csv-mode.el (82637B)
1 ;;; csv-mode.el --- Major mode for editing comma/char separated values -*- lexical-binding: t -*- 2 3 ;; Copyright (C) 2003-2023 Free Software Foundation, Inc 4 5 ;; Author: "Francis J. Wright" <F.J.Wright@qmul.ac.uk> 6 ;; Maintainer: emacs-devel@gnu.org 7 ;; Version: 1.23 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 1.21: 111 ;; - New command `csv-insert-column'. 112 ;; - New config var `csv-align-min-width' for `csv-align-mode'. 113 ;; - New option `csv-confirm-region'. 114 115 ;; Since 1.9: 116 ;; - `csv-align-mode' auto-aligns columns dynamically (on screen). 117 118 ;; Before that: 119 ;; Begun on 15 November 2003 to provide lexicographic sorting of 120 ;; simple CSV data by field and released as csv.el. Facilities to 121 ;; kill multiple fields and customize separator added on 9 April 2004. 122 ;; Converted to a major mode and renamed csv-mode.el on 10 April 2004, 123 ;; partly at the suggestion of Stefan Monnier <monnier at 124 ;; IRO.UMontreal.CA> to avoid conflict with csv.el by Ulf Jasper. 125 ;; Field alignment, comment support and CSV mode customization group 126 ;; added on 1 May 2004. Support for index ranges added on 6 June 127 ;; 2004. Multiple field separators added on 12 June 2004. 128 ;; Transposition added on 22 June 2004. Separator invisibility added 129 ;; on 23 June 2004. 130 131 ;;; To do (maybe): 132 133 ;; Make separators and quotes buffer-local and locally settable. 134 ;; Support (La)TeX tables: set separator and comment; support record 135 ;; end string. 136 ;; Convert comma-separated to space- or tab-separated. 137 138 ;;; Code: 139 140 (eval-when-compile 141 (require 'cl-lib) 142 (require 'subr-x)) 143 144 (defgroup CSV nil 145 "Major mode for editing files of comma-separated value type." 146 :group 'convenience) 147 148 (defvar csv-separator-chars nil 149 "Field separators as a list of character. 150 Set by customizing `csv-separators' -- do not set directly!") 151 152 (defvar csv-separator-regexp nil 153 "Regexp to match a field separator. 154 Set by customizing `csv-separators' -- do not set directly!") 155 156 (defvar csv--skip-chars nil 157 "Char set used by `skip-chars-forward' etc. to skip fields. 158 Set by customizing `csv-separators' -- do not set directly!") 159 160 (defvar csv-font-lock-keywords nil 161 "Font lock keywords to highlight the field separators in CSV mode. 162 Set by customizing `csv-separators' -- do not set directly!") 163 164 (defcustom csv-separators '("," "\t") 165 "Field separators: a list of *single-character* strings. 166 For example: (\",\"), the default, or (\",\" \";\" \":\"). 167 Neighbouring fields may be separated by any one of these characters. 168 The first is used when inserting a field separator into the buffer. 169 All must be different from the field quote characters, `csv-field-quotes'. 170 171 Changing this variable with `setq' won't affect the current Emacs 172 session. Use `customize-set-variable' instead if that is required." 173 ;; Suggested by Eckhard Neber <neber@mwt.e-technik.uni-ulm.de> 174 :type '(repeat string) 175 ;; FIXME: Character would be better, but in Emacs 21.3 does not display 176 ;; correctly in a customization buffer. 177 :set (lambda (variable value) 178 (mapc (lambda (x) 179 (if (/= (length x) 1) 180 (error "Non-single-char string %S" x)) 181 (if (and (boundp 'csv-field-quotes) 182 (member x csv-field-quotes)) 183 (error "%S is already a quote" x))) 184 value) 185 (custom-set-default variable value) 186 (setq csv-separator-chars (mapcar #'string-to-char value)) 187 (setq csv--skip-chars 188 (apply #'concat "^\n" 189 (mapcar (lambda (s) (concat "\\" s)) value))) 190 (setq csv-separator-regexp (regexp-opt value)) 191 (setq csv-font-lock-keywords 192 ;; NB: csv-separator-face variable evaluates to itself. 193 `((,csv-separator-regexp (0 'csv-separator-face)))))) 194 195 (defcustom csv-field-quotes '("\"") 196 "Field quotes: a list of *single-character* strings. 197 For example: (\"\\\"\"), the default, or (\"\\\"\" \"\\='\" \"\\=`\"). 198 A field can be delimited by a pair of any of these characters. 199 All must be different from the field separators, `csv-separators'." 200 :type '(repeat string) 201 ;; Character would be better, but in Emacs 21 does not display 202 ;; correctly in a customization buffer. 203 :set (lambda (variable value) 204 (mapc (lambda (x) 205 (if (/= (length x) 1) 206 (error "Non-single-char string %S" x)) 207 (if (member x csv-separators) 208 (error "%S is already a separator" x))) 209 value) 210 (when (boundp 'csv-mode-syntax-table) 211 ;; FIRST remove old quote syntax: 212 (with-syntax-table text-mode-syntax-table 213 (mapc (lambda (x) 214 (modify-syntax-entry 215 (string-to-char x) 216 (string (char-syntax (string-to-char x))) 217 ;; symbol-value to avoid compiler warning: 218 (symbol-value 'csv-mode-syntax-table))) 219 csv-field-quotes)) 220 ;; THEN set new quote syntax: 221 (csv-set-quote-syntax value)) 222 ;; BEFORE setting new value of `csv-field-quotes': 223 (custom-set-default variable value))) 224 225 (defun csv-set-quote-syntax (field-quotes) 226 "Set syntax for field quote characters FIELD-QUOTES to be \"string\". 227 FIELD-QUOTES should be a list of single-character strings." 228 (mapc (lambda (x) 229 (modify-syntax-entry 230 (string-to-char x) "\"" 231 ;; symbol-value to avoid compiler warning: 232 (symbol-value 'csv-mode-syntax-table))) 233 field-quotes)) 234 235 (defvar csv-comment-start nil 236 "String that starts a comment line, or nil if no comment syntax. 237 Such comment lines are ignored by CSV mode commands. 238 This variable is buffer local; its default value is that of 239 `csv-comment-start-default'. It is set by the function 240 `csv-set-comment-start' -- do not set it directly!") 241 242 (make-variable-buffer-local 'csv-comment-start) 243 244 (defcustom csv-comment-start-default "#" 245 "String that starts a comment line, or nil if no comment syntax. 246 Such comment lines are ignored by CSV mode commands. 247 Default value of buffer-local variable `csv-comment-start'. 248 Changing this variable does not affect any existing CSV mode buffer." 249 :type '(choice (const :tag "None" nil) string) 250 :set (lambda (variable value) 251 (custom-set-default variable value) 252 (setq-default csv-comment-start value))) 253 254 (defcustom csv-align-style 'left 255 "Aligned field style: one of `left', `centre', `right' or `auto'. 256 Alignment style used by `csv-align-mode' and `csv-align-fields'. 257 Auto-alignment means left align text and right align numbers." 258 :type '(choice (const left) (const centre) 259 (const right) (const auto))) 260 261 (defcustom csv-align-padding 1 262 "Aligned field spacing: must be a positive integer. 263 Number of spaces used by `csv-align-mode' and `csv-align-fields' 264 after separators." 265 :type 'integer) 266 267 (defcustom csv-header-lines 0 268 "Header lines to skip when setting region automatically." 269 :type 'integer) 270 271 (defcustom csv-invisibility-default t 272 "If non-nil, make separators in aligned records invisible." 273 :type 'boolean) 274 275 (defcustom csv-confirm-region t 276 "If non-nil, confirm that region is OK in interactive commands." 277 :type 'boolean) 278 279 (defface csv-separator-face 280 '((t :inherit escape-glyph)) 281 "CSV mode face used to highlight separators.") 282 283 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 284 ;;; Mode definition, key bindings and menu 285 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 286 287 288 (defconst csv-mode-line-format 289 '(csv-field-index-string ("" csv-field-index-string)) 290 "Mode line format string for CSV mode.") 291 292 (defvar csv-mode-map 293 (let ((map (make-sparse-keymap))) 294 (define-key map [(control ?c) (control ?v)] #'csv-toggle-invisibility) 295 (define-key map [(control ?c) (control ?t)] #'csv-transpose) 296 (define-key map [(control ?c) (control ?c)] #'csv-set-comment-start) 297 (define-key map [(control ?c) (control ?u)] #'csv-unalign-fields) 298 (define-key map [(control ?c) (control ?a)] #'csv-align-fields) 299 (define-key map [(control ?c) (control ?z)] #'csv-yank-as-new-table) 300 (define-key map [(control ?c) (control ?y)] #'csv-yank-fields) 301 (define-key map [(control ?c) (control ?k)] #'csv-kill-fields) 302 (define-key map [(control ?c) (control ?d)] #'csv-toggle-descending) 303 (define-key map [(control ?c) (control ?r)] #'csv-reverse-region) 304 (define-key map [(control ?c) (control ?n)] #'csv-sort-numeric-fields) 305 (define-key map [(control ?c) (control ?s)] #'csv-sort-fields) 306 (define-key map "\t" #'csv-tab-command) 307 (define-key map [backtab] #'csv-backtab-command) 308 map)) 309 310 ;;;###autoload 311 (define-derived-mode csv-mode text-mode "CSV" 312 "Major mode for editing files of comma-separated value type. 313 314 CSV mode is derived from `text-mode', and runs `text-mode-hook' before 315 running `csv-mode-hook'. It turns `auto-fill-mode' off by default. 316 CSV mode can be customized by user options in the CSV customization 317 group. The separators are specified by the value of `csv-separators'. 318 319 CSV mode commands ignore blank lines and comment lines beginning with 320 the value of `csv-comment-start', which delimit \"paragraphs\". 