r2h

#!/bin/bash

### This only works with bash. dash/ash does not support the text
# substitutions that we make. I should just rewrite this in perl
####!/bin/sh

### If this script is NOT a symlink, readlink should just
# return $0. If it is a symlink, readlink will dereference it.
#
# NOTE: if you want to install this script in a separate
# directory from bootstrap_rst2html then edit the code
# below and explicitly set INSTALL_DIR.
INSTALL_DIR=`dirname $(readlink -f $0)`
BOOTSTRAP_SCRIPT="$INSTALL_DIR/bootstrap_rst2html"

TEMPLATE_DIR="${INSTALL_DIR}/templates"
CSS_DIR="${INSTALL_DIR}/css"

### Defaults:
TEMPLATE="${TEMPLATE_DIR}/pygments_template.txt"
HIGHLIGHTER=pygments
STYLE=autumn

BLOGGER_STYLESHEET="${CSS_DIR}/blogger_compat.css"

BROWSER_CMD="x-www-browser {FILE}"
ACTION="print"

usage()
{
    script_name=`basename $0`
    cat <<END_USAGE

    Usage: $script_name [OPTIONS] [RST_FILE]

    Translates the contents of RST_FILE to html. 

    If RST_FILE is not specified, $script_name reads its input from STDIN.

    Options:

        -c, --clipboard             send output to your clipboard
                                    instead of to STDOUT (the default)

        -v, --validate              validate the markup. This will discard
                                    STDOUT and print only STDERR, which will
                                    print errors (if any) without printing
                                    the output html.

        -w, --preview               preview HTML output in a web browser
                                    instead of to STDOUT.

        -b, --browser               set the web browser command to use
                                    for --preview. You must include {FILE}
                                    as a placeholder for the temporary
                                    file that your browser will open if
                                    you define a custom command. Default is
                                    ${BROWSER_CMD}.

                                    Note that {FILE} will automatically be
                                    expressed in
                                    file:///path/to/auto/generated/tempfile
                                    URL notation.

        -l, --lib                   the name of the syntax highlighting library
                                    to use. These opt args are supported
                                    (case insensitive):

                                    sh | SyntaxHighlighter  = SyntaxHighlighter
                                    p | pygments            = Pygments

        -L, --list                  List syntax highlighting backends and
                                    styles.

        -s, --style                 The name of the syntax highlighting style
                                    to use. Look in css/[LIBRARY NAME]/*css
                                    for the list of css files. Specify an
                                    unambiguous substring of the css
                                    filename (no need for abs path).
                                        
        -t, --tmpfile               the basename of the file to use as a
                                    tempfile. If this is not supplied and
                                    we do a preview using STDIN as input,
                                    mktemp is used to generate a temp filename.
                                    If RST_FILE is defined (not reading STDIN),
                                    then the temp file will be
                                    \$TMPDIR/RST_FILE.html (TMPDIR defaults to
                                    /tmp if not defined)

    Notes:
        Some examples for --browser:

            --browser "chromium-browser --app={FILE} --temp-profile" 
            --browser "epiphany --new-window {FILE}" 
            --browser "iceweasel -safe-mode -new-window {FILE}" 
            --browser "gnome-terminal -e 'w3m {FILE}'" 

        BROWSER="rxvt-unicode --font"
        BROWSER="\$BROWSER '-misc-fixed-medium-r-normal--15-120-100-100-c-90-iso8859-1'"
        BROWSER="\$BROWSER -e w3m {FILE}"

            --browser="\$BROWSER"

    Tip: If you are editing an .rst file in vim and you want to 
    translate the current buffer to html and copy it to the clipboard
    in one step (so you can then paste it into, e.g., your 
    blog-publishing client), do:

        :!r2h -c %

    To alias the above as just R2hclip, put this in your ~/.vimrc:

        command R2hclip !r2h --clip

    Examples:
        # Preview using SyntaxHighlighter for code formatting with the Midnight theme:
        r2h --preview --style=midnight --lib=sh myfile.rst

        # Preview using Pygments for code formatting with the Manni theme:
        r2h --preview --style=manni --lib=pygments myfile.rst

END_USAGE
}

### The <body> tag is always followed by <div class="document", and
# the last 2 lines will close the div and the body... This is fragile,
# but it is also trivial to fix if changes to the output format in
# a future docutils version break it.
extract_document_body()
{
    sed -n '/^<body>/,/^<\/body>/p' | sed -e '1,2d' | head -n '-2'
}

cleanup_tmp_file()
{
    r2h_tmp_file="$1"
    #[ -f "$r2h_tmp_file" ] && rm "$r2h_tmp_file"
    echo "MUST CLEAN UP: $r2h_tmp_file"
}

### Copy STDIN to the Xwindows CLIPBOARD paste buffer
to_clip()
{
    xsel -bi
}

list_backends()
{
    pattern="%-22s %s\n"
    printf "$pattern" "----------------------" "------------------------" 
    printf "$pattern" "Highlighting Backend"    Theme
    printf "$pattern" "----------------------" "------------------------" 
    be=Pygments
    for f in `ls "$INSTALL_DIR"/css/pygments`; do
        f=`basename "$f"`
        theme=${f%.css}
        printf "$pattern" $be $theme
    done
    echo
    be=SyntaxHighlighter
    for f in `ls "$INSTALL_DIR"/css/syntax_highlighter |grep -v shCore`; do
        f=`basename "$f"`
        theme=${f%.css}
        theme=${theme#shTheme}
        printf "$pattern" $be $theme
    done
    printf "$pattern" "----------------------" "------------------------" 
    echo
}

