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         :lisp-stat-config
  34         ;;:cxml
  35         :lisp-stat-object-system
  36         :lisp-stat-compound-data
  37         :lisp-stat-matrix
  38         :lisp-stat-linalg
  39         :lisp-stat-sequence)
  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 ;;; Data Types:
  70 ;;;
  71 ;;; Data types are the representation of data from a computer-science
  72 ;;; perspective, i.e. what it is that they contain.  These types
  73 ;;; include particular forms of compound types (i.e. dataframe,
  74 ;;; relationdata are compounds of arrays of different types where the
  75 ;;; difference is row-wise, while array is a compound of elements of
  76 ;;; the same type.
  77
  78 ;;Examples:
  79 ;;  (defun equidimensional (a)
  80 ;;    (or (< (array-rank a) 2)
  81 ;;        (apply #'= (array-dimensions a)))) =>  EQUIDIMENSIONAL
  82 ;;  (deftype square-matrix (&optional type size)
  83 ;;    `(and (array ,type (,size ,size))
  84 ;;          (satisfies equidimensional))) =>  SQUARE-MATRIX
  85
  86 (deftype dt-scalar (&optional type)
  87   `(or integer double complex))
  88
  89 (deftype dt-array (&optional type)
  90   `(satisfies equal-type))
  91
  92 (deftype dt-dataframe ()
  93   `(satisfies equal-type-within-column))
  94
  95 (deftype dt-relationdata ()
  96   `(satisfies (foreach unit in relationalUnit
  97                (typep unit 'dt-dataframe))))
  98
  99
 100
 101
 102 ;;; Statistical Variable Classes
 103 ;;(deftype sv-nominal )
 104 ;;(deftype sv-ordinal )
 105 ;;(deftype sv-categorical (or 'sv-nominal 'sv-ordinal))
 106 ;;(deftype sv-integer )
 107 ;;(deftype sv-real )
 108 ;;(deftype sv-rational )
 109 ;;(deftype sv-complex )
 110 ;;(deftype sv-continuous (or 'sv-integer 'sv-real 'sv-rational 'sv-complex))
 111
 112
 113 ;;; Data I/O
 114
 115 ;; We can read 2 types of data -- those which are pure data, and those
 116 ;; which are imprue (lisp-enables).
 117
 118 (defparameter *lisp-stat-data-formats*
 119   '(csv tsv))
 120
 121 ;; (defgeneric data-read (srce frmt)
 122 ;;   "read data from stream srce, in format frmt.")
 123
 124 ;; (defgeneric data-write (srce frmt)
 125 ;;   "read data from stream srce, in format frmt.")
 126
 127 ;; (defmacro with-data (body)
 128 ;;   "Stream-handling, maintaining I/O through object typing.")
 129
 130 ;; design-wise should these be replaced with a "with-data" form?
 131
 132
 133 ;; DSV processing
 134
 135 ;; XML processing
 136
 137 ;;; Data Management
 138
 139 ;; the goal is to have 2 operations which can be used to create new
 140 ;; data formats out of old ones.
 141
 142 ;; (defgeneric data-subset (ds description)
 143 ;;   "Take a dataset and make it smaller.")
 144
 145 ;; (defgeneric data-relate (ds description)
 146 ;;   "Take 2 or more datasets, and grow them into a bigger one through
 147 ;; relating them (i.e. merge is one example).")
 148
 149 ;;; Data tools from "statistics.lsp"
 150
 151 ;;;;
 152 ;;;; Data File Reading
 153 ;;;;
 154
 155 (defun count-file-columns (fname)
 156 "Args: (fname)
 157 Returns the number of lisp items on the first nonblank line of file FNAME."
 158   (with-open-file (f fname)
 159     (if f
 160         (let ((line (do ((line (read-line f) (read-line f)))
 161                         ((or (null line) (< 0 (length line))) line))))
 162           (if line
 163               (with-input-from-string (s line)
 164                 (do ((n 0 (+ n 1)) (eof (gensym)))
 165                     ((eq eof (read s nil eof)) n))))))))
 166
 167 #+xlisp (defvar *xlisptable* *readtable*)
 168
 169 (if (not (fboundp 'open-file-dialog))
 170   #+dialogs
 171   (defun open-file-dialog (&optional set)
 172     (get-string-dialog "Enter a data file name:"))
 173   #-dialogs
 174   (defun open-file-dialog (&optional set)
 175     (error "You must provide a file name explicitly")))
 176
 177 (defun read-data-file (&optional (file (open-file-dialog t)))
 178 "Args:  (file)
 179 Returns a list of all lisp objects in FILE. FILE can be a string or a symbol,
 180 in which case the symbol'f print name is used."
 181   (if file
 182       (let ((eof (gensym)))
 183         (with-open-file (f file)
 184           (if f
 185               (do* ((r (read f nil eof) (read f nil eof))
 186                     (x (list nil))
 187                     (tail x (cdr tail)))
 188                    ((eq r eof) (cdr x))
 189                    (setf (cdr tail) (list r))))))))
 190
 191 ;;; New definition to avoid stack size limit in apply
 192 (defun read-data-columns (&optional (file (open-file-dialog t))
 193                                     (cols (if file
 194                                               (count-file-columns file))))
 195 "Args: (&optional file cols)
 196 Reads the data in FILE as COLS columns and returns a list of lists representing the columns."
 197   (if (and file cols)
 198       (transpose (split-list (read-data-file file) cols))))
 199
 200
 201 ;;; FIXME:AJR:  ALL THE FOLLOWING ARE SOLVED BY PLATFORM-INDEP PATHNAME WORK!
