csv-mode.el (84342B)
1 ;;; csv-mode.el --- Major mode for editing comma/char separated values -*- lexical-binding: t -*- 2 3 ;; Copyright (C) 2003-2024 Free Software Foundation, Inc 4 5 ;; Author: "Francis J. Wright" <F.J.Wright@qmul.ac.uk> 6 ;; Maintainer: emacs-devel@gnu.org 7 ;; Version: 1.27 8 ;; Package-Requires: ((emacs "27.1") (cl-lib "0.5")) 9 ;; Keywords: convenience 10 11 ;; This package is free software; you can redistribute it and/or modify 12 ;; it under the terms of the GNU General Public License as published by 13 ;; the Free Software Foundation; either version 3, or (at your option) 14 ;; any later version. 15 16 ;; This package is distributed in the hope that it will be useful, 17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of 18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 19 ;; GNU General Public License for more details. 20 21 ;; You should have received a copy of the GNU General Public License 22 ;; along with GNU Emacs. If not, see <https://www.gnu.org/licenses/>. 23 24 ;;; Commentary: 25 26 ;; This package implements CSV mode, a major mode for editing records 27 ;; in a generalized CSV (character-separated values) format. It binds 28 ;; files with prefix ".csv" to `csv-mode' (and ".tsv" to `tsv-mode') in 29 ;; `auto-mode-alist'. 30 31 ;; In CSV mode, the following commands are available: 32 33 ;; - C-c C-s (`csv-sort-fields') and C-c C-n (`csv-sort-numeric-fields') 34 ;; respectively sort lexicographically and numerically on a 35 ;; specified field or column. 36 37 ;; - C-c C-r (`csv-reverse-region') reverses the order. (These 38 ;; commands are based closely on, and use, code in `sort.el'.) 39 40 ;; - C-c C-k (`csv-kill-fields') and C-c C-y (`csv-yank-fields') kill 41 ;; and yank fields or columns, although they do not use the normal 42 ;; kill ring. C-c C-k can kill more than one field at once, but 43 ;; multiple killed fields can be yanked only as a fixed group 44 ;; equivalent to a single field. 45 46 ;; - `csv-align-mode' keeps fields visually aligned, on-the-fly. 47 ;; It truncates fields to a maximum width that can be changed per-column 48 ;; with `csv-align-set-column-width'. 49 ;; Alternatively, C-c C-a (`csv-align-fields') aligns fields into columns 50 ;; and C-c C-u (`csv-unalign-fields') undoes such alignment; 51 ;; separators can be hidden within aligned records (controlled by 52 ;; `csv-invisibility-default' and `csv-toggle-invisibility'). 53 54 ;; - C-c C-t (`csv-transpose') interchanges rows and columns. For 55 ;; details, see the documentation for the individual commands. 56 57 ;; - `csv-set-separator' sets the CSV separator of the current buffer, 58 ;; while `csv-guess-set-separator' guesses and sets the separator 59 ;; based on the current buffer's contents. 60 ;; `csv-guess-set-separator' can be useful to add to the mode hook 61 ;; to have CSV mode guess and set the separator automatically when 62 ;; visiting a buffer: 63 ;; 64 ;; (add-hook 'csv-mode-hook 'csv-guess-set-separator) 65 66 ;; CSV mode can recognize fields separated by any of several single 67 ;; characters, specified by the value of the customizable user option 68 ;; `csv-separators'. CSV data fields can be delimited by quote 69 ;; characters (and must if they contain separator characters). This 70 ;; implementation supports quoted fields, where the quote characters 71 ;; allowed are specified by the value of the customizable user option 72 ;; `csv-field-quotes'. By default, the both commas and tabs are considered 73 ;; as separators and the only field quote is a double quote. 74 ;; These user options can be changed ONLY by customizing them, e.g. via M-x 75 ;; customize-variable. 76 77 ;; CSV mode commands ignore blank lines and comment lines beginning 78 ;; with the value of the buffer local variable `csv-comment-start', 79 ;; which by default is #. The user interface is similar to that of 80 ;; the standard commands `sort-fields' and `sort-numeric-fields', but 81 ;; see the major mode documentation below. 82 83 ;; The global minor mode `csv-field-index-mode' provides display of 84 ;; the current field index in the mode line, cf. `line-number-mode' 85 ;; and `column-number-mode'. It is on by default. 86 87 ;;;; See also: 88 89 ;; the standard GNU Emacs 21 packages align.el, which will align 90 ;; columns within a region, and delim-col.el, which helps to prettify 91 ;; columns in a text region or rectangle; 92 93 ;; csv.el by Ulf Jasper <ulf.jasper at web.de>, which provides 94 ;; functions for reading/parsing comma-separated value files and is 95 ;; available at http://de.geocities.com/ulf_jasper/emacs.html (and in 96 ;; the gnu.emacs.sources archives). 97 98 ;;; Installation: 99 100 ;; Put this file somewhere that Emacs can find it (i.e. in one of the 101 ;; directories in your `load-path' such as `site-lisp'), optionally 102 ;; byte-compile it (recommended), and put this in your .emacs file: 103 ;; 104 ;; (add-to-list 'auto-mode-alist '("\\.[Cc][Ss][Vv]\\'" . csv-mode)) 105 ;; (autoload 'csv-mode "csv-mode" 106 ;; "Major mode for editing comma-separated value files." t) 107 108 ;;; News: 109 110 ;; Since 1.27: 111 ;; - `csv-end-of-field' no longer errors out in the presence of 112 ;; unclosed quotes. 113 114 ;; Since 1.26: 115 ;; - `csv-guess-separator' will no longer guess the comment-start 116 ;; character as a potential separator character. 117 118 ;; Since 1.25: 119 ;; - The ASCII control character 31 Unit Separator can now be 120 ;; recognized as a CSV separator by `csv-guess-separator'. 121 122 ;; Since 1.24: 123 ;; - New function `csv--unquote-value'. 124 ;; - New function `csv-parse-current-row'. 125 126 ;; Since 1.21: 127 ;; - New command `csv-insert-column'. 128 ;; - New config var `csv-align-min-width' for `csv-align-mode'. 129 ;; - New option `csv-confirm-region'. 130 131 ;; Since 1.20: 132 ;; - New command `csv-guess-set-separator' that automatically guesses 133 ;; and sets the CSV separator of the current buffer. 134 ;; - New command `csv-set-separator' for setting the CSV separator 135 ;; manually. 136 137 ;; Since 1.9: 138 ;; - `csv-align-mode' auto-aligns columns dynamically (on screen). 139 140 ;; Before that: 141 ;; Begun on 15 November 2003 to provide lexicographic sorting of 142 ;; simple CSV data by field and released as csv.el. Facilities to 143 ;; kill multiple fields and customize separator added on 9 April 2004. 144 ;; Converted to a major mode and renamed csv-mode.el on 10 April 2004, 145 ;; partly at the suggestion of Stefan Monnier <monnier at 146 ;; IRO.UMontreal.CA> to avoid conflict with csv.el by Ulf Jasper. 147 ;; Field alignment, comment support and CSV mode customization group 148 ;; added on 1 May 2004. Support for index ranges added on 6 June 149 ;; 2004. Multiple field separators added on 12 June 2004. 150 ;; Transposition added on 22 June 2004. Separator invisibility added 151 ;; on 23 June 2004. 152 153 ;;; To do (maybe): 154 155 ;; Make separators and quotes buffer-local and locally settable. 156 ;; Support (La)TeX tables: set separator and comment; support record 157 ;; end string. 158 ;; Convert comma-separated to space- or tab-separated. 159 160 ;;; Code: 161 162 (eval-when-compile 163 (require 'cl-lib) 164 (require 'subr-x)) 165 166 (defgroup CSV nil 167 "Major mode for editing files of comma-separated value type." 168 :group 'convenience) 169 170 (defvar csv-separator-chars nil 171 "Field separators as a list of character. 172 Set by customizing `csv-separators' -- do not set directly!") 173 174 (defvar csv-separator-regexp nil 175 "Regexp to match a field separator. 176 Set by customizing `csv-separators' -- do not set directly!") 177 178 (defvar csv--skip-chars nil 179 "Char set used by `skip-chars-forward' etc. to skip fields. 180 Set by customizing `csv-separators' -- do not set directly!") 181 182 (defvar csv-font-lock-keywords nil 183 "Font lock keywords to highlight the field separators in CSV mode. 184 Set by customizing `csv-separators' -- do not set directly!") 185 186 (defcustom csv-separators '("," "\t") 187 "Field separators: a list of *single-character* strings. 188 For example: (\",\"), the default, or (\",\" \";\" \":\"). 189 Neighbouring fields may be separated by any one of these characters. 190 The first is used when inserting a field separator into the buffer. 191 All must be different from the field quote characters, `csv-field-quotes'. 192 193 Changing this variable with `setq' won't affect the current Emacs 194 session. Use `customize-set-variable' instead if that is required." 195 ;; Suggested by Eckhard Neber <neber@mwt.e-technik.uni-ulm.de> 196 :type '(repeat string) 197 ;; FIXME: Character would be better, but in Emacs 21.3 does not display 198 ;; correctly in a customization buffer. 199 :set (lambda (variable value) 200 (mapc (lambda (x) 201 (if (/= (length x) 1) 202 (error "Non-single-char string %S" x)) 203 (if (and (boundp 'csv-field-quotes) 204 (member x csv-field-quotes)) 205 (error "%S is already a quote" x))) 206 value) 207 (custom-set-default variable value) 208 (setq csv-separator-chars (mapcar #'string-to-char value)) 209 (setq csv--skip-chars 210 (apply #'concat "^\n" 211 (mapcar (lambda (s) (concat "\\" s)) value))) 212 (setq csv-separator-regexp (regexp-opt value)) 213 (setq csv-font-lock-keywords 214 ;; NB: csv-separator-face variable evaluates to itself. 215 `((,csv-separator-regexp (0 'csv-separator-face)))))) 216 217 (defcustom csv-field-quotes '("\"") 218 "Field quotes: a list of *single-character* strings. 219 For example: (\"\\\"\"), the default, or (\"\\\"\" \"\\='\" \"\\=`\"). 220 A field can be delimited by a pair of any of these characters. 221 All must be different from the field separators, `csv-separators'." 222 :type '(repeat string) 223 ;; Character would be better, but in Emacs 21 does not display 224 ;; correctly in a customization buffer. 