1 ;;; refbib.el --- convert refer-style references to ones usable by Latex bib
3 ;; Copyright (C) 1989 Free Software Foundation, Inc.
5 ;; Author: Henry Kautz <kautz@research.att.com>
8 ;; This file is part of GNU Emacs.
10 ;; GNU Emacs is free software; you can redistribute it and/or modify
11 ;; it under the terms of the GNU General Public License as published by
12 ;; the Free Software Foundation; either version 2, or (at your option)
15 ;; GNU Emacs is distributed in the hope that it will be useful,
16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 ;; GNU General Public License for more details.
20 ;; You should have received a copy of the GNU General Public License
21 ;; along with GNU Emacs; see the file COPYING. If not, write to the
22 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
23 ;; Boston, MA 02111-1307, USA.
27 ;; Use: from a buffer containing the refer-style bibliography,
28 ;; M-x r2b-convert-buffer
29 ;; Program will prompt for an output buffer name, and will log
30 ;; warnings during the conversion process in the buffer *Log*.
35 ;; 9/88, created H.Kautz
36 ;; modified 1/19/89, allow books with editor but no author;
37 ;; added %O ordering field;
38 ;; appended illegal multiple fields, instead of
40 ;; added rule, a tech report whose %R number
41 ;; contains "ISBN" is really a book
42 ;; added rule, anything with an editor is a book
44 ;; added 'manual type, for items with institution
45 ;; but no author or editor
46 ;; fixed bug so trailing blanks are trimmed
47 ;; added 'proceedings type
48 ;; used "organization" field for proceedings
49 ;; modified 2/16/89, updated help messages
50 ;; modified 2/23/89, include capitalize stop words in r2b stop words,
51 ;; fixed problems with contractions (e.g. it's),
52 ;; caught multiple stop words in a row
53 ;; modified 3/1/89, fixed capitalize-title for first words all caps
54 ;; modified 3/15/89, allow use of " to delimit fields
55 ;; modified 4/18/89, properly "quote" special characters on output
59 ;**********************************************************
63 "Convert refer-style references to ones usable by Latex bib."
67 (defcustom r2b-trace-on nil
68 "*Non-nil means trace conversion."
72 (defcustom r2b-journal-abbrevs
75 "Abbreviation list for journal names.
76 If the car of an element matches a journal name exactly, it is replaced by
77 the cadr when output. Braces must be included if replacement is a
78 {string}, but not if replacement is a bibtex abbreviation. The cadr
79 may be eliminated if is exactly the same as the car.
80 Because titles are capitalized before matching, the abbreviation
81 for the journal name should be listed as beginning with a capital
82 letter, even if it really doesn't.
83 For example, a value of '((\"Aij\" \"{Artificial Intelligence}\")
84 \(\"Ijcai81\" \"ijcai7\")) would expand Aij to the text string
85 \"Artificial Intelligence\", but would replace Ijcai81 with the
86 BibTeX macro \"ijcai7\"."
87 :type
'(repeat (list string string
))
90 (defcustom r2b-booktitle-abbrevs
93 "Abbreviation list for book and proceedings names.
94 If the car of an element matches a title or booktitle exactly, it is
95 replaced by the cadr when output. Braces must be included if
96 replacement is a {string}, but not if replacement is a bibtex
97 abbreviation. The cadr may be eliminated if is exactly the same as
99 Because titles are capitalized before matching, the abbreviated title
100 should be listed as beginning with a capital letter, even if it doesn't.
101 For example, a value of '((\"Aij\" \"{Artificial Intelligence}\")
102 \(\"Ijcai81\" \"ijcai7\")) would expand Aij to the text string
103 \"Artificial Intelligence\", but would replace Ijcai81 with the
104 BibTeX macro \"ijcai7\"."
105 :type
'(repeat (list string string
))
108 (defcustom r2b-proceedings-list
110 "Assoc list of books or journals which are really conference proceedings,
111 but whose name and whose abbrev expansion (as defined in `r2b-journal-abbrevs'
112 and `r2b-booktitle-abbrevs') does not contain the words \"conference\" or
113 \"proceedings\". (Those cases are handled automatically.)
