\NeedsTeXFormat{LaTeX2e}[1994/12/01]
\ProvidesPackage{edfnotes}[2011/02/16 v0.6b
annotations to footnote lines (UL)]
%% Copyright (C) 2010 2011 Uwe Lueck,
%% http://www.contact-ednotes.sty.de.vu
%% -- author-maintained in the sense of LPPL below --
%%
%% This file can be redistributed and/or modified under
%% the terms of the LaTeX Project Public License; either
%% version 1.3c of the License, or any later version.
%% The latest version of this license is in
%% http://www.latex-project.org/lppl.txt
%% We did our best to help you, but there is NO WARRANTY.
%%
%% Please report bugs, problems, and suggestions via
%%
%% http://www.contact-ednotes.sty.de.vu
%%
%% {\it This work has been supported %% rm. `partially' 2011/02/14
%% by the Deutsche Forschungsgemeinschaft (DFG),
%% organized %% 2010/12/18 TODO!?
%% by Prof.~Dr.\ Dr.\ Christian Tapp
%% at Ruhr-Universit\"at Bochum, Germany.}
%% %% 2011/01/14:
%% Christian also participated immensely in development
%% by creating really small failure examples for debugging.
%% %% 2011/01/17:
%% Completing the work was possible thanks to a loan
%% from my mother, Mrs.~Renate L\"uck.
%%
%% === Known Issues ===
%% Please note the following difficulties you may experience
%% with 'edfnotes', and what to do about them:
%% \begin{enumerate}
%% \item In certain situations, ``\strong{hanging}" of the `latex'
%% run has been observed.
%% As to \emph{implementation}, %% 2011/02/14
%% it has not been analysed completely yet (TODO),
%% but it occurred with `\marginpar' and printing the second
%% part of a footnote before its first part---a well-known
%% \LaTeX\ bug,
%% ``\strong{footnote placement disaster}."
%% However, this situation usually does not
%% lead into ``hanging," rather replacing `=' by \verb+>+
%% in the definition of `\EFN@annot' with package version
%% v0.6 (to keep ``displaced" annotations from wandering
%% to the last page) seems to cause the hang when a
%% ``footnote placement disaster" occurs
%% (where the footnote has annotations).
%%
%% \strong{You must be careful with marginals anyway}---with
%% critical editions. There also is the problem with
%% placement in the wrong margin. Mistakes of the named kinds
%% occur rather \emph{usually} in critical editions.
%% With \emph{annotations to footnotes} things may become
%% even worse, since placement restrictions
%% (``start note on same page as target")
%% may become difficult or impossible to
%% satisfy.\footnote{These advices should appear in
%% 'ednotes' already---TODO!}%% 2011/02/10
%%
%% Often, just manual `\pagebreak's (in the last line of a
%% page---when near completion of the work) help.
%%
%% The \ctanpkgref{mparhack} package\urlpkgfoot{mparhack} at least
%% ensures that marginals appear in the appropriate margin.
%%
%% ``Footnote placement disasters" can be avoided (at some cost)
%% by the \ctanpkgref{tamefloats} package\urlpkgfoot{tamefloats}
%% (helped sometimes, did not help with
%% \textit{Paradoxien des Unendlichen}).
%% Another possibility for avoiding footnote placement disasters
%% due to `\marginpar' is using `\marginnote' from the
%% \ctanpkgref{marginnote} package\urlpkgfoot{marginnote}
%% instead.
%%
%% However, `\marginnote' behaves badly when, e.g., page breaks
%% in two sources appear in the same line of the edition.
%% On the other hand, `\marginpar' ``moves" the second marginal,
%% which is not much better. I would prefer a ``manual"
%% solution: inserting the page break symbols within the
%% line manually and produce a single `\marginpar' or
%% `\marginnote' entry for both of them.
%% (Actually, this could be automated within 'lineno'.)
%%
%% \item \strong{Page break oscillations:} \
%% 'ednotes.sty' provides a message about page break
%% oscillations that gives advice how to fix them.
%% With annotations to footnotes, that advice may help only
%% little. Besides fixing the page break in main text,
%% it may be needed to place `\pagebreak'
%% (or `\warningpagebreak') commands in footnotes too.
%% While the oscillation of main text lines is fixed
%% (for the page about the message reports),
%% page breaks of footnotes may keep oscillating.
%% I was successful in a test case
%% (that giant footnote in \textit{Paradoxien des Unendlichen})
%% by placing a `\pagebreak' in the footnote text quite
%% early first and then moved it down line by line,
%% until the result became bad. At that point, I chose
%% the previous position of `\pagebreak'.
%% The result is presented as `PdUsample.pdf'.
%% % (part of documentation, `docint/' directory). %% rm. 2011/02/16
%%
%% \item For \strong{other Limitations}, please see the section
%% in 'fnlineno.pdf' of this title.
%% \end{enumerate}
%%
%% === Installing and Calling ===
%% The file 'edfnotes.sty' is provided ready, installation only requires
%% putting it somewhere where \TeX\ finds it
%% (which may need updating the filename data
%% base).\footnote{\url{http://www.tex.ac.uk/cgi-bin/texfaq2html?label=inst-wlcf}}
%%
%% As usually, 'edfnotes.sty' is loaded by
%% % \begin{verbatim}
%% `\usepackage{edfnotes}'
%% % \end{verbatim}
%% below the `\documentclass' line %% 2011/01/13
%% (after loading 'ednotes')
%% and before `\begin{document}'.