225 :set (lambda (variable value) 226 (mapc (lambda (x) 227 (if (/= (length x) 1) 228 (error "Non-single-char string %S" x)) 229 (if (member x csv-separators) 230 (error "%S is already a separator" x))) 231 value) 232 (when (boundp 'csv-mode-syntax-table) 233 ;; FIRST remove old quote syntax: 234 (with-syntax-table text-mode-syntax-table 235 (mapc (lambda (x) 236 (modify-syntax-entry 237 (string-to-char x) 238 (string (char-syntax (string-to-char x))) 239 ;; symbol-value to avoid compiler warning: 240 (symbol-value 'csv-mode-syntax-table))) 241 csv-field-quotes)) 242 ;; THEN set new quote syntax: 243 (csv-set-quote-syntax value)) 244 ;; BEFORE setting new value of `csv-field-quotes': 245 (custom-set-default variable value))) 246 247 (defun csv-set-quote-syntax (field-quotes) 248 "Set syntax for field quote characters FIELD-QUOTES to be \"string\". 249 FIELD-QUOTES should be a list of single-character strings." 250 (mapc (lambda (x) 251 (modify-syntax-entry 252 (string-to-char x) "\"" 253 ;; symbol-value to avoid compiler warning: 254 (symbol-value 'csv-mode-syntax-table))) 255 field-quotes)) 256 257 (defvar csv-comment-start nil 258 "String that starts a comment line, or nil if no comment syntax. 259 Such comment lines are ignored by CSV mode commands. 260 This variable is buffer local; its default value is that of 261 `csv-comment-start-default'. It is set by the function 262 `csv-set-comment-start' -- do not set it directly!") 263 264 (make-variable-buffer-local 'csv-comment-start) 265 266 (defcustom csv-comment-start-default "#" 267 "String that starts a comment line, or nil if no comment syntax. 268 Such comment lines are ignored by CSV mode commands. 269 Default value of buffer-local variable `csv-comment-start'. 270 Changing this variable does not affect any existing CSV mode buffer." 271 :type '(choice (const :tag "None" nil) string) 272 :set (lambda (variable value) 273 (custom-set-default variable value) 274 (setq-default csv-comment-start value))) 275 276 (defcustom csv-align-style 'left 277 "Aligned field style: one of `left', `centre', `right' or `auto'. 278 Alignment style used by `csv-align-mode' and `csv-align-fields'. 279 Auto-alignment means left align text and right align numbers." 280 :type '(choice (const left) (const centre) 281 (const right) (const auto))) 282 283 (defcustom csv-align-padding 1 284 "Aligned field spacing: must be a positive integer. 285 Number of spaces used by `csv-align-mode' and `csv-align-fields' 286 after separators." 287 :type 'integer) 288 289 (defcustom csv-header-lines 0 290 "Header lines to skip when setting region automatically." 291 :type 'integer) 292 293 (defcustom csv-invisibility-default t 294 "If non-nil, make separators in aligned records invisible." 295 :type 'boolean) 296 297 (defcustom csv-confirm-region t 298 "If non-nil, confirm that region is OK in interactive commands." 299 :type 'boolean) 300 301 (defface csv-separator-face 302 '((t :inherit escape-glyph)) 303 "CSV mode face used to highlight separators.") 304 305 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 306 ;;; Mode definition, key bindings and menu 307 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 308 309 310 (defconst csv-mode-line-format 311 '(csv-field-index-string ("" csv-field-index-string)) 312 "Mode line format string for CSV mode.") 313 314 (defvar csv-mode-map 315 (let ((map (make-sparse-keymap))) 316 (define-key map [(control ?c) (control ?v)] #'csv-toggle-invisibility) 317 (define-key map [(control ?c) (control ?t)] #'csv-transpose) 318 (define-key map [(control ?c) (control ?c)] #'csv-set-comment-start) 319 (define-key map [(control ?c) (control ?u)] #'csv-unalign-fields) 320 (define-key map [(control ?c) (control ?a)] #'csv-align-fields) 321 (define-key map [(control ?c) (control ?z)] #'csv-yank-as-new-table) 322 (define-key map [(control ?c) (control ?y)] #'csv-yank-fields) 323 (define-key map [(control ?c) (control ?k)] #'csv-kill-fields) 324 (define-key map [(control ?c) (control ?d)] #'csv-toggle-descending) 325 (define-key map [(control ?c) (control ?r)] #'csv-reverse-region) 326 (define-key map [(control ?c) (control ?n)] #'csv-sort-numeric-fields) 327 (define-key map [(control ?c) (control ?s)] #'csv-sort-fields) 328 (define-key map "\t" #'csv-tab-command) 329 (define-key map [backtab] #'csv-backtab-command) 330 map)) 331 332 ;;;###autoload 333 (define-derived-mode csv-mode text-mode "CSV" 334 "Major mode for editing files of comma-separated value type. 335 336 CSV mode is derived from `text-mode', and runs `text-mode-hook' before 337 running `csv-mode-hook'. It turns `auto-fill-mode' off by default. 338 CSV mode can be customized by user options in the CSV customization 339 group. The separators are specified by the value of `csv-separators'. 340 341 CSV mode commands ignore blank lines and comment lines beginning with 342 the value of `csv-comment-start', which delimit \"paragraphs\". 343 \"Sexp\" is re-interpreted to mean \"field\", so that `forward-sexp' 344 \(\\[forward-sexp]), `kill-sexp' (\\[kill-sexp]), etc. all apply to fields. 345 Standard comment commands apply, such as `comment-dwim' (\\[comment-dwim]). 346 347 If `font-lock-mode' is enabled then separators, quoted values and 348 comment lines are highlighted using respectively `csv-separator-face', 349 `font-lock-string-face' and `font-lock-comment-face'. 350 351 The user interface (UI) for CSV mode commands is similar to that of 352 the standard commands `sort-fields' and `sort-numeric-fields', except 353 that if there is no prefix argument then the UI prompts for the field 354 index or indices. In `transient-mark-mode' only: if the region is not 355 set then the UI attempts to set it to include all consecutive CSV 356 records around point, and prompts for confirmation; if there is no 357 prefix argument then the UI prompts for it, offering as a default the 358 index of the field containing point if the region was not set 359 explicitly. The region set automatically is delimited by blank lines 360 and comment lines, and the number of header lines at the beginning of 361 the region given by the value of `csv-header-lines' are skipped. 362 363 Sort order is controlled by `csv-descending'. 364 365 CSV mode provides the following specific keyboard key bindings: 366 367 \\{csv-mode-map}" 368 :group 'CSV 369 ;; We used to `turn-off-auto-fill' here instead, but that's not very 370 ;; effective since text-mode-hook is run afterwards anyway! 371 (setq-local normal-auto-fill-function nil) 372 ;; Set syntax for field quotes: 373 (csv-set-quote-syntax csv-field-quotes) 374 ;; Make sexp functions apply to fields: 375 (set (make-local-variable 'forward-sexp-function) #'csv-forward-field) 376 (csv-set-comment-start csv-comment-start) 377 ;; Font locking -- separator plus syntactic: 378 (setq font-lock-defaults '(csv-font-lock-keywords)) 379 (setq-local jit-lock-contextually nil) ;Each line should be independent. 380 (if csv-invisibility-default (add-to-invisibility-spec 'csv)) 381 ;; Mode line to support `csv-field-index-mode': 382 (set (make-local-variable 'mode-line-position) 383 (pcase mode-line-position 384 (`(,(or (pred consp) (pred stringp)) . ,_) 385 `(,@mode-line-position ,csv-mode-line-format)) 386 (_ `("" ,mode-line-position ,csv-mode-line-format)))) 387 (set (make-local-variable 'truncate-lines) t) 388 ;; Enable or disable `csv-field-index-mode' (could probably do this 389 ;; a bit more efficiently): 390 (csv-field-index-mode (symbol-value 'csv-field-index-mode))) 391 392 (defun csv-set-comment-start (string) 393 "Set comment start for this CSV mode buffer to STRING. 394 It must be either a string or nil." 395 (interactive 396 (list (edit-and-eval-command 397 "Comment start (string or nil): " csv-comment-start))) 398 ;; Paragraph means a group of contiguous records: 399 (set (make-local-variable 'paragraph-separate) "[[:space:]]*$") ; White space. 400 (set (make-local-variable 'paragraph-start) "\n");Must include \n explicitly! 401 ;; Remove old comment-start/end if available 402 (with-syntax-table text-mode-syntax-table 403 (when comment-start 404 (modify-syntax-entry (string-to-char comment-start) 405 (string (char-syntax (string-to-char comment-start))) 406 csv-mode-syntax-table)) 407 (modify-syntax-entry ?\n 408 (string (char-syntax ?\n)) 409 csv-mode-syntax-table)) 410 (when string 411 (setq paragraph-separate (concat paragraph-separate "\\|" string) 412 paragraph-start (concat paragraph-start "\\|" string)) 413 (set (make-local-variable 'comment-start) string) 414 (modify-syntax-entry 415 (string-to-char string) "<" csv-mode-syntax-table) 416 (modify-syntax-entry ?\n ">" csv-mode-syntax-table)) 417 (setq csv-comment-start string)) 418 419 (defvar csv--set-separator-history nil) 420 421 (defun csv-set-separator (sep) 422 "Set the CSV separator in the current buffer to SEP." 423 (interactive (list (read-char-from-minibuffer 424 "Separator: " nil 'csv--set-separator-history))) 425 (when (and (boundp 'csv-field-quotes) 426 (member (string sep) csv-field-quotes)) 427 (error "%c is already a quote" sep)) 428 (setq-local csv-separators (list (string sep))) 429 (setq-local csv-separator-chars (list sep)) 430 (setq-local csv--skip-chars (format "^\n\\%c" sep)) 431 (setq-local csv-separator-regexp (regexp-quote (string sep))) 432 (setq-local csv-font-lock-keywords 433 `((,csv-separator-regexp (0 'csv-separator-face)))) 434 (font-lock-refresh-defaults)) 435 436 ;;;###autoload 437 (add-to-list 'auto-mode-alist '("\\.[Cc][Ss][Vv]\\'" . csv-mode)) 438 439 (defvar csv-descending nil 440 "If non-nil, CSV mode sort functions sort in order of descending sort key. 441 Usually they sort in order of ascending sort key.") 442 443 (defun csv-toggle-descending () 444 "Toggle `csv-descending'." 445 (interactive) 446 (setq csv-descending (not csv-descending)) 447 (message "Sort order is %sscending" (if csv-descending "de" "a"))) 448 449 (defun csv-toggle-invisibility () 450 ;; FIXME: Make it into a proper minor mode? 451 "Toggle `buffer-invisibility-spec'." 