/tools/relic/relic

#!/usr/bin/python

# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.

# Adapted from the 'lick' and 'ripl' Python scripts. (See:
# <http://bugzilla.mozilla.org/show_bug.cgi?id=98089>)

"""
    relic - RE-LICense a given file, set of files, or directory of files
            from the Mozilla source tree

    Usage:
        relic [options...] [files...]
        relic [options...] < files...

    Options to Select Mode (use one):
        <none>              List the licenses in each file.
        -s, --statistics    Should a summary table of licenses in each file.
                            The -x, --extended option may be added to show
                            some additional detail to the stats.
        -r, --relicense     Modify the given files to include to
                            appropriate Mozilla license, where
                            "appropriate" is either the NPL/GPL/LPGL
                            tri-license if was already under the NPL or
                            the MPL/LPGL/GPL license in all other cases.
        -R, --force-relicense
                            Relicenses files (as -r|--relicense), but
                            does NOT skip files that already appear to
                            have a complete license.
        -A, --add           Add a license to files that do not appear to
                            have one.
        -I, --initial-developers
                            Display initial developer for each file.

    General Options:
        -h, --help          dump this help and exit
        -V, --version       dump this script's version and exit
        -v, --verbose       verbose output
        -d, --debug         more verbose output

        -f, --force         Continue processing after an error. (Errors
                            are summarized at end.)
        -q, --quick         Quick scanning. Use only basic license checks
                            (only use in report mode).
        -M, --MPL           Replace NPL licenses with MPL ones.
        -a, --all           Check all files (only skip CVS directories).
        --dry-run           Go through motions but don't actually change
                            any files.
        --backup            Make backups of changes files with
                            relicensing. Backup filenames are the
                            original filename suffixed with a ~# where
                            "#" is the lowest number to avoid a file
                            conflict.
        -o <orig_code_is>   Provide fallback value for the "Original
                            Code is" block.
        -D <orig_code_date> Provide fallback value for the date
                            that is part of the "Original Code is" block.
        -i <initial_dev>    Provide fallback value for the "Initial
                            Developer of the Original Code is" block.
        -y <year>           Provide fallback value for "Initial
                            Developer" copyright year.
        --defaults          Use the following default fallback values:
                             original_code_is: "mozilla.org Code"
                             initial_copyright_date: "2001"
                             initial_developer: "Netscape Communications
                                                 Corporation"
                            Note: the "Original Code" date is generally
                            not required, so a default is not included
                            here.

    Examples:
        # List license in files under mozilla/js/src.
        relic mozilla/js/src        # list licenses in files
        relic -s mozilla/js/src     # show summary stats on licenses
        relic -r mozilla/js/src     # re-license files

"""

import os
import sys
import re
import getopt
import pprint
import shutil


class RelicError(Exception):
    pass


#---- setup logging

try:
    # This package will be std in Python 2.3, but many Python 2.2
    # installation will not have it.
    import logging
    logging.basicConfig()
except ImportError:
    # Local fallback logging module.
    try:
        import _logging as logging
    except ImportError:
        sys.stderr.write("Your Python installation does not have the logging "
                         "package, nor could the fallback _logging module be "
                         "found. One of the two is required to run this "
                         "script.\n\n")
        raise

log = logging.getLogger("relic")

#---- globals

_version_ = (0, 7, 2)

# When processing files, 'relic' skips files and directories according
# to these settings. Note: files identified in .cvsignore files are also
# skipped.
_g_skip_exts = [".mdp", ".order", ".dsp", ".dsw", ".uf"]
_g_skip_file_basenames = [
    # Used by CVS (and this script)
    ".cvsignore",
    
    # GPL with autoconf exception
    "config.guess",
    "config.sub",

    # Auto-generated from other files
    "configure",

    # license and readme files
    "license",
    "readme",
    "copyright",
    "LICENSE-MPL",
    "MPL-1.1.txt",
]
_g_skip_files = [
    # TODO: update with MPL block - or CVS remove (check history)
    "tools/wizards/templates/licenses/MPL/lic.mak",
    "tools/wizards/templates/licenses/MPL/lic.pl",

###########################################################################
# Everything in _g_skip_files below this line needs no further work.
###########################################################################
    
    # Files containing copies of licence text which confuses the script
    "LICENSE",
    "js2/COPYING",
    "security/svrcore/LICENSE",
    "extensions/xmlterm/doc/MPL",
    "gfx/cairo/cairo/COPYING-LGPL-2.1",
    "gfx/cairo/cairo/COPYING-MPL-1.1",
    
    # Files containing global licensing information
    "xpfe/global/resources/content/license.html",
    "toolkit/content/license.html",
    
    # Ben Bucksch - files are tri-licensed with an extra clause.
    "netwerk/streamconv/converters/mozTXTToHTMLConv.cpp",
    "netwerk/streamconv/converters/mozTXTToHTMLConv.h",
    "netwerk/streamconv/public/mozITXTToHTMLConv.idl",
    
    # GPLed build tools
    "config/preprocessor.pl",
    "intl/uconv/tools/parse-mozilla-encoding-table.pl",
    "intl/uconv/tools/gen-big5hkscs-2001-mozilla.pl",
    "js2/missing",
    
    # Files which the script doesn't handle well. All have been relicensed
    # manually.
    "xpinstall/wizard/windows/builder/readme.txt",
    "xpfe/bootstrap/icons/windows/readme.txt",
    "embedding/qa/testembed/README.TXT",
    "security/nss/lib/freebl/ecl/README.FP",
    "nsprpub/pkg/linux/sun-nspr.spec",
    "security/nss/pkg/linux/sun-nss.spec",
    "security/jss/pkg/linux/sun-jss.spec",
    "security/nss/lib/freebl/mpi/utils/README",
    "security/nss/lib/freebl/ecl/README",
    "security/nss/lib/freebl/mpi/README",
    "lib/mac/UserInterface/Tables/TableClasses.doc",
    "parser/htmlparser/tests/html/bug23680.html",
    "security/nss/lib/freebl/mpi/montmulfv9.s",
    "tools/performance/pageload/base/lxr.mozilla.org/index.html",
    "testing/performance/win32/page_load_test/" +\
                                            "base/lxr.mozilla.org/index.html",
    "testing/performance/win32/page_load_test/" +\
                                    "base/lxr.mozilla.org/20001028.html.orig",
    
    # Not sure what to do with this...
    "gfx/cairo/stdint.diff",
    
    # GPL with autoconf exception (same license as files distributed with)
    "build/autoconf/codeset.m4",
    "toolkit/airbag/airbag/autotools/depcomp",
    "toolkit/airbag/airbag/autotools/missing",
    "toolkit/airbag/airbag/autotools/ltmain.sh",
    "js/tamarin/pcre/ltmain.sh",
    "security/svrcore/compile",
    "security/svrcore/ltmain.sh",
    "security/svrcore/missing",
    "security/svrcore/depcomp",
    "security/svrcore/aclocal.m4",
    
    # Public domain or equivalent
    "nsprpub/config/nspr.m4",
    "toolkit/airbag/airbag/aclocal.m4",
    "security/nss/lib/freebl/mpi/mp_comba_amd64_sun.s",
    
    # GSSAPI has BSD-like licence requiring some attribution
    "extensions/auth/gssapi.h",

    # This script
    "tools/relic/relic",
]
_g_skip_dir_basenames = [
    "CVS",
]

_g_skip_dir_basenames_cvs_only = [
    "CVS",
]

# Complete path from mozilla dir to a dir to skip.
_g_skip_dirs = [
     # Test files for this script, which cause it to crash!
    "tools/relic/test",
    
    # License template files (TODO: this directory may disappear)
    "tools/wizards/templates/licenses",
        
    # As per the "New Original Source Files" section of:
    #   http://www.mozilla.org/MPL/license-policy.html
    #   with obsolete or now-relicensed directories removed
    "apache",             # Obsolete mod_gzip code
    "cck",                # mkaply's baby; not core code anyway.
    "dbm",
    "js/rhino",           # Currently MPL/GPL - may end up BSD
    "webtools",           # Various MPLed webtools
    
    # These could be done, but no-one's clamouring for it, and it's a hassle
    # sorting it all out, so let sleeping dogs lie.
    "msgsdk",
    "java",
    "privacy",