321 \"Sexp\" is re-interpreted to mean \"field\", so that `forward-sexp' 322 \(\\[forward-sexp]), `kill-sexp' (\\[kill-sexp]), etc. all apply to fields. 323 Standard comment commands apply, such as `comment-dwim' (\\[comment-dwim]). 324 325 If `font-lock-mode' is enabled then separators, quoted values and 326 comment lines are highlighted using respectively `csv-separator-face', 327 `font-lock-string-face' and `font-lock-comment-face'. 328 329 The user interface (UI) for CSV mode commands is similar to that of 330 the standard commands `sort-fields' and `sort-numeric-fields', except 331 that if there is no prefix argument then the UI prompts for the field 332 index or indices. In `transient-mark-mode' only: if the region is not 333 set then the UI attempts to set it to include all consecutive CSV 334 records around point, and prompts for confirmation; if there is no 335 prefix argument then the UI prompts for it, offering as a default the 336 index of the field containing point if the region was not set 337 explicitly. The region set automatically is delimited by blank lines 338 and comment lines, and the number of header lines at the beginning of 339 the region given by the value of `csv-header-lines' are skipped. 340 341 Sort order is controlled by `csv-descending'. 342 343 CSV mode provides the following specific keyboard key bindings: 344 345 \\{csv-mode-map}" 346 :group 'CSV 347 ;; We used to `turn-off-auto-fill' here instead, but that's not very 348 ;; effective since text-mode-hook is run afterwards anyway! 349 (setq-local normal-auto-fill-function nil) 350 ;; Set syntax for field quotes: 351 (csv-set-quote-syntax csv-field-quotes) 352 ;; Make sexp functions apply to fields: 353 (set (make-local-variable 'forward-sexp-function) #'csv-forward-field) 354 (csv-set-comment-start csv-comment-start) 355 ;; Font locking -- separator plus syntactic: 356 (setq font-lock-defaults '(csv-font-lock-keywords)) 357 (setq-local jit-lock-contextually nil) ;Each line should be independent. 358 (if csv-invisibility-default (add-to-invisibility-spec 'csv)) 359 ;; Mode line to support `csv-field-index-mode': 360 (set (make-local-variable 'mode-line-position) 361 (pcase mode-line-position 362 (`(,(or (pred consp) (pred stringp)) . ,_) 363 `(,@mode-line-position ,csv-mode-line-format)) 364 (_ `("" ,mode-line-position ,csv-mode-line-format)))) 365 (set (make-local-variable 'truncate-lines) t) 366 ;; Enable or disable `csv-field-index-mode' (could probably do this 367 ;; a bit more efficiently): 368 (csv-field-index-mode (symbol-value 'csv-field-index-mode))) 369 370 (defun csv-set-comment-start (string) 371 "Set comment start for this CSV mode buffer to STRING. 372 It must be either a string or nil." 373 (interactive 374 (list (edit-and-eval-command 375 "Comment start (string or nil): " csv-comment-start))) 376 ;; Paragraph means a group of contiguous records: 377 (set (make-local-variable 'paragraph-separate) "[[:space:]]*$") ; White space. 378 (set (make-local-variable 'paragraph-start) "\n");Must include \n explicitly! 379 ;; Remove old comment-start/end if available 380 (with-syntax-table text-mode-syntax-table 381 (when comment-start 382 (modify-syntax-entry (string-to-char comment-start) 383 (string (char-syntax (string-to-char comment-start))) 384 csv-mode-syntax-table)) 385 (modify-syntax-entry ?\n 386 (string (char-syntax ?\n)) 387 csv-mode-syntax-table)) 388 (when string 389 (setq paragraph-separate (concat paragraph-separate "\\|" string) 390 paragraph-start (concat paragraph-start "\\|" string)) 391 (set (make-local-variable 'comment-start) string) 392 (modify-syntax-entry 393 (string-to-char string) "<" csv-mode-syntax-table) 394 (modify-syntax-entry ?\n ">" csv-mode-syntax-table)) 395 (setq csv-comment-start string)) 396 397 (defvar csv--set-separator-history nil) 398 399 (defun csv-set-separator (sep) 400 "Set the CSV separator in the current buffer to SEP." 401 (interactive (list (read-char-from-minibuffer 402 "Separator: " nil 'csv--set-separator-history))) 403 (when (and (boundp 'csv-field-quotes) 404 (member (string sep) csv-field-quotes)) 405 (error "%c is already a quote" sep)) 406 (setq-local csv-separators (list (string sep))) 407 (setq-local csv-separator-chars (list sep)) 408 (setq-local csv--skip-chars (format "^\n\\%c" sep)) 409 (setq-local csv-separator-regexp (regexp-quote (string sep))) 410 (setq-local csv-font-lock-keywords 411 `((,csv-separator-regexp (0 'csv-separator-face)))) 412 (font-lock-refresh-defaults)) 413 414 ;;;###autoload 415 (add-to-list 'auto-mode-alist '("\\.[Cc][Ss][Vv]\\'" . csv-mode)) 416 417 (defvar csv-descending nil 418 "If non-nil, CSV mode sort functions sort in order of descending sort key. 419 Usually they sort in order of ascending sort key.") 420 421 (defun csv-toggle-descending () 422 "Toggle `csv-descending'." 423 (interactive) 424 (setq csv-descending (not csv-descending)) 425 (message "Sort order is %sscending" (if csv-descending "de" "a"))) 426 427 (defun csv-toggle-invisibility () 428 ;; FIXME: Make it into a proper minor mode? 429 "Toggle `buffer-invisibility-spec'." 430 (interactive) 431 (if (memq 'csv buffer-invisibility-spec) 432 (remove-from-invisibility-spec 'csv) 433 (add-to-invisibility-spec 'csv)) 434 (message "Separators in aligned records will be %svisible \ 435 \(after re-aligning if soft)" 436 (if (memq 'csv buffer-invisibility-spec) "in" "")) 437 (redraw-frame (selected-frame))) 438 439 (easy-menu-define 440 csv-menu 441 csv-mode-map 442 "CSV major mode menu keymap" 443 '("CSV" 444 ["Sort By Field Lexicographically" csv-sort-fields :active t 445 :help "Sort lines in region lexicographically by the specified field"] 446 ["Sort By Field Numerically" csv-sort-numeric-fields :active t 447 :help "Sort lines in region numerically by the specified field"] 448 ["Reverse Order of Lines" csv-reverse-region :active t 449 :help "Reverse the order of the lines in the region"] 450 ["Use Descending Sort Order" csv-toggle-descending :active t 451 :style toggle :selected csv-descending 452 :help "If selected, use descending order when sorting"] 453 "--" 454 ["Kill Fields (Columns)" csv-kill-fields :active t 455 :help "Kill specified fields of each line in the region"] 456 ["Yank Fields (Columns)" csv-yank-fields :active t 457 :help "Yank killed fields as specified field of each line in region"] 458 ["Yank As New Table" csv-yank-as-new-table :active t 459 :help "Yank killed fields as a new table at point"] 460 ["Align Fields into Columns" csv-align-fields :active t 461 :help "Align the start of every field of each line in the region"] 462 ["Unalign Columns into Fields" csv-unalign-fields :active t 463 :help "Undo soft alignment and optionally remove redundant white space"] 464 ["Transpose Rows and Columns" csv-transpose :active t 465 :help "Rewrite rows (which may have different lengths) as columns"] 466 "--" 467 ["Forward Field" forward-sexp :active t 468 :help "Move forward across one field; with ARG, do it that many times"] 469 ["Backward Field" backward-sexp :active t 470 :help "Move backward across one field; with ARG, do it that many times"] 471 ["Kill Field Forward" kill-sexp :active t 472 :help "Kill field following cursor; with ARG, do it that many times"] 473 ["Kill Field Backward" backward-kill-sexp :active t 474 :help "Kill field preceding cursor; with ARG, do it that many times"] 475 "--" 476 ("Alignment Style" 477 ["Left" (setq csv-align-style 'left) :active t 478 :style radio :selected (eq csv-align-style 'left) 479 :help "If selected, `csv-align' left aligns fields"] 480 ["Centre" (setq csv-align-style 'centre) :active t 481 :style radio :selected (eq csv-align-style 'centre) 482 :help "If selected, `csv-align' centres fields"] 483 ["Right" (setq csv-align-style 'right) :active t 484 :style radio :selected (eq csv-align-style 'right) 485 :help "If selected, `csv-align' right aligns fields"] 486 ["Auto" (setq csv-align-style 'auto) :active t 487 :style radio :selected (eq csv-align-style 'auto) 488 :help "\ 489 If selected, `csv-align' left aligns text and right aligns numbers"] 490 ) 491 ["Set header line" csv-header-line :active t] 492 ["Auto-(re)align fields" csv-align-mode 493 :style toggle :selected csv-align-mode] 494 ["Show Current Field Index" csv-field-index-mode :active t 495 :style toggle :selected csv-field-index-mode 496 :help "If selected, display current field index in mode line"] 497 ["Make Separators Invisible" csv-toggle-invisibility :active t 498 :style toggle :selected (memq 'csv buffer-invisibility-spec) 499 :visible (not (tsv--mode-p)) 500 :help "If selected, separators in aligned records are invisible"] 501 ["Set Buffer's Comment Start" csv-set-comment-start :active t 502 :help "Set comment start string for this buffer"] 503 ["Customize CSV Mode" (customize-group 'CSV) :active t 504 :help "Open a customization buffer to change CSV mode options"] 505 )) 506 507 (require 'sort) 508 509 (defsubst csv-not-looking-at-record () 510 "Return t if looking at blank or comment line, nil otherwise. 511 Assumes point is at beginning of line." 512 (looking-at paragraph-separate)) 513 514 (defun csv-interactive-args (&optional type) 515 "Get arg or field(s) and region interactively, offering sensible defaults. 516 Signal an error if the buffer is read-only. 517 If TYPE is noarg then return a list (beg end). 