452 (interactive) 453 (if (memq 'csv buffer-invisibility-spec) 454 (remove-from-invisibility-spec 'csv) 455 (add-to-invisibility-spec 'csv)) 456 (message "Separators in aligned records will be %svisible \ 457 \(after re-aligning if soft)" 458 (if (memq 'csv buffer-invisibility-spec) "in" "")) 459 (redraw-frame (selected-frame))) 460 461 (easy-menu-define 462 csv-menu 463 csv-mode-map 464 "CSV major mode menu keymap" 465 '("CSV" 466 ["Sort By Field Lexicographically" csv-sort-fields :active t 467 :help "Sort lines in region lexicographically by the specified field"] 468 ["Sort By Field Numerically" csv-sort-numeric-fields :active t 469 :help "Sort lines in region numerically by the specified field"] 470 ["Reverse Order of Lines" csv-reverse-region :active t 471 :help "Reverse the order of the lines in the region"] 472 ["Use Descending Sort Order" csv-toggle-descending :active t 473 :style toggle :selected csv-descending 474 :help "If selected, use descending order when sorting"] 475 "--" 476 ["Kill Fields (Columns)" csv-kill-fields :active t 477 :help "Kill specified fields of each line in the region"] 478 ["Yank Fields (Columns)" csv-yank-fields :active t 479 :help "Yank killed fields as specified field of each line in region"] 480 ["Yank As New Table" csv-yank-as-new-table :active t 481 :help "Yank killed fields as a new table at point"] 482 ["Align Fields into Columns" csv-align-fields :active t 483 :help "Align the start of every field of each line in the region"] 484 ["Unalign Columns into Fields" csv-unalign-fields :active t 485 :help "Undo soft alignment and optionally remove redundant white space"] 486 ["Transpose Rows and Columns" csv-transpose :active t 487 :help "Rewrite rows (which may have different lengths) as columns"] 488 "--" 489 ["Forward Field" forward-sexp :active t 490 :help "Move forward across one field; with ARG, do it that many times"] 491 ["Backward Field" backward-sexp :active t 492 :help "Move backward across one field; with ARG, do it that many times"] 493 ["Kill Field Forward" kill-sexp :active t 494 :help "Kill field following cursor; with ARG, do it that many times"] 495 ["Kill Field Backward" backward-kill-sexp :active t 496 :help "Kill field preceding cursor; with ARG, do it that many times"] 497 "--" 498 ("Alignment Style" 499 ["Left" (setq csv-align-style 'left) :active t 500 :style radio :selected (eq csv-align-style 'left) 501 :help "If selected, `csv-align' left aligns fields"] 502 ["Centre" (setq csv-align-style 'centre) :active t 503 :style radio :selected (eq csv-align-style 'centre) 504 :help "If selected, `csv-align' centres fields"] 505 ["Right" (setq csv-align-style 'right) :active t 506 :style radio :selected (eq csv-align-style 'right) 507 :help "If selected, `csv-align' right aligns fields"] 508 ["Auto" (setq csv-align-style 'auto) :active t 509 :style radio :selected (eq csv-align-style 'auto) 510 :help "\ 511 If selected, `csv-align' left aligns text and right aligns numbers"] 512 ) 513 ["Set header line" csv-header-line :active t] 514 ["Auto-(re)align fields" csv-align-mode 515 :style toggle :selected csv-align-mode] 516 ["Show Current Field Index" csv-field-index-mode :active t 517 :style toggle :selected csv-field-index-mode 518 :help "If selected, display current field index in mode line"] 519 ["Make Separators Invisible" csv-toggle-invisibility :active t 520 :style toggle :selected (memq 'csv buffer-invisibility-spec) 521 :visible (not (tsv--mode-p)) 522 :help "If selected, separators in aligned records are invisible"] 523 ["Set Buffer's Comment Start" csv-set-comment-start :active t 524 :help "Set comment start string for this buffer"] 525 ["Customize CSV Mode" (customize-group 'CSV) :active t 526 :help "Open a customization buffer to change CSV mode options"] 527 )) 528 529 (require 'sort) 530 531 (defsubst csv-not-looking-at-record () 532 "Return t if looking at blank or comment line, nil otherwise. 533 Assumes point is at beginning of line." 534 (looking-at paragraph-separate)) 535 536 (defun csv-interactive-args (&optional type) 537 "Get arg or field(s) and region interactively, offering sensible defaults. 538 Signal an error if the buffer is read-only. 539 If TYPE is noarg then return a list (beg end). 540 Otherwise, return a list (arg beg end), where arg is: 541 the raw prefix argument by default; 542 a single field index if TYPE is single; 543 a list of field indices or index ranges if TYPE is multiple. 544 Field defaults to the current prefix arg; if not set, prompt user. 545 546 A field index list consists of positive or negative integers or ranges, 547 separated by any non-integer characters. A range has the form m-n, 548 where m and n are positive or negative integers, m < n, and n defaults 549 to the last field index if omitted. 550 551 In transient mark mode, if the mark is not active then automatically 552 select and highlight CSV records around point, and query user. 553 The default field when read interactively is the current field." 554 ;; Must be run interactively to activate mark! 555 (let* ((arg current-prefix-arg) (default-field 1) 556 (region 557 (if (not (use-region-p)) 558 ;; Set region automatically: 559 (save-excursion 560 (if arg 561 (beginning-of-line) 562 (let ((lbp (line-beginning-position))) 563 (while (re-search-backward csv-separator-regexp lbp 1) 564 ;; Move as far as possible, i.e. to beginning of line. 565 (setq default-field (1+ default-field))))) 566 (if (csv-not-looking-at-record) 567 (error "Point must be within CSV records")) 568 (let ((startline (point))) 569 ;; Set mark at beginning of region: 570 (while (not (or (bobp) (csv-not-looking-at-record))) 571 (forward-line -1)) 572 (if (csv-not-looking-at-record) (forward-line 1)) 573 ;; Skip header lines: 574 (forward-line csv-header-lines) 575 (set-mark (point)) ; OK since in save-excursion 576 ;; Move point to end of region: 577 (goto-char startline) 578 (beginning-of-line) 579 (while (not (or (eobp) (csv-not-looking-at-record))) 580 (forward-line 1)) 581 ;; Show mark briefly if necessary: 582 (unless (and (pos-visible-in-window-p) 583 (pos-visible-in-window-p (mark))) 584 (exchange-point-and-mark) 585 (sit-for 1) 586 (exchange-point-and-mark)) 587 (when csv-confirm-region 588 (or (y-or-n-p "Region OK? ") 589 (error "Action aborted by user")) 590 (message nil)) ; clear y-or-n-p message 591 (list (region-beginning) (region-end)))) 592 ;; Use region set by user: 593 (list (region-beginning) (region-end))))) 594 (setq default-field (number-to-string default-field)) 595 (cond 596 ((eq type 'multiple) 597 (if arg 598 ;; Ensure that field is a list: 599 (or (consp arg) 600 (setq arg (list (prefix-numeric-value arg)))) 601 ;; Read field interactively, ignoring non-integers: 602 (setq arg 603 (mapcar 604 (lambda (x) 605 (if (string-match "-" x 1) ; not first character 606 ;; Return a range as a pair - the cdr may be nil: 607 (let ((m (substring x 0 (match-beginning 0))) 608 (n (substring x (match-end 0)))) 609 (cons (car (read-from-string m)) 610 (and (not (string= n "")) 611 (car (read-from-string n))))) 612 ;; Return a number as a number: 613 (car (read-from-string x)))) 614 (split-string 615 (read-string 616 "Fields (sequence of integers or ranges): " default-field) 617 "[^-+0-9]+"))))) 618 ((eq type 'single) 619 (if arg 620 (setq arg (prefix-numeric-value arg)) 621 (while (not (integerp arg)) 622 (setq arg (eval-minibuffer "Field (integer): " default-field)))))) 623 (if (eq type 'noarg) region (cons arg region)))) 624 625 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 626 ;;; Sorting by field 627 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 628 629 (defun csv-nextrecfun () 630 "Called by `csv-sort-fields-1' with point at end of previous record. 631 It moves point to the start of the next record. 632 It should move point to the end of the buffer if there are no more records." 633 (forward-line) 634 (while (and (not (eobp)) (csv-not-looking-at-record)) 635 (forward-line))) 636 637 (defun csv-sort-fields-1 (field beg end startkeyfun endkeyfun) 638 "Modified version of `sort-fields-1' that skips blank or comment lines. 639 640 FIELD is a single field index, and BEG and END specify the region to 641 sort. 642 643 STARTKEYFUN moves from the start of the record to the start of the key. 644 It may return either a non-nil value to be used as the key, or 645 else the key is the substring between the values of point after 646 STARTKEYFUN and ENDKEYFUN are called. If STARTKEYFUN is nil, the key 647 starts at the beginning of the record. 648 649 ENDKEYFUN moves from the start of the sort key to the end of the sort key. 650 ENDKEYFUN may be nil if STARTKEYFUN returns a value or if it would be the 651 same as ENDRECFUN." 652 (let ((tbl (syntax-table))) 653 (if (zerop field) (setq field 1)) 654 (unwind-protect 655 (save-excursion 656 (save-restriction 657 (narrow-to-region beg end) 658 (goto-char (point-min)) 659 (set-syntax-table sort-fields-syntax-table) 660 (sort-subr csv-descending 661 'csv-nextrecfun 'end-of-line 662 startkeyfun endkeyfun))) 663 (set-syntax-table tbl)))) 664 665 (defun csv-sort-fields (field beg end) 666 "Sort lines in region lexicographically by the ARGth field of each line. 667 If not set, the region defaults to the CSV records around point. 668 Fields are separated by `csv-separators' and null fields are allowed anywhere. 669 Field indices increase from 1 on the left or decrease from -1 on the right. 670 A prefix argument specifies a single field, otherwise prompt for field index. 671 Ignore blank and comment lines. The variable `sort-fold-case' 672 determines whether alphabetic case affects the sort order. 673 When called non-interactively, FIELD is a single field index; 674 BEG and END specify the region to sort." 675 ;; (interactive "*P\nr") 676 (interactive (csv-interactive-args 'single)) 677 (barf-if-buffer-read-only) 678 (csv-sort-fields-1 field beg end 679 (lambda () (csv-sort-skip-fields field) nil) 680 (lambda () (skip-chars-forward csv--skip-chars)))) 681 682 (defun csv-sort-numeric-fields (field beg end) 683 "Sort lines in region numerically by the ARGth field of each line. 684 If not set, the region defaults to the CSV records around point. 685 Fields are separated by `csv-separators'. 686 Null fields are allowed anywhere and sort as zeros. 687 Field indices increase from 1 on the left or decrease from -1 on the right. 688 A prefix argument specifies a single field, otherwise prompt for field index. 