    # These have their own BSD-like license
    "media/libjpeg",
    
    # The following are not supposed to be relicensed, but they do have a
    # few files in we care about (like makefiles)
    "media/libpng",
    "modules/zlib",
    "gc/boehm",
    "other-licenses",
    
    # Copy of GPLed tool
    "tools/buildbot",
    
    # Other directories we want to exclude
    "embedding/tests",     # Agreed as BSD
    "calendar/libical",    # LGPL/MPL
    "gfx/cairo/cairo/src", # LGPL/MPL
]

_g_basename_to_comment_info = {
    "configure": (["dnl"], ),

    "Makefile": (["#"], ),
    "makefile": (["#"], ),
    "nfspwd": (["#"], ),
    "typemap": (["#"], ),
    "xmplflt.conf": (["#"], ),
    "ldapfriendly": (["#"], ),
    "ldaptemplates.conf": (["#"], ),
    "ldapsearchprefs.conf": (["#"], ),
    "ldapfilter.conf": (["#"], ),
    "README.configure": (["#"], ),
    "Options.txt": (["#"], ),
    "fdsetsize.txt": (["#"], ),
    "prototype": (["#"], ),
    "prototype_i386": (["#"], ),
    "prototype3_i386": (["#"], ),
    "prototype_com": (["#"], ),
    "prototype3_com": (["#"], ),
    "prototype_sparc": (["#"], ),
    "prototype3_sparc": (["#"], ),
    "nglayout.mac": (["#"], ),
    "pkgdepend": (["#"], ),
    "Maketests": (["#"], ),
    "depend": (["#"], ),
    "csh-aliases": (["#"], ),
    "csh-env": (["#"], ),
    ".cshrc": (["#"], ),
    "MANIFEST": (["#"], ),
    "mozconfig": (["#"], ),
    "makecommon": (["#"], ),
    "bld_awk_pkginfo": (["#"], ),
    "prototype_i86pc": (["#"], ),
    "pkgdepend_5_6": (["#"], ),
    "awk_pkginfo-i386": (["#"], ),
    "awk_pkginfo-sparc": (["#"], ),
    "pkgdepend_64bit": (["#"], ),
    "WIN32": (["#"], ),
    "WIN16": (["#"], ),
    "Makefile.linux": (["#"], ),

    "README": ([""], ["#"]),
    "copyright": ([""], ),

    "xptcstubs_asm_ppc_darwin.s.m4": (["/*", "*", "*/"], ),
    "xptcstubs_asm_mips.s.m4": (["/*", "*", "*/"], ),

    "nsIDocCharsetTest.txt": (["<!--", "-", "-->"], ),
    "nsIFontListTest.txt": (["<!--", "-", "-->"], ),
    "ComponentListTest.txt": (["<!--", "-", "-->"], ),
    "nsIWebBrowserPersistTest1.txt": (["<!--", "-", "-->"], ),
    "nsIWebBrowserPersistTest2.txt": (["<!--", "-", "-->"], ),
    "nsIWebBrowserPersistTest3.txt": (["<!--", "-", "-->"], ),
    "plugins.txt": (["<!--", "-", "-->"], ),
    "NsISHistoryTestCase1.txt": (["<!--", "-", "-->"], ),
    "EmbedSmokeTest.txt": (["<!--", "-", "-->"], ),

    "lineterm_LICENSE": (["/*", "*", "*/"], ),
    "XMLterm_LICENSE": (["/*", "*", "*/"], ),
    "BrowserView.cpp.mod": (["/*", "*", "*/"], ),
    "header_template": (["/*", "*", "*/"], ),
    "cpp_template": (["/*", "*", "*/"], ),

    "abcFormat470.txt": (["//"], ),
    "opcodes.tbl": (["//"], ),
}

_g_ext_to_comment_info = {
    ".txt":   (["##", "#", ], ["#"]),
    ".TXT":   (["##", "#", ]),

    ".doc":   (["", ]),
    ".build": (["", ]),
    ".1st":   (["", ]),
    ".lsm":   (["", ]),
    ".FP":    (["", ]),
    ".spec":  (["", ]),

    ".CPP":    (["/*", "*", "*/"], ),
    ".cpp":    (["/*", "*", "*/"], ),
    ".H":      (["/*", "*", "*/"], ),
    ".h":      (["/*", "*", "*/"], ),
    ".hxx":    (["/*", "*", "*/"], ),
    ".c":      (["/*", "*", "*/"], ),
    ".css":    (["/*", "*", "*/"], ['#']),
    ".js":     (["/*", "*", "*/"], ['#']),
    ".idl":    (["/*", "*", "*/"], ),
    ".ut":     (["/*", "*", "*/"], ),
    ".rc":     (["/*", "*", "*/"], ),
    ".rc2":    (["/*", "*", "*/"], ),
    ".RC":     (["/*", "*", "*/"], ),
    ".Prefix": (["/*", "*", "*/"], ),
    ".prefix": (["/*", "*", "*/"], ),
    ".cfg":    (["/*", "*", "*/"], ["#"]),
    ".cp":     (["/*", "*", "*/"], ),
    ".cs":     (["/*", "*", "*/"], ),
    ".java":   (["/*", "*", "*/"], ),
    ".jst":    (["/*", "*", "*/"], ),
    ".tbl":    (["/*", "*", "*/"], ),
    ".tab":    (["/*", "*", "*/"], ),
    ".cc":     (["/*", "*", "*/"], ),
    ".msg":    (["/*", "*", "*/"], ),
    ".y":      (["/*", "*", "*/"], ),
    ".r":      (["/*", "*", "*/"], ),
    ".mm":     (["/*", "*", "*/"], ),
    ".x-ccmap":(["/*", "*", "*/"], ),
    ".ccmap":  (["/*", "*", "*/"], ),
    ".sql":    (["/*", "*", "*/"], ),
    ".pch++":  (["/*", "*", "*/"], ),
    ".xpm":    (["/*", "*", "*/"], ),
    ".uih":    (["/*", "*", "*/"], ),
    ".uil":    (["/*", "*", "*/"], ),
    ".ccmap":  (["/*", "*", "*/"], ),
    ".map":    (["/*", "*", "*/"], ),
    ".win98":  (["/*", "*", "*/"], ),
    ".php":    (["/*", "*", "*/"], ),
    ".m":      (["/*", "*", "*/"], ),
    ".jnot":   (["/*", "*", "*/"], ),
    ".l":      (["/*", "*", "*/"], ),
    ".htp":    (["/*", "*", "*/"], ),
    ".xs":     (["/*", "*", "*/"], ),
    ".as":     (["/*", "*", "*/"], ),
    ".api":    (["/*", "*", "*/"], ['#']),
    ".applescript": (["(*", "*", "*)"], ["--"], ["#"]),

    ".html": (["<!--", "-", "-->"], ["#"]),
    ".xml":  (["<!--", "-", "-->"], ["#"]),
    ".xbl":  (["<!--", "-", "-->"], ["#"]),
    ".xsl":  (["<!--", "-", "-->"], ),
    ".xul":  (["<!--", "-", "-->"], ["#"]),
    ".dtd":  (["<!--", "-", "-->"], ["#"]),
    ".rdf":  (["<!--", "-", "-->"], ["#"]),
    ".htm":  (["<!--", "-", "-->"], ),
    ".out":  (["<!--", "-", "-->"], ),
    ".resx": (["<!--", "-", "-->"], ),
    ".bl":   (["<!--", "-", "-->"], ),
    ".xif":  (["<!--", "-", "-->"], ),
    ".xhtml":(["<!--", "-", "-->"], ["#"]),

    ".inc":  (["<!--", "-", "-->"], 
              ["#"],
              ["@!"],
              ["/*", "*", "*/"]),