518 Otherwise, return a list (arg beg end), where arg is: 519 the raw prefix argument by default; 520 a single field index if TYPE is single; 521 a list of field indices or index ranges if TYPE is multiple. 522 Field defaults to the current prefix arg; if not set, prompt user. 523 524 A field index list consists of positive or negative integers or ranges, 525 separated by any non-integer characters. A range has the form m-n, 526 where m and n are positive or negative integers, m < n, and n defaults 527 to the last field index if omitted. 528 529 In transient mark mode, if the mark is not active then automatically 530 select and highlight CSV records around point, and query user. 531 The default field when read interactively is the current field." 532 ;; Must be run interactively to activate mark! 533 (let* ((arg current-prefix-arg) (default-field 1) 534 (region 535 (if (not (use-region-p)) 536 ;; Set region automatically: 537 (save-excursion 538 (if arg 539 (beginning-of-line) 540 (let ((lbp (line-beginning-position))) 541 (while (re-search-backward csv-separator-regexp lbp 1) 542 ;; Move as far as possible, i.e. to beginning of line. 543 (setq default-field (1+ default-field))))) 544 (if (csv-not-looking-at-record) 545 (error "Point must be within CSV records")) 546 (let ((startline (point))) 547 ;; Set mark at beginning of region: 548 (while (not (or (bobp) (csv-not-looking-at-record))) 549 (forward-line -1)) 550 (if (csv-not-looking-at-record) (forward-line 1)) 551 ;; Skip header lines: 552 (forward-line csv-header-lines) 553 (set-mark (point)) ; OK since in save-excursion 554 ;; Move point to end of region: 555 (goto-char startline) 556 (beginning-of-line) 557 (while (not (or (eobp) (csv-not-looking-at-record))) 558 (forward-line 1)) 559 ;; Show mark briefly if necessary: 560 (unless (and (pos-visible-in-window-p) 561 (pos-visible-in-window-p (mark))) 562 (exchange-point-and-mark) 563 (sit-for 1) 564 (exchange-point-and-mark)) 565 (when csv-confirm-region 566 (or (y-or-n-p "Region OK? ") 567 (error "Action aborted by user")) 568 (message nil)) ; clear y-or-n-p message 569 (list (region-beginning) (region-end)))) 570 ;; Use region set by user: 571 (list (region-beginning) (region-end))))) 572 (setq default-field (number-to-string default-field)) 573 (cond 574 ((eq type 'multiple) 575 (if arg 576 ;; Ensure that field is a list: 577 (or (consp arg) 578 (setq arg (list (prefix-numeric-value arg)))) 579 ;; Read field interactively, ignoring non-integers: 580 (setq arg 581 (mapcar 582 (lambda (x) 583 (if (string-match "-" x 1) ; not first character 584 ;; Return a range as a pair - the cdr may be nil: 585 (let ((m (substring x 0 (match-beginning 0))) 586 (n (substring x (match-end 0)))) 587 (cons (car (read-from-string m)) 588 (and (not (string= n "")) 589 (car (read-from-string n))))) 590 ;; Return a number as a number: 591 (car (read-from-string x)))) 592 (split-string 593 (read-string 594 "Fields (sequence of integers or ranges): " default-field) 595 "[^-+0-9]+"))))) 596 ((eq type 'single) 597 (if arg 598 (setq arg (prefix-numeric-value arg)) 599 (while (not (integerp arg)) 600 (setq arg (eval-minibuffer "Field (integer): " default-field)))))) 601 (if (eq type 'noarg) region (cons arg region)))) 602 603 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 604 ;;; Sorting by field 605 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 606 607 (defun csv-nextrecfun () 608 "Called by `csv-sort-fields-1' with point at end of previous record. 609 It moves point to the start of the next record. 610 It should move point to the end of the buffer if there are no more records." 611 (forward-line) 612 (while (and (not (eobp)) (csv-not-looking-at-record)) 613 (forward-line))) 614 615 (defun csv-sort-fields-1 (field beg end startkeyfun endkeyfun) 616 "Modified version of `sort-fields-1' that skips blank or comment lines. 617 618 FIELD is a single field index, and BEG and END specify the region to 619 sort. 620 621 STARTKEYFUN moves from the start of the record to the start of the key. 622 It may return either a non-nil value to be used as the key, or 623 else the key is the substring between the values of point after 624 STARTKEYFUN and ENDKEYFUN are called. If STARTKEYFUN is nil, the key 625 starts at the beginning of the record. 626 627 ENDKEYFUN moves from the start of the sort key to the end of the sort key. 628 ENDKEYFUN may be nil if STARTKEYFUN returns a value or if it would be the 629 same as ENDRECFUN." 630 (let ((tbl (syntax-table))) 631 (if (zerop field) (setq field 1)) 632 (unwind-protect 633 (save-excursion 634 (save-restriction 635 (narrow-to-region beg end) 636 (goto-char (point-min)) 637 (set-syntax-table sort-fields-syntax-table) 638 (sort-subr csv-descending 639 'csv-nextrecfun 'end-of-line 640 startkeyfun endkeyfun))) 641 (set-syntax-table tbl)))) 642 643 (defun csv-sort-fields (field beg end) 644 "Sort lines in region lexicographically by the ARGth field of each line. 645 If not set, the region defaults to the CSV records around point. 646 Fields are separated by `csv-separators' and null fields are allowed anywhere. 647 Field indices increase from 1 on the left or decrease from -1 on the right. 648 A prefix argument specifies a single field, otherwise prompt for field index. 649 Ignore blank and comment lines. The variable `sort-fold-case' 650 determines whether alphabetic case affects the sort order. 651 When called non-interactively, FIELD is a single field index; 652 BEG and END specify the region to sort." 653 ;; (interactive "*P\nr") 654 (interactive (csv-interactive-args 'single)) 655 (barf-if-buffer-read-only) 656 (csv-sort-fields-1 field beg end 657 (lambda () (csv-sort-skip-fields field) nil) 658 (lambda () (skip-chars-forward csv--skip-chars)))) 659 660 (defun csv-sort-numeric-fields (field beg end) 661 "Sort lines in region numerically by the ARGth field of each line. 662 If not set, the region defaults to the CSV records around point. 663 Fields are separated by `csv-separators'. 664 Null fields are allowed anywhere and sort as zeros. 665 Field indices increase from 1 on the left or decrease from -1 on the right. 666 A prefix argument specifies a single field, otherwise prompt for field index. 667 Specified non-null field must contain a number in each line of the region, 668 which may begin with \"0x\" or \"0\" for hexadecimal and octal values. 669 Otherwise, the number is interpreted according to sort-numeric-base. 670 Ignore blank and comment lines. 671 When called non-interactively, FIELD is a single field index; 672 BEG and END specify the region to sort." 673 ;; (interactive "*P\nr") 674 (interactive (csv-interactive-args 'single)) 675 (barf-if-buffer-read-only) 676 (csv-sort-fields-1 field beg end 677 (lambda () 678 (csv-sort-skip-fields field) 679 (let* ((case-fold-search t) 680 (base 681 (if (looking-at "\\(0x\\)[0-9a-f]\\|\\(0\\)[0-7]") 682 (cond ((match-beginning 1) 683 (goto-char (match-end 1)) 684 16) 685 ((match-beginning 2) 686 (goto-char (match-end 2)) 687 8) 688 (t nil))))) 689 (string-to-number (buffer-substring (point) 690 (save-excursion 691 (forward-sexp 1) 692 (point))) 693 (or base sort-numeric-base)))) 694 nil)) 695 696 (defun csv-reverse-region (beg end) 697 "Reverse the order of the lines in the region. 698 This is just a CSV-mode style interface to `reverse-region', which is 699 the function that should be used non-interactively. It takes two 700 point or marker arguments, BEG and END, delimiting the region." 701 ;; (interactive "*P\nr") 702 (interactive (csv-interactive-args 'noarg)) 703 (barf-if-buffer-read-only) 704 (reverse-region beg end)) 705 706 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 707 ;;; Moving by field 708 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 709 710 (defun csv-end-of-field () 711 "Skip forward over one field." 712 (skip-chars-forward " ") 713 ;; If the first character is a double quote, then we have a quoted 714 ;; value. 715 (when (eq (char-syntax (following-char)) ?\") 716 (forward-char) 717 (let ((ended nil)) 718 (while (not ended) 719 (cond ((not (eq (char-syntax (following-char)) ?\")) 720 (forward-char 1)) 721 ;; According to RFC-4180 (sec 2.7), quotes inside quoted strings 722 ;; are quoted by doubling the quote char: a,"b""c,",d 723 ;; FIXME: Maybe we should handle this via syntax-propertize? 724 ((let ((c (char-after (1+ (point))))) 725 (and c (eq (char-syntax c) ?\"))) 726 (forward-char 2)) 727 (t 728 (setq ended t)))))) 729 (skip-chars-forward csv--skip-chars)) 730 731 (defun csv--bof-p () 732 (or (bolp) 733 (memq (preceding-char) csv-separator-chars))) 734 735 (defun csv--eof-p () 736 (or (eolp) 737 (memq (following-char) csv-separator-chars))) 738 739 (defun csv-beginning-of-field () 740 "Skip backward over one field." 741 (skip-syntax-backward " ") 742 (if (eq (char-syntax (preceding-char)) ?\") 743 (goto-char (scan-sexps (point) -1))) 744 (skip-chars-backward csv--skip-chars)) 745 746 (defun csv-forward-field (arg) 747 "Move forward across one field, cf. `forward-sexp'. 748 With ARG, do it that many times. Negative arg -N means 749 move backward across N fields." 