`headache`
                               **********
                            Vincent Simonet
                            ===============
                             November, 2002
                             ==============

  This manual is also available in plain text (1), PostScript (2) and
PDF (3).


1 Overview
*=*=*=*=*=


  It is a common usage to put at the beginning of source code files a
short header giving, for instance, some copyright informations.
`headache` is a simple and lightweight tool for managing easily these
headers. Among its functionalities, one may mention: 

   - Headers must generally be generated as comments in source code
   files. `headache` deals with different file types and generates for
   each of them headers in an appropriate format. 
   - `headache` automatically detects existing headers and removes them.
   Thus, you can use it to update headers in a set of files. 

  `headache` is distributed under the terms of the GNU Library General
Public License. See file `LICENSE` of the distribution for more
information.


2 Compilation and installation
*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=


  Building `headache` requires Objective Caml (available at
http://caml.inria.fr/) and GNU Make. In addition, from version
1.03-utf8, the build requires the Unicode library Camomile and, from
version 1.04, Dune.

Instructions

  `headache` is available through OPAM (available at
http://opam.ocaml.org/), the OCaml Package Manager. This is the
preferred installation method. Then the following sequence of commands
should install the package: 
<<
    opam init
    opam install headache
>>

  Alternatively, you can use these commands (Dune must be installed):
<<
    make && sudo make INSTALLDIR=/usr/local/bin install
>>
   Build the executable and install it into the specified directory.


3 Usage
*=*=*=*


  Let us illustrate the use of this tool with a small example. Assume
you have a small project mixing C and Caml code consisting in three
files `foo.c`, `bar.ml` and `bar.mli`, and you want to equip them with
some header. First of all, write a header file, i.e. a plain text file
including the information headers must mention. An example of such a
file is given in figure 1. In the following, we assume this file is
named `myheader` and is in the same directory as source files.
  Then, in order to generate headers, just run the command: 
<<
    headache -h myheader foo.c bar.ml bar.mli
>>
   Each file is equipped with a header including the text given in the
header file `myheader`, surrounded by some extra characters depending on
its format making it a comment (e.g. `(*` and `*)` in `.ml` files). If
you update informations in the header file `myheader`, you simply need
to re-run the above command to update headers in source code files:
existing ones are automatically removed.
  Similarly, running: 
<<
    headache -r foo.c bar.ml bar.mli
>>
   removes any existing in files `foo.c`, `bar.ml` and `bar.mli`. Files
which do not have a header are kept unchanged.
  The current headers of files can be extracted: 
<<
    headache -e foo.c bar.ml bar.mli
>>
   prints on the standard output the current headers of the files
`foo.c`, `bar.ml` and `bar.mli`. All files are kept unchanged.
   --------------------------------------------------------------------

   <<
                                  Headache
                    Automatic generation of files headers

             Vincent Simonet, Projet Cristal, INRIA Rocquencourt

     Copyright 2002
     Institut National de Recherche en Informatique et en Automatique.
     All rights reserved.  This file is distributed under the terms of
     the GNU Library General Public License.

     Vincent.Simonet@inria.fr          
   http://cristal.inria.fr/~simonet/
   >>

     Figure 1: An example of header file

   --------------------------------------------------------------------



4 Configuration file
*=*=*=*=*=*=*=*=*=*=


  File types and format of header may be specified by a configuration
file. By default, the default builtin configuration file given in
figure 2 is used. You can also use your own configuration file thanks to
the `-c` option: 
<<
    headache -c myconfig -h myheader foo.c bar.ml bar.mli
>>

  In order to write your own configuration, you can follow the example
given in figure 2. A configuration file consists in a list of entries
separated by the character `|`. Each of them is made of two parts
separated by an `->`: 

   - The first one is a regular expression. Regular expression are
   enclosed within double quotes and have the same syntax as in Gnu
   Emacs. `headache` determines file types according to file basenames;
   thus, each file is dealt with using the first line its name matches. 
   - The second one describes the format of headers for files of this
   type. It consists of the name of a model (e.g.  `frame`), possibly
   followed by a list of arguments. Arguments are named: `open:"(*"`
   means that the value of the argument `open` is `(*`. 
   `headache` currently supports three models: 