    ".properties": (["#"], ),
    ".win":        (["#"], ),
    ".dsp":        (["#"], ),
    ".exp":        (["#"], ),
    ".mk":         (["#"], ),
    ".mn":         (["#"], ),
    ".mak":        (["#"], ),
    ".MAK":        (["#"], ),
    ".perl":       (["#"], ),
    ".pl":         (["#"], ),
    ".PL":         (["#"], ),
    ".sh":         (["#"], ),
    ".dsw":        (["#"], ),
    ".cgi":        (["#"], ),
    ".pm":         (["#"], ),
    ".pod":        (["#"], ),
    ".src":        (["#"], ),
    ".csh":        (["#"], ),
    ".DLLs":       (["#"], ),
    ".ksh":        (["#"], ),
    ".toc":        (["#"], ),
    ".am":         (["#"], ),
    ".df":         (["#"], ),
    ".client":     (["#"], ),
    ".ref":        (["#"], ), # all of them "Makefile.ref"
    ".ldif":       (["#"], ),
    ".ex":         (["#"], ),
    ".reg":        (["#"], ),
    ".py":         (["#"], ),
    ".adb":        (["#"], ),
    ".dtksh":      (["#"], ),
    ".pkg":        (["#"], ),
    ".et":         (["#"], ),
    ".stub":       (["#"], ),
    ".nss":        (["#"], ),
    ".os2":        (["#"], ),
    ".Solaris":    (["#"], ),
    ".rep":        (["#"], ),
    ".NSS":        (["#"], ),
    ".server":     (["#"], ),
    ".awk":        (["#"], ),
    ".targ":       (["#"], ),
    ".gnuplot":    (["#"], ),
    ".bash":       (["#"], ),
    ".tmpl":       (["#"], ),
    ".com":        (["#"], ),
    ".dat":        (["#"], ),
    ".rpm":        (["#"], ),
    ".nsi":        (["#"], ),
    ".nsh":        (["#"], ),
    ".template":   (["#"], ),
    ".ldkd":       (["#"], ),
    ".ldku":       (["#"], ),
    ".arm":        (["#"], ),

    ".tdf":  ([";"], ),
    ".def":  ([";+#"], [";"]),
    ".DEF":  ([";+#"], [";"]),
    ".ini":  ([";"], ),
    ".it":   ([";"], ),
    ".lisp": ([";;;"], ),

    ".cmd": (["rem"], ["REM"]),
    ".bat": (["rem"], ["REM"]),

    ".tex":  (["%"], ),
    ".texi": (["%"], ),

    ".m4":  (["dnl"], ),

    ".asm": ([";"], ),
    ".vbs": (["'"], ),
    ".il":  (["!"], ),
    ".ad":  (["!"], ),

    ".script": (["(*", " *", "*)"], ),

    ".3x":  (['.\\"'], ),
	    
    # What a mess...
    ".s": (["#"], ["//"], ["/*", "*", "*/"], ["!"], [";"], ["/"]),
}
_g_shebang_pattern_to_comment_info = [
    (re.compile(ur'\A#!.*/bin/(ba)?sh.*$', re.IGNORECASE), (["#"], )),
    (re.compile(ur'\A#!.*perl.*$', re.IGNORECASE), (["#"], )),
    (re.compile(ur'\A#!.*php.*$', re.IGNORECASE), (["#"], )),
    (re.compile(ur'\A#!.*python.*$', re.IGNORECASE), (["#"], )),
    (re.compile(ur'\A#!.*ruby.*$', re.IGNORECASE), (["#"], )),
    (re.compile(ur'\A#!.*tclsh.*$', re.IGNORECASE), (["#"], )),
    (re.compile(ur'\A#!.*wish.*$', re.IGNORECASE), (["#"], )),
    (re.compile(ur'\A#!.*expect.*$', re.IGNORECASE), (["#"], )),
]


_g_trilicense_parts = {
    "mpl": """\
***** BEGIN LICENSE BLOCK *****
Version: MPL 1.1/GPL 2.0/LGPL 2.1

The contents of this file are subject to the Mozilla Public License Version
1.1 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.mozilla.org/MPL/

Software distributed under the License is distributed on an "AS IS" basis,
WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
for the specific language governing rights and limitations under the
License.

""",
    "npl": """\
***** BEGIN LICENSE BLOCK *****
Version: NPL 1.1/GPL 2.0/LGPL 2.1

The contents of this file are subject to the Netscape Public License
Version 1.1 (the "License"); you may not use this file except in
compliance with the License. You may obtain a copy of the License at
http://www.mozilla.org/NPL/

Software distributed under the License is distributed on an "AS IS" basis,
WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
for the specific language governing rights and limitations under the
License.

""",
    "original_code_is": """\
The Original Code is %(original_code_is)s.

""",
    "original_code_is_with_date": """\
The Original Code is %(original_code_is)s, released
%(original_code_date)s.

""",
    "initial_developer": """\
The Initial Developer of the Original Code is
%(initial_developer)s.
Portions created by the Initial Developer are Copyright (C) %(initial_copyright_date)s
the Initial Developer. All Rights Reserved.

""",
    "contributors": """\
Contributor(s):
%s
""",
    "gpl for mpl": """\
Alternatively, the contents of this file may be used under the terms of
the GNU General Public License Version 2 or later (the "GPL"), in which
case the provisions of the GPL are applicable instead of those above. If
you wish to allow use of your version of this file only under the terms of
the GPL and not to allow others to use your version of this file under the
MPL, indicate your decision by deleting the provisions above and replacing
them with the notice and other provisions required by the GPL. If you do
not delete the provisions above, a recipient may use your version of this
file under either the MPL or the GPL.

***** END LICENSE BLOCK *****""",
    "gpl for npl": """\
Alternatively, the contents of this file may be used under the terms of
the GNU General Public License Version 2 or later (the "GPL"), in which
case the provisions of the GPL are applicable instead of those above. If
you wish to allow use of your version of this file only under the terms of
the GPL and not to allow others to use your version of this file under the
NPL, indicate your decision by deleting the provisions above and replacing
them with the notice and other provisions required by the GPL. If you do
not delete the provisions above, a recipient may use your version of this
file under either the NPL or the GPL.

***** END LICENSE BLOCK *****""",
    "gpl/lgpl for mpl": """\
Alternatively, the contents of this file may be used under the terms of
either the GNU General Public License Version 2 or later (the "GPL"), or
the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
in which case the provisions of the GPL or the LGPL are applicable instead
of those above. If you wish to allow use of your version of this file only
under the terms of either the GPL or the LGPL, and not to allow others to
use your version of this file under the terms of the MPL, indicate your
decision by deleting the provisions above and replace them with the notice
and other provisions required by the GPL or the LGPL. If you do not delete
the provisions above, a recipient may use your version of this file under
the terms of any one of the MPL, the GPL or the LGPL.

***** END LICENSE BLOCK *****""",
    "gpl/lgpl for npl": """\
Alternatively, the contents of this file may be used under the terms of
either the GNU General Public License Version 2 or later (the "GPL"), or
the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
in which case the provisions of the GPL or the LGPL are applicable instead
of those above. If you wish to allow use of your version of this file only
under the terms of either the GPL or the LGPL, and not to allow others to
use your version of this file under the terms of the NPL, indicate your
decision by deleting the provisions above and replace them with the notice
and other provisions required by the GPL or the LGPL. If you do not delete
the provisions above, a recipient may use your version of this file under
the terms of any one of the NPL, the GPL or the LGPL.

***** END LICENSE BLOCK *****""",
}

_g_dry_run = 0 # iff true, don't modify any files
_g_force = 0
_g_check_all = 0


#---- internal support routines

def _is_binary(filename):
    """Return true iff the given filename is binary.

