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