# ws.opt - rbmake options and rules for converting Baen WebScription books
#
# You use this file like this (once for each ISBN book number):
#
#    rbmake -L ISBN -l ws.opt
#
# All comments in the options section must occupy a full line.

# These substitution rules allow the user to specify the ISBN number
# either with or without dashes, and we handle the rest.  They cause
# the options that follow to get tweaked according to these rules.

"Set-Info":
{
    s/(ISBN=\d)(\d{3})(\d{5})([\dX])/$1-$2-$3-$4/;
    s/(isbn-\d+)-(\d+)-(\d+)-([\dX]+)/$1$2$3$4/;
}

"Book-Filename":
"Cover-Image":
"Input-File":
"Exclude-URLs-Matching":
"Menu-Item":
{
    s/(\d+)-(\d+)-(\d+)-([\dX]+)/$1$2$3$4/;
}

"Cover-Image":
{
    s/0671062000\.jpg//;
}

-- End of optional substitution section ---

# Note that the _Put_Title_Here_ filename causes rbmake magic to happen
# (i.e. the actual title grabbed from the HTML is used for the filename).
# If you want to use the ISBN number for the book's name, change it to $1.
Book-Filename: _Put_Title_Here_

Set-Info: URL=ebook:isbn-$1
Set-Info: ISBN=$1

Cover-Image: $1.jpg
Input-File: $1_toc.htm
Input-File: $1__c_.htm

Follow-Links: yes
Include-Images: yes
Include-Audio: no

Accept-URLs-Matching:
Auto-Accept-Input-File-Dirs: yes

Exclude-URLs-Matching: */$1.jpg
Exclude-URLs-Matching: */$1.htm
Exclude-URLs-Matching: */order_btn.gif

Menu-Item: Contents=$1_toc.htm

Use-Book-Paragraphs: yes
Enhance-Punctuation: "'-
Allow-Old-Style-Page-Breaks: no
Page-Joining: all
Image-Edge-Enhancement: 7  (none, 1 - 9)
Image-Edge-Enhancement: 9 = *_m[\d_].jpg

Substitution-Rule-File: -

--- Substitution-Rule Data Follows ---
# This rule data will tweak some things in a Baen WebScription book before
# the rbmake HTML parser takes over and turns it into an .rb file.  If a
# page matches more than one section, the rules are guaranteed to be run
# in order.

# Fix some typos (some of which have already been fixed in the latest
# webscription versions -- re-download your older books when in doubt).
"*/0671578871_toc.htm":
{
    s/>Sentry Peek</>Sentry Peak</g;
}

"*/0671578510_toc.htm":
"*/0671578510__c_.htm":
{
    s/Telzey Amberson/Telzey Amberdon/g;
}

"*/0671319566_toc.htm":
"*/0671319566__c_.htm":
{
    s/Teh Sacred Pool/The Sacred Pool/g;
}

"*/0671319531__10.htm":
"*/0671578855__c_.htm":
{
    s/&npsp;/&nbsp;/g;
}

"*/0671578723_toc.htm":
{
    s/Sheild/Shield/g;
}

"*/0671578774_toc.htm":
{
    s/Seer' Blood/Seer's Blood/g;
}

# Add a missing author/copyright meta tag.
"*/0671319566_toc.htm":
{
    s%(</title>)%$1\n<meta name="Copyright" content="&copy; 2000 by L. Warren Douglas" />%;
}

# Fix the scene breaks in 1632's chapter files.
m"/0671578499_+\d+\.htm$":
{
    s/(<div align="center">\*\s+\*\s+\*)t\s+t\s+t/$1/g;
}

# Fix the scene breaks in The Legend That Was Earth's chapter files
# plus chapter 32 of The Vlad Tapes.
m"/0671319450_+\d+\.htm$":
"*/0671578782__32.htm":
{
    s/(<div align="center">\*\s+\*\s+\*)\?/$1/g;
}

# Fix one scene break in The Philosophical Strangler's chapter 12 file.
"*/0671319868__12.htm":
{
    s/(<div align="center">\*\s+\*\s+\*)u\s+u\s+u/$1/g;
}

# Fix one scene break in Seer's Blood's chapter 22 file.
"*/0671578774__22.htm":
{
    s/(<div align="center">\*\s+\*\s+\*)Nn/$1/g;
}

# Fix one scene break in The Fata Morgana's chapter 9 file.
"*/0671578766___9.htm":
{
    s%<i>(\*\s+\*\s+\*)</i>%$1%g;
}

# These rules affect all the HTML files.
"*.htm":
{
    # Get rid of some useless NAME/ID tags.
    s%<a id="(p?|Heading)\d+" name="(p?|Heading)\d+"></a>%%ig;

    # Make sure ellipses stay as separate chars & don't break across lines.
    s{\. \. \.}    {.\xA0.\xA0.}g;
    s{\.\xA0\. \.} {.\xA0.\xA0.}g;