750 (interactive "p") 751 (if (< arg 0) 752 (csv-backward-field (- arg)) 753 (while (>= (setq arg (1- arg)) 0) 754 (if (or (bolp) 755 (when (and (not (eobp)) (eolp)) (forward-char) t)) 756 (while (and (not (eobp)) (csv-not-looking-at-record)) 757 (forward-line 1))) 758 (if (memq (following-char) csv-separator-chars) (forward-char)) 759 (csv-end-of-field)))) 760 761 (defun csv-backward-field (arg) 762 "Move backward across one field, cf. `backward-sexp'. 763 With ARG, do it that many times. Negative arg -N means 764 move forward across N fields." 765 (interactive "p") 766 (if (< arg 0) 767 (csv-forward-field (- arg)) 768 (while (>= (setq arg (1- arg)) 0) 769 (when (or (eolp) 770 (when (and (not (bobp)) (bolp)) (backward-char) t)) 771 (while (progn 772 (beginning-of-line) 773 (csv-not-looking-at-record)) 774 (backward-char)) 775 (end-of-line)) 776 (if (memq (preceding-char) csv-separator-chars) (backward-char)) 777 (csv-beginning-of-field)))) 778 779 (defun csv-tab-command () 780 "Skip to the next field on the same line. 781 Create a new field at end of line, if needed." 782 (interactive) 783 (skip-chars-forward csv--skip-chars) 784 (if (eolp) 785 (insert (car csv-separators)) 786 (forward-char 1))) 787 788 (defun csv-backtab-command () 789 "Skip to the beginning of the previous field." 790 (interactive) 791 (skip-chars-backward csv--skip-chars) 792 (forward-char -1) 793 (skip-chars-backward csv--skip-chars)) 794 795 (defun csv-sort-skip-fields (n &optional yank) 796 "Position point at the beginning of field N on the current line. 797 Fields are separated by `csv-separators'; null terminal field allowed. 798 Assumes point is initially at the beginning of the line. 799 YANK non-nil allows N to be greater than the number of fields, in 800 which case extend the record as necessary." 801 (if (> n 0) 802 ;; Skip across N - 1 fields. 803 (let ((i (1- n))) 804 (while (> i 0) 805 (csv-end-of-field) 806 (if (eolp) 807 (if yank 808 (if (> i 1) (insert (car csv-separators))) 809 (error "Line has too few fields: %s" 810 (buffer-substring 811 (save-excursion (beginning-of-line) (point)) 812 (save-excursion (end-of-line) (point))))) 813 (forward-char)) ; skip separator 814 (setq i (1- i)))) 815 (end-of-line) 816 ;; Skip back across -N - 1 fields. 817 (let ((i (1- (- n)))) 818 (while (> i 0) 819 (csv-beginning-of-field) 820 (if (bolp) 821 (error "Line has too few fields: %s" 822 (buffer-substring 823 (save-excursion (beginning-of-line) (point)) 824 (save-excursion (end-of-line) (point))))) 825 (backward-char) ; skip separator 826 (setq i (1- i))) 827 ;; Position at the front of the field 828 ;; even if moving backwards. 829 (csv-beginning-of-field)))) 830 831 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 832 ;;; Field index mode 833 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 834 835 ;; Based partly on paren.el 836 837 (defcustom csv-field-index-delay 0.125 838 "Time in seconds to delay before updating field index display." 839 :type '(number :tag "seconds")) 840 841 (defvar csv-field-index-idle-timer nil) 842 843 (defvar csv-field-index-string nil) 844 (make-variable-buffer-local 'csv-field-index-string) 845 846 (defvar csv-field-index-old nil) 847 (make-variable-buffer-local 'csv-field-index-old) 848 849 (define-minor-mode csv-field-index-mode 850 "Toggle CSV-Field-Index mode. 851 With prefix ARG, turn CSV-Field-Index mode on if and only if ARG is positive. 852 Returns the new status of CSV-Field-Index mode (non-nil means on). 853 When CSV-Field-Index mode is enabled, the current field index appears in 854 the mode line after `csv-field-index-delay' seconds of Emacs idle time." 855 :global t 856 :init-value t ; for documentation, since default is t 857 ;; This macro generates a function that first sets the mode 858 ;; variable, then runs the following code, runs the mode hooks, 859 ;; displays a message if interactive, updates the mode line and 860 ;; finally returns the variable value. 861 862 ;; First, always disable the mechanism (to avoid having two timers): 863 (when csv-field-index-idle-timer 864 (cancel-timer csv-field-index-idle-timer) 865 (setq csv-field-index-idle-timer nil)) 866 ;; Now, if the mode is on and any buffer is in CSV mode then 867 ;; re-initialize and enable the mechanism by setting up a new timer: 868 (if csv-field-index-mode 869 (if (memq t (mapcar (lambda (buffer) 870 (with-current-buffer buffer 871 (when (derived-mode-p 'csv-mode) 872 (setq csv-field-index-string nil 873 csv-field-index-old nil) 874 t))) 875 (buffer-list))) 876 (setq csv-field-index-idle-timer 877 (run-with-idle-timer csv-field-index-delay t 878 #'csv-field-index))) 879 ;; but if the mode is off then remove the display from the mode 880 ;; lines of all CSV buffers: 881 (mapc (lambda (buffer) 882 (with-current-buffer buffer 883 (when (derived-mode-p 'csv-mode) 884 (setq csv-field-index-string nil 885 csv-field-index-old nil) 886 (force-mode-line-update)))) 887 (buffer-list)))) 888 889 (defun csv--field-index () 890 (save-excursion 891 (let ((start (point)) 892 (field 0)) 893 (beginning-of-line) 894 (while (and (<= (point) start) 895 (not (eolp))) 896 (csv-end-of-field) 897 (unless (eolp) 898 (forward-char 1)) 899 (setq field (1+ field))) 900 field))) 901 902 (defun csv-field-index () 903 "Construct `csv-field-index-string' to display in mode line. 904 Called by `csv-field-index-idle-timer'." 905 (if (derived-mode-p 'csv-mode) 906 (let ((field (csv--field-index))) 907 (when (not (eq field csv-field-index-old)) 908 (setq csv-field-index-old field 909 csv-field-index-string 910 (and field (format "F%d" field))) 911 (force-mode-line-update))))) 912 913 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 914 ;;; Killing and yanking fields 915 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 916 917 (defvar csv-killed-fields nil 918 "A list of the fields or sub-records last killed by `csv-kill-fields'.") 919 920 (defun csv-kill-fields (fields beg end) 921 "Kill specified fields of each line in the region. 922 If not set, the region defaults to the CSV records around point. 923 Fields are separated by `csv-separators' and null fields are allowed anywhere. 924 Field indices increase from 1 on the left or decrease from -1 on the right. 925 The fields are stored for use by `csv-yank-fields'. Fields can be 926 specified in any order but are saved in increasing index order. 927 Ignore blank and comment lines. 928 929 When called interactively, a prefix argument specifies a single field, 930 otherwise prompt for a field list, which may include ranges in the form 931 m-n, where m < n and n defaults to the last field index if omitted. 932 933 When called non-interactively, FIELDS is a single field index or a 934 list of field indices, with ranges specified as (m.n) or (m), and BEG 935 and END specify the region to process." 936 ;; (interactive "*P\nr") 937 (interactive (csv-interactive-args 'multiple)) 938 (barf-if-buffer-read-only) 939 ;; Kill the field(s): 940 (setq csv-killed-fields nil) 941 (save-excursion 942 (save-restriction 943 (narrow-to-region beg end) 944 (goto-char (point-min)) 945 (if (or (cdr fields) (consp (car fields))) 946 (csv-kill-many-columns fields) 947 (csv-kill-one-column (car fields))))) 948 (setq csv-killed-fields (nreverse csv-killed-fields))) 949 950 (defun csv-kill-one-field (field) 951 "Kill field with index FIELD in current line. 952 Return killed text. Assumes point is at beginning of line." 953 ;; Move to start of field to kill: 954 (csv-sort-skip-fields field) 955 ;; Kill to end of field (cf. `kill-region'): 956 (prog1 (delete-and-extract-region 957 (point) 958 (progn (csv-end-of-field) (point))) 959 (if (eolp) 960 (unless (bolp) (delete-char -1)) ; Delete trailing separator at eol 961 (delete-char 1)))) ; or following separator otherwise. 962 963 (defun csv-kill-one-column (field) 964 "Kill field with index FIELD in all lines in (narrowed) buffer. 965 Save killed fields in `csv-killed-fields'. 966 Assumes point is at `point-min'. Called by `csv-kill-fields'. 967 Ignore blank and comment lines." 968 (while (not (eobp)) 969 (or (csv-not-looking-at-record) 970 (push (csv-kill-one-field field) csv-killed-fields)) 971 (forward-line))) 972 973 (defun csv-insert-column (field) 974 "Insert an empty column at point." 975 (interactive 976 (let ((cur (csv--field-index))) 977 (list (if (and (csv--eof-p) (not (csv--bof-p))) (1+ cur) cur)))) 978 (save-excursion 979 (goto-char (point-min)) 980 (while (not (eobp)) 981 (or (csv-not-looking-at-record) 982 (progn 983 (csv-sort-skip-fields field t) 984 (insert (car csv-separators)))) 985 (forward-line 1)) 986 (csv--jit-flush-columns))) 987 988 (defun csv-kill-many-columns (fields) 989 "Kill several fields in all lines in (narrowed) buffer. 990 FIELDS is an unordered list of field indices. 991 Save killed fields in increasing index order in `csv-killed-fields'. 992 Assumes point is at `point-min'. Called by `csv-kill-fields'. 993 Ignore blank and comment lines." 994 (if (eolp) (error "First record is empty")) 995 ;; Convert non-positive to positive field numbers: 996 (let ((last 1) (f fields)) 997 (csv-end-of-field) 998 (while (not (eolp)) 999 (forward-char) ; skip separator 1000 (csv-end-of-field) 1001 (setq last (1+ last))) ; last = # fields in first record 1002 (while f 1003 (cond ((consp (car f)) 1004 ;; Expand a field range: (m.n) -> m m+1 ... n-1 n. 1005 ;; If n is nil then it defaults to the number of fields. 