 202
 203 #+unix
 204 (defun load-data (file)
 205 "Args: (file)
 206 Read in data file from the data examples library."
 207   (if (load (format nil "~aData/~a" *default-path* file))
 208       t
 209       (load (format nil "~aExamples/~a" *default-path* file))))
 210
 211 #+unix
 212 (defun load-example (file)
 213 "Args: (file)
 214 Read in lisp example file from the examples library."
 215   (if (load (format nil "~aExamples/~a" *default-path* file))
 216       t
 217       (load (format nil "~aData/~a" *default-path* file))))
 218 #+macintosh
 219 (defun load-data (s) (require s (concatenate 'string ":Data:" s)))
 220 #+macintosh
 221 (defun load-example (s) (require s (concatenate 'string ":Examples:" s)))
 222
 223 #+msdos
 224 (defun load-data (file)
 225 "Args: (file)
 226 Read in data file from the data examples library."
 227   (load (format nil "~aData\\~a" *default-path* file)))
 228
 229 #+msdos
 230 (defun load-example (file)
 231 "Args: (file)
 232 Read in lisp example file from the examples library."
 233   (load (format nil "~aExamples\\~a" *default-path* file)))
 234
 235 ;;;;
 236 ;;;; Listing and Saving Variables and Functions
 237 ;;;;
 238
 239 (defvar *variables* nil)
 240 (defvar *ask-on-redefine* nil)
 241
 242 (defmacro def (symbol value)
 243 "Syntax: (def var form)
 244 VAR is not evaluated and must be a symbol.  Assigns the value of FORM to
 245 VAR and adds VAR to the list *VARIABLES* of def'ed variables. Returns VAR.
 246 If VAR is already bound and the global variable *ASK-ON-REDEFINE*
 247 is not nil then you are asked if you want to redefine the variable."
 248   `(unless (and *ask-on-redefine*
 249                 (boundp ',symbol)
 250                 (not (y-or-n-p "Variable has a value. Redefine?")))
 251            (pushnew ',symbol *variables*)
 252            (setf ,symbol ,value)
 253            ',symbol))
 254
 255 (defun variables-list ()
 256     (mapcar #'intern (sort-data (mapcar #'string *variables*))))
 257
 258 (defun variables ()
 259 "Args:()
 260 Returns a list of the names of all def'ed variables to STREAM"
 261   (if *variables*
 262       (mapcar #'intern (sort-data (mapcar #'string *variables*)))))
 263
 264 (defun savevar (vars file)
 265 "Args: (vars file-name-root)
 266 VARS is a symbol or a list of symbols. FILE-NAME-ROOT is a string (or a symbol
 267 whose print name is used) not endinf in .lsp. The VARS and their current values
 268 are written to the file FILE-NAME-ROOT.lsp in a form suitable for use with the
 269 load command."
 270   (with-open-file (f (concatenate 'string (namestring file) ".lsp")
 271                      :direction :output)
 272     (let ((vars (if (consp vars) vars (list vars))))
 273       (flet ((save-one (x)
 274                (let ((v (symbol-value x)))
 275                  (if (objectp v)
 276                      (format f "(def ~s ~s)~%" x (send v :save))
 277                    (format f "(def ~s '~s)~%" x v)))))
 278             (mapcar #'save-one vars))
 279       vars)))
 280
 281 (defun undef (v)
 282 "Args: (v)
 283 If V is the symbol of a defined variable the variable it is unbound and
 284 removed from the list of defined variables. If V is a list of variable
 285 names each is unbound and removed. Returns V."
 286   (dolist (s (if (listp v) v (list v)))
 287           (when (member s *variables*)
 288                 (setq *variables* (delete s *variables*))
 289                 (makunbound s)))
 290   v)
 291
 292 ;;;;
 293 ;;;; Miscellaneous Routines
 294 ;;;;
 295
 296 (defun split-list (x n)
 297 "Args: (list cols)
 298 Returns a list of COLS lists of equal length of the elements of LIST.
 299 Example: (split-list '(1 2 3 4 5 6) 2) returns ((1 2 3) (4 5 6))"
 300   (check-one-fixnum n)
 301   (if (/= (rem (length x) n) 0) (error "length not divisible by ~a" n))
 302   (flet ((next-split ()
 303            (let ((result nil)
 304                  (end nil))
 305              (dotimes (i n result)
 306                (declare (fixnum i))
 307                (let ((c-elem (list (first x))))
 308                  (cond ((null result)
 309                         (setf result c-elem)
 310                         (setf end result))
 311                        (t
 312                         (setf (rest end) c-elem)
 313                         (setf end (rest end)))))
 314                (setf x (rest x))))))
 315     (let ((result nil)
 316           (end nil)
 317           (k (/ (length x) n)))
 318       (declare (fixnum k))
 319       (dotimes (i k result)
 320         (declare (fixnum i))
 321         (let ((c-sub (list (next-split))))
 322           (cond ((null result)
 323                  (setf result c-sub)
 324                  (setf end result))
 325                 (t
 326                  (setf (rest end) c-sub)
 327                  (setf end (rest end)))))))))