Commit b098d0fd authored by Nikodemus Siivola's avatar Nikodemus Siivola
Browse files

copy docstrings.lisp from SBCL

  Easier to work with like this. Note: the code has already mutated a bit.
parent 01a5b9bf
......@@ -6,11 +6,10 @@ clean:
# Hook into the super sekrit texinfo generator in the SBCL tree -- this is just a quick way
# to bootrap documentation for now.
sbcl --eval '(progn (require :asdf) (require :alexandria) (rename-package :alexandria :alexandria))' \
--eval '(load (merge-pathnames "doc/manual/docstrings.lisp" (posix-getenv "SBCL_SOURCE_ROOT")))' \
--eval '(sb-texinfo:generate-includes "include/" :alexandria)' \
sbcl --eval '(progn (require :asdf) (require :alexandria))' \
--load docstrings.lisp \
--eval '(sb-texinfo:generate-includes "include/" (list :alexandria) :base-package :alexandria)' \
--eval '(quit)'
mv include/fun-alexandria-type=.texinfo include/fun-alexandria-type-equal.texinfo
pdf: include
texi2pdf alexandria.texinfo
;;; -*- lisp -*-
;;;; A docstring extractor for the sbcl manual. Creates
;;;; @include-ready documentation from the docstrings of exported
;;;; symbols of specified packages.
;;;; This software is part of the SBCL software system. SBCL is in the
;;;; public domain and is provided with absolutely no warranty. See
;;;; the COPYING file for more information.
;;;; Written by Rudi Schlatte <>, mangled
;;;; by Nikodemus Siivola.
;;;; TODO
;;;; * Verbatim text
;;;; * Quotations
;;;; * Method documentation untested
;;;; * Method sorting, somehow
;;;; * Index for macros & constants?
;;;; * This is getting complicated enough that tests would be good
;;;; * Nesting (currently only nested itemizations work)
;;;; * doc -> internal form -> texinfo (so that non-texinfo format are also
;;;; easily generated)
;;;; FIXME: The description below is no longer complete. This
;;;; should possibly be turned into a contrib with proper documentation.
;;;; Formatting heuristics (tweaked to format SAVE-LISP-AND-DIE sanely):
;;;; Formats SYMBOL as @code{symbol}, or @var{symbol} if symbol is in
;;;; the argument list of the defun / defmacro.
;;;; Lines starting with * or - that are followed by intented lines
;;;; are marked up with @itemize.
;;;; Lines containing only a SYMBOL that are followed by indented
;;;; lines are marked up as @table @code, with the SYMBOL as the item.
(eval-when (:compile-toplevel :load-toplevel :execute)
(require 'sb-introspect))
(defpackage :sb-texinfo
(:use :cl :sb-mop)
(:shadow #:documentation)
(:export #:generate-includes #:document-package)
"Tools to generate TexInfo documentation from docstrings."))
(in-package :sb-texinfo)
;;;; various specials and parameters
(defvar *texinfo-output*)
(defvar *texinfo-variables*)
(defvar *documentation-package*)
(defvar *base-package*)
(defparameter *undocumented-packages* '(sb-pcl sb-int sb-kernel sb-sys sb-c))
(defparameter *documentation-types*
;;structure ; also handled by `type'
"A list of symbols accepted as second argument of `documentation'")
(defparameter *character-replacements*
'((#\* . "star") (#\/ . "slash") (#\+ . "plus")
(#\< . "lt") (#\> . "gt")
(#\= . "equals"))
"Characters and their replacement names that `alphanumize' uses. If
the replacements contain any of the chars they're supposed to replace,
you deserve to lose.")
(defparameter *characters-to-drop* '(#\\ #\` #\')
"Characters that should be removed by `alphanumize'.")
(defparameter *texinfo-escaped-chars* "@{}"
"Characters that must be escaped with #\@ for Texinfo.")
(defparameter *itemize-start-characters* '(#\* #\-)
"Characters that might start an itemization in docstrings when
at the start of a line.")