1006 (let* ((range (car f)) (cdrf (cdr f)) 1007 (m (car range)) (n (cdr range))) 1008 (if (< m 0) (setq m (+ m last 1))) 1009 (if n 1010 (if (< n 0) (setq n (+ n last 1))) 1011 (setq n last)) 1012 (setq range (list n)) 1013 (while (> n m) (push (setq n (1- n)) range)) 1014 (setcar f (car range)) 1015 (setcdr f (cdr range)) 1016 (setcdr (setq f (last range)) cdrf))) 1017 ((zerop (car f)) (setcar f 1)) 1018 ((< (car f) 0) (setcar f (+ f last 1)))) 1019 (setq f (cdr f)))) 1020 (goto-char (point-min)) 1021 ;; Kill from right to avoid miscounting: 1022 (setq fields (sort fields #'>)) 1023 (while (not (eobp)) 1024 (or (csv-not-looking-at-record) 1025 (let ((fields fields) killed-fields field) 1026 (while fields 1027 (setq field (car fields) 1028 fields (cdr fields)) 1029 (beginning-of-line) 1030 (push (csv-kill-one-field field) killed-fields)) 1031 (push (mapconcat #'identity killed-fields (car csv-separators)) 1032 csv-killed-fields))) 1033 (forward-line))) 1034 1035 (defun csv-yank-fields (field beg end) 1036 "Yank fields as the ARGth field of each line in the region. 1037 ARG may be arbitrarily large and records are extended as necessary. 1038 If not set, the region defaults to the CSV records around point; 1039 if point is not in a CSV record then offer to yank as a new table. 1040 The fields yanked are those last killed by `csv-kill-fields'. 1041 Fields are separated by `csv-separators' and null fields are allowed anywhere. 1042 Field indices increase from 1 on the left or decrease from -1 on the right. 1043 A prefix argument specifies a single field, otherwise prompt for field index. 1044 Ignore blank and comment lines. When called non-interactively, FIELD 1045 is a single field index; BEG and END specify the region to process." 1046 ;; (interactive "*P\nr") 1047 (interactive (condition-case err 1048 (csv-interactive-args 'single) 1049 (error (list nil nil err)))) 1050 (barf-if-buffer-read-only) 1051 (if (null beg) 1052 (if (y-or-n-p (concat (error-message-string end) 1053 ". Yank as a new table? ")) 1054 (csv-yank-as-new-table) 1055 (error (error-message-string end))) 1056 (if (<= field 0) (setq field (1+ field))) 1057 (save-excursion 1058 (save-restriction 1059 (narrow-to-region beg end) 1060 (goto-char (point-min)) 1061 (let ((fields csv-killed-fields)) 1062 (while (not (eobp)) 1063 (unless (csv-not-looking-at-record) 1064 ;; Yank at start of specified field if possible, 1065 ;; otherwise yank at end of record: 1066 (if (zerop field) 1067 (end-of-line) 1068 (csv-sort-skip-fields field 'yank)) 1069 (and (eolp) (insert (car csv-separators))) 1070 (when fields 1071 (insert (car fields)) 1072 (setq fields (cdr fields))) 1073 (or (eolp) (insert (car csv-separators)))) 1074 (forward-line))))))) 1075 1076 (defun csv-yank-as-new-table () 1077 "Yank fields as a new table starting at point. 1078 The fields yanked are those last killed by `csv-kill-fields'." 1079 (interactive "*") 1080 (let ((fields csv-killed-fields)) 1081 (while fields 1082 (insert (car fields) ?\n) 1083 (setq fields (cdr fields))))) 1084 1085 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1086 ;;; Aligning fields 1087 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1088 1089 (defun csv--make-overlay (beg end &optional buffer front-advance rear-advance props) 1090 (let ((o (make-overlay beg end buffer front-advance rear-advance))) 1091 (overlay-put o 'csv t) 1092 (while props 1093 (overlay-put o (pop props) (pop props))) 1094 o)) 1095 1096 (defun csv--delete-overlay (o) 1097 (and (overlay-get o 'csv) (delete-overlay o))) 1098 1099 (defun csv--column-widths (beg end) 1100 "Return a list of two lists (COLUMN-WIDTHS FIELD-WIDTHS). 1101 COLUMN-WIDTHS is a list of elements (WIDTH START END) 1102 indicating the widths of the columns after point (and the position of the 1103 widest field that determined the overall width). 1104 FIELD-WIDTHS contains the widths of each individual field after 1105 point." 1106 (let ((column-widths '()) 1107 (field-widths '())) 1108 (goto-char beg) 1109 ;; Construct list of column widths: 1110 (while (< (point) end) ; for each record... 1111 (or (csv-not-looking-at-record) 1112 (let ((w column-widths) 1113 (col (current-column)) 1114 (beg (point)) 1115 field-width) 1116 (while (not (eolp)) 1117 (csv-end-of-field) 1118 (setq field-width (- (current-column) col)) 1119 (push field-width field-widths) 1120 (if w 1121 (if (> field-width (caar w)) 1122 (setcar w (list field-width beg (point)))) 1123 (setq w (list (list field-width beg (point))) 1124 column-widths (nconc column-widths w))) 1125 (or (eolp) (forward-char)) ; Skip separator. 1126 (setq w (cdr w) col (current-column) beg (point))))) 1127 (forward-line)) 1128 (list column-widths (nreverse field-widths)))) 1129 1130 (defun csv-align-fields (hard beg end) 1131 "Align all the fields in the region to form columns. 1132 The alignment style is specified by `csv-align-style'. The number of 1133 spaces specified by `csv-align-padding' appears after each separator. 1134 Use soft alignment done by displaying virtual white space after the 1135 separators unless invoked with an argument, in which case insert real 1136 space characters into the buffer after the separators. 1137 Unalign first (see `csv-unalign-fields'). Ignore blank and comment lines. 1138 1139 In hard-aligned records, separators become invisible whenever 1140 `buffer-invisibility-spec' is non-nil. In soft-aligned records, make 1141 separators invisible if and only if `buffer-invisibility-spec' is 1142 non-nil when the records are aligned; this can be changed only by 1143 re-aligning. \(Unaligning always makes separators visible.) 1144 1145 When called non-interactively, use hard alignment if HARD is non-nil; 1146 BEG and END specify the region to align. 1147 If there is no selected region, default to the whole buffer." 1148 (interactive (cons current-prefix-arg 1149 (if (use-region-p) 1150 (list (region-beginning) (region-end)) 1151 (list (point-min) (point-max))))) 1152 ;; FIXME: Use csv--jit-align when applicable! 1153 (setq end (copy-marker end)) 1154 (csv-unalign-fields hard beg end) ; If hard then barfs if buffer read only. 1155 (save-excursion 1156 (pcase-let ((`(,column-widths ,field-widths) (csv--column-widths beg end))) 1157 (save-restriction 1158 (narrow-to-region beg end) 1159 (set-marker end nil) 1160 1161 ;; Align fields: 1162 (goto-char (point-min)) 1163 (while (not (eobp)) ; for each record... 1164 (unless (csv-not-looking-at-record) 1165 (let ((w column-widths) 1166 (column 0)) ;Desired position of left-side of this column. 1167 (while (and w (not (eolp))) 1168 (let* ((beg (point)) 1169 (align-padding (if (bolp) 0 csv-align-padding)) 1170 (left-padding 0) (right-padding 0) 1171 (field-width (pop field-widths)) 1172 (column-width (car (pop w))) 1173 (x (- column-width field-width))) ; Required padding. 1174 (csv-end-of-field) 1175 (set-marker end (point)) ; End of current field. 1176 ;; beg = beginning of current field 1177 ;; end = (point) = end of current field 1178 1179 ;; Compute required padding: 1180 (cond 1181 ((eq csv-align-style 'left) 1182 ;; Left align -- pad on the right: 1183 (setq left-padding align-padding 1184 right-padding x)) 1185 ((eq csv-align-style 'right) 1186 ;; Right align -- pad on the left: 1187 (setq left-padding (+ align-padding x))) 1188 ((eq csv-align-style 'auto) 1189 ;; Auto align -- left align text, right align numbers: 1190 (if (string-match "\\`[-+.[:digit:]]+\\'" 1191 (buffer-substring beg (point))) 1192 ;; Right align -- pad on the left: 1193 (setq left-padding (+ align-padding x)) 1194 ;; Left align -- pad on the right: 1195 (setq left-padding align-padding 1196 right-padding x))) 1197 ((eq csv-align-style 'centre) 1198 ;; Centre -- pad on both left and right: 1199 (let ((y (/ x 2))) ; truncated integer quotient 1200 (setq left-padding (+ align-padding y) 1201 right-padding (- x y))))) 1202 1203 (cond 1204 (hard ;; Hard alignment... 1205 (when (> left-padding 0) ; Pad on the left. 1206 ;; Insert spaces before field: 1207 (if (= beg end) ; null field 1208 (insert (make-string left-padding ?\ )) 1209 (goto-char beg) ; beginning of current field 1210 (insert (make-string left-padding ?\ )) 1211 (goto-char end))) ; end of current field 1212 (unless (eolp) 1213 (if (> right-padding 0) ; pad on the right 1214 ;; Insert spaces after field: 1215 (insert (make-string right-padding ?\ ))) 1216 ;; Make separator (potentially) invisible; 1217 ;; in Emacs 21.3, neighbouring overlays 1218 ;; conflict, so use the following only 1219 ;; with hard alignment: 1220 (csv--make-overlay (point) (1+ (point)) nil t nil 1221 '(invisible csv evaporate t)) 1222 (forward-char))) ; skip separator 1223 1224 ;; Soft alignment... 1225 ((or (memq 'csv buffer-invisibility-spec) 1226 ;; For TSV, hidden or not doesn't make much difference, 1227 ;; but the behavior is slightly better when we "hide" 1228 ;; the TABs with a `display' property than if we add 1229 ;; before/after-strings. 1230 (tsv--mode-p)) 1231 1232 ;; Hide separators... 1233 ;; Merge right-padding from previous field 1234 ;; with left-padding from this field: 1235 (if (zerop column) 1236 (when (> left-padding 0) 1237 ;; Display spaces before first field 1238 ;; by overlaying first character: 1239 (csv--make-overlay 1240 beg (1+ beg) nil nil nil 1241 `(before-string ,(make-string left-padding ?