#!/bin/sh

# paginate 0.1.1 (2001-Apr-14-Sat)
# Adam M. Costello <amc@cs.berkeley.edu>

# paginate [ -L <pagelength> ] [ -n ] [ <file>... ]
#
# Concatenates the named files (or stdin if none are specified),
# removing all formfeed characters, removing trailing whitespace from
# every line, and occasionally replacing contiguous blank lines with a
# formfeed so that no page exceeds <pagelength> lines (defaults to 64).
# When no blank lines are found soon enough, a formfeed is inserted.
# If -n is supplied, a newline follows every formfeed.
#
# Tries not to end the page with a paragraph in which fewer than two
# lines contain letters, unless that paragraph is followed by more than
# one blank line.

maxpage=64
newline=0

while :; do
  case "$1" in
    -L) maxpage=$2
        shift 2
        ;;

    -n) newline=1
        shift
        ;;

    -*) echo "unknown option $1" >&2
        exit 1
        ;;

     *) break
        ;;
  esac
done

# paralen is the number of lines in the current paragraph.
# gaplen is the number of blanks lines preceeding the current paragraph.
# pagelen is the number of lines on the current page: the number of
#   lines already printed plus gaplen plus paralen.
# numlwl is the number of lines with letters seen since the last blank
#   line.
# hardblank is a flag indicating that the current paragraph ends with
#   a blank line.

{ cat ${1+"$@"}; echo; echo; } | tr -d '\014' |
awk '
  BEGIN {
    paralen = 0;
    gaplen = 0;
    pagelen = 0;
    numlwl = 0;
    hardblank = 0;
  }

  NR == 1 {
    if (newline) formfeed = "\f\n";
    else formfeed = "\f";
  }

  { line = $0;
    n = match(line, /[ \t]*$/);
    line = substr(line, 1, n - 1);
    ++pagelen;

    if (line != "") {
      paragraph[++paralen] = line;
      hardblank = 0;

      if (match(line, /[a-zA-Z]/))
        ++numlwl;

      if (paralen > maxpage) {
        prefixlen = maxpage - (pagelen - paralen);

        if (prefixlen < 0) {
          printf formfeed;
          gaplen = 0;
          pagelen = paralen;
          prefixlen = maxpage;
        }

        for (i = 0;  i < gaplen;  ++i)
          print "";
        for (i = 1;  i <= prefixlen;  ++i)
          print paragraph[i];
        printf formfeed;
        gaplen = 0;
        for (i = prefixlen + 1;  i <= paralen;  ++i)
          paragraph[i - prefixlen] = paragraph[i];
        paralen -= prefixlen;
        pagelen = paralen;
      }
    }
    else {
      if (paralen == 0) {
        ++gaplen;
      }
      else {
        if (! hardblank && numlwl < 2) {
          paragraph[++paralen] = "";
          hardblank = 1;
        }
        else {
          if (pagelen - 1 - hardblank > maxpage) {
            printf formfeed;
            gaplen = 0;
            pagelen = paralen + 1;
          }

          for (i = 0;  i < gaplen;  ++i)
            print "";
          for (i = 1;  i <= paralen - hardblank;  ++i)
            print paragraph[i];
          paralen = 0;
          gaplen = hardblank + 1;
          hardblank = 0;
        }
      }

      numlwl = 0;
    }
  }
' maxpage=$maxpage newline=$newline