    Raises an EnvironmentError if the file does not exist or cannot be
    accessed.
    """
    fin = open(filename, 'rb')
    try:
        CHUNKSIZE = 1024
        while 1:
            chunk = fin.read(CHUNKSIZE)
            if '\0' in chunk: # found null byte
                return 1
            if len(chunk) < CHUNKSIZE:
                break # done
    finally:
        fin.close()

    return 0


_g_cvsignore_cache = {} # optimization: keep a cache of .cvsignore content
def _should_skip_according_to_cvsignore(path):
    dirname, basename = os.path.split(path)
    cvsignore = os.path.join(dirname, ".cvsignore")
    if not os.path.exists(cvsignore):
        return 0
    elif cvsignore not in _g_cvsignore_cache:
        fin = open(cvsignore, 'r')
        to_ignore = []
        try:
            for f in fin:
                if f[-1] == "\n": f = f[:-1] # chomp
                if not f: continue # skip empty lines
                to_ignore.append(f)
        finally:
            fin.close()
        _g_cvsignore_cache[cvsignore] = to_ignore

    # At this point .cvsignore exists and its contents are in the cache.
    to_ignore = _g_cvsignore_cache[cvsignore]
    if basename in to_ignore:
        return 1
    else:
        return 0

_g_backup_pattern = re.compile("~\d+$")
def _should_skip_file(path):
    log.debug("_should_skip_file(path='%s')", path)
    if _g_check_all:
        return 0
    ext = os.path.splitext(path)[1]
    if ext in _g_skip_exts:
        log.info("Skipping '%s' (according to '_g_skip_exts').", path)
        return 1
    xpath = '/'.join(path.split(os.sep)) # use same sep as in _g_skip_files
    for sf in _g_skip_files:
        if xpath.endswith(sf):
            log.info("Skipping '%s' (according to '_g_skip_files').", path)
            return 1
    if os.path.basename(path) in _g_skip_file_basenames:
        log.info("Skipping '%s' (according to '_g_skip_file_basenames').", path)
        return 1
    if _should_skip_according_to_cvsignore(path):
        log.info("Skipping '%s' (according to .cvsignore).", path)
        return 1
    if _g_backup_pattern.search(path):
        log.info("Skipping '%s' (looks like backup file).", path)
        return 1
    return 0

def _should_skip_dir(path):
    log.debug("_should_skip_dir(path='%s')", path)
    if _g_check_all:
        if os.path.basename(path) in _g_skip_dir_basenames_cvs_only:
            return 1
        return 0
    if os.path.basename(path) in _g_skip_dir_basenames:
        log.info("Skipping '%s' (according to _g_skip_dir_basenames).", path)
        return 1
    xpath = '/'.join(path.split(os.sep)) # use same sep as in _g_skip_dirs
    # These could do with being a proper path canonicalisation function...
    if xpath[-1] == '/': xpath = xpath[:-1]  # treat "calendar/" the same as "calendar"
    if xpath[0:2] == './': xpath = xpath[2:]  # treat "./calendar" the same as "calendar"
    for sd in _g_skip_dirs:
        # Changed by gerv to make skip_dirs require whole path
        if xpath == sd:
            log.info("Skipping '%s' (according to _g_skip_dirs).", path)
            return 1
    if _should_skip_according_to_cvsignore(path):
        log.info("Skipping '%s' (according to .cvsignore).", path)
        return 1
    return 0


def _get_license_info(filename, show_initial=0, quick=0):
    """Return license block information for the given file.

        "filename" is the path to the file to scan.
        "show_initial" is a boolean that indicates if initial developer info
          should be displayed.
        "quick" is a boolean that can be set for a quick scan. In this
            case, only the "parts" field of the return dictionary will
            be filled out.
    
    Returns a dictionary adequately describing the license block in the
    given file for the purpose of determining whether to patch the
    license block and how. Returns a dictionary of the following form:
        {"parts": <list of zero or more of "mpl", "npl", "gpl", "lgpl",
                   "unknown", "block_begin", "block_end" in the
                   order in which they were found>,
         # if necessary, the following keys are included as well
         "begin_line": <(0-based) index at which license block starts>,
         "end_line": <(0-based) index at which license block ends>,
         "first_prefix": <prefix to use for new license block first line>,
         "subsequent_prefix": <prefix to use for subsequent lines>,
         "last_suffix": <suffix to use for last line>,
         # The following fields are correspond to the file specific
         # portions of the license template as described here:
         #   http://www.mozilla.org/MPL/relicensing-faq.html#new-license
         # If the associated block is not found, then the value is None.
         "original_code_is": ...,
         "original_code_date": ...,
         "initial_developer": ...,
         "initial_copyright_date": ...,
         "contributors": ...,
        }

    precondition: should not be called on binary files
    """
    lic_info = {
        "parts": [],
    }

    fin = open(filename, 'r')
    try:
        content = fin.read()
    finally:
        fin.close()
    
    # Help me find filena
    log.info("Next file is: %s", filename)

    # do quick search to see if any of the desired licenses is in here
    # - if it looks like all the parts are there, good, done
    # - if some but not all parts, continue
    parts_pattern = re.compile("""(
        (?P<block_begin>\*\*\*\*\*\ BEGIN\ LICENSE\ BLOCK\ \*\*\*\*\*)
        | (?P<mpl>The\ contents\ of\ this\ file\ are\ subject\ to\ the\ Mozilla)
        | (?P<npl>The\ contents\ of\ this\ file\ are\ subject\ to\ the\ Netscape)
        | (?P<gpl>GNU\ (General\ )?Public\ License)
        | (?P<lgpl>(Library|Lesser)\ General\ Public\ License)
        | (?P<block_end>\*\*\*\*\*\ END\ LICENSE\ BLOCK\ \*\*\*\*\*)
        )""",
        re.VERBOSE)
    parts = [] # found license parts in this file
    start = 0
    blocks = 0
    while 1:
        match = parts_pattern.search(content, start)
        if match:
            # Skip this block, if the last license block is more than 10 lines
            # away (file is probably used for autogeneration of files then).
            if blocks == 1 and (match.start()-start) > 10:
                break
            else:
                parts = match.groupdict()
                for part in parts:
                    if parts[part]:
                        lic_info["parts"].append(part)
                        log.info("%s license/delimeter found", part)
                        start = match.end()
                        if part == "block_end":
                            blocks = blocks + 1
                        else:
                            blocks = 0
                        break
                else:
                    raise RelicError("unexpected license part: %r" % parts)
        else:
            break

    # no license block at all
    if not parts:
        # - if not, check to see if License or Copyright shows up in the
        #   file; if so, then error out; if not, skip out
        any_lic_pattern = re.compile("(Copyright|Licen[sc]e)", re.IGNORECASE)
        match = any_lic_pattern.search(content)
        if match:
            lic_info["parts"].append("unknown")
            log.info("unknown license found: %r",
                     content[max(match.start()-20,0):match.end()+20])
        else:
            log.info("no license found")
        return lic_info

    # license block with non-tri-license version headers
    elif lic_info["parts"] == ["block_begin", "block_end"]:
        lic_info["parts"].append("unknown")
        log.info("unknown license found (license block with non-tri-license)")
        return lic_info

    # license block with tri-license version headers   
    elif (lic_info["parts"] == ["block_begin", "mpl", "gpl", "lgpl", "block_end"] or
          lic_info["parts"] == ["block_begin", "npl", "gpl", "lgpl", "block_end"]):
        log.info("license looks good, no changes necessary")
    if quick:
        return lic_info

    # Otherwise, the license needs to be fixed, so gather more detailed
    # information. Here is the algorithm we will use:
    # - find first license line
    # - find the end of this comment block (assumption: from the first
    #   license line to the end of the comment block is the full
    #   license block)
    #   This is a bad assumption in two cases and steps have been taken
    #   to try to deal with those cases:
    #   - There could be a trailing part bit of comment that is
    #     NOT part of the license but is part of the same comment
    #     block. A common example are the:
    #       This Original Code has been modified by IBM...
    #     files (about 130 of them in the moz tree).
    #     (c.f. test_relicense_ibm_copyright_suffix.c)
    #   - Some files have split up the license paragraphs into
    #     multiple comment blocks, e.g.
    #     "mozilla/build/unix/abs2rel.pl":
    #        # The contents of this file are subject to the
    #        # ...
    #        # the License at http://www.mozilla.org/MPL/
    #
    #        # The Initial Developer of the Original Code
    #        # ...
    #        # Rights Reserved.
    #     (c.f. test_relicense_separated_license_comment_blocks.pl)
    # - these are the lines to replace
    # - gather embedded lic data
    # - use second line to determine line prefix
    # ? Should we only allow processing of unknown-delimiter-files with
    #   an option?

