src/data/import.lisp

   1 ;;; -*- mode: lisp -*-
   2 ;;; Copyright (c) 2008, 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 ;;; Time-stamp: <2009-12-06 10:08:34 tony>
   7 ;;; Creation:   <2008-09-03 08:10:00 tony>
   8 ;;; File:       import.lisp
   9 ;;; Author:     AJ Rossini <blindglobe@gmail.com>
  10 ;;; Copyright:  (c)2007--2009, AJ Rossini.  GPLv2
  11 ;;; Purpose:    base structures for importing data into CLS
  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 (in-package :lisp-stat-data)
  18
  19 ;;; Data I/O
  20
  21 ;; We can read 2 types of data -- those which are non-lisp-native
  22 ;; data, and those which are lisp-native (lisp-enabled, an extension
  23 ;; of lisp-serialized, i.e. data as program as data thingy's).
  24
  25 ;; of the non-native, there could be raw sources (ascii file formats),
  26 ;; xml sources (xml -> lisp, possible with some preprocessing.
  27
  28 ;;; Reading from DSV files:
  29
  30 ;;; consider either the cybertyggyr-dsv package, or the rsm.string
  31 ;;; package.  The latter seems to actually work a bit at what we need
  32 ;;; to acccomplish, but the former is a challenge to get right when we
  33 ;;; need to think abut what it is that we need to get done.  The
  34 ;;; latter is also better licensed. i.e. BSD-style.
  35
  36
  37 (defparameter *lisp-stat-data-external-source-formats*
  38   '(csv tsv xml ;; ex of text-based (UTF, ASCII, or similar) formats
  39     sql ;; ex of RDBMS call
  40     fcs affy))  ;; ex of binary formats
  41
  42 (defparameter *lisp-stat-data-import-referencing-type*
  43   '(lisp-data-structure reference lisp-function))
  44
  45 (defgeneric data-import (source source-format referencing-type)
  46   (:documentation "read data from stream srce, in format srce-frmt;
  47                    return a reftype, which could be a
  48                    lisp-data-structure, a reference to such, or a lisp
  49                    function which can be evaluated to generate
  50                    either."))
  51
  52 (defgeneric data-export (data target-format target-referencing-type)
  53   (:documentation "write data from stream srce, in format srce-frmt;
  54                    return a reftype, which could be a
  55                    lisp-data-structure, a reference to such, or a lisp
  56                    function which can be evaluated to generate
  57                    either."))
  58
  59
  60
  61
  62 ;;; Potentially useful functions
  63
  64 ;; the following belongs here if we are working externally, but might
  65 ;; belong with data if we are working internlly
  66
  67 ;; (defmacro with-data (body)
  68 ;;   "Stream-handling, maintaining I/O through object typing.")
  69
  70
  71 ;;;
  72 ;;; Related to data file reading
  73 ;;;
  74
  75 (defun count-file-columns (fname)
  76 "Args: (fname)
  77 Returns the number of lisp items on the first nonblank line of file FNAME."
  78   (with-open-file (f fname)
  79     (if f
  80         (let ((line (do ((line (read-line f) (read-line f)))
  81                         ((or (null line) (< 0 (length line))) line))))
  82           (if line
  83               (with-input-from-string (s line)
  84                 (do ((n 0 (+ n 1)) (eof (gensym)))
  85                     ((eq eof (read s nil eof)) n))))))))
  86
  87 (if (not (fboundp 'open-file-dialog))
  88   #+dialogs
  89   (defun open-file-dialog () ;; why?(&optional set)
  90     (get-string-dialog "Enter a data file name:"))
  91   #-dialogs
  92   (defun open-file-dialog () ;; why? (&optional set)
  93     (error "You must provide a file name explicitly")))
  94
  95 (defun read-data-file (&optional (file (open-file-dialog)))
  96 "Args:  (file)
  97 Returns a list of all lisp objects in FILE. FILE can be a string or a symbol,
  98 in which case the symbol'f print name is used."