%% 'edfnotes' does not offer any package options.
%%
%% This should suffice just to get the advertised functionality.
%% There are \emph{two} user commands |\ForceFootnoteAnnotations|
%% and |\clearfootnoteannotations|
%% described in section~\ref{sec:enddoc}.
%%
%% %% 2011/02/16:
%% Minimum functionality of 'edfnotes' requires that
%% \CtanPkgRef{ednotes}{ednotes.sty},
%% \CtanPkgRef{lineno}{fnlineno.sty}, and
%% \CtanPkgRef{lineno}{lineno.sty} are installed and ``visible."
%% For more functionality, see the descriptions of both
%% bundles.\urlpkgfoot{ednotes}\urlpkgfoot{lineno}.
%%
%%
%% \pagebreak %% 2011/02/16
%% == Implementation ==
%% === Loading Required Package ===
%% The package builds on 'fnlineno''s ability to
%% number footnote lines and processing `\linelabel' there.
%% So we load it. %%% (TODO or `\AtBeginDocument'?):
%% We need a version without earlier tries at supporting
%% 'edfnotes.sty' v0.2:
\RequirePackage{fnlineno}[2011/01/07]
%%
%% === Overview of Processing Levels ===
%% %% new 2011/01/18
%% 'ednotes'' and 'edfnotes'' apparatus entries
%% are handled by the 'manyfoot' package.
%% An entry <insert> for apparatus level <note-fam>
%% is passed to 'manyfoot' by 'manyfoot''s command
%% |\Footnotetext<note-fam>\@empty{<insert>}|.
%%
%% In terms of \TeX's hard-wired mechanisms,
%% such a `\Foonotetext' command issues an `\insert'
%% command. `\insert' is a so-called ``primitive,"
%% hard-wired command. It typesets the entry <insert>
%% and stores the result internally, and in choosing
%% the next page break, \TeX\ will try to start
%% the typesetting result of <insert> on the ``current" page.
%%
%% \TeX\ is not so well-prepared (\strong{hard-wired})
%% for placing a footnote <ff> (its beginning)
%% intended to refer to a passage
%% in another footnote <f> on the same page as that passage
%% (starts).
%% Our \strong{hard-weird} macros build an own mechanism
%% using the `.aux' file in order to manage this.
%% A footnote <ff> to a footnote <f> is not directly passed to
%% 'manyfoot', such ``secondary" footnotes <ff> are rather
%% withhold and collected in a list <list> from which
%% 'edfnotes' will later decide what to do with them.
%% The goal is to find a good point for finally passing
%% <ff> to 'manyfoot' such that \TeX\ will be able to place
%% <f> and <ff> appropriately.
%%
%% In the most basic application of 'edfnotes',
%% there is \emph{one} critical apparatus referring both
%% to the main text and to the footnotes of an edited text.
%% We want that, in the apparatus,
%% annotations to \emph{main text} lines
%% precede annotations to \emph{footnote} lines.
%% To achieve this, the idea is to issue the
%% `\Footnotetext' commands for annotations to footnotes
%% only at \emph{last main text lines} of pages.
%%
%% Unfortunately, sometimes the `.aux' file provides
%% either \emph{no} or \emph{wrong} information for
%% identifying that last line. Therefore, we sometimes
%% \emph{skip} the item-by-item procedure that tries to pair
%% <ff> with <f>, and instead somewhere pass \emph{all} the
%% <ff> items in that list <list> to 'manyfoot'
%% and \TeX's primitive `\insert' command.
%% This will be just a kind of ``emergency" routine,
%% considered necessary temporarily, while hoping that
%% the required and correct information will appear
%% in the `.aux' file some runs later.
%%
%% \pagebreak %% 2011/02/16
%% === Postponing Annotations ===
%% ==== Goal and Strategy ====
%% Version v0.2 of 'edfnotes' used
%% `\@EN@hookfn{<insert>}'---that is called from 'ednotes''
%% `\@EN@putdown'---in order to intercept %%% and collect
%% 'manyfoot''s `\FootnotetextA' etc. commands that
%% execute \TeX's `\insert' commands for 'ednotes'' annotations.
%% `\@EN@hookfn' has the trivial meaning `\@firstofone' in 'ednotes'
%% and is only activated by 'edtable.sty' ('lineno' bundle)
%% in order to collect annotations to a tabular environment
%% at the latter's ``footnote hook" that finally executes
%% the `\insert' commands. These annotations
%% are inserted to the \emph{current} page, i.e.,
%% all the annotations collected there
%% start on the same page.
%%
%% This behaviour (starting on the same page)
%% is in general not adequate with annotations to footnotes
%% when the latter are split and continued at later pages.
%% If a footnote starts at page~$p$, is continued on
%% page~$p+1$, and a passage of it on page~$p+1$ gets a
%% critical annotation, that annotation should start on
%% page~$p+1$, not on page~$p$. The footnote may also get
%% an annotation to its page~$p$ part, that one should start
%% at page $p$ indeed. The mechanism we have to introduce here
%% must be able to tell these two cases apart.
%% For each annotation to a footnote, it must retrieve the
%% page which the annotation refers to.
%%
%% Interception at 'ednotes'' `\@EN@hookfn' now becomes too clumsy.
%% We redefine
%% \[|\@EN@putdown{<id>}{<note-fam>}{<lemma-tag>}{<note>}|\]
%% entirely, replacing its former `\@EN@hookfn' section
%% by a new macro
%% \[|\@EN@appentry{<id>}{<note-fam>}{<lemma-tag>}{<note>}|\]
%% i.e., taking the same arguments as `\@EN@putdown'.