    # Get comment delimiter info for this file.
    comment_delim_sets = _get_comment_delim_sets(filename)

    # - find first license line (and determine which set of comment
    #   delimiters are in use)
    lines = content.splitlines()
    for comment_delims in comment_delim_sets:
        if len(comment_delims) == 3:
            # Note: allow for whitespace before continuation character
            prefix_pattern = "%s|\s*%s|" % (re.escape(comment_delims[0]),
                                            re.escape(comment_delims[1]))
            suffix_pattern = "%s" % re.escape(comment_delims[2])
        elif len(comment_delims) == 2:
            prefix_pattern = "%s|" % re.escape(comment_delims[0])
            suffix_pattern = "%s" % re.escape(comment_delims[1])
        elif len(comment_delims) == 1:
            prefix_pattern = re.escape(comment_delims[-1])
            suffix_pattern = ""
        else: # len(comment_delims) == 0
            prefix_pattern = ""
            suffix_pattern = ""

        lic_begin_pattern = re.compile("""
            ^(?P<prefix>%s)
            (?P<space>\s*)
            (\*+\ BEGIN\ LICENSE\ BLOCK\ \*+
             |\-+\ BEGIN\ LICENSE\ BLOCK\ \-+
             | Version:\ MPL\ \d+\.\d+/GPL\ \d+\.\d+/LGPL\ \d+\.\d+
             | The\ contents\ of\ this\ file\ are\ subject\ to\ the\ Mozilla[\w ]*
             | The\ contents\ of\ this\ file\ are\ subject\ to\ the\ Netscape[\w ]*
             | Alternatively,\ the\ contents\ of\ this\ file\ may\ be\ used\ under\ the[\w ]*)
            (?P<suffix>%s|)\s*?$
            """ % (prefix_pattern, suffix_pattern), re.VERBOSE)

        for i in range(len(lines)):
            match = lic_begin_pattern.search(lines[i])
            if match:
                beginline = {
                    "content": lines[i],
                    "linenum": i,
                    "prefix": match.group("prefix"),
                    "space": match.group("space"),
                    "suffix": match.group("suffix")
                }
                # Optimization: If the line before the "beginline" is simply
                # a block comment open the include that line in parsed out
                # license block. E.g.,
                #    <!--
                #       - ***** BEGIN LICENSE BLOCK *****
                #       ...
                if (len(comment_delims) > 1  # only for block comments
                    and beginline["prefix"] != comment_delims[0]
                    and i-1 >= 0
                    and lines[i-1].strip() == comment_delims[0]):
                    beginline["linenum"] -= 1
                    beginline["prefix"] = comment_delims[0]
                break
        if match: break
    else:
        raise RelicError("couldn't find start line with this pattern (even "
                         "though it looks like there is a license block in "
                         "%s): %s" % (filename, lic_begin_pattern.pattern))
    log.info("comment delimiters: %s", comment_delims)
    log.debug("beginline dict: %s", beginline)
    lic_info["comment_delims"] = comment_delims
    lic_info["begin_line"] = beginline["linenum"]
    lic_info["first_prefix"] = beginline["prefix"]
    log.info("prefix for first line: '%s'", beginline["prefix"])

    # - get second license line
    lic_middle_pattern = re.compile("""
        ^(?P<prefix>%s|)
        (?P<space>\s*)
        (?P<content>.*)
        (?P<suffix>%s|)\s*?$
        """ % (prefix_pattern, suffix_pattern),
        re.VERBOSE)
    # skip empty lines which might result in bogus scanning later, e.g.:
    #   mozilla/layout/html/tests/table/marvin/x_thead_align_center.xml
    second_linenum = beginline["linenum"]+1
    while second_linenum < len(lines):
        if lines[second_linenum].strip():
            break
        log.debug("skip blank 'second' line: %d", second_linenum)
        second_linenum +=1
    else:
        raise RelicError("all lines after the first license block line (%d) "
                         "were empty" % (beginline["linenum"]+1))
    match = lic_middle_pattern.search(lines[second_linenum])
    if match:
        secondline = {
            "content": lines[second_linenum],
            "linenum": second_linenum,
            "prefix": match.group("prefix"),
            "space": match.group("space"),
            "suffix": match.group("suffix")
        }
    else:
        raise RelicError("didn't find second line with pattern: %s"
                         % lic_middle_pattern.pattern)
    log.debug("secondline dict: %s", secondline)
    lic_info["subsequent_prefix"] = secondline["prefix"]
    log.info("prefix for subsequent lines: '%s'", secondline["prefix"])

    # - find block comment end
    orig_code_modified_pattern = re.compile("This Original Code has been "
                                            "modified", re.I)
    non_lic_content_in_same_comment_block = 0
    if len(comment_delims) == 1:
        # line-style comments: The comment block "end" is defined as the
        # last line before a line NOT using the block comment delimiter.
        #XXX:BUG: This is not good enough for:
        #           test/inputs/separated_license_comment_blocks.pl
        if comment_delims[0] == "":
            raise RelicError(
                "Don't know how to find the end of a line-style comment "
                "block when the delimiter is the empty string. (Basically "
                "this script cannot handle this type of file.)")
        for i in range(beginline["linenum"]+1, len(lines)):
            if not lines[i].startswith(comment_delims[0]):
                endlinenum = i-1
                break
            elif lines[i].find("END LICENSE BLOCK") != -1:
                endlinenum = i
                break
            # As per "test_relicense_trailing_orig_code_modified.pl", a
            # paragraph starting with:
            #     This Original Code has been modified
            # is deemed to be OUTside the license block, i.e. it is not
            # replaced for relicensing.
            if orig_code_modified_pattern.search(lines[i]):
                non_lic_content_in_same_comment_block = 1
                # The endline is the first non-blank line before this one.
                endlinenum = i-1
                while 1:
                    line = lines[endlinenum]
                    match = lic_middle_pattern.search(line)
                    if not match:
                        raise RelicError("Line did not match lic_middle_pattern "
                                         "unexpectedly: %r" % line)
                    if match.group("content").strip(): # non-empty line
                        break
                    endlinenum -= 1
                break
        else:
            raise RelicError("Could not find license comment block end "
                             "line in '%s'." % filename)
    elif len(comment_delims) >= 2:   # block-style comments
        for i in range(beginline["linenum"]+1, len(lines)):
            if lines[i].find(comment_delims[-1]) != -1:
                endlinenum = i
                break
            elif lines[i].find("END LICENSE BLOCK") != -1:
                endlinenum = i
                non_lic_content_in_same_comment_block = 1
                break
            # As per "test_relicense_ibm_copyright_suffix.c", a
            # paragraph starting with:
            #     This Original Code has been modified
            # is deemed to be OUTside the license block, i.e. it is not
            # replaced for relicensing.
            if orig_code_modified_pattern.search(lines[i]):
                non_lic_content_in_same_comment_block = 1
                # The endline is the first non-blank line before this one.
                endlinenum = i-1
                while 1:
                    line = lines[endlinenum]
                    match = lic_middle_pattern.search(line)
                    if not match:
                        raise RelicError("Line did not match lic_middle_pattern "
                                         "unexpectedly: %r" % line)
                    if match.group("content").strip(): # non-empty line
                        break
                    endlinenum -= 1
                break
        else:
            raise RelicError("Could not find license comment block end "
                             "line in '%s'." % filename)
        if not non_lic_content_in_same_comment_block\
           and not lines[endlinenum].strip().endswith(comment_delims[-1]):
            raise RelicError(
                "There is text AFTER the license block comment end "
                "delimiter, but on the SAME LINE. This is unexpected. "
                "Bailing.\n%s:%s:%r"
                % (filename, endlinenum, lines[endlinenum]))
    else: # len(comment_delims) == 0
        # For files without a comment character to help out, we ONLY
        # successfully break one the full correct "END LICENSE BLOCK"
        # token.
        for i in range(beginline["linenum"]+1, len(lines)):
            if lines[i].find("END LICENSE BLOCK") != -1:
                endlinenum = i
                break
            elif i > beginline["linenum"]+1+50:
                raise RelicError("Haven't found 'END LICENSE BLOCK' marker "
                                 "within 50 lines of the start of the "
                                 "license block on line %d. Aborting." 
                                 % (beginline["linenum"]+1))
            # As per "test_relicense_trailing_orig_code_modified.pl", a
            # paragraph starting with:
            #     This Original Code has been modified
            # is deemed to be OUTside the license block, i.e. it is not
            # replaced for relicensing.
            if orig_code_modified_pattern.search(lines[i]):
                non_lic_content_in_same_comment_block = 1
                # The endline is the first non-blank line before this one.
                endlinenum = i-1
                while 1:
                    line = lines[endlinenum]
                    match = lic_middle_pattern.search(line)
                    if not match:
                        raise RelicError("Line did not match lic_middle_pattern "
                                         "unexpectedly: %r" % line)
                    if match.group("content").strip(): # non-empty line
                        break
                    endlinenum -= 1
                break
        else:
            raise RelicError("Could not find license comment block end "
                             "line in '%s'." % filename)