689 Specified non-null field must contain a number in each line of the region, 690 which may begin with \"0x\" or \"0\" for hexadecimal and octal values. 691 Otherwise, the number is interpreted according to sort-numeric-base. 692 Ignore blank and comment lines. 693 When called non-interactively, FIELD is a single field index; 694 BEG and END specify the region to sort." 695 ;; (interactive "*P\nr") 696 (interactive (csv-interactive-args 'single)) 697 (barf-if-buffer-read-only) 698 (csv-sort-fields-1 field beg end 699 (lambda () 700 (csv-sort-skip-fields field) 701 (let* ((case-fold-search t) 702 (base 703 (if (looking-at "\\(0x\\)[0-9a-f]\\|\\(0\\)[0-7]") 704 (cond ((match-beginning 1) 705 (goto-char (match-end 1)) 706 16) 707 ((match-beginning 2) 708 (goto-char (match-end 2)) 709 8) 710 (t nil))))) 711 (string-to-number (buffer-substring (point) 712 (save-excursion 713 (forward-sexp 1) 714 (point))) 715 (or base sort-numeric-base)))) 716 nil)) 717 718 (defun csv-reverse-region (beg end) 719 "Reverse the order of the lines in the region. 720 This is just a CSV-mode style interface to `reverse-region', which is 721 the function that should be used non-interactively. It takes two 722 point or marker arguments, BEG and END, delimiting the region." 723 ;; (interactive "*P\nr") 724 (interactive (csv-interactive-args 'noarg)) 725 (barf-if-buffer-read-only) 726 (reverse-region beg end)) 727 728 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 729 ;;; Moving by field 730 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 731 732 (defun csv-end-of-field () 733 "Skip forward over one field." 734 (skip-chars-forward " ") 735 ;; If the first character is a double quote, then we have a quoted 736 ;; value. 737 (when (eq (char-syntax (following-char)) ?\") 738 (forward-char) 739 (let ((ended nil)) 740 (while (and (not ended) (not (eolp))) 741 (cond ((not (eq (char-syntax (following-char)) ?\")) 742 (forward-char 1)) 743 ;; According to RFC-4180 (sec 2.7), quotes inside quoted strings 744 ;; are quoted by doubling the quote char: a,"b""c,",d 745 ;; FIXME: Maybe we should handle this via syntax-propertize? 746 ((let ((c (char-after (1+ (point))))) 747 (and c (eq (char-syntax c) ?\"))) 748 (forward-char 2)) 749 (t 750 (setq ended t)))))) 751 (skip-chars-forward csv--skip-chars)) 752 753 (defun csv--bof-p () 754 (or (bolp) 755 (memq (preceding-char) csv-separator-chars))) 756 757 (defun csv--eof-p () 758 (or (eolp) 759 (memq (following-char) csv-separator-chars))) 760 761 (defun csv-beginning-of-field () 762 "Skip backward over one field." 763 (skip-syntax-backward " ") 764 (if (eq (char-syntax (preceding-char)) ?\") 765 (goto-char (scan-sexps (point) -1))) 766 (skip-chars-backward csv--skip-chars)) 767 768 (defun csv-forward-field (arg) 769 "Move forward across one field, cf. `forward-sexp'. 770 With ARG, do it that many times. Negative arg -N means 771 move backward across N fields." 772 (interactive "p") 773 (if (< arg 0) 774 (csv-backward-field (- arg)) 775 (while (>= (setq arg (1- arg)) 0) 776 (if (or (bolp) 777 (when (and (not (eobp)) (eolp)) (forward-char) t)) 778 (while (and (not (eobp)) (csv-not-looking-at-record)) 779 (forward-line 1))) 780 (if (memq (following-char) csv-separator-chars) (forward-char)) 781 (csv-end-of-field)))) 782 783 (defun csv-backward-field (arg) 784 "Move backward across one field, cf. `backward-sexp'. 785 With ARG, do it that many times. Negative arg -N means 786 move forward across N fields." 787 (interactive "p") 788 (if (< arg 0) 789 (csv-forward-field (- arg)) 790 (while (>= (setq arg (1- arg)) 0) 791 (when (or (eolp) 792 (when (and (not (bobp)) (bolp)) (backward-char) t)) 793 (while (progn 794 (beginning-of-line) 795 (csv-not-looking-at-record)) 796 (backward-char)) 797 (end-of-line)) 798 (if (memq (preceding-char) csv-separator-chars) (backward-char)) 799 (csv-beginning-of-field)))) 800 801 (defun csv-tab-command () 802 "Skip to the next field on the same line. 803 Create a new field at end of line, if needed." 804 (interactive) 805 (skip-chars-forward csv--skip-chars) 806 (if (eolp) 807 (insert (car csv-separators)) 808 (forward-char 1))) 809 810 (defun csv-backtab-command () 811 "Skip to the beginning of the previous field." 812 (interactive) 813 (skip-chars-backward csv--skip-chars) 814 (forward-char -1) 815 (skip-chars-backward csv--skip-chars)) 816 817 (defun csv-sort-skip-fields (n &optional yank) 818 "Position point at the beginning of field N on the current line. 819 Fields are separated by `csv-separators'; null terminal field allowed. 820 Assumes point is initially at the beginning of the line. 821 YANK non-nil allows N to be greater than the number of fields, in 822 which case extend the record as necessary." 823 (if (> n 0) 824 ;; Skip across N - 1 fields. 825 (let ((i (1- n))) 826 (while (> i 0) 827 (csv-end-of-field) 828 (if (eolp) 829 (if yank 830 (if (> i 1) (insert (car csv-separators))) 831 (error "Line has too few fields: %s" 832 (buffer-substring 833 (save-excursion (beginning-of-line) (point)) 834 (save-excursion (end-of-line) (point))))) 835 (forward-char)) ; skip separator 836 (setq i (1- i)))) 837 (end-of-line) 838 ;; Skip back across -N - 1 fields. 839 (let ((i (1- (- n)))) 840 (while (> i 0) 841 (csv-beginning-of-field) 842 (if (bolp) 843 (error "Line has too few fields: %s" 844 (buffer-substring 845 (save-excursion (beginning-of-line) (point)) 846 (save-excursion (end-of-line) (point))))) 847 (backward-char) ; skip separator 848 (setq i (1- i))) 849 ;; Position at the front of the field 850 ;; even if moving backwards. 851 (csv-beginning-of-field)))) 852 853 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 854 ;;; Field index mode 855 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 856 857 ;; Based partly on paren.el 858 859 (defcustom csv-field-index-delay 0.125 860 "Time in seconds to delay before updating field index display." 861 :type '(number :tag "seconds")) 862 863 (defvar csv-field-index-idle-timer nil) 864 865 (defvar csv-field-index-string nil) 866 (make-variable-buffer-local 'csv-field-index-string) 867 868 (defvar csv-field-index-old nil) 869 (make-variable-buffer-local 'csv-field-index-old) 870 871 (define-minor-mode csv-field-index-mode 872 "Toggle CSV-Field-Index mode. 873 With prefix ARG, turn CSV-Field-Index mode on if and only if ARG is positive. 874 Returns the new status of CSV-Field-Index mode (non-nil means on). 875 When CSV-Field-Index mode is enabled, the current field index appears in 876 the mode line after `csv-field-index-delay' seconds of Emacs idle time." 877 :global t 878 :init-value t ; for documentation, since default is t 879 ;; This macro generates a function that first sets the mode 880 ;; variable, then runs the following code, runs the mode hooks, 881 ;; displays a message if interactive, updates the mode line and 882 ;; finally returns the variable value. 883 884 ;; First, always disable the mechanism (to avoid having two timers): 885 (when csv-field-index-idle-timer 886 (cancel-timer csv-field-index-idle-timer) 887 (setq csv-field-index-idle-timer nil)) 888 ;; Now, if the mode is on and any buffer is in CSV mode then 889 ;; re-initialize and enable the mechanism by setting up a new timer: 890 (if csv-field-index-mode 891 (if (memq t (mapcar (lambda (buffer) 892 (with-current-buffer buffer 893 (when (derived-mode-p 'csv-mode) 894 (setq csv-field-index-string nil 895 csv-field-index-old nil) 896 t))) 897 (buffer-list))) 898 (setq csv-field-index-idle-timer 899 (run-with-idle-timer csv-field-index-delay t 900 #'csv-field-index))) 901 ;; but if the mode is off then remove the display from the mode 902 ;; lines of all CSV buffers: 903 (mapc (lambda (buffer) 904 (with-current-buffer buffer 905 (when (derived-mode-p 'csv-mode) 906 (setq csv-field-index-string nil 907 csv-field-index-old nil) 908 (force-mode-line-update)))) 909 (buffer-list)))) 910 911 (defun csv--field-index () 912 (save-excursion 913 (let ((start (point)) 914 (field 0)) 915 (beginning-of-line) 916 (while (and (<= (point) start) 917 (not (eolp))) 918 (csv-end-of-field) 919 (unless (eolp) 920 (forward-char 1)) 921 (setq field (1+ field))) 922 field))) 923 924 (defun csv-field-index () 925 "Construct `csv-field-index-string' to display in mode line. 926 Called by `csv-field-index-idle-timer'." 927 (if (derived-mode-p 'csv-mode) 928 (let ((field (csv--field-index))) 929 (when (not (eq field csv-field-index-old)) 930 (setq csv-field-index-old field 931 csv-field-index-string 932 (and field (format "F%d" field))) 933 (force-mode-line-update))))) 934 935 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 936 ;;; Killing and yanking fields 937 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 938 939 (defvar csv-killed-fields nil 940 "A list of the fields or sub-records last killed by `csv-kill-fields'.") 941 942 (defun csv-kill-fields (fields beg end) 943 "Kill specified fields of each line in the region. 944 If not set, the region defaults to the CSV records around point. 945 Fields are separated by `csv-separators' and null fields are allowed anywhere. 946 Field indices increase from 1 on the left or decrease from -1 on the right. 947 The fields are stored for use by `csv-yank-fields'. Fields can be 948 specified in any order but are saved in increasing index order. 949 Ignore blank and comment lines. 950 951 When called interactively, a prefix argument specifies a single field, 952 otherwise prompt for a field list, which may include ranges in the form 953 m-n, where m < n and n defaults to the last field index if omitted. 954 955 When called non-interactively, FIELDS is a single field index or a 956 list of field indices, with ranges specified as (m.n) or (m), and BEG 957 and END specify the region to process." 