Merge changes in branches/wrappers into trunk.

    [in trunk] svn merge -r105:HEAD \
        https://pandoc.googlecode.com/svn/branches/wrappers


git-svn-id: https://pandoc.googlecode.com/svn/trunk@177 788f1e2b-df1e-0410-8736-df70ead52e1b

12 files changed, 486 insertions, 0 deletions

diff --git a/src/wrappers/checkin.sh b/src/wrappers/checkin.sh
new file mode 100644
index 000000000..c9c564a23
--- /dev/null
+++ b/src/wrappers/checkin.sh
@@ -0,0 +1,7 @@
+# Check if input files exist.
+for f; do
+    if [ -n "$f" ] && ! [ -f "$f" ]; then
+        err "File '$f' not found."
+        exit 1
+    fi
+done
diff --git a/src/wrappers/common.sh b/src/wrappers/common.sh
new file mode 100644
index 000000000..99a83be50
--- /dev/null
+++ b/src/wrappers/common.sh
@@ -0,0 +1,56 @@
+THIS=${0##*/}
+
+NEWLINE='
+'
+WRAPPER_ARGS=
+WRAPPEE_ARGS=
+
+err ()  { echo "$*"   | fold -s -w ${COLUMNS:-110} >&2; }
+errn () { printf "$*" | fold -s -w ${COLUMNS:-110} >&2; }
+
+usage () {
+    synopsis="$@"
+    err "Usage:  $THIS $synopsis"
+    err "See $THIS(1) man file for details."
+}
+
+runpandoc () {
+    if [ -n "$WRAPPEE_ARGS" ]; then
+        # Unpack arguments that will be passed to pandoc.
+        oldifs="$IFS"; IFS="$NEWLINE"; set -- $WRAPPEE_ARGS "$@"; IFS="$oldifs"
+        case "$1" in --) shift;; esac # tolerate the existence of a leading '--'
+    fi
+
+    pandoc "$@"
+}
+
+# Portable which(1).
+pathfind () {
+    oldifs="$IFS"; IFS=':'
+    for _p in $PATH; do
+        if [ -x "$_p/$*" ] && [ -f "$_p/$*" ]; then
+            IFS="$oldifs"
+            return 0
+        fi
+    done
+    IFS="$oldifs"
+    return 1
+}
+
+HAVE_ICONV=
+if pathfind iconv; then
+    HAVE_ICONV=1
+    alias to_utf8='iconv -t utf-8'
+    alias from_utf8='iconv -f utf-8'
+else
+    err "Warning:  iconv not present.  Assuming UTF-8 character encoding."
+    alias to_utf8='cat'
+    alias from_utf8='cat'
+fi
+
+for p in pandoc $REQUIRED; do
+    pathfind $p || {
+        err "You need '$p' to use this program!"
+        exit 1
+    }
+done
diff --git a/src/wrappers/getopts.sh b/src/wrappers/getopts.sh
new file mode 100644
index 000000000..263263c07
--- /dev/null
+++ b/src/wrappers/getopts.sh
@@ -0,0 +1,12 @@
+if [ -z "$SYNOPSIS" ]; then
+    SYNOPSIS="[-h] [input_file]"
+    [ -n "$THIS_NARG" ] || SYNOPSIS="${SYNOPSIS}..."
+fi
+
+while getopts h opt; do
+    case $opt in
+    h|?) usage "$SYNOPSIS"; exit 2 ;;
+    esac
+done
+
+shift $(($OPTIND - 1))
diff --git a/src/wrappers/html2markdown.in b/src/wrappers/html2markdown.in
new file mode 100644
index 000000000..0fece3ccd
--- /dev/null
+++ b/src/wrappers/html2markdown.in
@@ -0,0 +1,134 @@
+#!/bin/sh -e
+# converts html to markdown
+# uses an available program to fetch URL and tidy to normalize it first
+
+REQUIRED=tidy
+
+### common.sh
+
+grab_url_with () {
+    url="${1:?internal error: grab_url_with: url required}"
+
+    shift
+    cmdline="$@"
+
+    prog=
+    prog_opts=
+    if [ -n "$cmdline" ]; then
+	eval "set -- $cmdline"
+	prog=$1
+	shift
+	prog_opts="$@"
+    fi
+
+    if [ -z "$prog" ]; then
+	# Locate a sensible web grabber (note the order).
+	for p in wget lynx w3m curl links w3c; do
+		if pathfind $p; then
+		    prog=$p
+		    break
+		fi
+	done
+
+	[ -n "$prog" ] || {
+            errn "$THIS:  Couldn't find a program to fetch the file from URL "
+	    err "(e.g. wget, w3m, lynx, w3c, or curl)."