    # Test case: test_relicense_separated_license_comment_blocks.pl
    # It is possible that a separate comment block immediately following
    # the license block we just parsed should be included in the license
    # block.
    if (not non_lic_content_in_same_comment_block
        and len(comment_delims) == 1): # only do this for line-style comments
        lic_indicators = [
            re.compile("^The content of this file are subject to", re.I),
            re.compile("^Software distributed under the License", re.I),
            re.compile("^The Original Code is", re.I),
            re.compile("^The Initial Developer", re.I),
            re.compile("^Contributor", re.I),
            re.compile("^Alternatively, the content of this file", re.I),
        ]
        comment_line_pattern = re.compile("""
            ^(?P<prefix>%s|)
            (?P<space>\s*)
            (?P<content>.*)
            (?P<suffix>%s|)\s*?$
            """ % (prefix_pattern, suffix_pattern),
            re.VERBOSE)
        i = endlinenum
        while i+1 < len(lines):
            i += 1; line = lines[i]
            comment_index = line.find(comment_delims[0])
            if comment_index != -1:
                content = line[:comment_index].strip()
                comment = line[comment_index+len(comment_delims[0]):].strip()
            else:
                content = line.strip()
                comment = ""
            if content: # if non-comment content, then skip out
                break
            if not comment:
                continue
            for indicator in lic_indicators:
                if indicator.search(comment):
                    # include this paragraph in the lic block
                    while i < len(lines):
                        i += 1; line = lines[i]
                        if not line.strip().startswith(comment_delims[0]):
                            break
                        if not line.strip()[len(comment_delims[0]):]:
                            break
                    endlinenum = i-1
                    break
            else:
                break # this is a non-lic-related comment
    
    # Get the end-line data.
    if non_lic_content_in_same_comment_block:
        lic_end_pattern = re.compile(
            "^(?P<prefix>%s)(?P<space>\s*).*?\s*?$"
            % prefix_pattern)
    else:
        lic_end_pattern = re.compile(
            "^(?P<prefix>%s)(?P<space>\s*).*?(?P<suffix>%s)\s*?$"
            % (prefix_pattern, suffix_pattern))
    match = lic_end_pattern.match(lines[endlinenum])
    if match:
        endline = {
            "content": lines[endlinenum],
            "linenum": endlinenum,
            "prefix": match.group("prefix"),
            "space": match.group("space"),
            "suffix": match.groupdict().get("suffix", ""),
        }
    else:
        raise RelicError("license block end line did not match: line='%s', "
                         "pattern='%s'"
                         % (lines[endlinenum], lic_end_pattern.pattern))
    log.debug("endline dict: %s", endline)
    lic_info["last_suffix"] = endline["suffix"]
    log.info("suffix for last line: '%s'", endline["suffix"])
    lic_info["end_line"] = endline["linenum"]
    log.info("license lines: %d-%d", beginline["linenum"], endline["linenum"])

    # So at this point we have the beginline, secondline, and endline
    # dicts describing and bounding the license block.
    
    # - gather embedded lic data
    # As described here:
    #   http://www.mozilla.org/MPL/relicensing-faq.html#new-license
    # we have to parse out the following possible fields:
    #   original_code_is
    #   original_code_date
    #   initial_developer
    #   initial_copyright_date
    #   contributors
    lic_line_pattern = re.compile( # regex to parse out the line _body_
        "^(?P<prefix>%s)(?P<space>\s*)(?P<body>.*?)(?P<suffix>%s|)\s*?$"
        % (prefix_pattern, suffix_pattern))

    original_code_is = None
    original_code_date = None
    # Parse out the "The Original Code is ..." paragraph _content_.
    paragraph = ""
    in_paragraph = 0
    for i in range(beginline["linenum"], endline["linenum"]+1):
        body = lic_line_pattern.match(lines[i]).group("body")
        if (not in_paragraph and body.startswith("The Original Code is")):
            in_paragraph = 1
        if in_paragraph:
            if not body.strip(): # i.e. a blank line, end of paragraph
                break
            # ensure one space btwn lines
            if paragraph: paragraph = paragraph.rstrip() + " "
            paragraph += body
    if paragraph:
        pattern1 = re.compile('^The Original Code is (.*), released (.*)\.')
        match = pattern1.search(paragraph)
        if match:
            original_code_is = match.group(1)
            original_code_date = match.group(2)
        else:
             pattern2 = re.compile('^The Original Code is (.*?)\.?$')
             match = pattern2.search(paragraph)
             if match:
                 original_code_is = match.group(1)
             else:
                raise RelicError( 
                    "%s: 'The Original Code is' paragraph did not match the "
                    "expected patterns. paragraph=\n\t%r\n"
                    "pattern1=\n\t%r\npattern2=\n\t%r"
                    % (filename, paragraph, pattern1.pattern, pattern2.pattern))
    lic_info["original_code_is"] = original_code_is
    lic_info["original_code_date"] = original_code_date
    log.info("original code is: %s", original_code_is)
    log.info("original_code_date: %s", original_code_date)

    initial_developer = None
    initial_copyright_date = None
    # Parse out the "The Initial Developer..." paragraph _content_.
    paragraph = ""
    in_paragraph = 0
    for i in range(beginline["linenum"], endline["linenum"]+1):
        body = lic_line_pattern.match(lines[i]).group("body")
        if (not in_paragraph and
            (body.startswith("The Initial Developer of") or
             body.startswith("The Initial Developers of"))):
            in_paragraph = 1
        if in_paragraph:
            if not body.strip(): # i.e. a blank line, end of paragraph
                # Catch the possible case where there is an empty line
                # but the paragraph picks up on the next line with
                # "Portions created by"
                # (test_relicense_no_period_after_origcodeis.cpp).
                try:
                    nextlinebody = lic_line_pattern.match(lines[i+1]).group("body")
                except:
                    nextlinebody = ""
                if not nextlinebody.startswith("Portions created by"):
                    break
            # ensure one space btwn lines
            if paragraph: paragraph = paragraph.rstrip() + " "
            paragraph += body
    if paragraph:
        pattern = re.compile("""^
            The\ Initial\ Developers?\ of\ 
            (the\ Original\ Code\ (is\ |are\ |is\.)|this\ code\ under\ the\ [MN]PL\ (is|are)\ )
            (?P<developer>.*?)
            \.?         # maybe a trailing period
            (
                \s+Portions\ created\ by\ .*?
                are\ Copyright\ \(C\)\[?\ (?P<date>[\d-]+)
                .*?     # maybe a trailing period
                (\s+All\ Rights\ Reserved\.)?
            )?
            $""", re.VERBOSE)
        match = pattern.search(paragraph)
        if not match:
            raise RelicError(
                "%s: 'This Initial Developer' paragraph did not match the "
                "expected pattern. paragraph=\n\t%r\npattern=\n\t%s"
                % (filename, paragraph, pattern.pattern))
        initial_developer = match.group("developer")
        initial_copyright_date = match.group("date")
    lic_info["initial_developer"] = initial_developer
    lic_info["initial_copyright_date"] = initial_copyright_date
    log.info("initial developer paragraph: %r", paragraph)
    log.info("initial developer: %r", initial_developer)
    log.info("initial copyright date: %r", initial_copyright_date)