### By default, pipe to cat, which will just print to STDOUT
OUTPUT_FILTER=cat

RST_ARGS='--link-stylesheet --no-doc-title'

### getopt boilerplate:
LONG_OPTS="help,clipboard,preview,list,validate,lib:,style:,browser:,tmpfile:"
OPTS=$(getopt --long "$LONG_OPTS" "hcwvLb:t:l:s:" "$@")
[ $? != 0 ] && ( usage; exit 1 )
eval set -- "$OPTS"

while true; do
  case "$1" in
    -h|--help)      usage; exit 0;;
    -c|--clip*)     OUTPUT_FILTER="to_clip"; shift;;
    -w|--prev*)     ACTION=preview; shift;;
    -v|--valid*)    ACTION=validate; shift;;
    -L|--list*)     ACTION=list; shift;;
    -b|--browser)   BROWSER_CMD="$2"; shift 2;;
    -l|--lib)       
        export SYNTAX_HIGHLIGHTING_FRAMEWORK="$2"
        is_sh=0
        for s in SH SYNTAXHIGHLIGHTER; do
            m=`expr match "${SYNTAX_HIGHLIGHTING_FRAMEWORK^^}" "$s"`
            is_sh=$((is_sh+m))
        done
        if [ $is_sh -gt 0 ]; then
            TEMPLATE="${TEMPLATE_DIR}/syntax_highlighter_template.txt"
            HIGHLIGHTER=syntax_highlighter
        fi
        shift 2
        ;;
    -s|--style)     STYLE="$2"; shift 2;;
    -t|--tmpfile)   TMP_FILE="$2"; shift 2;;
    --) shift; break ;;
    *) break ;;
  esac
done

RST_DOCUMENT="$1"

### If RST_DOCUMENT is specified, validate that it is a real file
if [ -n "$RST_DOCUMENT" ] && [ ! -f "$RST_DOCUMENT" ]; then
    usage; exit 1
fi

case "$ACTION" in
    validate) 
        $BOOTSTRAP_SCRIPT $RST_ARGS --exit-status=3 \
            $RST_DOCUMENT 1> /dev/null
        exit_status=$?
        [ $exit_status = 0 ] && echo "No Errors in Markup."
        ;;
    list)
        list_backends
        exit 0
        ;;
    print)
        $BOOTSTRAP_SCRIPT \
            $RST_ARGS --template="$TEMPLATE_DIR/template_clipboard.txt" \
            $RST_DOCUMENT \
            | $OUTPUT_FILTER
        ;;
    preview)
        if [ -f "$RST_DOCUMENT" ] || [ -n "$TMP_FILE" ]; then
            [ ! -d "$TMPDIR" ] && TMPDIR=/tmp
            if [ -z "$TMP_FILE" ]; then 
                TMP_FILE="$RST_DOCUMENT"
            fi
            fname=`basename "$TMP_FILE"`
            r2h_tmp_file="${TMPDIR}/${fname}.html"
        else
            r2h_tmp_file=`mktemp --suffix=.html`
        fi

        LIB_CSS_DIR="${CSS_DIR}/${HIGHLIGHTER}"
        if [ -z "$STYLE" ] && [ "$HIGHLIGHTER" = syntax_highlighter ]; then
            STYLE=Default
        fi

        STYLE_DEF="$BLOGGER_STYLESHEET"
        STYLE_FILE=`find "$LIB_CSS_DIR" -iregex ".*$STYLE.*" |head -n1`

        if [ ! -z "$STYLE_FILE" ]; then
            sf=`readlink -f "$STYLE_FILE"`
            STYLE_DEF="${STYLE_DEF},${sf}"
        fi

        $BOOTSTRAP_SCRIPT \
            $RST_ARGS --template="$TEMPLATE" \
            --stylesheet="$STYLE_DEF" \
            $RST_DOCUMENT > "$r2h_tmp_file"

        # ensure the file is cleaned up on EXIT, even if the user
        # sends a SIGINT. Below we have a hack for firefox/iceweasel, which
        # do not block when you call them on the command line if a current
        # browser process is running. Because firefox does not block, there
        # is a race-condition between the trap to delete the tmp file
        # and the loading of the file in FF. Our lame workaround is to
        # just sleep in the EXIT trap to allow FF to load the file before
        # we delete it. I also include a check for setsid
        # in case we want to run with, e.g., "setsid chromium-browser {FILE}".
        # setsid gives responsiveness, but b/c we have to sleep for the browser
        # to win the race condition (and we want to ensure that the file is
        # cleaned up on exit), using it makes no sense
        is_race_condition=0
        for s in firefox iceweasel setsid chromium-browser; do
            idx=`expr match "$BROWSER_CMD" "$s"`
            is_race_condition=$((is_race_condition+idx))
        done

        if [ $is_race_condition -gt 0 ]; then
            trap '[ -f "$r2h_tmp_file" ] && sleep 4 ; rm "$r2h_tmp_file"' EXIT
        else
            trap '[ -f "$r2h_tmp_file" ] && rm "$r2h_tmp_file"' EXIT
        fi

        ### This only works in bash (not, e.g., dash). I should this entire
        # script in perl or python
        ${BROWSER_CMD/\{FILE\}/file://$r2h_tmp_file} >/dev/null 2>&1
        ;;
    *) usage; exit 1;;
esac