data.lisp

   1 ;;; -*- mode: lisp -*-
   2 ;;; Copyright (c) 2005--2007, by A.J. Rossini <blindglobe@gmail.com>
   3 ;;; See COPYRIGHT file for any additional restrictions (BSD license).
   4 ;;; Since 1991, ANSI was finally finished.  Edited for ANSI Common Lisp.
   5
   6 ;;; File:       data.lisp
   7 ;;; Author:     AJ Rossini <blindglobe@gmail.com>
   8 ;;; Copyright:  (c)2007, AJ Rossini.  BSD, LLGPL, or GPLv2, depending on how it arrives.
   9 ;;; Purpose:    data package for lispstat
  10 ;;; Time-stamp: <2006-05-19 12:33:41 rossini>
  11 ;;; Creation:   <2006-05-17 21:34:07 rossini>
  12
  13 ;;; What is this talk of 'release'? Klingons do not make software
  14 ;;; 'releases'.  Our software 'escapes', leaving a bloody trail of
  15 ;;; designers and quality assurance people in its wake.
  16
  17 ;;; This organization and structure is new to the 21st Century
  18 ;;; version.
  19
  20 ;;; conside    that dataa has 3 genotypic chracteristrics.  The first
  21 ;;; would be form -- scalar, vector, array.   second would be
  22 ;;; datarep type.  in particular integer, real, string, symbol.  The last
  23 ;;; would be statistical type.  augmenting datarep type with use in a
  24 ;;; statistical context, i.e. that would include nominal, ordinal,
  25 ;;; integer, continous, interval (orderable subtypes)
  26
  27 (in-package :cl-user)
  28
  29 (defpackage :lisp-stat-data
  30   (:documentation "Data I/O, management, other data technologies.")
  31   (:nicknames :ls-data)
  32   (:use :common-lisp
  33         ;;:cxml
  34         :lisp-stat-config
  35         :lisp-stat-object-system
  36         :lisp-stat-types
  37         :lisp-stat-compound-data
  38         :lisp-stat-matrix
  39         :lisp-stat-linalg)
  40   (:shadowing-import-from :lisp-stat-object-system
  41                           slot-value call-method call-next-method)
  42   (:export
  43    ;;; from statistics.lsp
  44           open-file-dialog read-data-file read-data-columns load-data
  45           load-example *variables* *ask-on-redefine* def variables savevar
  46           undef split-list ))
  47
  48 (in-package :lisp-stat-data)
  49
  50 ;; (deftype dataformtype (list scalar vector sequence list array relation))
  51 ;; (deftype datareptype (list integer rational real complex string symbol=
  52 ;; (deftype stattzpe (list
  53
  54
  55
  56
  57 ;;; The purpose of this package is to manage data which will be
  58 ;;; processed by LispStat.  In particular, it willbe importnat to
  59 ;;; register variables, datasets, relational structures, and other
  60 ;;; objects which could be the target for statistical modeling and
  61 ;;; inference.
  62
  63 (defvar *lisp-stat-data-table* (make-hash-table)
  64   "Marks up the data the could be used by.")
  65
  66 (defvar *lisp-stat-data-count* 0
  67   "number of items currently recorded.")
  68
  69 #|
  70 ;;; Data Types:
  71 ;;;
  72 ;;; Data types are the representation of data from a computer-science
  73 ;;; perspective, i.e. what it is that they contain.  These types
  74 ;;; include particular forms of compound types (i.e. dataframe,
  75 ;;; relationdata are compounds of arrays of different types where the
  76 ;;; difference is row-wise, while array is a compound of elements of
  77 ;;; the same type.
