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 ;;; The purpose of this package is to manage data which will be
  49 ;;; processed by LispStat.  In particular, it willbe importnat to
  50 ;;; register variables, datasets, relational structures, and other
  51 ;;; objects which could be the target for statistical modeling and
  52 ;;; inference.
  53
  54 (defvar *lisp-stat-data-table* (make-hash-table)
  55   "Marks up the data the could be used by.")
  56
  57 (defvar *lisp-stat-data-count* 0
  58   "number of items currently recorded.")
  59
  60
  61 ;;; Data (storage) Types, dt-{.*}
  62 ;;;
  63 ;;; Data types are the representation of data from a computer-science
  64 ;;; perspective, i.e. what it is that they contain.  These types
  65 ;;; include particular forms of compound types (i.e. dataframe is
  66 ;;; array-like, but types differ, difference is row-wise, while array
  67 ;;; is a compound of elements of the same type.
  68 ;;;
  69
  70 ;;Examples:
  71 ;;  (defun equidimensional (a)
  72 ;;    (or (< (array-rank a) 2)
  73 ;;        (apply #'= (array-dimensions a)))) =>  EQUIDIMENSIONAL
  74 ;;  (deftype square-matrix (&optional type size)
  75 ;;    `(and (array ,type (,size ,size))
  76 ;;          (satisfies equidimensional))) =>  SQUARE-MATRIX
  77
  78 (defun array-of-equal-dt-scalar-type (x)
  79   ;; return dt-scalar-type which fits (more precise that works)
  80   (if x
  81       'integer
  82       nil))
  83
  84 (defun array-of-equal-dt-scalar-type-within-column (x)
  85   ;; return dt-scalar-type which fits (more precise that works)
  86   (if x
  87       'integer
  88       nil))
  89
  90
  91
  92 (deftype dt-scalar (&optional type)
  93   `(or integer double complex symbol))
  94
  95 (deftype dt-array (&optional ndim dimlist)
  96   `(satisfies array-of-equal-dt-scalar-type))
  97
  98 (deftype dt-dataframe (&optional )
  99   `(satisfies array-of-equal-dt-scalar-type-within-column))
 100
 101 ;(deftype dt-relationaldata ()
 102 ;  `(satisfies (foreach unit in relationalUnit
 103 ;              (typep unit 'dt-dataframe))))
 104
 105
 106 ;;; Statistical Variable Types, sv-{.*}
 107 ;;;
 108 ;;; Statistical variable types work to represent the statistical
 109 ;;; category represented by the variable, i.e. nominal, ordinal,
 110 ;;; integral, continous, ratio.   This metadata can be used to hint at
 111 ;;; appropriate analysis methods -- or perhaps more critically, to
 112 ;;; define how these methods will fail in the final interrpretation.
 113
 114 (deftype sv-nominal (&optional n)
 115   ())
 116
 117 (deftype sv-ordinal (ordering &optional n)
 118   ())
 119
 120 (deftype sv-categorical ()
 121   `(satisfies (or sv-nominal sv-ordinal)))
 122 ;;(deftype sv-integer )
 123 ;;(deftype sv-real )  ;; precision could be a secondary component of real, rational, complex.
 124 ;;(deftype sv-rational )
 125 ;;(deftype sv-complex )
 126 ;;(deftype sv-continuous (or 'sv-integer 'sv-real 'sv-rational 'sv-complex)) ;; perhaps, call it "mostly contin..."
 127
 128
 129 ;;; Data I/O
 130
 131 ;; We can read 2 types of data -- those which are pure data, and those
 132 ;; which are impure (lisp-enabled, data as program as data thingy's).
 133
 134 (defparameter *lisp-stat-data-formats*
 135   '(csv tsv))
 136
 137 ;; (defgeneric data-read (srce frmt)
 138 ;;   "read data from stream srce, in format frmt.")
 139
 140 ;; (defgeneric data-write (srce frmt)
 141 ;;   "read data from stream srce, in format frmt.")
 142
 143 ;; (defmacro with-data (body)
 144 ;;   "Stream-handling, maintaining I/O through object typing.")
 145
 146 ;; design-wise should these be replaced with a "with-data" form?
 147
 148 ;; DSV processing
 149
 150 ;; XML processing
 151
 152 ;;; Data Management
 153
 154 ;; the goal is to have 2 operations which can be used to create new
 155 ;; data formats out of old ones.
 156
 157 ;; (defgeneric data-subset (ds description)
 158 ;;   "Take a dataset and make it smaller.")
 159
 160 ;; (defgeneric data-relate (ds description)
 161 ;;   "Take 2 or more datasets, and grow them into a bigger one through
 162 ;; relating them (i.e. merge is one example).")