958 ;; (interactive "*P\nr") 959 (interactive (csv-interactive-args 'multiple)) 960 (barf-if-buffer-read-only) 961 ;; Kill the field(s): 962 (setq csv-killed-fields nil) 963 (save-excursion 964 (save-restriction 965 (narrow-to-region beg end) 966 (goto-char (point-min)) 967 (if (or (cdr fields) (consp (car fields))) 968 (csv-kill-many-columns fields) 969 (csv-kill-one-column (car fields))))) 970 (setq csv-killed-fields (nreverse csv-killed-fields))) 971 972 (defun csv-kill-one-field (field) 973 "Kill field with index FIELD in current line. 974 Return killed text. Assumes point is at beginning of line." 975 ;; Move to start of field to kill: 976 (csv-sort-skip-fields field) 977 ;; Kill to end of field (cf. `kill-region'): 978 (prog1 (delete-and-extract-region 979 (point) 980 (progn (csv-end-of-field) (point))) 981 (if (eolp) 982 (unless (bolp) (delete-char -1)) ; Delete trailing separator at eol 983 (delete-char 1)))) ; or following separator otherwise. 984 985 (defun csv-kill-one-column (field) 986 "Kill field with index FIELD in all lines in (narrowed) buffer. 987 Save killed fields in `csv-killed-fields'. 988 Assumes point is at `point-min'. Called by `csv-kill-fields'. 989 Ignore blank and comment lines." 990 (while (not (eobp)) 991 (or (csv-not-looking-at-record) 992 (push (csv-kill-one-field field) csv-killed-fields)) 993 (forward-line))) 994 995 (defun csv-insert-column (field) 996 "Insert an empty column at point." 997 (interactive 998 (let ((cur (csv--field-index))) 999 (list (if (and (csv--eof-p) (not (csv--bof-p))) (1+ cur) cur)))) 1000 (save-excursion 1001 (goto-char (point-min)) 1002 (while (not (eobp)) 1003 (or (csv-not-looking-at-record) 1004 (progn 1005 (csv-sort-skip-fields field t) 1006 (insert (car csv-separators)))) 1007 (forward-line 1)) 1008 (csv--jit-flush-columns))) 1009 1010 (defun csv-kill-many-columns (fields) 1011 "Kill several fields in all lines in (narrowed) buffer. 1012 FIELDS is an unordered list of field indices. 1013 Save killed fields in increasing index order in `csv-killed-fields'. 1014 Assumes point is at `point-min'. Called by `csv-kill-fields'. 1015 Ignore blank and comment lines." 1016 (if (eolp) (error "First record is empty")) 1017 ;; Convert non-positive to positive field numbers: 1018 (let ((last 1) (f fields)) 1019 (csv-end-of-field) 1020 (while (not (eolp)) 1021 (forward-char) ; skip separator 1022 (csv-end-of-field) 1023 (setq last (1+ last))) ; last = # fields in first record 1024 (while f 1025 (cond ((consp (car f)) 1026 ;; Expand a field range: (m.n) -> m m+1 ... n-1 n. 1027 ;; If n is nil then it defaults to the number of fields. 1028 (let* ((range (car f)) (cdrf (cdr f)) 1029 (m (car range)) (n (cdr range))) 1030 (if (< m 0) (setq m (+ m last 1))) 1031 (if n 1032 (if (< n 0) (setq n (+ n last 1))) 1033 (setq n last)) 1034 (setq range (list n)) 1035 (while (> n m) (push (setq n (1- n)) range)) 1036 (setcar f (car range)) 1037 (setcdr f (cdr range)) 1038 (setcdr (setq f (last range)) cdrf))) 1039 ((zerop (car f)) (setcar f 1)) 1040 ((< (car f) 0) (setcar f (+ f last 1)))) 1041 (setq f (cdr f)))) 1042 (goto-char (point-min)) 1043 ;; Kill from right to avoid miscounting: 1044 (setq fields (sort fields #'>)) 1045 (while (not (eobp)) 1046 (or (csv-not-looking-at-record) 1047 (let ((fields fields) killed-fields field) 1048 (while fields 1049 (setq field (car fields) 1050 fields (cdr fields)) 1051 (beginning-of-line) 1052 (push (csv-kill-one-field field) killed-fields)) 1053 (push (mapconcat #'identity killed-fields (car csv-separators)) 1054 csv-killed-fields))) 1055 (forward-line))) 1056 1057 (defun csv-yank-fields (field beg end) 1058 "Yank fields as the ARGth field of each line in the region. 1059 ARG may be arbitrarily large and records are extended as necessary. 1060 If not set, the region defaults to the CSV records around point; 1061 if point is not in a CSV record then offer to yank as a new table. 1062 The fields yanked are those last killed by `csv-kill-fields'. 1063 Fields are separated by `csv-separators' and null fields are allowed anywhere. 1064 Field indices increase from 1 on the left or decrease from -1 on the right. 1065 A prefix argument specifies a single field, otherwise prompt for field index. 1066 Ignore blank and comment lines. When called non-interactively, FIELD 1067 is a single field index; BEG and END specify the region to process." 1068 ;; (interactive "*P\nr") 1069 (interactive (condition-case err 1070 (csv-interactive-args 'single) 1071 (error (list nil nil err)))) 1072 (barf-if-buffer-read-only) 1073 (if (null beg) 1074 (if (y-or-n-p (concat (error-message-string end) 1075 ". Yank as a new table? ")) 1076 (csv-yank-as-new-table) 1077 (error (error-message-string end))) 1078 (if (<= field 0) (setq field (1+ field))) 1079 (save-excursion 1080 (save-restriction 1081 (narrow-to-region beg end) 1082 (goto-char (point-min)) 1083 (let ((fields csv-killed-fields)) 1084 (while (not (eobp)) 1085 (unless (csv-not-looking-at-record) 1086 ;; Yank at start of specified field if possible, 1087 ;; otherwise yank at end of record: 1088 (if (zerop field) 1089 (end-of-line) 1090 (csv-sort-skip-fields field 'yank)) 1091 (and (eolp) (insert (car csv-separators))) 1092 (when fields 1093 (insert (car fields)) 1094 (setq fields (cdr fields))) 1095 (or (eolp) (insert (car csv-separators)))) 1096 (forward-line))))))) 1097 1098 (defun csv-yank-as-new-table () 1099 "Yank fields as a new table starting at point. 1100 The fields yanked are those last killed by `csv-kill-fields'." 1101 (interactive "*") 1102 (let ((fields csv-killed-fields)) 1103 (while fields 1104 (insert (car fields) ?\n) 1105 (setq fields (cdr fields))))) 1106 1107 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1108 ;;; Aligning fields 1109 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1110 1111 (defun csv--make-overlay (beg end &optional buffer front-advance rear-advance props) 1112 (let ((o (make-overlay beg end buffer front-advance rear-advance))) 1113 (overlay-put o 'csv t) 1114 (while props 1115 (overlay-put o (pop props) (pop props))) 1116 o)) 1117 1118 (defun csv--delete-overlay (o) 1119 (and (overlay-get o 'csv) (delete-overlay o))) 1120 1121 (defun csv--column-widths (beg end) 1122 "Return a list of two lists (COLUMN-WIDTHS FIELD-WIDTHS). 1123 COLUMN-WIDTHS is a list of elements (WIDTH START END) 1124 indicating the widths of the columns after point (and the position of the 1125 widest field that determined the overall width). 1126 FIELD-WIDTHS contains the widths of each individual field after 1127 point." 1128 (let ((column-widths '()) 1129 (field-widths '())) 1130 (goto-char beg) 1131 ;; Construct list of column widths: 1132 (while (< (point) end) ; for each record... 1133 (or (csv-not-looking-at-record) 1134 (let ((w column-widths) 1135 (col (current-column)) 1136 (beg (point)) 1137 field-width) 1138 (while (not (eolp)) 1139 (csv-end-of-field) 1140 (setq field-width (- (current-column) col)) 1141 (push field-width field-widths) 1142 (if w 1143 (if (> field-width (caar w)) 1144 (setcar w (list field-width beg (point)))) 1145 (setq w (list (list field-width beg (point))) 1146 column-widths (nconc column-widths w))) 1147 (or (eolp) (forward-char)) ; Skip separator. 1148 (setq w (cdr w) col (current-column) beg (point))))) 1149 (forward-line)) 1150 (list column-widths (nreverse field-widths)))) 1151 1152 (defun csv-align-fields (hard beg end) 1153 "Align all the fields in the region to form columns. 1154 The alignment style is specified by `csv-align-style'. The number of 1155 spaces specified by `csv-align-padding' appears after each separator. 1156 Use soft alignment done by displaying virtual white space after the 1157 separators unless invoked with an argument, in which case insert real 1158 space characters into the buffer after the separators. 1159 Unalign first (see `csv-unalign-fields'). Ignore blank and comment lines. 1160 1161 In hard-aligned records, separators become invisible whenever 1162 `buffer-invisibility-spec' is non-nil. In soft-aligned records, make 1163 separators invisible if and only if `buffer-invisibility-spec' is 1164 non-nil when the records are aligned; this can be changed only by 1165 re-aligning. \(Unaligning always makes separators visible.) 1166 1167 When called non-interactively, use hard alignment if HARD is non-nil; 1168 BEG and END specify the region to align. 1169 If there is no selected region, default to the whole buffer." 1170 (interactive (cons current-prefix-arg 1171 (if (use-region-p) 1172 (list (region-beginning) (region-end)) 1173 (list (point-min) (point-max))))) 1174 ;; FIXME: Use csv--jit-align when applicable! 1175 (setq end (copy-marker end)) 1176 (csv-unalign-fields hard beg end) ; If hard then barfs if buffer read only. 1177 (save-excursion 1178 (pcase-let ((`(,column-widths ,field-widths) (csv--column-widths beg end))) 1179 (save-restriction 1180 (narrow-to-region beg end) 1181 (set-marker end nil) 1182 1183 ;; Align fields: 1184 (goto-char (point-min)) 1185 (while (not (eobp)) ; for each record... 1186 (unless (csv-not-looking-at-record) 1187 (let ((w column-widths) 1188 (column 0)) ;Desired position of left-side of this column. 1189 (while (and w (not (eolp))) 1190 (let* ((beg (point)) 1191 (align-padding (if (bolp) 0 csv-align-padding)) 1192 (left-padding 0) (right-padding 0) 1193 (field-width (pop field-widths)) 1194 (column-width (car (pop w))) 1195 (x (- column-width field-width))) ; Required padding. 1196 (csv-end-of-field) 1197 (set-marker end (point)) ; End of current field. 1198 ;; beg = beginning of current field 1199 ;; end = (point) = end of current field 1200 1201 ;; Compute required padding: 1202 (cond 1203 ((eq csv-align-style 'left) 1204 ;; Left align -- pad on the right: 1205 (setq left-padding align-padding 1206 right-padding x)) 1207 ((eq csv-align-style 'right) 1208 ;; Right align -- pad on the left: 1209 (setq left-padding (+ align-padding x))) 1210 ((eq csv-align-style 'auto) 1211 ;; Auto align -- left align text, right align numbers: 1212 (if (string-match "\\`[-+.[:digit:]]+\\'" 1213 (buffer-substring beg (point))) 1214 ;; Right align -- pad on the left: 1215 (setq left-padding (+ align-padding x)) 1216 ;; Left align -- pad on the right: 1217 (setq left-padding align-padding 1218 right-padding x))) 1219 ((eq csv-align-style 'centre) 1220 ;; Centre -- pad on both left and right: 1221 (let ((y (/ x 2))) ; truncated integer quotient 1222 (setq left-padding (+ align-padding y) 1223 right-padding (- x y))))) 1224 1225 (cond 1226 (hard ;; Hard alignment... 