\ )))) 1242 ;; Display separator as spaces: 1243 (with-silent-modifications 1244 (put-text-property 1245 (1- beg) beg 1246 'display `(space :align-to 1247 ,(+ left-padding column))))) 1248 (unless (eolp) (forward-char)) ; Skip separator. 1249 (setq column (+ column column-width align-padding))) 1250 1251 (t ;; Do not hide separators... 1252 (let ((overlay (csv--make-overlay beg (point) nil nil t))) 1253 (when (> left-padding 0) ; Pad on the left. 1254 ;; Display spaces before field: 1255 (overlay-put overlay 'before-string 1256 (make-string left-padding ?\ ))) 1257 (unless (eolp) 1258 (if (> right-padding 0) ; Pad on the right. 1259 ;; Display spaces after field: 1260 (overlay-put 1261 overlay 1262 'after-string (make-string right-padding ?\ ))) 1263 (forward-char)))) ; Skip separator. 1264 1265 ))))) 1266 (forward-line))))) 1267 (set-marker end nil)) 1268 1269 (defun csv-unalign-fields (hard beg end) 1270 "Undo soft alignment and optionally remove redundant white space. 1271 Undo soft alignment introduced by `csv-align-fields'. If invoked with 1272 an argument then also remove all spaces and tabs around separators. 1273 Also make all invisible separators visible again. 1274 Ignore blank and comment lines. When called non-interactively, remove 1275 spaces and tabs if HARD non-nil; BEG and END specify region to unalign. 1276 If there is no selected region, default to the whole buffer." 1277 (interactive (cons current-prefix-arg 1278 (if (use-region-p) 1279 (list (region-beginning) (region-end)) 1280 (list (point-min) (point-max))))) 1281 ;; Remove any soft alignment: 1282 (mapc #'csv--delete-overlay (overlays-in beg end)) 1283 (with-silent-modifications 1284 (remove-list-of-text-properties beg end '(display invisible))) 1285 (when hard 1286 (barf-if-buffer-read-only) 1287 ;; Remove any white-space padding around separators: 1288 (save-excursion 1289 (save-restriction 1290 (narrow-to-region beg end) 1291 (goto-char (point-min)) 1292 (while (not (eobp)) 1293 (or (csv-not-looking-at-record) 1294 (while (not (eolp)) 1295 ;; Delete horizontal white space forward: 1296 ;; (delete-horizontal-space) 1297 ;; This relies on left-to-right argument evaluation; 1298 ;; see info node (elisp) Function Forms. 1299 (delete-region (point) 1300 (+ (point) (skip-chars-forward " \t"))) 1301 (csv-end-of-field) 1302 ;; Delete horizontal white space backward: 1303 ;; (delete-horizontal-space t) 1304 (delete-region (point) 1305 (+ (point) (skip-chars-backward " \t"))) 1306 (or (eolp) (forward-char)))) 1307 (forward-line)))))) 1308 1309 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1310 ;;; Transposing rows and columns 1311 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1312 1313 (defun csv-transpose (beg end) 1314 "Rewrite rows (which may have different lengths) as columns. 1315 Null fields are introduced as necessary within records but are 1316 stripped from the ends of records. Preserve soft alignment. 1317 This function is its own inverse. Ignore blank and comment lines. 1318 When called non-interactively, BEG and END specify region to process." 1319 ;; (interactive "*P\nr") 1320 (interactive (csv-interactive-args 'noarg)) 1321 (barf-if-buffer-read-only) 1322 (save-excursion 1323 (save-restriction 1324 (narrow-to-region beg end) 1325 (goto-char (point-min)) 1326 ;; Delete rows and collect them as a reversed list of lists of 1327 ;; fields, skipping comment and blank lines: 1328 (let ((sep (car csv-separators)) 1329 (align (overlays-in beg end)) 1330 rows columns) 1331 ;; Remove soft alignment if necessary: 1332 (when align 1333 (mapc #'csv--delete-overlay align) 1334 (setq align t)) 1335 (while (not (eobp)) 1336 (if (csv-not-looking-at-record) 1337 ;; Skip blank and comment lines: 1338 (forward-line) 1339 (let ((lep (line-end-position))) 1340 (push 1341 (csv--collect-fields lep) 1342 rows) 1343 (delete-region (point) lep) 1344 (or (eobp) (delete-char 1))))) 1345 ;; Rows must have monotonic decreasing lengths to be 1346 ;; transposable, so ensure this by padding with null fields. 1347 ;; rows is currently a reversed list of field lists, which 1348 ;; must therefore have monotonic increasing lengths. 1349 (let ((oldlen (length (car rows))) newlen 1350 (r (cdr rows))) 1351 (while r 1352 (setq newlen (length (car r))) 1353 (if (< newlen oldlen) 1354 (nconc (car r) (make-list (- oldlen newlen) nil)) 1355 (setq oldlen newlen)) 1356 (setq r (cdr r)))) 1357 ;; Collect columns as a reversed list of lists of fields: 1358 (while rows 1359 (let (column (r rows) row) 1360 (while r 1361 (setq row (car r)) 1362 ;; Provided it would not be a trailing null field, push 1363 ;; field onto column: 1364 (if (or column (string< "" (car row))) 1365 (push (car row) column)) 1366 ;; Pop field off row: 1367 (setcar r (cdr row)) 1368 ;; If row is now empty then remove it: 1369 (or (car r) (setq rows (cdr rows))) 1370 (setq r (cdr r))) 1371 (push column columns))) 1372 ;; Insert columns into buffer as rows: 1373 (setq columns (nreverse columns)) 1374 (while columns 1375 (insert (mapconcat #'identity (car columns) sep) ?\n) 1376 (setq columns (cdr columns))) 1377 ;; Re-do soft alignment if necessary: 1378 (if align (csv-align-fields nil (point-min) (point-max))))))) 1379 1380 (defun csv--collect-fields (row-end-position) 1381 "Collect the fields of a row. 1382 Splits a row into fields, honoring quoted fields, and returns 1383 the list of fields. ROW-END-POSITION is the end-of-line position. 1384 point is assumed to be at the beginning of the line." 1385 (let ((csv-field-quotes-regexp (apply #'concat `("[" ,@csv-field-quotes "]"))) 1386 (row-text (buffer-substring-no-properties (point) row-end-position)) 1387 fields field-start) 1388 (if (not (string-match csv-field-quotes-regexp row-text)) 1389 (split-string row-text csv-separator-regexp) 1390 (save-excursion 1391 (while (< (setq field-start (point)) row-end-position) 1392 ;; csv-forward-field will skip a separator if point is on 1393 ;; it, and we'll miss an empty field 1394 (unless (memq (following-char) csv-separator-chars) 1395 (csv-forward-field 1)) 1396 (push 1397 (buffer-substring-no-properties field-start (point)) 1398 fields) 1399 (if (memq (following-char) csv-separator-chars) 1400 (forward-char))) 1401 (nreverse fields))))) 1402 1403 (defvar-local csv--header-line nil) 1404 (defvar-local csv--header-hscroll nil) 1405 (defvar-local csv--header-string nil) 1406 1407 (defun csv-header-line (&optional use-current-line) 1408 "Set/unset the header line. 1409 If the optional prefix arg USE-CURRENT-LINE is nil, use the first line 1410 as the header line. 1411 If there is already a header line, then unset the header line." 1412 (interactive "P") 1413 (if csv--header-line 1414 (progn 1415 (delete-overlay csv--header-line) 1416 (setq csv--header-line nil) 1417 (kill-local-variable 'header-line-format)) 1418 (save-excursion 1419 (unless use-current-line (goto-char (point-min))) 1420 (setq csv--header-line (make-overlay (line-beginning-position) 1421 (line-end-position) 1422 nil nil t)) 1423 (overlay-put csv--header-line 'modification-hooks 1424 '(csv--header-flush))) 1425 (csv--header-flush) 1426 ;; These are introduced in Emacs 29. 1427 (unless (boundp 'header-line-indent) 1428 (setq-local header-line-indent "" 1429 header-line-indent-width 0)) 1430 (setq header-line-format 1431 '("" header-line-indent (:eval (csv--header-string)))))) 1432 1433 (defun csv--header-flush (&rest _) 1434 ;; Force re-computation of the header-line. 1435 (setq csv--header-hscroll nil)) 1436 1437 (defun csv--header-string () 1438 ;; FIXME: Won't work with multiple windows showing that same buffer. 1439 (if (eql (window-hscroll) csv--header-hscroll) 1440 csv--header-string 1441 (setq csv--header-hscroll (window-hscroll)) 1442 (setq csv--header-string 1443 (csv--compute-header-string)))) 1444 1445 (defun csv--compute-header-string () 1446 (with-demoted-errors "csv--compute-header-string %S" 1447 (save-excursion 1448 (goto-char (overlay-start csv--header-line)) 1449 ;; Re-set the line-end-position, just in case. 1450 (move-overlay csv--header-line (point) (line-end-position)) 1451 (jit-lock-fontify-now (point) (line-end-position)) 1452 ;; Not sure why it is sometimes nil! 1453 (move-to-column (or csv--header-hscroll 0)) 1454 (let ((str (replace-regexp-in-string 1455 "%" "%%" (buffer-substring (point) (line-end-position)))) 1456 (i 0)) 1457 (while (and i (< i (length str))) 1458 (let ((prop (get-text-property i 'display str))) 1459 (and (eq (car-safe prop) 'space) 1460 (eq (car-safe (cdr prop)) :align-to) 1461 (let* ((x (nth 2 prop)) 1462 (nexti (next-single-property-change i 'display str)) 1463 (newprop 1464 `(space :align-to 1465 (+ ,(if (numberp x) 1466 (- x (or csv--header-hscroll 0)) 1467 `(- ,x csv--header-hscroll)) 1468 header-line-indent-width)))) 1469 (put-text-property i (or nexti (length str)) 1470 'display newprop str) 1471 (setq i nexti)))) 1472 (setq i (next-single-property-change i 'display str))) 1473 (concat (propertize " " 'display '((space :align-to 0))) str))))) 1474 1475 ;;; Auto-alignment 1476 1477 (defcustom csv-align-max-width 40 1478 "Maximum width of a column in `csv-align-mode'. 1479 This does not apply to the last column (for which the usual `truncate-lines' 1480 setting works better)." 