 163
 164 ;;; Data tools from "statistics.lsp"
 165
 166 ;;;;
 167 ;;;; Data File Reading
 168 ;;;;
 169
 170 (defun count-file-columns (fname)
 171 "Args: (fname)
 172 Returns the number of lisp items on the first nonblank line of file FNAME."
 173   (with-open-file (f fname)
 174     (if f
 175         (let ((line (do ((line (read-line f) (read-line f)))
 176                         ((or (null line) (< 0 (length line))) line))))
 177           (if line
 178               (with-input-from-string (s line)
 179                 (do ((n 0 (+ n 1)) (eof (gensym)))
 180                     ((eq eof (read s nil eof)) n))))))))
 181
 182 #+xlisp (defvar *xlisptable* *readtable*)
 183
 184 (if (not (fboundp 'open-file-dialog))
 185   #+dialogs
 186   (defun open-file-dialog () ;; why?(&optional set)
 187     (get-string-dialog "Enter a data file name:"))
 188   #-dialogs
 189   (defun open-file-dialog () ;; why? (&optional set)
 190     (error "You must provide a file name explicitly")))
 191
 192 (defun read-data-file (&optional (file (open-file-dialog t)))
 193 "Args:  (file)
 194 Returns a list of all lisp objects in FILE. FILE can be a string or a symbol,
 195 in which case the symbol'f print name is used."
 196   (if file
 197       (let ((eof (gensym)))
 198         (with-open-file (f file)
 199           (if f
 200               (do* ((r (read f nil eof) (read f nil eof))
 201                     (x (list nil))
 202                     (tail x (cdr tail)))
 203                    ((eq r eof) (cdr x))
 204                    (setf (cdr tail) (list r))))))))
 205
 206 ;;; New definition to avoid stack size limit in apply
 207 (defun read-data-columns (&optional (file (open-file-dialog t))
 208                                     (cols (if file
 209                                               (count-file-columns file))))
 210 "Args: (&optional file cols)
 211 Reads the data in FILE as COLS columns and returns a list of lists representing the columns."
 212   (if (and file cols)
 213       (transpose (split-list (read-data-file file) cols))))
 214
 215
 216 ;;; FIXME:AJR:  ALL THE FOLLOWING NEED TO BE SOLVED BY PLATFORM-INDEP PATHNAME WORK!
 217 ;;; FIXME:AJR: use either string or pathname.
 218
 219 (defun path-string-to-path (p s)
 220     (pathname (concatenate 'string (namestring p) s)))
 221
 222 (defun load-data (file)
 223 "Args: (file) as string
 224 Read in data file from the data examples library."
 225   (if (load (path-string-to-path *lispstat-data-dir* file))
 226       t
 227       (load (path-string-to-path *lispstat-examples-dir* file))))
 228
 229 (defun load-example (file)
 230   "Args: (file) as string
 231 Read in lisp example file from the examples library."
 232   (if (load (path-string-to-path *lispstat-examples-dir* file))
 233       t
 234       (load (path-string-to-path *lispstat-data-dir* file))))
 235
 236 ;;;;
 237 ;;;; Listing and Saving Variables and Functions
 238 ;;;;
 239
 240 (defvar *variables* nil)
 241 (defvar *ask-on-redefine* nil)
 242
 243 (defmacro def (symbol value)
 244 "Syntax: (def var form)
 245 VAR is not evaluated and must be a symbol.  Assigns the value of FORM to
 246 VAR and adds VAR to the list *VARIABLES* of def'ed variables. Returns VAR.
 247 If VAR is already bound and the global variable *ASK-ON-REDEFINE*
 248 is not nil then you are asked if you want to redefine the variable."
 249   `(unless (and *ask-on-redefine*
 250                 (boundp ',symbol)
 251                 (not (y-or-n-p "Variable has a value. Redefine?")))
 252     (if (boundp ',symbol)
 253         (setf ,symbol ,value)
 254         (defvar ,symbol ,value))
 255     (pushnew ',symbol *variables*)
 256     ',symbol))
 257
 258 (defun variables-list ()
 259     (mapcar #'intern (sort-data (mapcar #'string *variables*))))
 260
 261 (defun variables ()
 262 "Args:()
 263 Returns a list of the names of all def'ed variables to STREAM"
 264   (if *variables*
 265       (mapcar #'intern (sort-data (mapcar #'string *variables*)))))
 266
 267 (defun savevar (vars file)
 268 "Args: (vars file-name-root)
 269 VARS is a symbol or a list of symbols. FILE-NAME-ROOT is a string (or a symbol
 270 whose print name is used) not endinf in .lsp. The VARS and their current values
 271 are written to the file FILE-NAME-ROOT.lsp in a form suitable for use with the
 272 load command."
 273   (with-open-file (f (concatenate 'string (namestring file) ".lsp")
 274                      :direction :output)
 275     (let ((vars (if (consp vars) vars (list vars))))
 276       (flet ((save-one (x)
 277                (let ((v (symbol-value x)))
 278                  (if (objectp v)
 279                      (format f "(def ~s ~s)~%" x (send v :save))
 280                    (format f "(def ~s '~s)~%" x v)))))
 281             (mapcar #'save-one vars))
 282       vars)))
 283
 284 (defun undef (v)
 285 "Args: (v)
 286 If V is the symbol of a defined variable the variable it is unbound and
 287 removed from the list of defined variables. If V is a list of variable
 288 names each is unbound and removed. Returns V."
 289   (dolist (s (if (listp v) v (list v)))
 290           (when (member s *variables*)
 291                 (setq *variables* (delete s *variables*))
 292                 (makunbound s)))
 293   v)
 294