(defparameter *symbol-characters* "ABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890*:-+&#'"
"List of characters that make up symbols in a docstring.")
(defparameter *symbol-delimiters* " ,.!?;")
(defparameter *ordered-documentation-kinds*
'(package type structure condition class macro))
;;;; utilities
(defun flatten (list)
(cond ((null list)
((consp (car list))
(nconc (flatten (car list)) (flatten (cdr list))))
((null (cdr list))
(cons (car list) nil))
(cons (car list) (flatten (cdr list))))))
(defun whitespacep (char)
(find char #(#\tab #\space #\page)))
(defun setf-name-p (name)
(or (symbolp name)
(and (listp name) (= 2 (length name)) (eq (car name) 'setf))))
(defgeneric specializer-name (specializer))
(defmethod specializer-name ((specializer eql-specializer))
(list 'eql (eql-specializer-object specializer)))
(defmethod specializer-name ((specializer class))
(class-name specializer))
(defun ensure-class-precedence-list (class)
(unless (class-finalized-p class)
(finalize-inheritance class))
(class-precedence-list class))
(defun specialized-lambda-list (method)
;; courtecy of AMOP p. 61
(let* ((specializers (method-specializers method))
(lambda-list (method-lambda-list method))
(n-required (length specializers)))
(append (mapcar (lambda (arg specializer)
(if (eq specializer (find-class 't))
`(,arg ,(specializer-name specializer))))
(subseq lambda-list 0 n-required)
(subseq lambda-list n-required))))
(defun string-lines (string)
"Lines in STRING as a vector."
(coerce (with-input-from-string (s string)
(loop for line = (read-line s nil nil)
while line collect line))
(defun indentation (line)
"Position of first non-SPACE character in LINE."
(position-if-not (lambda (c) (char= c #\Space)) line))
(defun docstring (x doc-type)
(cl:documentation x doc-type))
(defun flatten-to-string (list)
(format nil "~{~A~^-~}" (flatten list)))
(defun alphanumize (original)
"Construct a string without characters like *`' that will f-star-ck
up filename handling. See `*character-replacements*' and
`*characters-to-drop*' for customization."
(let ((name (remove-if (lambda (x) (member x *characters-to-drop*))
(if (listp original)
(flatten-to-string original)
(string original))))
(chars-to-replace (mapcar #'car *character-replacements*)))
(flet ((replacement-delimiter (index)
(cond ((or (< index 0) (>= index (length name))) "")
((alphanumericp (char name index)) "-")
(t ""))))
(loop for index = (position-if #'(lambda (x) (member x chars-to-replace))
while index
do (setf name (concatenate 'string (subseq name 0 index)
(replacement-delimiter (1- index))
(cdr (assoc (aref name index)
(replacement-delimiter (1+ index))
(subseq name (1+ index))))))
;;;; generating various names
(defgeneric name (thing)
(:documentation "Name for a documented thing. Names are either
symbols or lists of symbols."))
(defmethod name ((symbol symbol))
(defmethod name ((cons cons))
(defmethod name ((package package))
(short-package-name package))
(defmethod name ((method method))
(generic-function-name (method-generic-function method))
(method-qualifiers method)
(specialized-lambda-list method)))
;;; Node names for DOCUMENTATION instances
(defgeneric name-using-kind/name (kind name doc))
(defmethod name-using-kind/name (kind (name string) doc)
(declare (ignore kind doc))
(defmethod name-using-kind/name (kind (name symbol) doc)
(declare (ignore kind))
(format nil "~@[~A:~]~A" (short-package-name (get-package doc)) name))
(defmethod name-using-kind/name (kind (name list) doc)
(declare (ignore kind))
(assert (setf-name-p name))
(format nil "(setf ~@[~A:~]~A)" (short-package-name (get-package doc)) (second name)))
(defmethod name-using-kind/name ((kind (eql 'method)) name doc)
(format nil "~A~{ ~A~} ~A"
(name-using-kind/name nil (first name) doc)
(second name)
(third name)))
(defun node-name (doc)
"Returns TexInfo node name as a string for a DOCUMENTATION instance."