%% <id> is a string that is used to write commands
%% `\newlabel{-<id>}{<l/p-b>}', `\newlabel{+<id>}{<l/p-e>}',
%% and `\newlabel{<id>}{<note-page>}' to the `.aux' file.
%% From these commands, the next run builds a macro containing
%% placement informations for that annotation with ID <id>.
%% <l/p-b> contains the page on which the annotated passage
%% starts. This is where the annotation \emph{should} start
%% as well. It just contains the ``public" (``relative") page number,
%% as opposed to ``absolute" page numbers that 'lineno.sty'
%% traces through `\c@LN@truepage'. Well, the minute chance
%% of failure arising from this difference should be
%% acceptable for the present attempt at equipping 'edfnotes'
%% with functionality adequate for real life.
%%
%% ==== Re-implementation of \cs{@EN@putdown} ====
%% |\@EN@putdown| ...
\long\def\@EN@putdown#1#2#3#4{%
% #1 label, #2 note family, #3 lemma tag, #4 note.
% \long note/lemma!?
\linelabel{-#1}% Ensures hmode.
%% `\insert' apparatus entry:
\@EN@appentry{#1}{#2}{#3}{#4}%
%% Lemma in main text:
\nobreak \hskip\z@skip %% 2006/01/12 cf. german.sty: \allowhyphens
\@EN@lemmaarg %% Should not end with space so \linelabel...
%% \allowhyphens not needed here (tested) 2006/01/12
\linelabel{+#1}%
% \csname #1rightmark\endcsname %% Suggested.
}
%% |\@EN@appentry| ...
\def \@EN@appentry #1#2#3#4{%
\csname Footnotetext#2\endcsname\@empty{%
\csname#2notefmt\endcsname
\let\@currentlabel\@empty
\label{#1}% Need page number of note.
\let\nopunct\@gobble % or by \if...
%% Print line numbers:
\@ifundefined{r@#1}%
{\differentlines\@EN@unknown\@EN@unknown}%
{\let\@EN@incomplete@ref\relax
\@EN@xpxpxp \@EN@extract\csname r@#1\endcsname
\@EN@incomplete@ref\@EN@incomplete@ref
\@EN@incomplete@ref\@EN@incomplete@ref
\ifx\@EN@elemp\@EN@Incomplete
\G@refundefinedtrue
\let\@EN@incomplete@ref\@EN@unknown
\fi
\@tempswatrue
\ifx\@EN@bleml\@EN@eleml \else \@tempswafalse \fi
%% Restart of line numbers / different page!?
\ifx\@EN@blemp\@EN@elemp \else \@tempswafalse \fi
\if@tempswa
\sameline{\@EN@plref\@EN@blemp\@EN@bleml}%
\else
\@EN@lastline@z@
\differentlines{\@EN@plref\@EN@blemp\@EN@bleml}%
{\ifx\@EN@blemp\@EN@elemp
\@EN@eleml
\else
\pageandline\@EN@elemp\@EN@eleml
\fi}%
\@EN@lastline@z@
\fi}%
\lemmafmt{{\@EN@lemmaexpands#3}}%
\notefmt{#4}%
}%
}%
%% A copy |\EFN@@appentry| for the case that
%% `\@EN@appentry' is redefined by `\FNLN@text':
\let \EFN@@appentry \@EN@appentry
%%
%% ==== Basic Changes ====
%% We patch the patch of `\@footnotetext' again ...
%% |\EFN@@text| stores 'fnlineno''s variant |\FNLN@text|
%% of \LaTeX's `\@footnotetext':
\let \EFN@@text \FNLN@text
%% A hook |\EFN@annot@hook| will collect and
%% handle annotations to the footnotes of a page.
%% It will be changed globally:
\global\let \EFN@annot@hook \@empty
%% |\EFN@move{<id>}{<note-fam>}{<lemma-tag>}{<note>}|
%% will move an annotation into the (next ...\,) hook:
\def \EFN@move #1#2#3#4{%
\g@addto@macro\EFN@annot@hook{\EFN@annot{#1}{#2}{#3}{#4}}}
%% The new |\FNLN@text| will redefine `\@EN@appentry'
%% in order to redirect the annotation:
\renewcommand\FNLN@text[1]{\EFN@@text{\let\@EN@appentry\EFN@move #1}}
%%
%% === Inserting Annotations to Footnotes ===
%% ==== Strategy ====
%% Version v0.2 of 'edfnotes.sty' aims at proper \strong{sorting}
%% of the annotations, in the sense that in the critical apparatus,
%% notes on \emph{main text} lines appear before all notes
%% to \emph{footnote} lines.
%%
%% David Kastrup's 'bigfoot' and 'perpage' packages
%% seem to deal with this task
%% (however, the documentation doesn't help much in
%% making use of it); on the other hand,
%% 'lineno.sty' offers its own nice tools ...
%%
%% Our idea is inserting all the \emph{annotations to footnotes}
%% right after the \emph{last main text line of the page}---this
%% way they should appear \emph{after} all annotations to main text lines.
%% (This idea grew with v0.2, and with v0.3 it became clear
%% this hook can be used for passing certain annotations
%% to the next page.)
%% % ; `samepage@hook' then turned into `annot@hook'.)
%% %% <- indeed 2011/01/17
%%
%% For version v0.3, this is refined a little.