+	    return 1
+	}
+    else
+	pathfind "$prog" || {
+	    err "$THIS:  No such web grabber '$prog' found; aborting."
+	    return 1
+	}
+    fi
+
+    # Setup proper base options for known grabbers.
+    base_opts=
+    case "$prog" in
+    wget)  base_opts="-O-" ;;
+    lynx)  base_opts="-source" ;;
+    w3m)   base_opts="-dump_source" ;;
+    curl)  base_opts="" ;;
+    links) base_opts="-source" ;;
+    w3c)   base_opts="-n -get" ;;
+    *)     err "$THIS:  unhandled web grabber '$prog'; hope it succeeds."
+    esac
+
+    err "$THIS: invoking '$prog $base_opts $prog_opts $url'..."
+    eval "set -- $base_opts $prog_opts"
+    $prog "$@" "$url"
+}
+
+encoding=
+grabber=
+nograb=
+while getopts e:g:nh opt; do
+    case $opt in
+    e) encoding="$OPTARG" ;;
+    g) grabber="$OPTARG" ;;
+    n) nograb=1 ;;
+    h|?)
+        usage "[-e encoding] [-g grabber_command] [-n] [-h] [input_file|url]"
+        exit 2 ;;
+    esac
+done
+
+shift $(($OPTIND - 1))
+
+### postopts.sh
+
+### singlearg.sh
+
+inurl=
+if [ -n "$1" ] && ! [ -f "$1" ]; then
+    if [ -n "$nograb" ]; then
+        err "'$1' not found; refusing to treat input as URL."
+        exit 1
+    fi
+    # Treat given argument as an URL.
+    inurl="$1"
+fi
+
+if [ -n "$inurl" ]; then
+    err "Attempting to fetch file from '$inurl'..."
+
+    ### tempdir.sh
+
+    grabber_out=$THIS_TEMPDIR/grabber.out
+    grabber_log=$THIS_TEMPDIR/grabber.log
+    if ! grab_url_with "$inurl" "$grabber" 1>$grabber_out \
+                                           2>$grabber_log; then
+        errn "grab_url_with failed"
+        if [ -f $grabber_log ]; then
+            err " with the following error log."
+            err
+            cat >&2 $grabber_log
+        else
+            err .
+        fi
+        exit 1
+    fi
+
+    set -- $grabber_out
+fi
+
+if [ -z "$encoding" ] && [ "x$@" != "x" ]; then
+    # Try to determine character encoding unless not specified
+    # and input is STDIN.
+    encoding=$(
+        head "$@" |
+        LC_ALL=C tr 'A-Z' 'a-z' |
+        sed -ne '/<meta .*content-type.*charset=/ {
+            s/.*charset=["'\'']*\([-a-zA-Z0-9]*\).*["'\'']*/\1/p
+        }'
+    )
+fi
+
+if [ -n "$encoding" ] && [ -n "$HAVE_ICONV" ]; then
+    alias to_utf8='iconv -f "$encoding" -t utf-8'
+elif [ -n "$inurl" ]; then # assume web pages are UTF-8
+    alias to_utf8='cat'
+fi # else just use local encoding
+
+to_utf8 "$@" | tidy -utf8 2>/dev/null |
+runpandoc -r html -w markdown -s | from_utf8
diff --git a/src/wrappers/latex2markdown.in b/src/wrappers/latex2markdown.in
new file mode 100644
index 000000000..e8cde8a97
--- /dev/null
+++ b/src/wrappers/latex2markdown.in
@@ -0,0 +1,14 @@
+#!/bin/sh -e
+# runs pandoc to convert latex to markdown
+
+### common.sh
+
+### getopts.sh
+
+### postopts.sh
+
+### singlearg.sh
+
+### checkin.sh
+
+to_utf8 "$@" | runpandoc -r latex -w markdown -s | from_utf8
diff --git a/src/wrappers/markdown2html.in b/src/wrappers/markdown2html.in
new file mode 100644
index 000000000..e255398d2
--- /dev/null
+++ b/src/wrappers/markdown2html.in
@@ -0,0 +1,12 @@
+#!/bin/sh -e
+# converts markdown to HTML
+
+### common.sh
+
+### getopts.sh
+
+### postopts.sh
+
+### checkin.sh
+
+to_utf8 "$@" | runpandoc | from_utf8
diff --git a/src/wrappers/markdown2latex.in b/src/wrappers/markdown2latex.in