  78
  79 ;;Examples:
  80 ;;  (defun equidimensional (a)
  81 ;;    (or (< (array-rank a) 2)
  82 ;;        (apply #'= (array-dimensions a)))) =>  EQUIDIMENSIONAL
  83 ;;  (deftype square-matrix (&optional type size)
  84 ;;    `(and (array ,type (,size ,size))
  85 ;;          (satisfies equidimensional))) =>  SQUARE-MATRIX
  86
  87 (deftype dt-scalar (&optional type)
  88   `(or integer double complex))
  89
  90 (deftype dt-array (&optional type)
  91   `(satisfies array-of-equal-type))
  92
  93 (deftype dt-dataframe ()
  94   `(satisfies array-of-equal-type-within-column))
  95
  96 (deftype dt-relationaldata ()
  97   `(satisfies (foreach unit in relationalUnit
  98                (typep unit 'dt-dataframe))))
  99
 100
 101
 102
 103 ;;; Statistical Variable Classes
 104
 105
 106 (deftype sv-nominal (&optional length)
 107   ())
 108
 109 (deftype sv-ordinal (ordering &optional length)
 110   ())
 111
 112 (deftype sv-categorical ()
 113   `(satisfies (or sv-nominal sv-ordinal)))
 114 ;;(deftype sv-integer )
 115 ;;(deftype sv-real )
 116 ;;(deftype sv-rational )
 117 ;;(deftype sv-complex )
 118 ;;(deftype sv-continuous (or 'sv-integer 'sv-real 'sv-rational 'sv-complex))
 119
 120 |#
 121
 122
 123 ;;; Data I/O
 124
 125 ;; We can read 2 types of data -- those which are pure data, and those
 126 ;; which are imprue (lisp-enables).
 127
 128 (defparameter *lisp-stat-data-formats*
 129   '(csv tsv))
 130
 131 ;; (defgeneric data-read (srce frmt)
 132 ;;   "read data from stream srce, in format frmt.")
 133
 134 ;; (defgeneric data-write (srce frmt)
 135 ;;   "read data from stream srce, in format frmt.")
 136
 137 ;; (defmacro with-data (body)
 138 ;;   "Stream-handling, maintaining I/O through object typing.")
 139
 140 ;; design-wise should these be replaced with a "with-data" form?
 141
 142
 143 ;; DSV processing
 144
 145 ;; XML processing
 146
 147 ;;; Data Management
 148
 149 ;; the goal is to have 2 operations which can be used to create new
 150 ;; data formats out of old ones.
 151
 152 ;; (defgeneric data-subset (ds description)
 153 ;;   "Take a dataset and make it smaller.")
 154
 155 ;; (defgeneric data-relate (ds description)
 156 ;;   "Take 2 or more datasets, and grow them into a bigger one through
 157 ;; relating them (i.e. merge is one example).")
 158
 159 ;;; Data tools from "statistics.lsp"
 160
 161 ;;;;
 162 ;;;; Data File Reading
 163 ;;;;
 164
 165 (defun count-file-columns (fname)
 166 "Args: (fname)
 167 Returns the number of lisp items on the first nonblank line of file FNAME."
 168   (with-open-file (f fname)
 169     (if f
 170         (let ((line (do ((line (read-line f) (read-line f)))
 171                         ((or (null line) (< 0 (length line))) line))))
 172           (if line
 173               (with-input-from-string (s line)
 174                 (do ((n 0 (+ n 1)) (eof (gensym)))
 175                     ((eq eof (read s nil eof)) n))))))))
 176
 177 #+xlisp (defvar *xlisptable* *readtable*)
 178
 179 (if (not (fboundp 'open-file-dialog))
 180   #+dialogs
 181   (defun open-file-dialog () ;; why?(&optional set)
 182     (get-string-dialog "Enter a data file name:"))
 183   #-dialogs
 184   (defun open-file-dialog () ;; why? (&optional set)
 185     (error "You must provide a file name explicitly")))
 186
 187 (defun read-data-file (&optional (file (open-file-dialog t)))
 188 "Args:  (file)
 189 Returns a list of all lisp objects in FILE. FILE can be a string or a symbol,
 190 in which case the symbol'f print name is used."
 191   (if file
 192       (let ((eof (gensym)))
 193         (with-open-file (f file)
 194           (if f
 195               (do* ((r (read f nil eof) (read f nil eof))
 196                     (x (list nil))
 197                     (tail x (cdr tail)))
 198                    ((eq r eof) (cdr x))
 199                    (setf (cdr tail) (list r))))))))
 200
 201 ;;; New definition to avoid stack size limit in apply
 202 (defun read-data-columns (&optional (file (open-file-dialog t))
 203                                     (cols (if file
 204                                               (count-file-columns file))))
 205 "Args: (&optional file cols)
 206 Reads the data in FILE as COLS columns and returns a list of lists representing the columns."
 207   (if (and file cols)
 208       (transpose (split-list (read-data-file file) cols))))
 209
 210
 211 ;;; FIXME:AJR:  ALL THE FOLLOWING NEED TO BE SOLVED BY PLATFORM-INDEP PATHNAME WORK!
 212 ;;; FIXME:AJR: use either string or pathname.
