((string-equal type "INTEGER") + (case *umls-sql-type* + (:mysql "INTEGER") + ((:postgresql :postgresql-socket) "INT4") + (:oracle "NUMBER(9,0)") + (t "INTEGER"))) + ((string-equal type "BIGINT") + (case *umls-sql-type* + (:mysql "BIGINT") + ((:postgresql :postgresql-socket) "INT8") + (:oracle "NUMBER(38,0)") + (t "INTEGER"))) + ((string-equal type "TEXT") + (case *umls-sql-type* + (:mysql "TEXT") + ((:postgresql :postgresql-socket) "TEXT") + (:oracle "VARCHAR2(3000)") + (t "VARCHAR(3000)"))) + ((string-equal type "VARCHAR") + (case *umls-sql-type* + (:mysql "VARCHAR") + ((:postgresql :postgresql-socket) "VARCHAR") + (:oracle "VARCHAR2") + (t "VARCHAR"))) + ((string-equal type "NUMERIC") + (case *umls-sql-type* + (:mysql "NUMERIC") + ((:postgresql :postgresql-socket) "NUMERIC") + (:oracle "NUMBER") + (t "NUMERIC"))) + (t + type))) + (defun ensure-ucol-datatype (col datatype) -"Add data type information to column" + "Add data type information to column" (setf (datatype col) datatype) (case datatype - (sql-u (setf (sqltype col) "INTEGER" - (parse-fun col) #'parse-ui - (quote-str col) "")) - (sql-s (setf (sqltype col) "SMALLINT" - (parse-fun col) #'parse-integer - (quote-str col) "")) - (sql-l (setf (sqltype col) "BIGINT" - (parse-fun col) #'parse-integer - (quote-str col) "")) - (sql-i (setf (sqltype col) "INTEGER" - (parse-fun col) #'parse-integer - (quote-str col) "")) - (sql-f (setf (sqltype col) "NUMERIC" - (parse-fun col) #'read-from-string - (quote-str col) "")) - (t ; Default column type, optimized text storage - (setf (parse-fun col) #'add-sql-quotes - (quote-str col) "'") + (sql-u (setf (sqltype col) (canonicalize-column-type "INTEGER") + (parse-fun col) #'parse-ui + (quote-str col) "")) + (sql-s (setf (sqltype col) (canonicalize-column-type "SMALLINT") + (parse-fun col) #'parse-integer + (quote-str col) "")) + (sql-l (setf (sqltype col) (canonicalize-column-type "BIGINT") + (parse-fun col) #'parse-integer + (quote-str col) "")) + (sql-i (setf (sqltype col) (canonicalize-column-type "INTEGER") + (parse-fun col) #'parse-integer + (quote-str col) "")) + (sql-t (setf (sqltype col) (canonicalize-column-type "TINYINT") + (parse-fun col) #'parse-integer + (quote-str col) "")) + (sql-f (setf (sqltype col) (canonicalize-column-type "NUMERIC") + (parse-fun col) #'read-from-string + (quote-str col) "")) + (t ; Default column type, optimized text storage + (setf (parse-fun col) #'add-sql-quotes + (quote-str col) "'") (when (and (cmax col) (av col)) (if (> (cmax col) 255) - (setf (sqltype col) "TEXT") - (if (< (- (cmax col) (av col)) 4) - (setf (sqltype col) "CHAR") ; if average bytes wasted < 4 - (setf (sqltype col) "VARCHAR"))))))) + (setf (sqltype col) (canonicalize-column-type "TEXT")) + (setf (sqltype col) (canonicalize-column-type "VARCHAR"))))))) (defun escape-column-name (name) (substitute #\_ #\/ name)) + +;; SQLNAME is required for collision of SQL reserved words (MYSQL 8: RANK) +;; and column names in UMLS (RANK in MRRANK) +(defvar *sql-reserved-names* '("RANK")) +(defmethod sqlname ((c ucol)) + (sqlname (col c))) +(defmethod sqlname ((name string)) + (if (find name *sql-reserved-names* :test #'string-equal) + (concatenate 'string "_" name) + name)) +(defmethod sqlname ((l list)) + (mapcar #'sqlname l)) +(defmethod sqlname ((s symbol)) + (sqlname (symbol-name s)))