new file mode 100644
index 000000000..c532b2f99
--- /dev/null
+++ b/src/wrappers/markdown2latex.in
@@ -0,0 +1,12 @@
+#!/bin/sh -e
+# converts markdown to latex
+
+### common.sh
+
+### getopts.sh
+
+### postopts.sh
+
+### checkin.sh
+
+to_utf8 "$@" | runpandoc -w latex -s | from_utf8
diff --git a/src/wrappers/markdown2pdf.in b/src/wrappers/markdown2pdf.in
new file mode 100644
index 000000000..838767224
--- /dev/null
+++ b/src/wrappers/markdown2pdf.in
@@ -0,0 +1,64 @@
+#!/bin/sh -e
+# converts markdown to latex, then uses latex to make a PDF
+
+REQUIRED=pdflatex
+
+### common.sh
+
+outfile=
+while getopts o:h opt; do
+    case $opt in
+    o) outfile="$OPTARG" ;;
+    h|?) usage "[-o output_file] [-h] [input_file]..."; exit 2 ;;
+    esac
+done
+
+shift $(($OPTIND - 1))
+
+### postopts.sh
+
+### checkin.sh
+
+if [ -z "$outfile" ]; then
+    if [ -n "$1" ]; then
+        outfile="${1%.*}"
+    else
+        outfile="stdin" # input is STDIN, since no argument given
+    fi
+fi
+case "$outfile" in
+*.*) ;; # skip appending extension if one is already present
+*)   outfile="${outfile%.*}.pdf";;
+esac
+
+### tempdir.sh
+
+# We should use a filename without white spaces for pdflatex.
+TEXNAME=$THIS
+
+to_utf8 "$@" | runpandoc -w latex -s >$THIS_TEMPDIR/$TEXNAME.tex
+(
+    cd $THIS_TEMPDIR
+    if ! pdflatex -interaction=batchmode $TEXNAME.tex >/dev/null 2>&1; then
+        err "LaTeX errors:"
+        from_utf8 $TEXNAME.log | sed -ne '/^!/,/^ *$/p' >&2
+        if grep -q "File \`ucs.sty' not found" $TEXNAME.log; then
+            err "Please install the 'unicode' package from ctan.org."
+        fi
+        exit 1
+    fi
+)
+
+is_target_exists=
+if [ -f "$outfile" ]; then
+    is_target_exists=1
+    mv -f "$outfile" "$outfile~"
+fi
+
+mv -f $THIS_TEMPDIR/$TEXNAME.pdf "$outfile"
+
+errn "Created '$outfile'"
+[ -z "$is_target_exists" ] || {
+    errn " (previous file has been backed up as '$outfile~')"
+}
+err .
diff --git a/src/wrappers/postopts.sh b/src/wrappers/postopts.sh
new file mode 100644
index 000000000..e0d015f41
--- /dev/null
+++ b/src/wrappers/postopts.sh
@@ -0,0 +1,17 @@
+# Parse wrapper and wrappee (pandoc) arguments by taking
+# into account that they may have space or tab characters.
+pick="WRAPPER_ARGS"
+while [ $# -gt 0 ]; do
+    if [ "$pick" = "WRAPPER_ARGS" ]; then
+        case "$1" in
+        -*) pick="WRAPPEE_ARGS" ;;
+        esac
+    fi
+    # Pack args with NEWLINE to preserve spaces,
+    # and put them into the picked variable.
+    eval "$pick=\"\$${pick}${NEWLINE}${1}\""
+    shift
+done
+
+# Unpack filename arguments.  Now "$@" will hold the filenames.
+oldifs="$IFS"; IFS="$NEWLINE"; set -- $WRAPPER_ARGS; IFS="$oldifs"
diff --git a/src/wrappers/singlearg.sh b/src/wrappers/singlearg.sh
new file mode 100644
index 000000000..f742d1383
--- /dev/null
+++ b/src/wrappers/singlearg.sh
@@ -0,0 +1,7 @@
+# Ensure to work with a single argument.
+if [ $# -gt 1 ]; then
+    first_arg="$1"
+    shift
+    err "Warning:  extra arguments '$@' will be ignored."
+    set -- $first_arg
+fi
diff --git a/src/wrappers/tempdir.sh b/src/wrappers/tempdir.sh
new file mode 100644
index 000000000..8d2754b1c
--- /dev/null
+++ b/src/wrappers/tempdir.sh
@@ -0,0 +1,10 @@
+# As a security measure refuse to proceed if mktemp is not available.