114 The entry must match the given data exactly.
115 Because titles are capitalized before matching, the items in this list
116 should begin with a capital letter.
117 For example, suppose the title \"Ijcai81\" is used for the proceedings of
118 a conference, and its expansion is the BibTeX macro \"ijcai7\". Then
119 `r2b-proceedings-list' should be '((\"Ijcai81\") ...). If instead its
120 expansion were \"Proceedings of the Seventh International Conference
121 on Artificial Intelligence\", then you would NOT need to include Ijcai81
122 in `r2b-proceedings-list' (although it wouldn't cause an error)."
123 :type
'(repeat (list string string
))
126 (defvar r2b-additional-stop-words
128 "Words not to be used to build the citation key.
129 This is in addition to the `r2b-capitalize-title-stop-words'.")
131 (defcustom r2b-delimit-with-quote t
132 "*If true, then use \" to delimit fields, otherwise use braces."
136 ;**********************************************************
139 (defvar r2b-capitalize-title-stop-words
141 "the\\|and\\|of\\|is\\|a\\|an\\|of\\|for\\|in\\|to\\|in\\|on\\|at\\|"
142 "by\\|with\\|that\\|its")
143 "Words not to be capitalized in a title (unless the first word).")
145 (defvar r2b-capitalize-title-stop-regexp
146 (concat "\\(" r2b-capitalize-title-stop-words
"\\)\\(\\b\\|'\\)"))
148 (defun r2b-capitalize-title-region (begin end
)
149 "Like `capitalize-region', but don't capitalize stop words, except the first."
151 (let ((case-fold-search nil
) (orig-syntax-table (syntax-table)))
154 (set-syntax-table text-mode-syntax-table
)
155 (narrow-to-region begin end
)
156 (goto-char (point-min))
157 (if (looking-at "[A-Z][a-z]*[A-Z]")
160 (while (re-search-forward "\\<" nil t
)
161 (if (looking-at "[A-Z][a-z]*[A-Z]")
163 (if (let ((case-fold-search t
))
164 (looking-at r2b-capitalize-title-stop-regexp
))
166 (capitalize-word 1)))
168 (set-syntax-table orig-syntax-table
))))
171 (defun r2b-capitalize-title (s)
172 "Like `capitalize', but don't capitalize stop words, except the first."
174 (set-buffer (get-buffer-create "$$$Scratch$$$"))
177 (r2b-capitalize-title-region (point-min) (point-max))
180 ;*********************************************************
182 "Unbind defvars, for debugging."
184 (makunbound 'r2b-journal-abbrevs
)
185 (makunbound 'r2b-booktitle-abbrevs
)
186 (makunbound 'r2b-proceedings-list
)
187 (makunbound 'r2b-capitalize-title-stop-words
)
188 (makunbound 'r2b-capitalize-title-stop-regexp
)
189 (makunbound 'r2b-additional-stop-words
)
190 (makunbound 'r2b-stop-regexp
))
192 (defvar r2b-stop-regexp
194 r2b-additional-stop-words
"\\|" r2b-capitalize-title-stop-words
195 "\\)\\('\\w*\\)?\\W+\\)*\\([A-Z0-9]+\\)"))
198 (defun r2b-trace (&rest args
)
201 (apply (function message
) args
)
204 (defun r2b-match (exp)
205 "Returns string matched in current buffer."
206 (buffer-substring (match-beginning exp
) (match-end exp
)))
208 (defcustom r2b-out-buf-name
"*Out*"
209 "*Name of buffer for output from refer-to-bibtex."
213 (defcustom r2b-log-name
"*Log*"
214 "*Name of buffer for logs errors from refer-to-bibtex."
218 (defvar r2b-in-buf nil
)
219 (defvar r2b-out-buf nil
)
222 (defvar r2b-error-found nil
)
224 (setq r2b-variables
'(
233 r2bv-title-first-word
256 (defun r2b-clear-variables ()
257 "Set all global vars used by r2b to nil."