1227 (when (> left-padding 0) ; Pad on the left. 1228 ;; Insert spaces before field: 1229 (if (= beg end) ; null field 1230 (insert (make-string left-padding ?\ )) 1231 (goto-char beg) ; beginning of current field 1232 (insert (make-string left-padding ?\ )) 1233 (goto-char end))) ; end of current field 1234 (unless (eolp) 1235 (if (> right-padding 0) ; pad on the right 1236 ;; Insert spaces after field: 1237 (insert (make-string right-padding ?\ ))) 1238 ;; Make separator (potentially) invisible; 1239 ;; in Emacs 21.3, neighbouring overlays 1240 ;; conflict, so use the following only 1241 ;; with hard alignment: 1242 (csv--make-overlay (point) (1+ (point)) nil t nil 1243 '(invisible csv evaporate t)) 1244 (forward-char))) ; skip separator 1245 1246 ;; Soft alignment... 1247 ((or (memq 'csv buffer-invisibility-spec) 1248 ;; For TSV, hidden or not doesn't make much difference, 1249 ;; but the behavior is slightly better when we "hide" 1250 ;; the TABs with a `display' property than if we add 1251 ;; before/after-strings. 1252 (tsv--mode-p)) 1253 1254 ;; Hide separators... 1255 ;; Merge right-padding from previous field 1256 ;; with left-padding from this field: 1257 (if (zerop column) 1258 (when (> left-padding 0) 1259 ;; Display spaces before first field 1260 ;; by overlaying first character: 1261 (csv--make-overlay 1262 beg (1+ beg) nil nil nil 1263 `(before-string ,(make-string left-padding ?\ )))) 1264 ;; Display separator as spaces: 1265 (with-silent-modifications 1266 (put-text-property 1267 (1- beg) beg 1268 'display `(space :align-to 1269 ,(+ left-padding column))))) 1270 (unless (eolp) (forward-char)) ; Skip separator. 1271 (setq column (+ column column-width align-padding))) 1272 1273 (t ;; Do not hide separators... 1274 (let ((overlay (csv--make-overlay beg (point) nil nil t))) 1275 (when (> left-padding 0) ; Pad on the left. 1276 ;; Display spaces before field: 1277 (overlay-put overlay 'before-string 1278 (make-string left-padding ?\ ))) 1279 (unless (eolp) 1280 (if (> right-padding 0) ; Pad on the right. 1281 ;; Display spaces after field: 1282 (overlay-put 1283 overlay 1284 'after-string (make-string right-padding ?\ ))) 1285 (forward-char)))) ; Skip separator. 1286 1287 ))))) 1288 (forward-line))))) 1289 (set-marker end nil)) 1290 1291 (defun csv-unalign-fields (hard beg end) 1292 "Undo soft alignment and optionally remove redundant white space. 1293 Undo soft alignment introduced by `csv-align-fields'. If invoked with 1294 an argument then also remove all spaces and tabs around separators. 1295 Also make all invisible separators visible again. 1296 Ignore blank and comment lines. When called non-interactively, remove 1297 spaces and tabs if HARD non-nil; BEG and END specify region to unalign. 1298 If there is no selected region, default to the whole buffer." 1299 (interactive (cons current-prefix-arg 1300 (if (use-region-p) 1301 (list (region-beginning) (region-end)) 1302 (list (point-min) (point-max))))) 1303 ;; Remove any soft alignment: 1304 (mapc #'csv--delete-overlay (overlays-in beg end)) 1305 (with-silent-modifications 1306 (remove-list-of-text-properties beg end '(display invisible))) 1307 (when hard 1308 (barf-if-buffer-read-only) 1309 ;; Remove any white-space padding around separators: 1310 (save-excursion 1311 (save-restriction 1312 (narrow-to-region beg end) 1313 (goto-char (point-min)) 1314 (while (not (eobp)) 1315 (or (csv-not-looking-at-record) 1316 (while (not (eolp)) 1317 ;; Delete horizontal white space forward: 1318 ;; (delete-horizontal-space) 1319 ;; This relies on left-to-right argument evaluation; 1320 ;; see info node (elisp) Function Forms. 1321 (delete-region (point) 1322 (+ (point) (skip-chars-forward " \t"))) 1323 (csv-end-of-field) 1324 ;; Delete horizontal white space backward: 1325 ;; (delete-horizontal-space t) 1326 (delete-region (point) 1327 (+ (point) (skip-chars-backward " \t"))) 1328 (or (eolp) (forward-char)))) 1329 (forward-line)))))) 1330 1331 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1332 ;;; Transposing rows and columns 1333 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1334 1335 (defun csv-transpose (beg end) 1336 "Rewrite rows (which may have different lengths) as columns. 1337 Null fields are introduced as necessary within records but are 1338 stripped from the ends of records. Preserve soft alignment. 1339 This function is its own inverse. Ignore blank and comment lines. 1340 When called non-interactively, BEG and END specify region to process." 1341 ;; (interactive "*P\nr") 1342 (interactive (csv-interactive-args 'noarg)) 1343 (barf-if-buffer-read-only) 1344 (save-excursion 1345 (save-restriction 1346 (narrow-to-region beg end) 1347 (goto-char (point-min)) 1348 ;; Delete rows and collect them as a reversed list of lists of 1349 ;; fields, skipping comment and blank lines: 1350 (let ((sep (car csv-separators)) 1351 (align (overlays-in beg end)) 1352 rows columns) 1353 ;; Remove soft alignment if necessary: 1354 (when align 1355 (mapc #'csv--delete-overlay align) 1356 (setq align t)) 1357 (while (not (eobp)) 1358 (if (csv-not-looking-at-record) 1359 ;; Skip blank and comment lines: 1360 (forward-line) 1361 (let ((lep (line-end-position))) 1362 (push 1363 (csv--collect-fields lep) 1364 rows) 1365 (delete-region (point) lep) 1366 (or (eobp) (delete-char 1))))) 1367 ;; Rows must have monotonic decreasing lengths to be 1368 ;; transposable, so ensure this by padding with null fields. 1369 ;; rows is currently a reversed list of field lists, which 1370 ;; must therefore have monotonic increasing lengths. 1371 (let ((oldlen (length (car rows))) newlen 1372 (r (cdr rows))) 1373 (while r 1374 (setq newlen (length (car r))) 1375 (if (< newlen oldlen) 1376 (nconc (car r) (make-list (- oldlen newlen) nil)) 1377 (setq oldlen newlen)) 1378 (setq r (cdr r)))) 1379 ;; Collect columns as a reversed list of lists of fields: 1380 (while rows 1381 (let (column (r rows) row) 1382 (while r 1383 (setq row (car r)) 1384 ;; Provided it would not be a trailing null field, push 1385 ;; field onto column: 1386 (if (or column (string< "" (car row))) 1387 (push (car row) column)) 1388 ;; Pop field off row: 1389 (setcar r (cdr row)) 1390 ;; If row is now empty then remove it: 1391 (or (car r) (setq rows (cdr rows))) 1392 (setq r (cdr r))) 1393 (push column columns))) 1394 ;; Insert columns into buffer as rows: 1395 (setq columns (nreverse columns)) 1396 (while columns 1397 (insert (mapconcat #'identity (car columns) sep) ?\n) 1398 (setq columns (cdr columns))) 1399 ;; Re-do soft alignment if necessary: 1400 (if align (csv-align-fields nil (point-min) (point-max))))))) 1401 1402 (defun csv--collect-fields (row-end-position) 1403 "Collect the fields of a row. 1404 Splits a row into fields, honoring quoted fields, and returns 1405 the list of fields. ROW-END-POSITION is the end-of-line position. 1406 point is assumed to be at the beginning of the line." 1407 (let ((csv-field-quotes-regexp (apply #'concat `("[" ,@csv-field-quotes "]"))) 1408 (row-text (buffer-substring-no-properties (point) row-end-position)) 1409 fields field-start) 1410 (if (not (string-match csv-field-quotes-regexp row-text)) 1411 (split-string row-text csv-separator-regexp) 1412 (save-excursion 1413 (while (< (setq field-start (point)) row-end-position) 1414 ;; csv-forward-field will skip a separator if point is on 1415 ;; it, and we'll miss an empty field 1416 (unless (memq (following-char) csv-separator-chars) 1417 (csv-forward-field 1)) 1418 (push 1419 (buffer-substring-no-properties field-start (point)) 1420 fields) 1421 (if (memq (following-char) csv-separator-chars) 1422 (forward-char))) 1423 (nreverse fields))))) 1424 1425 (defun csv--unquote-value (value) 1426 "Remove quotes around VALUE. 1427 If VALUE contains escaped quote characters, un-escape them. If 1428 VALUE is not quoted, return it unchanged." 1429 (save-match-data 1430 (let ((quote-regexp (apply #'concat `("[" ,@csv-field-quotes "]")))) 1431 (if-let (((string-match (concat "^\\(" quote-regexp "\\)\\(.*\\)\\(" quote-regexp "\\)$") value)) 1432 (quote-char (match-string 1 value)) 1433 ((equal quote-char (match-string 3 value))) 1434 (unquoted (match-string 2 value))) 1435 (replace-regexp-in-string (concat quote-char quote-char) quote-char unquoted) 1436 value)))) 1437 1438 (defun csv-parse-current-row () 1439 "Parse the current CSV line. 1440 Return the field values as a list." 1441 (save-mark-and-excursion 1442 (goto-char (line-beginning-position)) 1443 (mapcar #'csv--unquote-value (csv--collect-fields (line-end-position))))) 1444 1445 (defvar-local csv--header-line nil) 1446 (defvar-local csv--header-hscroll nil) 1447 (defvar-local csv--header-string nil) 1448 1449 (defun csv-header-line (&optional use-current-line) 1450 "Set/unset the header line. 1451 If the optional prefix arg USE-CURRENT-LINE is nil, use the first line 1452 as the header line. 1453 If there is already a header line, then unset the header line." 1454 (interactive "P") 1455 (if csv--header-line 1456 (progn 1457 (delete-overlay csv--header-line) 1458 (setq csv--header-line nil) 1459 (kill-local-variable 'header-line-format)) 1460 (save-excursion 1461 (unless use-current-line (goto-char (point-min))) 1462 (setq csv--header-line (make-overlay (line-beginning-position) 1463 (line-end-position) 1464 nil nil t)) 1465 (overlay-put csv--header-line 'modification-hooks 1466 '(csv--header-flush))) 1467 (csv--header-flush) 1468 ;; These are introduced in Emacs 29. 1469 (unless (boundp 'header-line-indent) 1470 (setq-local header-line-indent "" 1471 header-line-indent-width 0)) 1472 (setq header-line-format 1473 '("" header-line-indent (:eval (csv--header-string)))))) 1474 1475 (defun csv--header-flush (&rest _) 1476 ;; Force re-computation of the header-line. 1477 (setq csv--header-hscroll nil)) 1478 1479 (defun csv--header-string () 1480 ;; FIXME: Won't work with multiple windows showing that same buffer. 