%% The hook will not actually `\insert' all annotations,
%% it will rather choose some for immediate insertion
%% and collect the remaining annotations
%% in the hook for the next page.
%%
%% We use |\MakeLineNo| from 'lineno.sty' for hooking in here
%% (you may find the context relevant here in `fnlineno.pdf').
%% `\MakeLineNo'---in ``pagewise" mode
%% that we assume here---has access to the absolute number
%% of the last numbered line of the current page,
%% involving 'lineno.sty''s |\testLastNumberedPage|.
%% We modify the latter so that it records that number as
%% |\theLastLineNumber|.
%% % (I tried more efficient things
%% % involving a switch, failed terribly,
%% % no idea what was going on.) %% 2011/01/07
%% ==== Deciding ====
%% |\EFN@run@annot@hook| first expands `\EFN@annot@hook',
%% then empties it,
%% and while the version for the recent page runs,
%% the version of |\EFN@annot@hook| for the \emph{next} page
%% may be built, containing annotations to footnote passages
%% on later pages:
\def \EFN@run@annot@hook {%
\expandafter \global \expandafter \let
\expandafter \EFN@annot@hook \expandafter \@empty
\EFN@annot@hook}
%% |\EFN@annot@hook| (unless empty) is a list of commands
%% \[|\EFN@annot{<id>}{<note-fam>}{<lemma-tag>}{<note>}|\]
%% that were issued either by `\@EN@putdown' or by the previous
%% version of `\EFN@run@annot@hook' using `\EFN@move'.
%% `\EFN@annot' uses <id> to determine the page number where
%% the footnote passage the annotation refers to starts.
%% There are \emph{three} cases deciding about immediate `\insert'ion
%% vs.\ postponing. The annotation is `\insert'ed immediately
%% if \emph{either} <id> has not been used in the previous run
%% (it contains an annotation number exceeding the earlier
%% number of annotations---minus one) \emph{or}
%% if the annotation was placed on the ``current" page
%% in the run before. In these cases, the four arguments of
%% `\EFN@annot' are run by |\EFN@@appentry|. \emph{Otherwise}
%% the annotation is deferred by running the four arguments
%% with |\EFN@move|.
\def \EFN@annot #1{%
\let \EFN@next \EFN@@appentry
%% I.e., immediate `\insert'ion is the ``default."
%% `undefined' in the next line means there is no placement
%% information from the previous run
%% (stored as `\r@<id>'). If there is, it is
%% extracted by `\@EN@extract', `\@EN@blemp' will carry the
%% ``public" (relative) page number.
\@ifundefined{r@#1}\relax{%
\@EN@xpxpxp \@EN@extract \csname r@#1\endcsname
\relax \relax \relax \relax %% somewhat lazier than ednotes
%% If the page numbers don't match, we `move'.
%% With v0.6 however, we don't `move' when we have
%% ``missed" the (according to `.aux')
%% matching page. This happens when
%% the passage we are referring to moves to later pages.
% \ifnum \@EN@blemp=\c@page %% TODO \ifx\relax!? \c@page OK?
% \else
\ifnum \@EN@blemp>\c@page %% v0.6
\let \EFN@next \EFN@move
\fi }%
\EFN@next{#1}%
}
%%
%% ==== Normal Insertion ====
%% The ``normal" way of inserting annotations to footnotes
%% is a call from the extended `\MakeLineNo'
%% involving testing whether the line is the last
%% numbered one on the page. It seems not to have been
%% easily accessible with 'lineno.sty', we are extending
%% the latter's |\testLastNumberedPage| so it stores
%% that number as |\theLastLineNumber|.
%% Indeed, calculating the ``pagewise" line number
%% in `\MakeLineNo' invokes a series of tests
%% that finally calls \[`\testLastNumberedPage{<int>}'\]
%% where <int> is the least ``last" line number of a
%% page with numbered lines that is greater than or
%% equal to the current absolute line number.
\def\testLastNumberedPage#1{\ifnum#1<\c@linenumber
\let\firstLN\@gobble
\fi
%% both tests new 2011/01/15,
%% not sure about efficiency TODO:
% \if@FNLN@sw@ \else
% <- don't run with \getfoot...! 2011/01/19 ->
\ifx\c@linenumber\c@pagewiselinenumber
\ifnum\theLastLineNumber=#1\relax \else
\gdef\theLastLineNumber{#1}% %% mod. 2011/01/07!
\fi \fi }
%% 'lineno''s |\MakeLineNo| is extended using that
%% `add' macro from 'ednotes'. It will
%% test if the line just numbered is a main text line and
%% the last numbered one on the page, and only then
%% call the `\insert's from 'manyfoot' for the critical
%% apparatus.---With v0.5, we actually extend %% 2011/01/19
%% |\stepLineNumber| that `\MakeLineNo' invokes.
%% Appending `\insert'ions to `\MakeLineNo' creates
%% a strong risk that it is executed on the \emph{next}
%% page. (So this requires a certain version of 'lineno.sty',
%% and the latter might better provide an ``official" hook.
%% \emph{Prepending} to `\MakeLineNo' in v0.34 failed
%% because `\theLastLineNumber' may be wrong there.)