(let ((kind (get-kind doc)))
(format nil "~:(~A~) ~(~A~)" kind (name-using-kind/name kind (get-name doc) doc))))
(defun short-package-name (package)
(unless (eq package *base-package*)
(car (sort (copy-list (cons (package-name package) (package-nicknames package)))
#'< :key #'length))))
;;; Definition titles for DOCUMENTATION instances
(defgeneric title-using-kind/name (kind name doc))
(defmethod title-using-kind/name (kind (name string) doc)
(declare (ignore kind doc))
(defmethod title-using-kind/name (kind (name symbol) doc)
(declare (ignore kind))
(format nil "~@[~A:~]~A" (short-package-name (get-package doc)) name))
(defmethod title-using-kind/name (kind (name list) doc)
(declare (ignore kind))
(assert (setf-name-p name))
(format nil "(setf ~@[~A:~]~A)" (short-package-name (get-package doc)) (second name)))
(defmethod title-using-kind/name ((kind (eql 'method)) name doc)
(format nil "~{~A ~}~A"
(second name)
(title-using-kind/name nil (first name) doc)))
(defun title-name (doc)
"Returns a string to be used as name of the definition."
(string-downcase (title-using-kind/name (get-kind doc) (get-name doc) doc)))
(defun include-pathname (doc)
(let* ((kind (get-kind doc))
(name (nstring-downcase
(if (eq 'package kind)
(format nil "package-~A" (alphanumize (get-name doc)))
(format nil "~A-~A-~A"
(case (get-kind doc)
((function generic-function) "fun")
(structure "struct")
(variable "var")
(otherwise (symbol-name (get-kind doc))))
(alphanumize (let ((*base-package* nil))
(short-package-name (get-package doc))))
(alphanumize (get-name doc)))))))
(make-pathname :name name :type "texinfo")))
;;;; documentation class and related methods
(defclass documentation ()
((name :initarg :name :reader get-name)
(kind :initarg :kind :reader get-kind)
(string :initarg :string :reader get-string)
(children :initarg :children :initform nil :reader get-children)
(package :initform *documentation-package* :reader get-package)))
(defmethod print-object ((documentation documentation) stream)
(print-unreadable-object (documentation stream :type t)
(princ (list (get-kind documentation) (get-name documentation)) stream)))
(defgeneric make-documentation (x doc-type string))
(defmethod make-documentation ((x package) doc-type string)
(declare (ignore doc-type))
(make-instance 'documentation
:name (name x)
:kind 'package
:string string))
(defmethod make-documentation (x (doc-type (eql 'function)) string)
(declare (ignore doc-type))
(let* ((fdef (and (fboundp x) (fdefinition x)))
(name x)
(kind (cond ((and (symbolp x) (special-operator-p x))
((and (symbolp x) (macro-function x))
((typep fdef 'generic-function)
(assert (or (symbolp name) (setf-name-p name)))
(assert (or (symbolp name) (setf-name-p name)))
(children (when (eq kind 'generic-function)
(collect-gf-documentation fdef))))
(make-instance 'documentation
:name (name x)
:string string
:kind kind
:children children)))
(defmethod make-documentation ((x method) doc-type string)
(declare (ignore doc-type))
(make-instance 'documentation
:name (name x)
:kind 'method
:string string))
(defmethod make-documentation (x (doc-type (eql 'type)) string)
(make-instance 'documentation
:name (name x)
:string string
:kind (etypecase (find-class x nil)
(structure-class 'structure)
(standard-class 'class)
(sb-pcl::condition-class 'condition)
((or built-in-class null) 'type))))
(defmethod make-documentation (x (doc-type (eql 'variable)) string)
(make-instance 'documentation
:name (name x)
:string string
:kind (if (constantp x)
(defmethod make-documentation (x (doc-type (eql 'setf)) string)
(declare (ignore doc-type))
(make-instance 'documentation
:name (name x)
:kind 'setf-expander
:string string))
(defmethod make-documentation (x doc-type string)
(make-instance 'documentation
:name (name x)
:kind doc-type
:string string))
(defun maybe-documentation (x doc-type)
"Returns a DOCUMENTATION instance for X and DOC-TYPE, or NIL if
there is no corresponding docstring."