1481 (if (eql (window-hscroll) csv--header-hscroll) 1482 csv--header-string 1483 (setq csv--header-hscroll (window-hscroll)) 1484 (setq csv--header-string 1485 (csv--compute-header-string)))) 1486 1487 (defun csv--compute-header-string () 1488 (with-demoted-errors "csv--compute-header-string %S" 1489 (save-excursion 1490 (goto-char (overlay-start csv--header-line)) 1491 ;; Re-set the line-end-position, just in case. 1492 (move-overlay csv--header-line (point) (line-end-position)) 1493 (jit-lock-fontify-now (point) (line-end-position)) 1494 ;; Not sure why it is sometimes nil! 1495 (move-to-column (or csv--header-hscroll 0)) 1496 (let ((str (replace-regexp-in-string 1497 "%" "%%" (buffer-substring (point) (line-end-position)))) 1498 (i 0)) 1499 (while (and i (< i (length str))) 1500 (let ((prop (get-text-property i 'display str))) 1501 (and (eq (car-safe prop) 'space) 1502 (eq (car-safe (cdr prop)) :align-to) 1503 (let* ((x (nth 2 prop)) 1504 (nexti (next-single-property-change i 'display str)) 1505 (newprop 1506 `(space :align-to 1507 (+ ,(if (numberp x) 1508 (- x (or csv--header-hscroll 0)) 1509 `(- ,x csv--header-hscroll)) 1510 header-line-indent-width)))) 1511 (put-text-property i (or nexti (length str)) 1512 'display newprop str) 1513 (setq i nexti)))) 1514 (setq i (next-single-property-change i 'display str))) 1515 (concat (propertize " " 'display '((space :align-to 0))) str))))) 1516 1517 ;;; Auto-alignment 1518 1519 (defcustom csv-align-max-width 40 1520 "Maximum width of a column in `csv-align-mode'. 1521 This does not apply to the last column (for which the usual `truncate-lines' 1522 setting works better)." 1523 :type 'integer) 1524 1525 (defcustom csv-align-min-width 1 1526 "Minimum width of a column in `csv-align-mode'." 1527 :type 'integer) 1528 1529 (defvar-local csv--config-column-widths nil 1530 "Settings per column, stored as a list indexed by the column.") 1531 1532 (defun csv-align--set-column (column value) 1533 (let ((len (length csv--config-column-widths))) 1534 (if (< len column) 1535 (setq csv--config-column-widths 1536 (nconc csv--config-column-widths (make-list (- column len) nil)))) 1537 (setf (nth (1- column) csv--config-column-widths) value))) 1538 1539 (defun csv-align-set-column-width (column width) 1540 "Set the max WIDTH to use for COLUMN." 1541 (interactive 1542 (let* ((field (or (csv--field-index) 1)) 1543 (curwidth (nth (1- field) csv--config-column-widths))) 1544 (list field 1545 (cond 1546 ((numberp current-prefix-arg) 1547 current-prefix-arg) 1548 (current-prefix-arg 1549 (read-number (format "Column width (for field %d): " field) 1550 curwidth)) 1551 (t (if curwidth nil (csv--ellipsis-width))))))) 1552 (when (eql width csv-align-max-width) 1553 (setq width nil)) 1554 (csv-align--set-column column width) 1555 (jit-lock-refontify)) 1556 1557 (defvar-local csv--jit-columns nil) 1558 1559 (defun csv--jit-flush-columns () 1560 "Throw away all cached info about column widths." 1561 ;; FIXME: Maybe we should kill its overlays as well. 1562 (setq csv--jit-columns nil)) 1563 1564 (defun csv--jit-merge-columns (column-widths) 1565 ;; FIXME: The incremental update (delayed by jit-lock-context-time) of column 1566 ;; width is a bit jarring at times. It's OK while scrolling or when 1567 ;; extending a column, but not right when enabling the csv-align-mode or 1568 ;; when shortening the longest field (or deleting the line containing it), 1569 ;; because in that case we have *several* cascaded updates, e.g.: 1570 ;; - Remove the line with the longest field of column N. 1571 ;; - Edit some line: this line is updated as if its field was the widest, 1572 ;; hence its subsequent fields are too much to the left. 1573 ;; - The rest is updated starting from the first few lines (according 1574 ;; to jit-lock-chunk-size). 1575 ;; - After the first few lines, come the next set of few lines, 1576 ;; which may cause the previous few lines to need refresh again. 1577 ;; - etc.. until arriving again at the edited line which is re-aligned 1578 ;; again. 1579 ;; - etc.. until the end of the windows, potentially causing yet more 1580 ;; refreshes as we discover yet-wider fields for this column. 1581 (let ((old-columns csv--jit-columns) 1582 (changed nil)) 1583 (while (and old-columns column-widths) 1584 (when (or (> (caar column-widths) (caar old-columns)) 1585 ;; Apparently modification-hooks aren't run when the 1586 ;; whole text containing the overlay is deleted (e.g. 1587 ;; the whole line), so detect this case here. 1588 ;; It's a bit too late, but better than never. 1589 (null (overlay-buffer (cdar old-columns)))) 1590 (setq changed t) ;; Return non-nil if some existing column changed. 1591 (pcase-let ((`(,width ,beg ,end) (car column-widths))) 1592 (setf (caar old-columns) width) 1593 (move-overlay (cdar old-columns) beg end))) 1594 (setq old-columns (cdr old-columns)) 1595 (setq column-widths (cdr column-widths))) 1596 (when column-widths 1597 ;; New columns appeared. 1598 (setq csv--jit-columns 1599 (nconc csv--jit-columns 1600 (mapcar (lambda (x) 1601 (pcase-let* 1602 ((`(,width ,beg ,end) x) 1603 (ol (make-overlay beg end))) 1604 (overlay-put ol 'csv-width t) 1605 (overlay-put ol 'evaporate t) 1606 (overlay-put ol 'modification-hooks 1607 (list #'csv--jit-width-change)) 1608 (cons width ol))) 1609 column-widths)))) 1610 changed)) 1611 1612 (defun csv--jit-width-change (ol after _beg _end &optional len) 1613 (when (and after (> len 0)) 1614 ;; (let ((x (rassq ol csv--jit-columns))) 1615 ;; (when x (setf (car x) -1))) 1616 (delete-overlay ol))) 1617 1618 (defun csv--jit-unalign (beg end) 1619 (with-silent-modifications 1620 (remove-text-properties beg end 1621 '( display nil csv--jit nil invisible nil 1622 cursor-sensor-functions nil csv--revealed nil)) 1623 (remove-overlays beg end 'csv--jit t))) 1624 1625 (defun csv--jit-flush (beg end) 1626 "Cause all the buffer (except for the BEG...END region) to be re-aligned." 1627 (cl-assert (>= end beg)) 1628 ;; The buffer shouldn't have changed since beg/end were computed, 1629 ;; but just in case, let's make sure they're still sane. 1630 (when (< beg (point-min)) 1631 (setq beg (point-min) end (max end beg))) 1632 (when (< (point-max) end) 1633 (setq end (point-max) beg (min end beg))) 1634 (let ((pos (point-min))) 1635 (while (and (< pos beg) 1636 (setq pos (text-property-any pos beg 'csv--jit t))) 1637 (jit-lock-refontify 1638 pos (setq pos (or (text-property-any pos beg 'csv--jit nil) beg)))) 1639 (setq pos end) 1640 (while (and (< pos (point-max)) 1641 (setq pos (text-property-any pos (point-max) 'csv--jit t))) 1642 (jit-lock-refontify 1643 pos (setq pos (or (text-property-any pos (point-max) 'csv--jit nil) 1644 (point-max)))))) 1645 (csv--header-flush)) 1646 1647 (defun csv--ellipsis-width () 1648 (let ((ellipsis 1649 (when standard-display-table 1650 (display-table-slot standard-display-table 1651 'selective-display)))) 1652 (if ellipsis (length ellipsis) 3))) 1653 1654 (defun csv-align--cursor-truncated (window oldpos dir) 1655 ;; FIXME: Neither the `entered' nor the `left' event are guaranteed 1656 ;; to be sent, and for the `left' case, even when we do get called, 1657 ;; it may be unclear where the revealed text was (it's somewhere around 1658 ;; `oldpos', but that position can be stale). 1659 ;; Worse, if we have several windows displaying the buffer, when one 1660 ;; cursor leaves we may need to keep the text revealed because of 1661 ;; another window's cursor. 1662 (let* ((prop (if (eq dir 'entered) 'invisible 'csv--revealed)) 1663 (pos (cond 1664 ((eq dir 'entered) (window-point window)) 1665 (t (max (point-min) 1666 (min (point-max) 1667 (or oldpos (window-point window))))))) 1668 (start (cond 1669 ((and (> pos (point-min)) 1670 (eq (get-text-property (1- pos) prop) 'csv-truncate)) 1671 (or (previous-single-property-change pos prop) (point-min))) 1672 (t pos))) 1673 (end (if (eq (get-text-property pos prop) 'csv-truncate) 1674 (or (next-single-property-change pos prop) (point-max)) 1675 pos))) 1676 (unless (eql start end) 1677 (with-silent-modifications 1678 (put-text-property start end 1679 (if (eq dir 'entered) 'csv--revealed 'invisible) 1680 'csv-truncate) 1681 (remove-text-properties start end (list prop)))))) 1682 1683 (defun csv--jit-align (beg end) 1684 (save-excursion 1685 ;; This is run with inhibit-modification-hooks set, so the overlays' 1686 ;; modification-hook doesn't work :-( 1687 (and csv--header-line 1688 (<= beg (overlay-end csv--header-line)) 1689 (>= end (overlay-start csv--header-line)) 1690 (csv--header-flush)) 1691 ;; First, round up to a whole number of lines. 1692 (goto-char end) 1693 (unless (bolp) (forward-line 1) (setq end (point))) 1694 (goto-char beg) 1695 (unless (bolp) (forward-line 1) (setq beg (point))) 1696 (csv--jit-unalign beg end) 1697 (put-text-property beg end 'csv--jit t) 1698 1699 (pcase-let* ((`(,column-widths ,field-widths) (csv--column-widths beg end)) 1700 (changed (csv--jit-merge-columns column-widths)) 1701 (ellipsis-width (csv--ellipsis-width))) 1702 (when changed 1703 ;; Do it after the current redisplay is over. 1704 (run-with-timer jit-lock-context-time nil #'csv--jit-flush beg end)) 1705 1706 ;; Align fields: 1707 (goto-char beg) 1708 (while (< (point) end) 1709 (unless (csv-not-looking-at-record) 1710 (let ((w csv--jit-columns) 1711 (widths-config csv--config-column-widths) 1712 (column 0)) ;Desired position of left-side of this column. 1713 (while (and w (not (eolp))) 1714 (let* ((field-beg (point)) 1715 (width-config (pop widths-config)) 1716 (align-padding (if (bolp) 0 csv-align-padding)) 1717 (left-padding 0) (right-padding 0) 1718 (field-width (pop field-widths)) 1719 (column-width 1720 (min (max csv-align-min-width 1721 (car (pop w))) 1722 (or width-config 1723 ;; Don't apply csv-align-max-width 1724 ;; to the last field! 1725 (if w csv-align-max-width 1726 most-positive-fixnum)))) 1727 (x (- column-width field-width)) ; Required padding. 