\@EN@addtomacro \stepLineNumber {%
\if@FNLN@sw@ \else
\ifx\EFN@annot@hook\@empty \else %% 2011/01/14 efficient!? TODO
\advance\c@linenumber\m@ne %% local to \output 2011/01/16
\ifnum\theLastLineNumber=\c@pagewiselinenumber
\EFN@run@annot@hook
\fi \fi \fi }
%%
%% ==== Forced Insertion ====
%% \paragraph{1.} At an \strong{initial run}
%% of the document (no `.aux' present),
%% there isn't any information about line number ranges for pages.
%% We then `\insert' everything immediately so that an initial run
%% at least produces page breaks that are useful for ordering
%% annotations in the next run. This is achieved by choosing
%% a default definition for |\theLastLineNumber| that
%% trivially renders the `\ifnum' test in `\MakeLineNo' `\iftrue'.
\@ifdefinable \theLastLineNumber {%
\let \theLastLineNumber \c@pagewiselinenumber}
%% \paragraph{2.} Another problem may occur with the
%% \strong{very last line} of an edition.
%% When between two \LaTeX\ runs the number of
%% main text lines is reduced (e.g., one has removed
%% wrong text or has improved line breaking by a manual
%% change of hyphenation), the annotations for the last
%% page may get lost because the line number that the
%% insertions wait for doesn't occur any more. They will
%% then get lost, and their `\newlabel' entries in the
%% `.aux' file will be missing, perhaps resulting in
%% other errors. We provide a command
%% |\ForceFootnoteAnnotations| for manual use by authors
%% or for being added to certain hooks. It might be
%% added to `\nolinenumbers' and `\endlinenumbers',
%% but I am not sure ...
%%
%% v0.5 allows a much cleaner implementation of
%% `\ForceFootnoteAnnotations' than we had in v0.35 and v0.4:
\newcommand* \ForceFootnoteAnnotations {{%
\let\EFN@annot\EFN@@appentry
\EFN@run@annot@hook}}
%% \paragraph{3.} Finally it may occur that a
%% %% \strong 2011/01/19:
%% \strong{chapter ends} with a very
%% \strong{long footnote} with annotations
%% to all of its pages, and the editor decides
%% that the footnotes should fill the final pages
%% of the chapter without main text on their pages.
%% Then the annotations have problem with our
%% approach of v0.3.
%% %
%% A similar problem could be that a short end
%% of the last footnote of a chapter
%% gets remarks of several pages that the editor
%% doesn't want to see in the following chapter
%% (e.g., when there isn't any following chapter).
%%
%% We therefore provide a command
%% |\clearfootnoteannotations|
%% that creates dummy page entries for those final pages.
%% Of course, these dummy page entries must not be
%% numbered like lines ...
%%
%% The first version of the macro worked fine with a
%% two-page footnote, but crashed as soon as a main text line
%% was removed. (Infinitly many pages are created then ...\,) \
%% This is just the problem that was addressed by
%% |\ForceFootnoteAnnotations|. But it would be very wrong to use
%% `\ForceFootnoteAnnotations' for the present problem,
%% because this would move the annotations for the next
%% footnote-only pages to the next page with numbered
%% main text, or to the end of the entire document. %% tested 2011/01/17
%%
%% The problem has then been %% `en' 2011/01/19
%% solved by inserting `\EFN@run@annot@hook'; %% `;' 2011/01/19
%% however,
%% in reasoning about it, about former failures, about the `\par',
%% it seems to become clear that this is a solution for Christian's
%% `Test13.tex' as of 2011/01/16 after a footnote that closes a
%% paragraph and a main text page at the same time---only,
%% while ...
\newcommand* \clearfootnoteannotations {%
\par %% TODO!? the whole only via \MakeLineNo!?
%% doesn't work without \par
\EFN@run@annot@hook
\loop \ifx\EFN@annot@hook\@empty \else
\pagebreak %% TODO!? \newpage? \clearpage?
%% leave to user before?
\write\m@ne{}\vbox{}%% TODO!? from \clearpage
\EFN@run@annot@hook
\repeat
%% TODO: \clearpage!? may come from \chapter
}
%% TODO: What if pages are removed? What if annotations are inserted
%% or removed? So far, ...
%%
%% === 'ednotes'' &\newlabel\ variant ===
%% 'ednotes' changes \LaTeX's `\newlabel' so that three
%% `\newlabel' entries in the `.aux' file build only
%% \emph{one} macro containing the information needed
%% for typesetting \emph{one} annotation in the apparatus.
%% %% 2011/01/11:
%% Both that building mechanism and the change testing
%% at the final run of the new `.aux' file
%% don't work with 'edfnotes.sty'.
%% To repair this, I needed much time to understand
%% those mechanisms again. I found that they were quite
%% redundant and now am re-implementing them almost
%% entirely.
%%
%% ==== Suppressing the ``&\get\,..." Commands ====
%% %% reworded 2011/01/11:
%% 'lineno' must suppress expansion of |\getpagewiselinenumber|
%% when the `.aux' file is read.
%% With 'fnlineno.sty', |\getfootnotelinenumber|
%% must be prevented from expansion as well. Both macros are
%% activated at `\begin{document}' only
%% when reading the `.aux' has been finished.
%%
%% With 'fnlineno.sty', both macros are accessed as
%% |\getwiselinenumber<cs>|, %% 2011/01/03
%% so we just need to switch `\getwiselinenumber':
% \let\EFN@getlineno\getwiselinenumber
% \let\getwiselinenumber\@gobble
% \AtBeginDocument{\let\getwiselinenumber\EFN@getlineno}
%% %% 2011/01/03:
%% On the other hand, 'ednotes' deals with
%% `\getpagewiselinenumbers' already, so we only ...