(let ((docstring (docstring x doc-type)))
(when docstring
(make-documentation x doc-type docstring))))
(defun lambda-list (doc)
(case (get-kind doc)
((package constant variable type structure class condition nil)
(third (get-name doc)))
;; KLUDGE: Eugh.
;; believe it or not, the above comment was written before CSR
;; came along and obfuscated this. (2005-07-04)
(when (symbolp (get-name doc))
(labels ((clean (x &key optional key)
(typecase x
(atom x)
((cons (member &optional))
(cons (car x) (clean (cdr x) :optional t)))
((cons (member &key))
(cons (car x) (clean (cdr x) :key t)))
((cons cons)
(cond (key (if (consp (caar x))
(caaar x)
(caar x)))
(optional (caar x))
(t (clean (car x))))
(clean (cdr x) :key key :optional optional)))
(cond ((or key optional) (car x))
(t (clean (car x))))
(clean (cdr x) :key key :optional optional))))))
(clean (sb-introspect:function-lambda-list (get-name doc))))))))
(defun get-string-name (x)
(let ((name (get-name x)))
(cond ((symbolp name)
(symbol-name name))
((and (consp name) (eq 'setf (car name)))
(symbol-name (second name)))
((stringp name)
(error "Don't know which symbol to use for name ~S" name)))))
(defun documentation< (x y)
(let ((p1 (position (get-kind x) *ordered-documentation-kinds*))
(p2 (position (get-kind y) *ordered-documentation-kinds*)))
(if (or (not (and p1 p2)) (= p1 p2))
(string< (get-string-name x) (get-string-name y))
(< p1 p2))))
;;;; turning text into texinfo
(defun escape-for-texinfo (string &optional downcasep)
"Return STRING with characters in *TEXINFO-ESCAPED-CHARS* escaped
with #\@. Optionally downcase the result."
(let ((result (with-output-to-string (s)
(loop for char across string
when (find char *texinfo-escaped-chars*)
do (write-char #\@ s)
do (write-char char s)))))
(if downcasep (nstring-downcase result) result)))
(defun empty-p (line-number lines)
(and (< -1 line-number (length lines))
(not (indentation (svref lines line-number)))))
;;; line markups
(defvar *not-symbols* '("ANSI" "CLHS"))
(defun locate-symbols (line)
"Return a list of index pairs of symbol-like parts of LINE."
;; This would be a good application for a regex ...
(let (result)
(flet ((grab (start end)
(unless (member (subseq line start end) '("ANSI" "CLHS"))
(push (list start end) result))))
(do ((begin nil)
(maybe-begin t)
(i 0 (1+ i)))
((= i (length line))
;; symbol at end of line
(when (and begin (or (> i (1+ begin))
(not (member (char line begin) '(#\A #\I)))))
(grab begin i))
(nreverse result))
((and begin (find (char line i) *symbol-delimiters*))
;; symbol end; remember it if it's not "A" or "I"
(when (or (> i (1+ begin)) (not (member (char line begin) '(#\A #\I))))
(grab begin i))
(setf begin nil
maybe-begin t))
((and begin (not (find (char line i) *symbol-characters*)))
;; Not a symbol: abort
(setf begin nil))
((and maybe-begin (not begin) (find (char line i) *symbol-characters*))
;; potential symbol begin at this position
(setf begin i
maybe-begin nil))
((find (char line i) *symbol-delimiters*)
;; potential symbol begin after this position
(setf maybe-begin t))
;; Not reading a symbol, not at potential start of symbol
(setf maybe-begin nil)))))))
(defun texinfo-line (line)
"Format symbols in LINE texinfo-style: either as code or as
variables if the symbol in question is contained in symbols
(with-output-to-string (result)
(let ((last 0))
(dolist (symbol/index (locate-symbols line))
(write-string (subseq line last (first symbol/index)) result)
(let ((symbol-name (apply #'subseq line symbol/index)))
(format result (if (member symbol-name *texinfo-variables*
:test #'string=)
(string-downcase symbol-name)))
(setf last (second symbol/index)))
(write-string (subseq line last) result))))
;;; lisp sections
(defun lisp-section-p (line line-number lines)
"Returns T if the given LINE looks like start of lisp code --
ie. if it starts with whitespace followed by a paren or
semicolon, and the previous line is empty"
(let ((offset (indentation line)))
(and offset
(plusp offset)
(find (find-if-not #'whitespacep line) "(;")
(empty-p (1- line-number) lines))))
(defun collect-lisp-section (lines line-number)
(let ((lisp (loop for index = line-number then (1+ index)
for line = (and (< index (length lines)) (svref lines index))
while (indentation line)
collect line)))
(values (length lisp) `("@lisp" ,@lisp "@end lisp"))))
;;; itemized sections
(defun maybe-itemize-offset (line)
"Return NIL or the indentation offset if LINE looks like it starts
an item in an itemization."