258 (let ((vars r2b-variables
))
261 (setq vars
(cdr vars
)))))
263 (defun r2b-warning (&rest args
)
264 (setq r2b-error-found t
)
265 (princ (apply (function format
) args
) r2b-log
)
267 (princ "\n" r2b-out-buf
)
268 (princ "% " r2b-out-buf
)
269 (princ (apply (function format
) args
) r2b-out-buf
))
271 (defun r2b-get-field (var field
&optional unique required capitalize
)
272 "Set VAR to string value of FIELD, if any. If none, VAR is set to
273 nil. If multiple fields appear, then separate values with the
274 '\\nand\\t\\t', unless UNIQUE is non-nil, in which case log a warning
275 and just concatenate the values. Trim off leading blanks and tabs on
276 first line, and trailing blanks and tabs of every line. Log a warning
277 and set VAR to the empty string if REQUIRED is true. Capitalize as a
278 title if CAPITALIZE is true. Returns value of VAR."
279 (let (item val
(not-past-end t
))
280 (r2b-trace "snarfing %s" field
)
281 (goto-char (point-min))
282 (while (and not-past-end
284 (concat "^" field
"\\b[ \t]*\\(.*[^ \t\n]\\)[ \t]*") nil t
))
285 (setq item
(r2b-match 1))
286 (while (and (setq not-past-end
(zerop (forward-line 1)))
287 (not (looking-at "[ \t]*$\\|%")))
288 (looking-at "\\(.*[^ \t\n]\\)[ \t]*$")
289 (setq item
(concat item
"\n" (r2b-match 1)))
295 (r2b-warning "*Illegal multiple field %s %s" field item
)
296 (setq val
(concat val
"\n" item
))
298 (setq val
(concat val
"\n\t\tand " item
))
302 (if (and val capitalize
)
303 (setq val
(r2b-capitalize-title val
)))
305 (if (and (null val
) required
)
309 (defun r2b-set-match (var n regexp string
)
310 "Set VAR to the Nth subpattern in REGEXP matched by STRING, or nil if none."
312 (if (and (stringp string
) (string-match regexp string
))
313 (substring string
(match-beginning n
) (match-end n
))
318 (defvar r2b-month-abbrevs
319 '(("jan") ("feb") ("mar") ("apr") ("may") ("jun") ("jul") ("aug")
320 ("sep") ("oct") ("nov") ("dec")))
322 (defun r2b-convert-month ()
323 "Try to convert `r2bv-month' to a standard 3 letter name."
325 (let ((months r2b-month-abbrevs
))
326 (if (string-match "[^0-9]" r2bv-month
)
328 (while (and months
(not (string-match (car (car months
))
330 (setq months
(cdr months
)))
332 (setq r2bv-month
(car (car months
)))))
334 (setq months
(car (read-from-string r2bv-month
)))
335 (if (and (numberp months
)
338 (setq r2bv-month
(car (nth months r2b-month-abbrevs
)))
340 (r2b-warning "* Ridiculous month")
341 (setq r2bv-month nil
))
347 (defun r2b-snarf-input ()
348 "Parse buffer into global variables."
349 (let ((case-fold-search t
))
350 (r2b-trace "snarfing...")