  99   (if file
 100       (let ((eof (gensym)))
 101         (with-open-file (f file)
 102           (if f
 103               (do* ((r (read f nil eof) (read f nil eof))
 104                     (x (list nil))
 105                     (tail x (cdr tail)))
 106                    ((eq r eof) (cdr x))
 107                    (setf (cdr tail) (list r))))))))
 108
 109 ;;; New definition to avoid stack size limit in apply
 110 #|
 111  (defun read-data-columns (&optional (file (open-file-dialog))
 112                                     (cols (if file
 113                                               (count-file-columns file))))
 114 "Args: (&optional file cols)
 115 Reads the data in FILE as COLS columns and returns a list of lists representing the columns."
 116   (if (and file cols)
 117       (transpose (split-list (read-data-file file) cols))))
 118 |#
 119
 120 ;;; FIXME:AJR:  ALL THE FOLLOWING NEED TO BE SOLVED BY PLATFORM-INDEP PATHNAME WORK!
 121 ;;; FIXME:AJR: use either string or pathname.
 122
 123 (defun path-string-to-path (p s)
 124     (pathname (concatenate 'string (namestring p) s)))
 125
 126 (defun load-data (file)
 127 "Args: (file) as string
 128 Read in data file from the data examples library."
 129   (if (load (path-string-to-path *cls-data-dir* file))
 130       t
 131       (load (path-string-to-path *cls-data-dir* file))))
 132
 133 (defun load-example (file)
 134   "Args: (file) as string
 135 Read in lisp example file from the examples library."
 136   (if (load (path-string-to-path *cls-examples-dir* file))
 137       t
 138       (load (path-string-to-path *cls-examples-dir* file))))
 139
 140 ;;;
 141 ;;; Saving Variables and Functions
 142 ;;;
 143
 144 (defun savevar (vars file)
 145 "Args: (vars file-name-root)
 146 VARS is a symbol or a list of symbols. FILE-NAME-ROOT is a string (or a symbol
 147 whose print name is used) not endinf in .lsp. The VARS and their current values
 148 are written to the file FILE-NAME-ROOT.lsp in a form suitable for use with the
 149 load command."
 150   (with-open-file (f (concatenate 'string (namestring file) ".lsp")
 151                      :direction :output)
 152     (let ((vars (if (consp vars) vars (list vars))))
 153       (flet ((save-one (x)
 154                (let ((v (symbol-value x)))
 155                  (if (objectp v)
 156                      (format f "(def ~s ~s)~%" x (send v :save))
 157                    (format f "(def ~s '~s)~%" x v)))))
 158             (mapcar #'save-one vars))
 159       vars)))
 160
 161
 162
 163 ;;; General modification approaches.
 164
 165 (defgeneric importData (source featureList)
 166   (:documentation "command to get data into CLS.  Specific methods
 167   will need to handle pathnames, internal data structures, and
 168   external services such as DBMS's.  We would like to be able to do
 169   thinks like:
 170      (importData MyPathName '(:formattype 'csvString))
 171      (importData '(sqlConnection :server host.domain.net :port 666)
 172                  '(:formattype 'table
 173   and so on."))
 174
 175
 176 (defun pathname-example (name)
 177   (let ((my-path (parse-namestring name)))
 178     (values (pathname-name my-path :case :common)
 179             (pathname-name my-path :case :local))))
 180
 181 (defvar sourceTypes (list 'csv 'lisp 'tsv 'special)
 182   "list of possible symbols.
 183
 184 Thsees are used to specify source formats that might be supported for
 185 input.  CSV and TSV are standard, LISP refers to forms, and SPECIAL
 186 refers to a FUNCTION which parses as appropriately.")
 187
 188 #|
 189 ;;; WRONG LOGIC.
 190  (defmethod importData ((fileHandle pathname)
 191                        (fmt list)) ;sourceTypes))
 192   "File-based input for data.
 193 Usually used by:
 194  (importData (parse-namestring 'path/to/file')
 195              (list :format 'csv))
 196
 197  (importData myPathName (list :format 'lisp))
 198 ."
 199   (let* ((fmtType (getf fmt :format))
 200          (newData (getDataAsLists fileHandle fmtType)))
 201     (case fmtType
 202       ('csv (  ))
 203       ('tsv (  ))
 204       ('lisp ( ))
 205       ('special (let ((parserFcn (getf fmt :special-parser)))))
 206       (:default (error "no standard default importData format")))))
 207 |#
 208
 209 (defmethod importData ((ds array) (fmt list))
 210   "mapping arrays into CLS data.")
 211
 212 #|
 213  (defmethod importData ((dsSpec DBMSandSQLextract)
 214                        (fmt mappingTypes))
 215   "mapping DBMS into CLS data.")
 216
 217 |#
 218
 219 ;;(defmacro with-dataframe (env &rest progn)
 220 ;;  "Compute using variable names with with.data.frame type semantics.")
 221
 222