1728 (truncate nil)) 1729 (csv-end-of-field) 1730 ;; beg = beginning of current field 1731 ;; end = (point) = end of current field 1732 (when (< x 0) 1733 (setq truncate (max column 1734 (+ column column-width 1735 align-padding (- ellipsis-width)))) 1736 (setq x 0)) 1737 ;; Compute required padding: 1738 (pcase csv-align-style 1739 ('left 1740 ;; Left align -- pad on the right: 1741 (setq left-padding align-padding 1742 right-padding x)) 1743 ('right 1744 ;; Right align -- pad on the left: 1745 (setq left-padding (+ align-padding x))) 1746 ('auto 1747 ;; Auto align -- left align text, right align numbers: 1748 (if (string-match "\\`[-+.[:digit:]]+\\'" 1749 (buffer-substring field-beg (point))) 1750 ;; Right align -- pad on the left: 1751 (setq left-padding (+ align-padding x)) 1752 ;; Left align -- pad on the right: 1753 (setq left-padding align-padding 1754 right-padding x))) 1755 ('centre 1756 ;; Centre -- pad on both left and right: 1757 (let ((y (/ x 2))) ; truncated integer quotient 1758 (setq left-padding (+ align-padding y) 1759 right-padding (- x y))))) 1760 1761 (cond 1762 1763 ((or (memq 'csv buffer-invisibility-spec) 1764 ;; For TSV, hidden or not doesn't make much difference, 1765 ;; but the behavior is slightly better when we "hide" 1766 ;; the TABs with a `display' property than if we add 1767 ;; before/after-strings. 1768 (tsv--mode-p)) 1769 1770 ;; Hide separators... 1771 ;; Merge right-padding from previous field 1772 ;; with left-padding from this field: 1773 (if (zerop column) 1774 (when (> left-padding 0) 1775 ;; Display spaces before first field 1776 ;; by overlaying first character: 1777 (csv--make-overlay 1778 field-beg (1+ field-beg) nil nil nil 1779 `(before-string ,(make-string left-padding ?\ ) 1780 csv--jit t))) 1781 ;; Display separator as spaces: 1782 (with-silent-modifications 1783 (put-text-property 1784 (1- field-beg) field-beg 1785 'display `(space :align-to 1786 ,(+ left-padding column)))))) 1787 1788 (t ;; Do not hide separators... 1789 (let ((overlay (csv--make-overlay field-beg (point) 1790 nil nil t 1791 '(csv--jit t)))) 1792 (when (> left-padding 0) ; Pad on the left. 1793 ;; Display spaces before field: 1794 (overlay-put overlay 'before-string 1795 (make-string left-padding ?\ ))) 1796 (unless (eolp) 1797 (if (> right-padding 0) ; Pad on the right. 1798 ;; Display spaces after field: 1799 (overlay-put 1800 overlay 1801 'after-string (make-string right-padding ?\ ))))))) 1802 (setq column (+ column column-width align-padding)) 1803 ;; Do it after applying the property, so `move-to-column' can 1804 ;; take it into account. 1805 (when truncate 1806 (let ((trunc-pos 1807 (save-excursion 1808 ;; ¡¡ BIG UGLY HACK !! 1809 ;; `current-column' and `move-to-column' count 1810 ;; text hidden with an ellipsis "as if" it were 1811 ;; fully visible, which is completely wrong here, 1812 ;; so circumvent this by temporarily pretending 1813 ;; that `csv-truncate' is fully invisible (which 1814 ;; isn't quite right either, but should work 1815 ;; just well enough for us here). 1816 (let ((buffer-invisibility-spec 1817 buffer-invisibility-spec)) 1818 (add-to-invisibility-spec 'csv-truncate) 1819 (move-to-column truncate)) 1820 (point)))) 1821 (put-text-property trunc-pos (point) 1822 'invisible 'csv-truncate) 1823 (when (> (- (point) trunc-pos) 1) 1824 ;; Arrange to temporarily untruncate the string when 1825 ;; cursor moves into it. 1826 ;; FIXME: This only works if 1827 ;; `global-disable-point-adjustment' is non-nil! 1828 ;; Arguably this should be fixed by making 1829 ;; point-adjustment code pay attention to 1830 ;; cursor-sensor-functions! 1831 (put-text-property 1832 (1+ trunc-pos) (point) 1833 'cursor-sensor-functions 1834 (list #'csv-align--cursor-truncated))))) 1835 (unless (eolp) (forward-char)) ; Skip separator. 1836 )))) 1837 (forward-line))) 1838 `(jit-lock-bounds ,beg . ,end))) 1839 1840 (define-minor-mode csv-align-mode 1841 "Align columns on the fly." 1842 :global nil 1843 (csv-unalign-fields nil (point-min) (point-max)) ;Just in case. 1844 (cond 1845 (csv-align-mode 1846 (add-to-invisibility-spec '(csv-truncate . t)) 1847 (kill-local-variable 'csv--jit-columns) 1848 (cursor-sensor-mode 1) 1849 (when (fboundp 'header-line-indent-mode) 1850 (header-line-indent-mode)) 1851 (jit-lock-register #'csv--jit-align) 1852 (jit-lock-refontify)) 1853 (t 1854 (remove-from-invisibility-spec '(csv-truncate . t)) 1855 (jit-lock-unregister #'csv--jit-align) 1856 (csv--jit-unalign (point-min) (point-max)))) 1857 (csv--header-flush)) 1858 1859 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1860 ;;; Separator guessing 1861 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1862 1863 (defvar csv--preferred-separators 1864 '(?, ?\; ?\t) 1865 "Preferred separator characters in case of a tied score.") 1866 1867 (defun csv-guess-set-separator () 1868 "Guess and set the CSV separator of the current buffer. 1869 1870 Add it to the mode hook to have CSV mode guess and set the 1871 separator automatically when visiting a buffer: 1872 1873 (add-hook \\='csv-mode-hook \\='csv-guess-set-separator)" 1874 (interactive) 1875 (let ((sep (csv-guess-separator 1876 (buffer-substring-no-properties 1877 (point-min) 1878 ;; We're probably only going to look at the first 2048 1879 ;; or so chars, but take more than we probably need to 1880 ;; minimize the chance of breaking the input in the 1881 ;; middle of a (long) row. 1882 (min 8192 (point-max))) 1883 2048))) 1884 (when sep 1885 (csv-set-separator sep)))) 1886 1887 (defun csv-guess-separator (text &optional cutoff) 1888 "Return a guess of which character is the CSV separator in TEXT." 1889 (let ((best-separator nil) 1890 (best-score 0)) 1891 (dolist (candidate (csv--separator-candidates text cutoff)) 1892 (let ((candidate-score 1893 (csv--separator-score candidate text cutoff))) 1894 (when (or (> candidate-score best-score) 1895 (and (= candidate-score best-score) 1896 (member candidate csv--preferred-separators))) 1897 (setq best-separator candidate) 1898 (setq best-score candidate-score)))) 1899 best-separator)) 1900 1901 (defun csv--separator-candidates (text &optional cutoff) 1902 "Return a list of candidate CSV separators in TEXT. 1903 When CUTOFF is passed, look only at the first CUTOFF number of characters." 1904 (let ((chars (make-hash-table))) 1905 (dolist (c (string-to-list 1906 (if cutoff 1907 (substring text 0 (min cutoff (length text))) 1908 text))) 1909 (when (and (not (gethash c chars)) 1910 (or (= c ?\t) 1911 (= c ?\C-_) 1912 (and (not (member c '(?. ?/ ?\" ?'))) 1913 (not (= c (string-to-char csv-comment-start))) 1914 (not (member (get-char-code-property c 'general-category) 1915 '(Lu Ll Lt Lm Lo Nd Nl No Ps Pe Cc Co)))))) 1916 (puthash c t chars))) 1917 (hash-table-keys chars))) 1918 1919 (defun csv--separator-score (separator text &optional cutoff) 1920 "Return a score on how likely SEPARATOR is a separator in TEXT. 1921 1922 When CUTOFF is passed, stop the calculation at the next whole 1923 line after having read CUTOFF number of characters. 1924 1925 The scoring is based on the idea that most CSV data is tabular, 1926 i.e. separators should appear equally often on each line. 1927 Furthermore, more commonly appearing characters are scored higher 1928 than those who appear less often. 1929 1930 Adapted from the paper \"Wrangling Messy CSV Files by Detecting 1931 Row and Type Patterns\" by Gerrit J.J. van den Burg , Alfredo 1932 Nazábal, and Charles Sutton: https://arxiv.org/abs/1811.11242." 1933 (let ((groups 1934 (with-temp-buffer 1935 (csv-set-separator separator) 1936 (save-excursion 1937 (insert text)) 1938 (let ((groups (make-hash-table)) 1939 (chars-read 0)) 1940 (while (and (/= (point) (point-max)) 1941 (or (not cutoff) 1942 (< chars-read cutoff))) 1943 (let* ((lep (line-end-position)) 1944 (nfields (length (csv--collect-fields lep)))) 1945 (cl-incf (gethash nfields groups 0)) 1946 (cl-incf chars-read (- lep (point))) 1947 (goto-char (+ lep 1)))) 1948 groups))) 1949 (sum 0)) 1950 (maphash 1951 (lambda (length num) 1952 (cl-incf sum (* num (/ (- length 1) (float length))))) 1953 groups) 1954 (let ((unique-groups (hash-table-count groups))) 1955 (if (= 0 unique-groups) 1956 0 1957 (/ sum unique-groups))))) 1958 1959 ;;; TSV support 1960 1961 ;; Since "the" CSV format is really a bunch of different formats, it includes 1962 ;; TSV as a subcase, but this subcase is sufficiently interesting that it has 1963 ;; its own mime-type and mostly standard file extension, also it suffers 1964 ;; less from the usual quoting problems of CSV (because the only problematic 1965 ;; chars are LF and TAB, really, which are much less common inside fields than 1966 ;; commas, space, and semi-colons) so it's "better behaved". 1967 1968 (defvar tsv-mode-syntax-table 1969 ;; Inherit from `text-mode-syntax-table' rather than from 1970 ;; `csv-mode-syntax-table' so as not to inherit the 1971 ;; `csv-field-quotes' settings. 1972 (let ((st (make-syntax-table text-mode-syntax-table))) 1973 st)) 1974 1975 (defvar tsv-mode-map 1976 (let ((map (make-sparse-keymap))) 1977 ;; In `tsv-mode', the `csv-invisibility-default/csv-toggle-invisibility' 1978 ;; business doesn't make much sense. 1979 (define-key map [remap csv-toggle-invisibility] #'undefined) 1980 map)) 1981 1982 ;;;###autoload 1983 (add-to-list 'auto-mode-alist '("\\.tsv\\'" . tsv-mode)) 1984 1985 (defun tsv--mode-p () 1986 (equal csv-separator-chars '(?\t))) 1987 1988 ;;;###autoload 1989 (define-derived-mode tsv-mode csv-mode "TSV" 1990 "Major mode for editing files of tab-separated value type." 1991 :group 'CSV 1992 ;; In TSV we know TAB is the only possible separator. 1993 (setq-local csv-separators '("\t")) 1994 ;; FIXME: Copy&pasted from the `:set'ter of csv-separators! 1995 (setq-local csv-separator-chars '(?\t)) 1996 (setq-local csv--skip-chars "^\n\t") 1997 (setq-local csv-separator-regexp "\t") 1998 (setq-local csv-font-lock-keywords 1999 ;; NB: csv-separator-face variable evaluates to itself. 2000 `((,csv-separator-regexp (0 'csv-separator-face)))) 2001 2002 ;; According to wikipedia, TSV doesn't use quotes but uses backslash escapes 2003 ;; of the form \n, \t, \r, and \\ instead. 2004 (setq-local csv-field-quotes nil)) 2005 2006 2007 (provide 'csv-mode) 2008 2009 ;;; csv-mode.el ends here