%% saves one expansion for each footnote line!
\let \EFN@getfnlineno \getfootnotelinenumber
\let \getfootnotelinenumber \relax
\AtBeginDocument{\let \getfootnotelinenumber \EFN@getfnlineno}
%% By analogy to 'ednotes'' `\AtEnd'\,...\,:
\AtEndDocument{\let\getfootnotelinenumber\relax}
%% ==== &\newlabel\ Building Info Macros ====
%% %% 2011/01/10:
%% For version v0.3 of 'edfnotes', it turns out that
%% 'ednotes'' (v1.1--1.3) mechanism for building the
%% note info macros `\r@EN@l<id>' assumes that
%% \[`\newlabel{-EN@l<id>}{<start-lemma-place>}'\]
%% will appear before
%% \[`\newlabel{+EN@l<id>}{<end-lemma-place>}'\]
%% and
%% \[`\newlabel{EN@l<id>}{<note-place>}'\]
%% in the `.aux' file.
%% With 'edfnotes' v0.3, an annotation to the second page
%% of a long footnote will be `\insert'ed on the previous
%% page in a ``first" run (no `.aux' present) due to the
%% missing placement information.
%% 'ednotes'' `\newlabel' mechanism then looses the
%% note placement information, and a ``missing number"
%% error occurs.
%%
%% To fix this, we modify 'ednotes''
%% |\@EN@addtolabeldef| so that at the first step of
%% the recursive building of `\r@EN@l<number>',
%% the latter is not used as input (as a part).
%% Then the tricky `\@gobble' in the first case of 'ednotes''
%% redefinition of |\newlabel| (see the recent presentation
%% of the code---currently [2011/01/10] `ednotesc.pdf')
%% can be replaced by the obvious empty argument.
%%
%% With the former implementation in 'ednotes',
%% I had already wondered if I couldn't postpone
%% reading the second argument of
%% |\newlabel{<label>}{<info>}|.
%% Now this is what I am doing indeed. %% 2011/01/11
%% 'ednotes' stores the original `\newlabel' as
%% `\@EN@newlabel', and this will be called for the default
%% treatment of <label> and <info>:
%% For re-implementing the `\newlabel' mechanisms,
%% first we don't use \LaTeX's `\in@', but our own test
%% whether `EN@l' is part of a `\newlabel' <label>:
\renewcommand* \newlabel [1] {%
\def\@EN@next{\@EN@newlabel{#1}}%
\@EN@test@noteid #1\@empty EN@l\@empty\@nil
\@EN@next}
%% Here are macros that `\newlabel' invokes:
%%
%% |\@EN@labels| chooses a ``name space" for labels.
%% The intention is that an `.aux' entry
%% `\newlabel{<label>}{<info>}' contributes note placement
%% information \emph{iff} <label> has form
%% `<sign>EN@l<digits>' where <sign> is nothing, `-', or `+'.
%% However, we won't really (soon) check if we have digits indeed ...
\newcommand* \@EN@labels {EN@l}
%% The next line defines a macro |\EN@l| calling
%% `\@EN@addtolabeldef' one way ...
\@namedef {\@EN@labels}{\@EN@addtolabeldef{{\@EN@blemp}}{}}
%% The next two lines define a macro |\-EN@l| calling
%% `\@EN@addtolabeldef' another way ...
\@namedef{-\@EN@labels}{%
\@EN@addtolabeldef{}{{\@EN@bleml}{\@EN@blemp}}}
%% The next two lines render |\+EN@l| an alias for `\-EN@l' ...
\expandafter \let \csname +\@EN@labels \expandafter\endcsname
\csname -\@EN@labels\endcsname
%% |\@EN@test@noteid<split1>EN@l<split2>\@empty<split3>\@nil| \
%% is meant for application as
%% `\@EN@test@noteid<label>\@empty EN@l\@empty\@nil'.
%% (I \emph{think} other control sequences than `\@empty'
%% would do as well.)
%% When the macro considers <label> belonging to the note info system,
%% `\@tempa' equals one out of `\EN@l', `\-EN@l', and `\+EN@l',
%% and `\@EN@tempa' expands to `\r@<split1>EN@l<split2>'.
\def \@EN@test@noteid #1EN@l#2\@empty#3\@nil{%
\ifx\\#2\\\else
%% This should mean that #2 is \emph{not} %% corr. 2011/01/12
%% empty (I \emph{think} the control sequence `\\'
%% can't appear in a label ...\,), %% TODO
%% and this is the case \emph{iff} <label> contains
%% `EN@l' and something more to the \emph{right.}
%% #1 (=~<split1>) then is the part of <label> to the
%% \emph{left} %% right %% corr. 2011/01/12
%% of that occurrence of `EN@l'. Further conditions for <label>
%% belonging to a note are
%% (\textit{i})~#1 is empty, `-', or `+'---we check this by comparing
%% `\<split1>EN@l' to `\EN@l', `\-EN@l', and `\+EN@l'---, and
%% (\textit{ii})~#2 is some <digits>---that we won't check (soon) ...