   - `frame`. With this model, headers are generated in a frame. This
   model requires three arguments: `open` and `close` (the opening and
   closing sequences for comments) and `line` (the character used to
   make the horizontal lines of the frame). Two optional arguments may
   be used: `margin` (a string printed between the left and right side
   of the frame and the border, by default two spaces) and `width` (the
   width of the inside of the frame, default is 68). 
   - `lines`. Headers are typeset between two lines. Three arguments
   must be provided: `open` and `close` (the opening and closing
   sequences for comments), `line` (the character used to make the
   horizontal lines). Three optional arguments are allowed: `begin` (a
   string typeset at the beginning of each line, by default two spaces),
   `last` (a string typeset at the beginning of the last line) and
   `width` (the width of the lines, default is 70). 
   - `no`. This model generates no header and has no argument. 

  It is possible to change the default builtin configuration file at
compile time. For this, just edit the file `config_builtin.txt` present
in the source distribution before building the software.
   --------------------------------------------------------------------

   <<
     # Objective Caml source
       ".*\\.ml[il]?" -> frame open:"(*" line:"*" close:"*)"
     | ".*\\.fml[i]?" -> frame open:"(*" line:"*" close:"*)"
     | ".*\\.mly"     -> frame open:"/*" line:"*" close:"*/"
     # C source
     | ".*\\.[chy]"    -> frame open:"/*" line:"*" close:"*/"
     # Latex
     | ".*\\.tex"     -> frame open:"%"  line:"%" close:"%"
     # Misc
     | ".*Makefile.*" -> frame open:"#"  line:"#" close:"#"
     | ".*README.*"   -> frame open:"*"  line:"*" close:"*"
     | ".*LICENSE.*"  -> frame open:"*"  line:"*" close:"*"
   >>

     Figure 2: The default builtin configuration file

   --------------------------------------------------------------------

  It is also possible to add entries into your own configuration file
that specify when the initial lines of the processed file have to be
skipped. As previously, these entries are separated by the character `|`
and each of them is made of two parts separated by an `->`: 

   - Again, the first part is a regular expression used by `headache` to
   determine the file type. But here, it is according to its full
   filename (including the pathname). 
   - The second part specifies when the initial lines must be skipped.
   It consists of the keyword `skip` followed by one of the named
   arguments `multiline_match:` or `match:`, then a regular expression.
   As long as the lines match a `multiline_match` parameter, `headache`
   skips them and checks the next line. If the current line matches only
   a `match` parameter, `headache` skips the current line and breaks the
   iteration there (of course, if nothing matches, `headache` puts the
   header before the current line).

   --------------------------------------------------------------------

   <<# Script file
      | ".*\\.sh" -> frame open:"#"  line:"#" close:"#"
      | ".*\\.sh" -> skip match:"#!.*"
   >>

     Figure 3: Example of a configuration file for skipping the shebang
   line of shell scripts

   --------------------------------------------------------------------

  Figure 3 shows an example of configuration file that can used to skip
the shebang line of shell scripts: when the first line of `.sh` files
starts with `#!`, `headache` does not modify that line and considers
that the header must start at the second line.
   --------------------------------------------------------------------

   <<# SWI Prolog file
      | ".*\\.pl" -> frame open:"%"  line:"%" close:"%"
      | ".*\\.pl" -> skip multiline_match:"#!.*" multiline_match:":-.*"
   >>

     Figure 4: Example of a configuration file for skipping the shebang
   line, as well as lines containing Prolog directives, such as Unicode
   usage.

   --------------------------------------------------------------------

  Figure 4 shows an example of configuration file that can used for `SWI
prolog` files: for a `.pl` file starting with the following three first
lines, `headache` considers that the header must start just after the
first two lines: 
<<
    #!/usr/bin/env swipl
    :- encoding(utf8).
    % remainder of the file, that can be after the header
>>

-----------------------------------------------------------------------
   This document was translated from LaTeX by HeVeA (4).
-----------------------------------


 (1) manual.txt
 (2) manual.ps.gz
 (3) manual.pdf
 (4) http://hevea.inria.fr/index.html