    contributors = []
    normal_leading_space = None
    in_contributors_block = 0
    contrib_end = endline["linenum"]
    # If line-style comment, include the last line in the block in the
    # range we examine; if block-style comment, we only allow it if the
    # comment-block doesn't end on the endline. On top of these
    # conditions we don't search the last line if it includes the
    # special end-of-license marker.
    if len(comment_delims) == 1 or not endline["suffix"]:
        if endline["content"].find("END LICENSE BLOCK") == -1:
            contrib_end += 1
    for i in range(beginline["linenum"], contrib_end):
        match = lic_line_pattern.match(lines[i])
        body = match.group("body")
        space = match.group("space").replace('\t', ' '*8)
        if not in_contributors_block \
           and body.startswith("Contributor"):
            in_contributors_block = 1
            normal_leading_space = space
            # Try to pickup "foo@bar.org" as a contributor for a
            # possible line like this:
            #   Contributor(s): foo@bar.org
            pivot = body.find(':')
            if pivot != -1:
                remainder = body[pivot+1:].strip()
                if remainder:
                    contributors.append(remainder)
        elif in_contributors_block:
            if not body.strip():
                # i.e. a blank line, end of paragraph
                #XXX:BUG This condition causes the latter two
                #        contributor lines to be lost from, e.g.,
                #        test/x_thead_align_center.xml.
                break
            if len(space) <= len(normal_leading_space):
                # A line in the "Contributor(s) paragraph is not
                # indented. This is considered an error.  Likely this is
                # a (not indented) contributor, but it might also be the
                # start of another paragraph (i.e. no blank line
                # terminating the "Contributor(s):" paragraph). We could
                # just error out here, but this is very common in the
                # Moz tree (~500) so lets try to deal with it.
                # - Heuristic #1: if the line contains what looks like
                #   an email address then this it is a contributor.
                # - Heuristic #2 (to accomodate js/rhino): if the line
                #   looks like just a person's name.
                # Otherwise, error out.
                words = body.split()
                if '@' in body:
                    lic_info["unindented_contributor_lines"] = 1
                elif (2 <= len(words) <= 3 and
                      words == [word[0].upper()+word[1:] for word in words]):
                    # Try to accept the following names:
                    #    Norris Boyd
                    #    Mike McCabe
                    #    George C. Scott
                    lic_info["unindented_contributor_lines"] = 1
                else:
                    raise RelicError("This line is part of the "
                        "'Contributor(s):' paragraph but (1) is not indented "
                        "and (2) does not look like it contains an email "
                        "address: %s:%s: %r" % (filename, i, lines[i]))
            contributors.append(body.strip())
    log.info("contributors: %s", contributors)
    lic_info["contributors"] = contributors

    ## Optimization: The only content in the remain license block lines
    ## (i.e. after the contributors block) should really be the GPL/LGPL
    ## or nothing. Trapping this will avoid losing the latter two
    ## contributor lines in test/x_thead_align_center.xml.
    #gpl_lgpl_lines = _g_trilicense_parts["gpl/lgpl"].splitlines(0)
    #gpl_lgpl = " ".join(gpl_lgpl_lines)
    #for i in range(i, endline["linenum"]):
    #    match = lic_line_pattern.match(lines[i])
    #    body = match.group("body")
    #    space = match.group("space").replace('\t', ' '*8)
    #    if not body.strip():
    #        continue
    #    #XXX This test is no robust enough to use.
    #    if (gpl_lgpl.find(body) == -1 and
    #        body.find(gpl_lgpl) == -1):
    #        print "QQQ: bogus following text: %r" % body

    return lic_info


def _report_on_file(path, (results, switch_to_mpl, show_initial, quick, _errors)):
    log.debug("_report_on_file(path='%s', results)", path)
    output = path + "\n"
    lic_info = {}
    
    if _is_binary(path):
        output += "... binary, skipping this file\n"
    else:
        try:
            lic_info = _get_license_info(path, show_initial, quick)
        except RelicError, ex:
            return _relicensing_error(ex, path, _errors)
    
        if log.isEnabledFor(logging.DEBUG):
            pprint.pprint(lic_info)
        parts = lic_info["parts"]
        if not parts:
            output += "... no license found\n"
        elif "unknown" in parts:
            output += "... unknown license (possibly) found\n"
        elif ((parts == ["block_begin", "mpl", "gpl", "lgpl", "block_end"] or
               parts == ["block_begin", "npl", "gpl", "lgpl", "block_end"]) and
              not lic_info.get("unindented_contributor_lines")):
            if (switch_to_mpl and
                parts == ["block_begin", "npl", "gpl", "lgpl", "block_end"]):
                output += "... %s found (looks complete, but is not MPL)"\
                          % "/".join(parts) + "\n"
            else:
                output += "... %s found (looks complete)"\
                          % "/".join(parts) + "\n"
        else:
            output += "... %s found" % "/".join(parts) + "\n"

        if not quick:
            if "begin_line" in lic_info and "end_line" in lic_info:
                output += "... license block lines: %(begin_line)d-%(end_line)d"\
    	          % lic_info + "\n"
            if "original_code_is" in lic_info:
                output += "... original code is: %(original_code_is)s"\
    		      % lic_info + "\n"
            if "original_code_date" in lic_info:
                output += "... original code date: %(original_code_date)s"\
    	              % lic_info + "\n"
            if "initial_developer" in lic_info:
                output += "... initial developer: %(initial_developer)s"\
    	              % lic_info + "\n"
            if "initial_copyright_date" in lic_info:
                output += "... initial copyright date: %(initial_copyright_date)s"\
    	              % lic_info + "\n"
            if "contributors" in lic_info:
                output += "... contributors: %s"\
    		      % ", ".join(lic_info["contributors"]) + "\n"
            if lic_info.get("unindented_contributor_lines"):
                output += "... one or more contributor lines were not indented properly"\
    	              + "\n"

    if show_initial:
      if "initial_developer" in lic_info:
        print lic_info["initial_developer"]
    else:
      print output;

def _gather_info_on_file(path, (results, _errors)):
    log.debug("_gather_info_on_file(path='%s', results)", path)
    # Skip binary files.
    try:
        if _is_binary(path):
            log.debug("Skipping binary file '%s'.", path)
            return
    except Exception, ex:
        return _relicensing_error(
            "error determining if file is binary: %s" % ex,
            path, _errors)

    try:
        results[path] = _get_license_info(path)
    except RelicError, ex:
        return _relicensing_error(ex, path, _errors, 1)


def _make_backup_path(path):
    for n in range(100):
        backup_path = "%s~%d" % (path, n)
        if not os.path.exists(backup_path):
            return backup_path
    raise RelicError("Could not find an unused backup path for '%s'." % path)


def _relicensing_error(err, path, cache=None, quiet=1):
    """Handle an error during relicensing.
    
        "err" may be an error string or an exception instance.
        "path" is the path of the file on which this error occured.
        "cache" is a mapping of path to errors on which errors may be
            stored for later reporting.
        "quiet" optionally allows one to silence the stdout output when
            force is in effect.

    If the --force option is in-effect then errors may be remembered and
    processing continues, rather than halting the whole process.
    """
    if _g_force:
        if not quiet:
            print "...", err
        if cache is not None:
            cache[path] = err
    elif isinstance(err, Exception):
        raise
    else:
        raise RelicError("%s: %s" % (path, err))


def _get_comment_delim_sets(filename):
    comment_delims = None
    if os.path.splitext(filename)[1] == ".in":
        # "<foo>.in" is generally a precursor for a filetype
        # identifiable without the ".in". Drop it.
        xfilename = os.path.splitext(filename)[0]
    else:
        xfilename = filename
    # special cases for some basenames
    basename = os.path.basename(xfilename)
    try:
        comment_delims = _g_basename_to_comment_info[basename]
    except KeyError:
        pass
    if not comment_delims: # use the file extension
        ext = os.path.splitext(xfilename)[1]
        try:
            comment_delims = _g_ext_to_comment_info[ext]
        except KeyError:
            pass
    if not comment_delims: # try to use the shebang line, if any
        fin = open(filename, 'r')
        firstline = fin.readline()
        fin.close()
        if firstline.startswith("#!"):
            for pattern, cds in _g_shebang_pattern_to_comment_info:
                if pattern.match(firstline):
                    comment_delims = cds
                    break
    if not comment_delims:
        raise RelicError("%s: couldn't determine file type (and "
            "comment delimiter info) from basename '%s' or "
            "extension '%s'): you may need to add to "
            "'_g_basename_to_comment_info', "
            "'_g_ext_to_comment_info', "
            "'_g_shebang_pattern_to_comment_info' "
            "or one of the '_g_skip_*' globals"
            % (filename, basename, ext))
    return comment_delims
    

def _relicense_file(original_path,
                    (fallback_initial_copyright_date,
                     fallback_initial_developer,
                     fallback_original_code_is,
                     fallback_original_code_date,
                     switch_to_mpl,
                     backup,
                     results,
                     force_relicensing,
                     _errors)):
    """Relicense the given file.