1481 :type 'integer) 1482 1483 (defcustom csv-align-min-width 1 1484 "Minimum width of a column in `csv-align-mode'." 1485 :type 'integer) 1486 1487 (defvar-local csv--config-column-widths nil 1488 "Settings per column, stored as a list indexed by the column.") 1489 1490 (defun csv-align--set-column (column value) 1491 (let ((len (length csv--config-column-widths))) 1492 (if (< len column) 1493 (setq csv--config-column-widths 1494 (nconc csv--config-column-widths (make-list (- column len) nil)))) 1495 (setf (nth (1- column) csv--config-column-widths) value))) 1496 1497 (defun csv-align-set-column-width (column width) 1498 "Set the max WIDTH to use for COLUMN." 1499 (interactive 1500 (let* ((field (or (csv--field-index) 1)) 1501 (curwidth (nth (1- field) csv--config-column-widths))) 1502 (list field 1503 (cond 1504 ((numberp current-prefix-arg) 1505 current-prefix-arg) 1506 (current-prefix-arg 1507 (read-number (format "Column width (for field %d): " field) 1508 curwidth)) 1509 (t (if curwidth nil (csv--ellipsis-width))))))) 1510 (when (eql width csv-align-max-width) 1511 (setq width nil)) 1512 (csv-align--set-column column width) 1513 (jit-lock-refontify)) 1514 1515 (defvar-local csv--jit-columns nil) 1516 1517 (defun csv--jit-flush-columns () 1518 "Throw away all cached info about column widths." 1519 ;; FIXME: Maybe we should kill its overlays as well. 1520 (setq csv--jit-columns nil)) 1521 1522 (defun csv--jit-merge-columns (column-widths) 1523 ;; FIXME: The incremental update (delayed by jit-lock-context-time) of column 1524 ;; width is a bit jarring at times. It's OK while scrolling or when 1525 ;; extending a column, but not right when enabling the csv-align-mode or 1526 ;; when shortening the longest field (or deleting the line containing it), 1527 ;; because in that case we have *several* cascaded updates, e.g.: 1528 ;; - Remove the line with the longest field of column N. 1529 ;; - Edit some line: this line is updated as if its field was the widest, 1530 ;; hence its subsequent fields are too much to the left. 1531 ;; - The rest is updated starting from the first few lines (according 1532 ;; to jit-lock-chunk-size). 1533 ;; - After the first few lines, come the next set of few lines, 1534 ;; which may cause the previous few lines to need refresh again. 1535 ;; - etc.. until arriving again at the edited line which is re-aligned 1536 ;; again. 1537 ;; - etc.. until the end of the windows, potentially causing yet more 1538 ;; refreshes as we discover yet-wider fields for this column. 1539 (let ((old-columns csv--jit-columns) 1540 (changed nil)) 1541 (while (and old-columns column-widths) 1542 (when (or (> (caar column-widths) (caar old-columns)) 1543 ;; Apparently modification-hooks aren't run when the 1544 ;; whole text containing the overlay is deleted (e.g. 1545 ;; the whole line), so detect this case here. 1546 ;; It's a bit too late, but better than never. 1547 (null (overlay-buffer (cdar old-columns)))) 1548 (setq changed t) ;; Return non-nil if some existing column changed. 1549 (pcase-let ((`(,width ,beg ,end) (car column-widths))) 1550 (setf (caar old-columns) width) 1551 (move-overlay (cdar old-columns) beg end))) 1552 (setq old-columns (cdr old-columns)) 1553 (setq column-widths (cdr column-widths))) 1554 (when column-widths 1555 ;; New columns appeared. 1556 (setq csv--jit-columns 1557 (nconc csv--jit-columns 1558 (mapcar (lambda (x) 1559 (pcase-let* 1560 ((`(,width ,beg ,end) x) 1561 (ol (make-overlay beg end))) 1562 (overlay-put ol 'csv-width t) 1563 (overlay-put ol 'evaporate t) 1564 (overlay-put ol 'modification-hooks 1565 (list #'csv--jit-width-change)) 1566 (cons width ol))) 1567 column-widths)))) 1568 changed)) 1569 1570 (defun csv--jit-width-change (ol after _beg _end &optional len) 1571 (when (and after (> len 0)) 1572 ;; (let ((x (rassq ol csv--jit-columns))) 1573 ;; (when x (setf (car x) -1))) 1574 (delete-overlay ol))) 1575 1576 (defun csv--jit-unalign (beg end) 1577 (with-silent-modifications 1578 (remove-text-properties beg end 1579 '( display nil csv--jit nil invisible nil 1580 cursor-sensor-functions nil csv--revealed nil)) 1581 (remove-overlays beg end 'csv--jit t))) 1582 1583 (defun csv--jit-flush (beg end) 1584 "Cause all the buffer (except for the BEG...END region) to be re-aligned." 1585 (cl-assert (>= end beg)) 1586 ;; The buffer shouldn't have changed since beg/end were computed, 1587 ;; but just in case, let's make sure they're still sane. 1588 (when (< beg (point-min)) 1589 (setq beg (point-min) end (max end beg))) 1590 (when (< (point-max) end) 1591 (setq end (point-max) beg (min end beg))) 1592 (let ((pos (point-min))) 1593 (while (and (< pos beg) 1594 (setq pos (text-property-any pos beg 'csv--jit t))) 1595 (jit-lock-refontify 1596 pos (setq pos (or (text-property-any pos beg 'csv--jit nil) beg)))) 1597 (setq pos end) 1598 (while (and (< pos (point-max)) 1599 (setq pos (text-property-any pos (point-max) 'csv--jit t))) 1600 (jit-lock-refontify 1601 pos (setq pos (or (text-property-any pos (point-max) 'csv--jit nil) 1602 (point-max)))))) 1603 (csv--header-flush)) 1604 1605 (defun csv--ellipsis-width () 1606 (let ((ellipsis 1607 (when standard-display-table 1608 (display-table-slot standard-display-table 1609 'selective-display)))) 1610 (if ellipsis (length ellipsis) 3))) 1611 1612 (defun csv-align--cursor-truncated (window oldpos dir) 1613 ;; FIXME: Neither the `entered' nor the `left' event are guaranteed 1614 ;; to be sent, and for the `left' case, even when we do get called, 1615 ;; it may be unclear where the revealed text was (it's somewhere around 1616 ;; `oldpos', but that position can be stale). 1617 ;; Worse, if we have several windows displaying the buffer, when one 1618 ;; cursor leaves we may need to keep the text revealed because of 1619 ;; another window's cursor. 1620 (let* ((prop (if (eq dir 'entered) 'invisible 'csv--revealed)) 1621 (pos (cond 1622 ((eq dir 'entered) (window-point window)) 1623 (t (max (point-min) 1624 (min (point-max) 1625 (or oldpos (window-point window))))))) 1626 (start (cond 1627 ((and (> pos (point-min)) 1628 (eq (get-text-property (1- pos) prop) 'csv-truncate)) 1629 (or (previous-single-property-change pos prop) (point-min))) 1630 (t pos))) 1631 (end (if (eq (get-text-property pos prop) 'csv-truncate) 1632 (or (next-single-property-change pos prop) (point-max)) 1633 pos))) 1634 (unless (eql start end) 1635 (with-silent-modifications 1636 (put-text-property start end 1637 (if (eq dir 'entered) 'csv--revealed 'invisible) 1638 'csv-truncate) 1639 (remove-text-properties start end (list prop)))))) 1640 1641 (defun csv--jit-align (beg end) 1642 (save-excursion 1643 ;; This is run with inhibit-modification-hooks set, so the overlays' 1644 ;; modification-hook doesn't work :-( 1645 (and csv--header-line 1646 (<= beg (overlay-end csv--header-line)) 1647 (>= end (overlay-start csv--header-line)) 1648 (csv--header-flush)) 1649 ;; First, round up to a whole number of lines. 1650 (goto-char end) 1651 (unless (bolp) (forward-line 1) (setq end (point))) 1652 (goto-char beg) 1653 (unless (bolp) (forward-line 1) (setq beg (point))) 1654 (csv--jit-unalign beg end) 1655 (put-text-property beg end 'csv--jit t) 1656 1657 (pcase-let* ((`(,column-widths ,field-widths) (csv--column-widths beg end)) 1658 (changed (csv--jit-merge-columns column-widths)) 1659 (ellipsis-width (csv--ellipsis-width))) 1660 (when changed 1661 ;; Do it after the current redisplay is over. 1662 (run-with-timer jit-lock-context-time nil #'csv--jit-flush beg end)) 1663 1664 ;; Align fields: 1665 (goto-char beg) 1666 (while (< (point) end) 1667 (unless (csv-not-looking-at-record) 1668 (let ((w csv--jit-columns) 1669 (widths-config csv--config-column-widths) 1670 (column 0)) ;Desired position of left-side of this column. 1671 (while (and w (not (eolp))) 1672 (let* ((field-beg (point)) 1673 (width-config (pop widths-config)) 1674 (align-padding (if (bolp) 0 csv-align-padding)) 1675 (left-padding 0) (right-padding 0) 1676 (field-width (pop field-widths)) 1677 (column-width 1678 (min (max csv-align-min-width 1679 (car (pop w))) 1680 (or width-config 1681 ;; Don't apply csv-align-max-width 1682 ;; to the last field! 1683 (if w csv-align-max-width 1684 most-positive-fixnum)))) 1685 (x (- column-width field-width)) ; Required padding. 1686 (truncate nil)) 1687 (csv-end-of-field) 1688 ;; beg = beginning of current field 1689 ;; end = (point) = end of current field 1690 (when (< x 0) 1691 (setq truncate (max column 1692 (+ column column-width 1693 align-padding (- ellipsis-width)))) 1694 (setq x 0)) 1695 ;; Compute required padding: 1696 (pcase csv-align-style 1697 ('left 1698 ;; Left align -- pad on the right: 1699 (setq left-padding align-padding 1700 right-padding x)) 1701 ('right 1702 ;; Right align -- pad on the left: 1703 (setq left-padding (+ align-padding x))) 1704 ('auto 1705 ;; Auto align -- left align text, right align numbers: 1706 (if (string-match "\\`[-+.