352 (set-buffer r2b-in-buf
)
353 (goto-char (point-min))
355 (princ (buffer-substring (point) (progn (end-of-line) (point))) r2b-log
)
358 (r2b-get-field 'r2bv-author
"%A")
359 (r2b-get-field 'r2bv-editor
"%E")
362 (r2b-set-match 'r2bv-primary-author
1
363 "\\b\\(\\w+\\)[ \t]*\\($\\|,\\)" r2bv-author
)
366 (r2b-set-match 'r2bv-primary-author
1
367 "\\b\\(\\w+\\)[ \t]*\\($\\|,\\)" r2bv-editor
)
370 (setq r2bv-primary-author
"")
374 (r2b-get-field 'r2bv-date
"%D" t t
)
375 (r2b-set-match 'r2bv-year
0 "[12][0-9][0-9][0-9]" r2bv-date
)
376 (and (null r2bv-year
)
377 (r2b-set-match 'r2bv-year
1 "[^0-9]\\([0-9][0-9]\\)$" r2bv-date
)
378 (setq r2bv-year
(concat "19" r2bv-year
)))
379 (r2b-set-match 'r2bv-decade
1 "..\\(..\\)" r2bv-year
)
380 (r2b-set-match 'r2bv-month
0
381 "[0-9]+/\\|[a-zA-Z]+" r2bv-date
)
382 (if (and (stringp r2bv-month
) (string-match "\\(.*\\)/$" r2bv-month
))
383 (setq r2bv-month
(substring r2bv-month
0 (match-end 1))))
386 (r2b-get-field 'r2bv-title
"%T" t t t
)
387 (r2b-set-match 'r2bv-title-first-word
4
391 (r2b-get-field 'r2bv-annote
"%X" t
)
392 (r2b-get-field 'r2bv-tr
"%R" t
)
393 (r2b-get-field 'r2bv-address
"%C" t
)
394 (r2b-get-field 'r2bv-institution
"%I" t
)
395 (r2b-get-field 'r2bv-keywords
"%K")
396 (r2b-get-field 'r2bv-booktitle
"%B" t nil t
)
397 (r2b-get-field 'r2bv-journal
"%J" t nil t
)
398 (r2b-get-field 'r2bv-volume
"%V" t
)
399 (r2b-get-field 'r2bv-number
"%N" t
)
400 (r2b-get-field 'r2bv-pages
"%P" t
)
401 (r2b-get-field 'r2bv-where
"%W" t
)
402 (r2b-get-field 'r2bv-ordering
"%O" t
)
407 (defun r2b-put-field (field data
&optional abbrevs
)
408 "Print bibtex FIELD = {DATA} if DATA not null; precede
409 with a comma and newline; if ABBREVS list is given, then
410 try to replace the {DATA} with an abbreviation."
412 (let (match nodelim multi-line index
)
414 ((and abbrevs
(setq match
(assoc data abbrevs
)))
415 (if (null (cdr match
))
416 (setq data
(car match
))
417 (setq data
(car (cdr match
))))
419 ((and (not (equal data
""))
420 (not (string-match "[^0-9]" data
)))
424 (while (string-match "[\\~^]" data index
)
425 (setq data
(concat (substring data
0 (match-beginning 0))
427 (substring data
(match-beginning 0) (match-end 0))
429 (substring data
(match-end 0))))
430 (setq index
(+ (match-end 0) 7)))
432 (while (string-match "[$&%#_{}]" data index
)
433 (setq data
(concat (substring data
0 (match-beginning 0))
435 (substring data
(match-beginning 0))))
436 (setq index
(+ (match-end 0) 1)))
438 (if r2b-delimit-with-quote
439 (while (string-match "\"" data index
)
440 (setq data
(concat (substring data
0 (match-beginning 0))
442 (substring data
(match-end 0))))
443 (setq index
(+ (match-end 0) 2))))
449 (if r2b-delimit-with-quote
452 (string-match ".*" data
)
453 (if (> (match-end 0) 59)
457 (if r2b-delimit-with-quote
464 (defun r2b-require (vars)
465 "If any of VARS is null, set to empty string and log error."
468 ((listp vars
) (r2b-require (car vars
)) (r2b-require (cdr vars
)))
470 (if (null (symbol-value vars
))
472 (r2b-warning "*Missing value for field %s" vars
)
479 (defmacro r2b-moveq
(new old
)
480 "Set NEW to OLD and set OLD to nil."
481 (list 'progn
(list 'setq new old
) (list 'setq old
'nil
)))
483 (defun r2b-isa-proceedings (name)
484 "Return t if NAME is the name of proceedings."
488 (string-match "proceedings\\|conference" name
)
489 (assoc name r2b-proceedings-list
)
490 (let ((match (assoc name r2b-booktitle-abbrevs
)))
492 (string-match "proceedings\\|conference" (car (cdr match
)))))
495 (defun r2b-isa-university (name)
496 "Return t if NAME is a university or similar organization,
497 but not a publisher."
500 (string-match "university" name
)
501 (not (string-match "press" name
))
505 (defun r2b-barf-output ()
506 "Generate bibtex based on global variables."
