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