[:digit:]]+\\'" 1707 (buffer-substring field-beg (point))) 1708 ;; Right align -- pad on the left: 1709 (setq left-padding (+ align-padding x)) 1710 ;; Left align -- pad on the right: 1711 (setq left-padding align-padding 1712 right-padding x))) 1713 ('centre 1714 ;; Centre -- pad on both left and right: 1715 (let ((y (/ x 2))) ; truncated integer quotient 1716 (setq left-padding (+ align-padding y) 1717 right-padding (- x y))))) 1718 1719 (cond 1720 1721 ((or (memq 'csv buffer-invisibility-spec) 1722 ;; For TSV, hidden or not doesn't make much difference, 1723 ;; but the behavior is slightly better when we "hide" 1724 ;; the TABs with a `display' property than if we add 1725 ;; before/after-strings. 1726 (tsv--mode-p)) 1727 1728 ;; Hide separators... 1729 ;; Merge right-padding from previous field 1730 ;; with left-padding from this field: 1731 (if (zerop column) 1732 (when (> left-padding 0) 1733 ;; Display spaces before first field 1734 ;; by overlaying first character: 1735 (csv--make-overlay 1736 field-beg (1+ field-beg) nil nil nil 1737 `(before-string ,(make-string left-padding ?\ ) 1738 csv--jit t))) 1739 ;; Display separator as spaces: 1740 (with-silent-modifications 1741 (put-text-property 1742 (1- field-beg) field-beg 1743 'display `(space :align-to 1744 ,(+ left-padding column)))))) 1745 1746 (t ;; Do not hide separators... 1747 (let ((overlay (csv--make-overlay field-beg (point) 1748 nil nil t 1749 '(csv--jit t)))) 1750 (when (> left-padding 0) ; Pad on the left. 1751 ;; Display spaces before field: 1752 (overlay-put overlay 'before-string 1753 (make-string left-padding ?\ ))) 1754 (unless (eolp) 1755 (if (> right-padding 0) ; Pad on the right. 1756 ;; Display spaces after field: 1757 (overlay-put 1758 overlay 1759 'after-string (make-string right-padding ?\ ))))))) 1760 (setq column (+ column column-width align-padding)) 1761 ;; Do it after applying the property, so `move-to-column' can 1762 ;; take it into account. 1763 (when truncate 1764 (let ((trunc-pos 1765 (save-excursion 1766 ;; ¡¡ BIG UGLY HACK !! 1767 ;; `current-column' and `move-to-column' count 1768 ;; text hidden with an ellipsis "as if" it were 1769 ;; fully visible, which is completely wrong here, 1770 ;; so circumvent this by temporarily pretending 1771 ;; that `csv-truncate' is fully invisible (which 1772 ;; isn't quite right either, but should work 1773 ;; just well enough for us here). 1774 (let ((buffer-invisibility-spec 1775 buffer-invisibility-spec)) 1776 (add-to-invisibility-spec 'csv-truncate) 1777 (move-to-column truncate)) 1778 (point)))) 1779 (put-text-property trunc-pos (point) 1780 'invisible 'csv-truncate) 1781 (when (> (- (point) trunc-pos) 1) 1782 ;; Arrange to temporarily untruncate the string when 1783 ;; cursor moves into it. 1784 ;; FIXME: This only works if 1785 ;; `global-disable-point-adjustment' is non-nil! 1786 ;; Arguably this should be fixed by making 1787 ;; point-adjustment code pay attention to 1788 ;; cursor-sensor-functions! 1789 (put-text-property 1790 (1+ trunc-pos) (point) 1791 'cursor-sensor-functions 1792 (list #'csv-align--cursor-truncated))))) 1793 (unless (eolp) (forward-char)) ; Skip separator. 1794 )))) 1795 (forward-line))) 1796 `(jit-lock-bounds ,beg . ,end))) 1797 1798 (define-minor-mode csv-align-mode 1799 "Align columns on the fly." 1800 :global nil 1801 (csv-unalign-fields nil (point-min) (point-max)) ;Just in case. 1802 (cond 1803 (csv-align-mode 1804 (add-to-invisibility-spec '(csv-truncate . t)) 1805 (kill-local-variable 'csv--jit-columns) 1806 (cursor-sensor-mode 1) 1807 (when (fboundp 'header-line-indent-mode) 1808 (header-line-indent-mode)) 1809 (jit-lock-register #'csv--jit-align) 1810 (jit-lock-refontify)) 1811 (t 1812 (remove-from-invisibility-spec '(csv-truncate . t)) 1813 (jit-lock-unregister #'csv--jit-align) 1814 (csv--jit-unalign (point-min) (point-max)))) 1815 (csv--header-flush)) 1816 1817 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1818 ;;; Separator guessing 1819 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1820 1821 (defvar csv--preferred-separators 1822 '(?, ?\; ?\t) 1823 "Preferred separator characters in case of a tied score.") 1824 1825 (defun csv-guess-set-separator () 1826 "Guess and set the CSV separator of the current buffer. 1827 1828 Add it to the mode hook to have CSV mode guess and set the 1829 separator automatically when visiting a buffer: 1830 1831 (add-hook \\='csv-mode-hook \\='csv-guess-set-separator)" 1832 (interactive) 1833 (let ((sep (csv-guess-separator 1834 (buffer-substring-no-properties 1835 (point-min) 1836 ;; We're probably only going to look at the first 2048 1837 ;; or so chars, but take more than we probably need to 1838 ;; minimize the chance of breaking the input in the 1839 ;; middle of a (long) row. 1840 (min 8192 (point-max))) 1841 2048))) 1842 (when sep 1843 (csv-set-separator sep)))) 1844 1845 (defun csv-guess-separator (text &optional cutoff) 1846 "Return a guess of which character is the CSV separator in TEXT." 1847 (let ((best-separator nil) 1848 (best-score 0)) 1849 (dolist (candidate (csv--separator-candidates text cutoff)) 1850 (let ((candidate-score 1851 (csv--separator-score candidate text cutoff))) 1852 (when (or (> candidate-score best-score) 1853 (and (= candidate-score best-score) 1854 (member candidate csv--preferred-separators))) 1855 (setq best-separator candidate) 1856 (setq best-score candidate-score)))) 1857 best-separator)) 1858 1859 (defun csv--separator-candidates (text &optional cutoff) 1860 "Return a list of candidate CSV separators in TEXT. 1861 When CUTOFF is passed, look only at the first CUTOFF number of characters." 1862 (let ((chars (make-hash-table))) 1863 (dolist (c (string-to-list 1864 (if cutoff 1865 (substring text 0 (min cutoff (length text))) 1866 text))) 1867 (when (and (not (gethash c chars)) 1868 (or (= c ?\t) 1869 (and (not (member c '(?. ?/ ?\" ?'))) 1870 (not (member (get-char-code-property c 'general-category) 1871 '(Lu Ll Lt Lm Lo Nd Nl No Ps Pe Cc Co)))))) 1872 (puthash c t chars))) 1873 (hash-table-keys chars))) 1874 1875 (defun csv--separator-score (separator text &optional cutoff) 1876 "Return a score on how likely SEPARATOR is a separator in TEXT. 1877 1878 When CUTOFF is passed, stop the calculation at the next whole 1879 line after having read CUTOFF number of characters. 1880 1881 The scoring is based on the idea that most CSV data is tabular, 1882 i.e. separators should appear equally often on each line. 1883 Furthermore, more commonly appearing characters are scored higher 1884 than those who appear less often. 1885 1886 Adapted from the paper \"Wrangling Messy CSV Files by Detecting 1887 Row and Type Patterns\" by Gerrit J.J. van den Burg , Alfredo 1888 Nazábal, and Charles Sutton: https://arxiv.org/abs/1811.11242." 1889 (let ((groups 1890 (with-temp-buffer 1891 (csv-set-separator separator) 1892 (save-excursion 1893 (insert text)) 1894 (let ((groups (make-hash-table)) 1895 (chars-read 0)) 1896 (while (and (/= (point) (point-max)) 1897 (or (not cutoff) 1898 (< chars-read cutoff))) 1899 (let* ((lep (line-end-position)) 1900 (nfields (length (csv--collect-fields lep)))) 1901 (cl-incf (gethash nfields groups 0)) 1902 (cl-incf chars-read (- lep (point))) 1903 (goto-char (+ lep 1)))) 1904 groups))) 1905 (sum 0)) 1906 (maphash 1907 (lambda (length num) 1908 (cl-incf sum (* num (/ (- length 1) (float length))))) 1909 groups) 1910 (let ((unique-groups (hash-table-count groups))) 1911 (if (= 0 unique-groups) 1912 0 1913 (/ sum unique-groups))))) 1914 1915 ;;; TSV support 1916 1917 ;; Since "the" CSV format is really a bunch of different formats, it includes 1918 ;; TSV as a subcase, but this subcase is sufficiently interesting that it has 1919 ;; its own mime-type and mostly standard file extension, also it suffers 1920 ;; less from the usual quoting problems of CSV (because the only problematic 1921 ;; chars are LF and TAB, really, which are much less common inside fields than 1922 ;; commas, space, and semi-colons) so it's "better behaved". 1923 1924 (defvar tsv-mode-syntax-table 1925 ;; Inherit from `text-mode-syntax-table' rather than from 1926 ;; `csv-mode-syntax-table' so as not to inherit the 1927 ;; `csv-field-quotes' settings. 1928 (let ((st (make-syntax-table text-mode-syntax-table))) 1929 st)) 1930 1931 (defvar tsv-mode-map 1932 (let ((map (make-sparse-keymap))) 1933 ;; In `tsv-mode', the `csv-invisibility-default/csv-toggle-invisibility' 1934 ;; business doesn't make much sense. 1935 (define-key map [remap csv-toggle-invisibility] #'undefined) 1936 map)) 1937 1938 ;;;###autoload 1939 (add-to-list 'auto-mode-alist '("\\.tsv\\'" . tsv-mode)) 1940 1941 (defun tsv--mode-p () 1942 (equal csv-separator-chars '(?\t))) 1943 1944 ;;;###autoload 1945 (define-derived-mode tsv-mode csv-mode "TSV" 1946 "Major mode for editing files of tab-separated value type." 1947 :group 'CSV 1948 ;; In TSV we know TAB is the only possible separator. 1949 (setq-local csv-separators '("\t")) 1950 ;; FIXME: Copy&pasted from the `:set'ter of csv-separators! 1951 (setq-local csv-separator-chars '(?\t)) 1952 (setq-local csv--skip-chars "^\n\t") 1953 (setq-local csv-separator-regexp "\t") 1954 (setq-local csv-font-lock-keywords 1955 ;; NB: csv-separator-face variable evaluates to itself. 1956 `((,csv-separator-regexp (0 'csv-separator-face)))) 1957 1958 ;; According to wikipedia, TSV doesn't use quotes but uses backslash escapes 1959 ;; of the form \n, \t, \r, and \\ instead. 1960 (setq-local csv-field-quotes nil)) 1961 1962 1963 (provide 'csv-mode) 1964 1965 ;;; csv-mode.el ends here