507 (let ((standard-output r2b-out-buf
) (case-fold-search t
) match
)
509 (r2b-trace "...barfing")
511 (set-buffer r2b-out-buf
)
513 (setq r2bv-kn
(concat r2bv-primary-author r2bv-decade
514 r2bv-title-first-word
))
516 (setq r2bv-entry-kind
518 ((r2b-isa-proceedings r2bv-journal
)
519 (r2b-moveq r2bv-booktitle r2bv-journal
)
520 (if (r2b-isa-university r2bv-institution
)
521 (r2b-moveq r2bv-organization r2bv-institution
)
522 (r2b-moveq r2bv-publisher r2bv-institution
))
523 (r2b-moveq r2bv-note r2bv-tr
)
524 (r2b-require 'r2bv-author
)
526 ((r2b-isa-proceedings r2bv-booktitle
)
527 (if (r2b-isa-university r2bv-institution
)
528 (r2b-moveq r2bv-organization r2bv-institution
)
529 (r2b-moveq r2bv-publisher r2bv-institution
))
530 (r2b-moveq r2bv-note r2bv-tr
)
531 (r2b-require 'r2bv-author
)
533 ((and r2bv-tr
(string-match "phd" r2bv-tr
))
534 (r2b-moveq r2bv-school r2bv-institution
)
535 (r2b-require 'r2bv-school
)
536 (r2b-require 'r2bv-author
)
538 ((and r2bv-tr
(string-match "master" r2bv-tr
))
539 (r2b-moveq r2bv-school r2bv-institution
)
540 (r2b-require 'r2bv-school
)
541 (r2b-require 'r2bv-author
)
543 ((and r2bv-tr
(string-match "draft\\|unpublish" r2bv-tr
))
544 (r2b-moveq r2bv-note r2bv-institution
)
545 (r2b-require 'r2bv-author
)
548 (r2b-require 'r2bv-author
)
551 (r2b-moveq r2bv-publisher r2bv-institution
)
552 (r2b-moveq r2bv-note r2bv-tr
)
553 (r2b-require 'r2bv-publisher
)
554 (r2b-require 'r2bv-author
)
558 (string-match "\\`personal communication\\'" r2bv-title
))
560 ((r2b-isa-proceedings r2bv-title
)
561 (if (r2b-isa-university r2bv-institution
)
562 (r2b-moveq r2bv-organization r2bv-institution
)
563 (r2b-moveq r2bv-publisher r2bv-institution
))
564 (r2b-moveq r2bv-note r2bv-tr
)
570 (string-match "\\bisbn\\b" r2bv-tr
))))
571 (r2b-moveq r2bv-publisher r2bv-institution
)
572 (r2b-moveq r2bv-note r2bv-tr
)
573 (r2b-require 'r2bv-publisher
)
574 (if (null r2bv-editor
)
575 (r2b-require 'r2bv-author
))
578 (r2b-require 'r2bv-institution
)
580 "\\`\\(\\(.\\|\n\\)+\\)[ \t\n]+\\([^ \t\n]\\)+\\'"
583 (setq r2bv-type
(substring r2bv-tr
0 (match-end 1)))
584 (setq r2bv-number
(substring r2bv-tr
585 (match-beginning 3)))
587 (r2b-moveq r2bv-number r2bv-tr
))
588 (r2b-require 'r2bv-author
)
591 (r2b-moveq r2bv-organization r2bv-institution
)
597 (r2b-require '( r2bv-year
))
600 (princ "\n% Warning -- Errors During Conversion Next Entry\n"))
603 (princ r2bv-entry-kind
)
607 (r2b-put-field "author" r2bv-author
)
608 (r2b-put-field "title" r2bv-title r2b-booktitle-abbrevs
)
609 (r2b-put-field "year" r2bv-year
)
611 (r2b-put-field "month" r2bv-month r2b-month-abbrevs
)
612 (r2b-put-field "journal" r2bv-journal r2b-journal-abbrevs
)
613 (r2b-put-field "volume" r2bv-volume
)
614 (r2b-put-field "type" r2bv-type
)
615 (r2b-put-field "number" r2bv-number
)
616 (r2b-put-field "booktitle" r2bv-booktitle r2b-booktitle-abbrevs
)
617 (r2b-put-field "editor" r2bv-editor
)
618 (r2b-put-field "publisher" r2bv-publisher
)
619 (r2b-put-field "institution" r2bv-institution
)
620 (r2b-put-field "organization" r2bv-organization
)
621 (r2b-put-field "school" r2bv-school
)
622 (r2b-put-field "pages" r2bv-pages
)
623 (r2b-put-field "address" r2bv-address
)
624 (r2b-put-field "note" r2bv-note
)
625 (r2b-put-field "keywords" r2bv-keywords
)
626 (r2b-put-field "where" r2bv-where
)
627 (r2b-put-field "ordering" r2bv-ordering
)
628 (r2b-put-field "annote" r2bv-annote
)
635 (defun r2b-convert-record (output-name)
636 "Transform current bib entry and append to buffer OUTPUT;
637 do \"M-x r2b-help\" for more info."