    # Dump a couple things that just clutter up the page.
    s%<img [^>]*src *= *"06\d\d\d\d\d\d\d[\dX]\.jpg[^>]*>\s*(?:<br />)?%%i;
    s%<a href=[^>]+>(Back|Next|Framed|Contents)</a>\s*(\| *|<br />)?%%ig;
    s%<p align="?right"?>\s*</p>%%ig;

    # Put the maps in the "Go To" menu.
    s{(<img(?: \S+)* +src="06\d\d\d\d\d\d\d[\dX]_?_m([\d_])\.[^>]*>)}
     {<A NAME="Map$2"></A>$1<META NAME="rocket-menu" CONTENT="Map #$2=#Map$2">\n}sig;

    # If there's only one map, don't call it "Map #1".
    s/CONTENT="Map #[1_]=/CONTENT="Map=/;
    s/(CONTENT="Map)(=.*CONTENT="Map #2=)/$1 #1$2/s;
}

# These rules are only for the chapter files.
/_[p\d]+\.htm$/:
{
    # Remove the trailing per-chapter copyright or just the body-ending stuff.
    s%(?:<p [^>]*>(?:&nbsp;)?</p>\s*)*\s*<hr />.*%%is;
    s%\n?<\/body>.*%\n%is;

    # Transform the scene breaks into <HR SCENE-BREAK> tags
    s{\n?<p (onmouseover=|align="?(left|justify|center)"?)[^>]*>(\s+|&nbsp;|\xA0)*(\*+(\s+|&nbsp;|\xA0)*)*(</p>)?\s*\n}
     {<HR SCENE-BREAK>\n}igs;
    s{\n?<center>(\s+|&nbsp;|\xA0)*(\*+(\s+|&nbsp;|\xA0)*)*</center>\s*} {<HR SCENE-BREAK>\n}igs;
    s{\n?<div align="?center"?>(\s+|&nbsp;|\xA0)*(\*+(\s+|&nbsp;|\xA0)*)*</div>\s*}
     {<HR SCENE-BREAK>\n}igs;
    s%(<HR SCENE-BREAK>\n?){2,}%<HR SCENE-BREAK>\n%g;
    s%(</(h[1-6]|blockquote)>\n?)<HR SCENE-BREAK>%$1%ig;
    s%<HR SCENE-BREAK>(\n?<(h[1-6]|blockquote))%$1%ig;
    s%(<HR SCENE-BREAK>)\n?<br>\n?%$1\n%ig;
    s%\n?<HR SCENE-BREAK>\s*$%\n%is;

    # If you like your paragraphs fully justified, uncomment this line.
    #s%<p\s+onmouseover[^>]+>%<p align=justify>%ig;

    s%&([a-z]+=)%&amp;$1%g;
}

# Try to fix the extraneous-BR-at-chapter-start bug in a few books.
m"/0671(319639|578405|578545|578758|578855)_+\d+\.htm$":
{
    s%(</h1>\s*<p [^>]*>[^<]+)<br />%$1%s;
}

# Fix a few things in Change of Command.
"*/0671319639___4.htm":
"*/0671578405___4.htm":
{
    s%(sapphire).*?(seas)%$1 $2%s;
}

"*/0671319639__10.htm":
"*/0671578405__10.htm":
{
    s%(Yes\.\s+)if( it were)%$1If$2%;
}

"*/0671319639__13.htm":
"*/0671578405__13.htm":
{
    s%posssibilities%possibilities%;
}

"*/0671319639__15.htm":
"*/0671578405__15.htm":
{
    s%(Jurowski), (but Martin-Jehore)%$1--$2%;
}

"*/0671319639__19.htm":
"*/0671578405__19.htm":
{
    s%(All )though( the questions that)%$1through$2%;
}

# Improve the names of the "Go To" menu items in Ashes of Victory.
m"/0671578545__(c_|49|50)\.htm$":
{
    s/Map #1/Diagram #1/;
    s/Map #2/Diagram #2/;
    s/Map #3/Map/;

    # Kludge around a broken link (in an older version of the book).
    s/(0671578545)(_m3\.)/$1_$2/;
}

# Rewrite the Table of Contents page.
"*_toc.htm":
{
    # Put common prefixes (The, A, An) at the end of the title
    s%(<title>)(An?) (.*?)(?=</title>)%$1$3, $2%i;
    s%(<title>)The (.*?)(?=</title>)%$1$2, The%i;

    s%(<title>Apocalypse Troll)(?=</title>)%$1, The%i;

    # Generate an author META tag so that rbmake knows who the author is.
    s{(<meta name="Copyright" content="\S+ \d+ (?:by |edited by |created by |to the Estate of ))([^"]+)}
     {<meta name="author" content="$2">\n$1$2}i;
}

# Rewrite the copyright page.
"*_c_.htm":
{
    # Get rid of all the CENTER tags, since they are often bogus.
    s%</?center>%%g;

    # This fixes some malformed tags.
    s%(</h[1-6])(\n<)%$1>$2%s;
    s%(</td)(\s*\n)%$1>$2%s;
}