+pathfind mktemp || { err "Couldn't find 'mktemp'; aborting."; exit 1;  }
+
+THIS_TEMPDIR=
+THIS_TEMPDIR="$(mktemp -d -t $THIS.XXXXXXXX)" || exit 1
+readonly THIS_TEMPDIR
+
+trap 'exitcode=$?
+      [ -z "$THIS_TEMPDIR" ] || rm -rf "$THIS_TEMPDIR"
+      exit $exitcode' 0 1 2 3 13 15
diff --git a/src/wrappers/testwrapper.in b/src/wrappers/testwrapper.in
new file mode 100644
index 000000000..b9e10b450
--- /dev/null
+++ b/src/wrappers/testwrapper.in
@@ -0,0 +1,141 @@
+#!/bin/sh
+
+THIS=$1
+
+ASH="ash -s"
+BASH="bash --posix -s"
+DASH="dash -s"
+KSH="ksh -s"
+POSH="posh -s"
+ZSH="zsh -s"
+
+ERROR=""
+
+wrapper () {
+    $SH -- "$@" <<-'EOF'
+### common.sh
+
+outfile=
+while getopts o: opt; do
+    case $opt in
+    o) outfile="$OPTARG" ;;
+    esac
+done
+
+shift $(($OPTIND - 1))
+
+### postopts.sh
+
+echo "Options passed to wrapper:"
+[ -z "$outfile" ] || echo "|$outfile|"
+
+echo "Arguments passed to wrapper:"
+for arg; do
+    echo "|$arg|"
+done
+
+pandoc () {
+    echo "Arguments passed to wrappee:"
+    for arg; do
+        echo "|$arg|"
+    done
+}
+runpandoc
+EOF
+}
+
+# Portable which(1).
+pathfind () {
+    oldifs="$IFS"; IFS=':'
+    for _p in $PATH; do
+        if [ -x "$_p/$*" ] && [ -f "$_p/$*" ]; then
+            IFS="$oldifs"
+            return 0
+        fi
+    done
+    IFS="$oldifs"
+    return 1
+}
+
+check_results () {
+        if [ "$1" = "$2" ]; then
+            echo >&2 ok
+            return 0
+        else
+            echo >&2 failed
+            sed "s/^/\t/" >&2 <<EOF
+Command line: '$3'
+===> Expected:
+$2
+<=== Got:
+$1
+EOF
+            return 1
+        fi
+}
+
+for SH in "$BASH" "$DASH" "$KSH" "$ZSH"; do
+    CMD=${SH%% *}
+    echo >&2 "  Testing with $CMD..."
+    if pathfind "$CMD"; then
+        if [ "$CMD" = "zsh" ]; then
+            # Zsh needs to be called as 'sh' to enable POSIX mode.
+            ln -s $(which zsh) ./sh
+            SH="./sh ${SH#* }"
+            trap 'err=$?; rm -f ./sh; exit $err' 0 1 2 3 13 15
+        fi
+
+        set -e
+
+        # Test 1
+        printf >&2 "    test case 1... "
+        actual=$(wrapper -o "output file" "foo bar" -A "quux baz" -B)
+        expected=$(cat <<'EOF'
+Options passed to wrapper:
+|output file|
+Arguments passed to wrapper:
+|foo bar|
+Arguments passed to wrappee:
+|-A|
+|quux baz|
+|-B|
+EOF
+)
+        check_results "$actual" "$expected" \
+            'wrapper -o "output file" "foo bar" -A "quux baz" -B'
+
+        # Test 2
+        printf >&2 "    test case 2... "
+        actual=$(wrapper -- -A "foo bar")
+        expected=$(cat <<'EOF'
+Options passed to wrapper:
+Arguments passed to wrapper:
+Arguments passed to wrappee:
+|-A|
+|foo bar|
+EOF
+)
+        check_results "$actual" "$expected" 'wrapper -- -A "foo bar"'
+
+        # Test 3 (Test 1 with a redundant '--')
+        printf >&2 "    test case 4... "
+        actual=$(wrapper -o "output file" "foo bar" -- -A "quux baz" -B)
+        expected=$(cat <<'EOF'
+Options passed to wrapper:
+|output file|
+Arguments passed to wrapper:
+|foo bar|
+Arguments passed to wrappee:
+|-A|
+|quux baz|
+|-B|
+EOF
+)
+        check_results "$actual" "$expected" \
+            'wrapper -o "output file" "foo bar" -- -A "quux baz" -B'
+    else
+        echo >&2 "Warning: cannot verify correctness with $CMD; shell not available"
+    fi
+done
+
+exit 0