\expandafter \let \expandafter \@tempa
\csname #1\@EN@labels\endcsname
\ifx\@tempa\relax \else
\let \@EN@next \@tempa
%% This should be the same `\@EN@tempa' as
%% `\@EN@extractcslp' from 'ednotes' v1.3 produced it:
\edef \@EN@tempa {\expandafter\noexpand
\csname r@\@EN@labels#2\endcsname}%
\expandafter
%% The remaining part of former `\@EN@extractcslp'
%% should be called by the new version of
%% `\@EN@addtolabeldef'. The following hook will be
%% used in testing for cross-reference changes:
\@EN@testnote@hook
\fi
\fi
}
%% 'ednotes'' `\@EN@addtolabeldef' even gets a new syntax:
%% \[|\@EN@addtolabeldef{<prepend>}{<append>}{<line/page>}|\]
\renewcommand* \@EN@addtolabeldef [3] {%
\@EN@extract@lp#3\@nil
\expandafter \xdef \@EN@tempa {%
#1%
\expandafter
\ifx\@EN@tempa\relax \else \@EN@tempa \fi
#2}%
\@EN@testnote@h@@k}
%% ... considering 'ednotes'' `\protected'\,... irrelevant.
%% 'ednotes'' `\@EN@extractcslp' is replaced by
%% |\@EN@extract@lp<line/page>\@nil|:
\let \@EN@extractcslp \@undefined %% must be renamed:
\def \@EN@extract@lp #1#2#3\@nil {%
\def\@EN@bleml{#1}\def\@EN@blemp{#2}}
%% The |\@EN@testnote@hook| and |\@EN@testnote@h@@k|
%% will only be used at `\end{document}':
\let \@EN@testnote@hook \relax
\let \@EN@testnote@h@@k \relax
%%
%% ==== &\newlabel\ Testing Cross-reference Changes ====
%% We won't use |\@EN@testlabel| and must override 'ednotes''
%% command to use it in the testing run of the `.aux'
%% (maybe we should care that 'ednotes' has been loaded before!?).
%% %% <- TODO!? 2011/01/12
%% Instead, the hook |\@EN@testnote@hook| will be activated.
%% For the overriding, we need to store our present redefinition
%% of `\newlabel'.
%% (When our re-implementation of `\newlabel' is merged into
%% 'ednotes', there simply shouldn't any change of `\newlabel'
%% at `\end{document}'.)
\@onlypreamble\@EN@testlabel
\let \EFN@newlabel \newlabel
\AtEndDocument{%
\let \newlabel \EFN@newlabel
\def \@EN@testnote@hook {%
%% From 'ednotes'' `\@EN@testlabel':
\expandafter
\ifx\@EN@tempa\relax
%% 2011/01/13:
%% This means that the note info macro that `\@EN@tempa'
%% expands to---call it <id> for the discussion---has been used
%% in the present run of the document, but not in the run before
%% (provided ...\,---see below).
%% So certainly this ``label has changed," and
\@tempswatrue
%% should issue \LaTeX's according final warning.
%% Moreover, the <info> argument of the current
%% `\newlabel' entry should now simply be ignored
%% (and this way the same `\@EN@tempa' will expand
%% to a `\relax' alias next time again):
\let\@EN@next\@gobble
\else
%% This means that the note info macro <id>
%% has been used both in the present run
%% and in the run before, so we want to find out
%% whether the info changed or not.
%% We will find three `\newlabel' entries for the same
%% annotation and must build its new info macro---the string
%% <new-info> that <id> will expand to after reading the `.aux'
%% in the next document run---,
%% we must store the old one---<id> still expands to
%% <old-info>---for comparing,
%% we must be able to detect if we have collected
%% all the three entries that we need for <new-info>,
%% and we must not increase the number of control
%% sequences much. In 'ednotes' we used \emph{one} separate
%% ``testing list" for all the info macros to be tested.
%% The situation is now quite different
%% and we try a new way: We collect all the relevant
%% informations in that info macro <id> and take them
%% apart when we have all of them.
%%
%% We are building <new-info>, at first it is empty.
%% We will insert
%% \[|\@EN@record<count>{<old-info>}|,\]
%% together with the contribution from the first `\newlabel'.
%% Let us see if the <new-info> from the current <id>
%% already contains such a `\@EN@record',
%% using |\@EN@test@record|:
\@EN@xpxpxp \@EN@test@record \@EN@tempa \@EN@record0\@nil
\fi}%
}
\def \@EN@test@record #1\@EN@record#2#3\@nil{%
\if #20%
%% `record' is missing (this conclusion is justified by the rest
%% of the macro), this is the first contribution to a note
%% info macro. We are preparing the insertion that
%% `\@EN@addtolabeldef' will execute---`record 1{<old-info>}':
\let\@EN@record\relax
\expandafter\edef\@EN@tempa{\@EN@record 1{\@EN@tempa}}%
\else
\if #21%
%% Here we have found a \emph{second} contribution.
\def\@EN@record##1{\noexpand\@EN@record 2}%
\else
%% This is the case where the \emph{third} `\newlabel' for the same
%% annotation has been encountered. When its <info> argument
%% has been processed, we can compare the resulting
%% <new-info> to the <old-info> that we are now putting
%% from #3 into `\@tempa':
\@EN@xpxpxp \@EN@extract@record \@EN@tempa \@nil
%% Now `\@tempa' should expand to <old-info>.
\let\@EN@record\@gobbletwo
\fi
\fi
}
%% |\@EN@extract@record| prepares comparison of <old-info> and
%% <new-info>: %% doc. completed 2011/01/17
\def \@EN@extract@record #1\@EN@record 2#2#3\@nil {\def\@tempa{#2}}
%% This is the active |\@EN@testnote@h@@k| providing
%% testing and warning about annotation placement changes:
\AtEndDocument{%
\def \@EN@testnote@h@@k {%
%% When another reference has changed, we may omit this info
%% (probably different to 'ednotes'~v1.3).