 213
 214 (defun path-string-to-path (p s)
 215     (pathname (concatenate 'string (namestring p) s)))
 216
 217 (defun load-data (file)
 218 "Args: (file) as string
 219 Read in data file from the data examples library."
 220   (if (load (path-string-to-path *lispstat-data-dir* file))
 221       t
 222       (load (path-string-to-path *lispstat-examples-dir* file))))
 223
 224 (defun load-example (file)
 225   "Args: (file) as string
 226 Read in lisp example file from the examples library."
 227   (if (load (path-string-to-path *lispstat-examples-dir* file))
 228       t
 229       (load (path-string-to-path *lispstat-data-dir* file))))
 230
 231 ;;;;
 232 ;;;; Listing and Saving Variables and Functions
 233 ;;;;
 234
 235 (defvar *variables* nil)
 236 (defvar *ask-on-redefine* nil)
 237
 238 (defmacro def (symbol value)
 239 "Syntax: (def var form)
 240 VAR is not evaluated and must be a symbol.  Assigns the value of FORM to
 241 VAR and adds VAR to the list *VARIABLES* of def'ed variables. Returns VAR.
 242 If VAR is already bound and the global variable *ASK-ON-REDEFINE*
 243 is not nil then you are asked if you want to redefine the variable."
 244   `(unless (and *ask-on-redefine*
 245                 (boundp ',symbol)
 246                 (not (y-or-n-p "Variable has a value. Redefine?")))
 247            (pushnew ',symbol *variables*)
 248            (setf ,symbol ,value)
 249            ',symbol))
 250
 251 (defun variables-list ()
 252     (mapcar #'intern (sort-data (mapcar #'string *variables*))))
 253
 254 (defun variables ()
 255 "Args:()
 256 Returns a list of the names of all def'ed variables to STREAM"
 257   (if *variables*
 258       (mapcar #'intern (sort-data (mapcar #'string *variables*)))))
 259
 260 (defun savevar (vars file)
 261 "Args: (vars file-name-root)
 262 VARS is a symbol or a list of symbols. FILE-NAME-ROOT is a string (or a symbol
 263 whose print name is used) not endinf in .lsp. The VARS and their current values
 264 are written to the file FILE-NAME-ROOT.lsp in a form suitable for use with the
 265 load command."
 266   (with-open-file (f (concatenate 'string (namestring file) ".lsp")
 267                      :direction :output)
 268     (let ((vars (if (consp vars) vars (list vars))))
 269       (flet ((save-one (x)
 270                (let ((v (symbol-value x)))
 271                  (if (objectp v)
 272                      (format f "(def ~s ~s)~%" x (send v :save))
 273                    (format f "(def ~s '~s)~%" x v)))))
 274             (mapcar #'save-one vars))
 275       vars)))
 276
 277 (defun undef (v)
 278 "Args: (v)
 279 If V is the symbol of a defined variable the variable it is unbound and
 280 removed from the list of defined variables. If V is a list of variable
 281 names each is unbound and removed. Returns V."
 282   (dolist (s (if (listp v) v (list v)))
 283           (when (member s *variables*)
 284                 (setq *variables* (delete s *variables*))
 285                 (makunbound s)))
 286   v)
 287
 288 ;;;;
 289 ;;;; Miscellaneous Routines
 290 ;;;;
 291
 292 (defun split-list (x n)
 293 "Args: (list cols)
 294 Returns a list of COLS lists of equal length of the elements of LIST.
 295 Example: (split-list '(1 2 3 4 5 6) 2) returns ((1 2 3) (4 5 6))"
 296   (check-one-fixnum n)
 297   (if (/= (rem (length x) n) 0) (error "length not divisible by ~a" n))
 298   (flet ((next-split ()
 299            (let ((result nil)
 300                  (end nil))
 301              (dotimes (i n result)
 302                (declare (fixnum i))
 303                (let ((c-elem (list (first x))))
 304                  (cond ((null result)
 305                         (setf result c-elem)
 306                         (setf end result))
 307                        (t
 308                         (setf (rest end) c-elem)
 309                         (setf end (rest end)))))
 310                (setf x (rest x))))))
 311     (let ((result nil)
 312           (end nil)
 313           (k (/ (length x) n)))
 314       (declare (fixnum k))
 315       (dotimes (i k result)
 316         (declare (fixnum i))
 317         (let ((c-sub (list (next-split))))
 318           (cond ((null result)
 319                  (setf result c-sub)
 320                  (setf end result))
 321                 (t
 322                  (setf (rest end) c-sub)
 323                  (setf end (rest end)))))))))