639 (list (read-string "Output to buffer: " r2b-out-buf-name
)))
640 (let (rec-end rec-begin not-done
)
641 (setq r2b-out-buf-name output-name
)
642 (setq r2b-out-buf
(get-buffer-create output-name
))
643 (setq r2b-in-buf
(current-buffer))
644 (set-buffer r2b-out-buf
)
645 (goto-char (point-max))
646 (setq r2b-log
(get-buffer-create r2b-log-name
))
648 (goto-char (point-max))
649 (set-buffer r2b-in-buf
)
650 (setq not-done
(re-search-forward "[^ \t\n]" nil t
))
653 (re-search-backward "^[ \t]*$" nil
2)
654 (re-search-forward "^%")
655 (beginning-of-line nil
)
656 (setq rec-begin
(point))
657 (re-search-forward "^[ \t]*$" nil
2)
658 (setq rec-end
(point))
659 (narrow-to-region rec-begin rec-end
)
660 (r2b-clear-variables)
663 (set-buffer r2b-in-buf
)
672 (defun r2b-convert-buffer (output-name)
673 "Transform current buffer and append to buffer OUTPUT.
674 Do `M-x r2b-help' for more info."
676 (list (read-string "Output to buffer: " r2b-out-buf-name
)))
678 (setq r2b-log
(get-buffer-create r2b-log-name
))
682 (goto-char (point-min))
683 (message "Working, please be patient...")
685 (while (r2b-convert-record output-name
) t
)
686 (message "Done, results in %s, errors in %s"
687 r2b-out-buf-name r2b-log-name
)
690 (defvar r2b-help-message
691 " Refer to Bibtex Bibliography Conversion
693 A refer-style database is of the form:
696 %T Great Thoughts I've Thought
700 This utility converts these kind of databases to bibtex form, for
701 users of TeX and LaTex. Instructions:
702 1. Visit the file containing the refer-style database.
704 M-x r2b-convert-buffer
705 converts the entire buffer, appending its output by default in a
706 buffer named *Out*, and logging progress and errors in a buffer
707 named *Log*. The original file is never modified.
708 Note that results are appended to *Out*, so if that buffer
709 buffer already exists and contains material you don't want to
710 save, you should kill it first.
711 3. Switch to the buffer *Out* and save it as a named file.
712 4. To convert a single refer-style entry, simply position the cursor
713 at the entry and enter
714 M-x r2b-convert-record
715 Again output is appended to *Out* and errors are logged in *Log*.
717 This utility is very robust and pretty smart about determining the
718 type of the entry. It includes facilities for expanding refer macros
719 to text, or substituting bibtex macros. Do M-x describe-variable on
721 r2b-booktitle-abbrevs
723 for information on these features.
725 Please send bug reports and suggestions to
727 kautz@research.att.com
732 "Print help describing the `refbib' package."
734 (with-output-to-temp-buffer "*Help*"
735 (princ r2b-help-message
)
737 (set-buffer standard-output
)
741 (provide 'refer-to-bibtex
)
743 ;;; refbib.el ends here