\if@tempswa \else
\ifx\@EN@record\@gobbletwo
\expandafter
\ifx \@EN@tempa \@tempa \else
\@tempswatrue %% for LaTeX's warning TODO!?
\@EN@linechange@warning
\fi
\fi
\fi}%
}
\newcommand* \@EN@linechange@warning{%
%% Now copying 'ednotes' v1.3 (with different code line breaks):
\expandafter\@EN@takepagesto\@tempa\@tempa
\@EN@xpxpxp \@EN@takepagesto\@EN@tempa\@EN@tempa
%% The message may come when line numbers have changed, not page
%% numbers. This may puzzle some users, but should be harmless.
\typeout{^^JPackage `ednotes.sty':^^J%
*** A lemma or note moved. Page numbers \@tempa\space turned
into \@EN@tempa.^^J*** Rerun and watch whether they come to
rest and references get right.^^J*** If they don't, use
\string\pagebreak\space or \string\warningpagebreak\space
to force^^J*** the earlier of oscillating page breaks.^^J***
(Cf. package documentation on `oscillating page breaks'.)}%
\let\newlabel\@EN@newlabel
\def\@newl@bel{\@gobblefour\relax}% Cf. `lblchng1.sty'.
}
%% |\@EN@prepare@fromtl| is obsolete:
\@onlypreamble \@EN@prepare@fromtl
%%
%% ==== Finish Typesetting before Testing ====
%% \label{sec:enddoc}
%% When the numbering macros are disabled for testing before
%% finishing typesetting, absolute line numbers instead of
%% relative ones are printed. The disabling commands are
%% issued using `\AtEndDocument' which appends the commands
%% to the `\@enddocumenthook'. We \emph{prepend} a `\par'
%% token to this hook so typesetting the numbered paragraph
%% happens early enough, likewise we may force final
%% annotations whose ``last" line numbers have been lost:
\toks@\expandafter{\@enddocumenthook}
\edef\@enddocumenthook{%
\noexpand\par
\noexpand\ForceFootnoteAnnotations
\the\toks@}
%% We might have used Heiko Oberdiek's 'atveryend' instead
%% of `\AtEndDocument'; and I thought of adding another
%% `\clearpage', but this may conflict with other packages.
%% These matters are described comprehensively in the
%% documentation of the 'lastpage' package.
%%
%% === Leaving the Package File ===
\endinput
%%
%% === VERSION HISTORY ===
v0.1 2010/01/01 starting, trying to change \FootnotetextA
2011/01/01 discovered \@EN@hookfn, exploited
2011/01/02 another task: ednotes' \newlabel
2011/01/03 disabling \getfoot... for .aux already;
added \par to \enddocument;
extended `Limitations', add-macros -> cs;
removed debugging code
TO CHRISTIAN as part of EDFN-r0.4
v0.2 2011/01/04 simplified hook stuff, moving to fnlineno
2011/01/07 <- moving back, insert annotations at
last line of page, restructuring;
\theLastLineNumber
TO CHRISTIAN as part of EDFN-r0.5 night to 2011/01/08
v0.3 2011/01/08 concept
2011/01/09 samepage -> page@end;
details for concept, implementation
2011/01/10 correction of ednotes' \newlabel
v0.31 2011/01/10 JUST STORING: first approach at
repairing \newlabel
v0.32 2011/01/11 restructuring and explanation for
re-implementation of \newlabel,
implementation of building
JUST STORING -- seems to work so far, but there is
old code that just is overridden
v0.33 2011/01/12 corr. doc. mistakes on test@noteid;
hard work at both versions of \newlabel ...
2011/01/13 untidy, but seems to work perfectly
v0.34 2011/01/14 rough and ready fix: \insert at *start*
of \MakeLineNo
TO CHRISTIAN without any other files
v0.35 2011/01/14 acc. Christian debugging
2011/01/15 assignment of \theLastLineNumber conditional,
another change of \MakeLineNo: insertion hook
*after* \makeLineNumber as in v0.33,
\c@linenumber stepped back
2011/01/16 rm. some \show...; \if@FNLN@sw@ replaces \ifx,
\EFN@insert@try, \ForceFootnoteAnnotations;
\par before \newlabel tests, note on
\clearpage and related packages;
restructured inserting section;
tidied \newlabel section
TO CHRISTIAN as part of EDFN r0.55
v0.4 2011/01/17 \EFN@page@end -> annot, \fi \fi ...;
\EFN@insert@try merged into \MakeLineNo again,
new \EFN@run@annot@hook, rm. \EFN@trivialize@last;
\paragraph's in "Forced", compl. doc.
\EFN@extract@record; \clearfootnoteannotations
v0.5 2011/01/18 re-impl. \@EN@putdown etc. entirely, much new doc.
2011/01/19 \testLastNumberedPage uses \ifx\c@linenumber...;
appended hook run to \stepLineNumber;
\ForceFootnoteAnnotations re-implemented;
doc. fixes \paragraph{3.}
v0.6 2011/02/02 \EFN@move in \EFN@annot only *before* lemma page,
copyright updated, DFG reduced, user command*s*
v0.6a 2011/02/10 "Known Issues"
v0.6b 2011/02/14 modifications for Christian's 2011/02/11;
mention `PdUsample.pdf'; history *sub*section;
`tamefloats' link corrected
2011/02/16 install: required files, \pagebreaks