(let* ((offset (indentation line))
(char (when offset (char line offset))))
(and offset
(member char *itemize-start-characters* :test #'char=)
(char= #\Space (find-if-not (lambda (c) (char= c char))
line :start offset))
(defun collect-maybe-itemized-section (lines starting-line)
;; Return index of next line to be processed outside
(let ((this-offset (maybe-itemize-offset (svref lines starting-line)))
(result nil)
(lines-consumed 0))
(loop for line-number from starting-line below (length lines)
for line = (svref lines line-number)
for indentation = (indentation line)
for offset = (maybe-itemize-offset line)
do (cond
((not indentation)
;; empty line -- inserts paragraph.
(push "" result)
(incf lines-consumed))
((and offset (> indentation this-offset))
;; nested itemization -- handle recursively
;; FIXME: tables in itemizations go wrong
(multiple-value-bind (sub-lines-consumed sub-itemization)
(collect-maybe-itemized-section lines line-number)
(when sub-lines-consumed
(incf line-number (1- sub-lines-consumed)) ; +1 on next loop
(incf lines-consumed sub-lines-consumed)
(setf result (nconc (nreverse sub-itemization) result)))))
((and offset (= indentation this-offset))
;; start of new item
(push (format nil "@item ~A"
(texinfo-line (subseq line (1+ offset))))
(incf lines-consumed))
((and (not offset) (> indentation this-offset))
;; continued item from previous line
(push (texinfo-line line) result)
(incf lines-consumed))
;; end of itemization
;; a single-line itemization isn't.
(if (> (count-if (lambda (line) (> (length line) 0)) result) 1)
(values lines-consumed `("@itemize" ,@(reverse result) "@end itemize"))
;;; table sections
(defun tabulation-body-p (offset line-number lines)
(when (< line-number (length lines))
(let ((offset2 (indentation (svref lines line-number))))
(and offset2 (< offset offset2)))))
(defun tabulation-p (offset line-number lines direction)
(let ((step (ecase direction
(:backwards (1- line-number))
(:forwards (1+ line-number)))))
(when (and (plusp line-number) (< line-number (length lines)))
(and (eql offset (indentation (svref lines line-number)))
(or (when (eq direction :backwards)
(empty-p step lines))
(tabulation-p offset step lines direction)
(tabulation-body-p offset step lines))))))
(defun maybe-table-offset (line-number lines)
"Return NIL or the indentation offset if LINE looks like it starts
an item in a tabulation. Ie, if it is (1) indented, (2) preceded by an
empty line, another tabulation label, or a tabulation body, (3) and
followed another tabulation label or a tabulation body."
(let* ((line (svref lines line-number))
(offset (indentation line))
(prev (1- line-number))
(next (1+ line-number)))
(when (and offset (plusp offset))
(and (or (empty-p prev lines)
(tabulation-body-p offset prev lines)