        "original_path" is the file to relicense
        "fallback_initial_copyright_date"
        "fallback_initial_developer"
        "fallback_original_code_is"
        "fallback_original_code_date"
            User-specified fallback values to use for the new license
            block if they cannot be found in the original.
        "switch_to_mpl" is a boolean indicating if an NPL-based license
            should be converted to MPL.
        "backup" (optional, default false) is a boolean indicating if
            backups should be made
        "results" is a dictionary in which to store statistics and errors.
            See relicense() for schema.
        "force_relicensing" is a boolean indicating if relicensing
            should be done even if the license block looks complete.
        "_errors" is a dictionary on which errors are reported
            (keyed by file path) when the force option is in effect.

    The function does not return anything.
    """
    log.debug("_relicense_file(original_path='%s')", original_path)
    print original_path

    # Ensure can access file.
    if not os.access(original_path, os.R_OK|os.W_OK):
        return _relicensing_error("cannot access", original_path, _errors)
    else:
        log.info("have read/write access")

    # Skip binary files.
    try:
        if _is_binary(original_path):
            print "... binary, skipping this file"
            results["binary"] += 1
            return
    except Exception, ex:
        return _relicensing_error(
            "error determining if file is binary: %s" % ex,
            original_path, _errors)

    try:
        lic_info = _get_license_info(original_path, 0)
    except RelicError, ex:
        return _relicensing_error(ex, original_path, _errors)

    # Load fallback info if necessary.
    if not lic_info.get("initial_copyright_date"):
        lic_info["initial_copyright_date"] = fallback_initial_copyright_date
    if not lic_info.get("initial_developer"):
        lic_info["initial_developer"] = fallback_initial_developer
    if not lic_info.get("original_code_is"):
        lic_info["original_code_is"] = fallback_original_code_is
    if not lic_info.get("original_code_date"):
        lic_info["original_code_date"] = fallback_original_code_date

    # Return/abort if cannot or do not need to re-license.
    parts = lic_info["parts"]
    if not parts:
        results["no license"] += 1
        print "... no license found, skipping this file"
        return
    elif "unknown" in parts:
        return _relicensing_error("unknown license (possibly) found",
                                  original_path, _errors)
    elif parts.count("block_begin") > 1: # sanity check
        return _relicensing_error(
            "'BEGIN LICENSE BLOCK' delimiter found more than once",
            original_path, _errors)
    elif parts.count("block_end") > 1: # sanity check
        return _relicensing_error(
            "'END LICENSE BLOCK' delimiter found more than once",
            original_path, _errors)
    elif not lic_info["initial_developer"]:
        return _relicensing_error(
            "no 'Initial Developer' section was found -- use "
            "the -i option to specify your own",
            original_path, _errors)
    elif not lic_info["initial_copyright_date"]:
        return _relicensing_error(
            "no initial copyright year was found -- use "
            "the -y option to specify your own",
            original_path, _errors)
    elif not lic_info["original_code_is"]:
        return _relicensing_error(
            "no 'Original Code is' section was found -- use "
            "the -o option to specify your own",
            original_path, _errors)
    elif ((parts == ["block_begin", "mpl", "gpl", "lgpl", "block_end"] or
           parts == ["block_begin", "npl", "gpl", "lgpl", "block_end"]) and
          not lic_info.get("unindented_contributor_lines")):
        #XXX Should add an option to relicense anyway because matching
        #    is not super-strict. E.g. nsWidgetFactory.cpp.
        if (switch_to_mpl and
            parts == ["block_begin", "npl", "gpl", "lgpl", "block_end"]):
            print "... %s found (looks complete, but is not MPL)"\
                  % "/".join(parts)
        elif force_relicensing:
            print "... %s found (looks complete, but forcing relicensing)"\
                  % "/".join(parts)
        else:
            results["good"] += 1
            print "... %s found (looks complete), nothing to do"\
                  % "/".join(parts)
            return

    # We need to re-license this file.
    print "... %s found, need to relicense" % "/".join(parts)
    if lic_info["original_code_is"]:
        print "... original code is: %(original_code_is)s" % lic_info
    if lic_info["original_code_date"]:
        print "... original code date: %(original_code_date)s" % lic_info
    if lic_info["initial_developer"]:
        print "... initial developer: %(initial_developer)s" % lic_info
    if lic_info["initial_copyright_date"]:
        print "... initial copyright date: %(initial_copyright_date)s" % lic_info
    if lic_info["contributors"]:
        print "... contributors: %s" % ", ".join(lic_info["contributors"])

    # Put the license block together.
    # - build up the license block from the appropriate parts
    trilicense = ""
    if (not switch_to_mpl) and ( "npl" in parts ):
        trilicense_name = "NPL/GPL/LGPL"
        trilicense += _g_trilicense_parts["npl"]
    else:
        trilicense_name = "MPL/GPL/LGPL"
        trilicense += _g_trilicense_parts["mpl"]
    print "... replacing lines %d-%d with %s tri-license"\
          % (lic_info["begin_line"], lic_info["end_line"], trilicense_name)
    if lic_info["original_code_is"] is not None:
        if lic_info["original_code_date"] is not None:
            trilicense += _g_trilicense_parts["original_code_is_with_date"] % lic_info
        else:
            trilicense += _g_trilicense_parts["original_code_is"] % lic_info
    #else:
    #    raise RelicError("Gerv, how should the new license block handle no "
    #                     "'Originial Code is...' information? --TM")
    if (lic_info["initial_developer"] is not None
        and lic_info["initial_copyright_date"] is not None):
        trilicense += _g_trilicense_parts["initial_developer"] % lic_info
    #else:
    #    raise RelicError("Gerv, how should the new license block handle no "
    #                     "'Initial Developer is...' information? --TM")
    if lic_info["contributors"]:
        contributors = "  " + "\n  ".join(lic_info["contributors"]) + "\n"
    else:
        contributors = ""
    trilicense += _g_trilicense_parts["contributors"] % contributors
    if trilicense_name == "NPL/GPL/LGPL":
        trilicense += _g_trilicense_parts["gpl/lgpl for npl"]
    else: # trilicense_name == "MPL/GPL/LGPL"
        trilicense += _g_trilicense_parts["gpl/lgpl for mpl"]

    # get fallback comment subsequent prefix
    fallback_prefix = _get_comment_delim_sets(original_path)
    
    # - add the comment delimiters
    lines = trilicense.splitlines()
    for i in range(len(lines)):
        if i == 0:
            prefix = lic_info["first_prefix"]
        else:
            if lic_info["subsequent_prefix"]:
                prefix = lic_info["subsequent_prefix"]
            else:
                prefix = fallback_prefix[0][1]
        if lines[i]:
            if len(lic_info["comment_delims"]) == 0:
                lines[i] = prefix + lines[i]
            else:
                lines[i] = prefix + ' ' + lines[i]
        else: # don't add trailing whitespace
            lines[i] = prefix
    if lic_info["last_suffix"]: # don't add that ' ' if there is no suffix
        lines[-1] += ' ' + lic_info["last_suffix"]
    for i in range(len(lines)): lines[i] += '\n'
    trilicense_lines = lines

    ##### uncomment to debug license block
    # pprint.pprint(lines)
    # return

    # Skip out now if doing a dry-run.
    if _g_dry_run:
        results["relicensed"] += 1
        return

    # Make a backup.
    if backup:
        backup_path = _make_backup_path(original_path)
        print "... backing up to '%s'" % backup_path
        try:
            shutil.copy(original_path, backup_path)
        except EnvironmentError, ex:
            return _relicensing_error(ex, original_path, _errors)

    # Re-license the file.
    try:
        fin = open(original_path, "r")
        try:
            lines = fin.readlines()
        finally:
            fin.close()

        lines[lic_info["begin_line"]:lic_info["end_line"]+1] = trilicense_lines

        fout = open(original_path, "w")
        try:
            fout.write(''.join(lines))
        finally:
            fout.close()
        results["relicensed"] += 1
        print "... done relicensing '%s'" % original_path
    except:
        if backup:
            print "... error relicensing, restoring original"
            if os.path.exists(original_path):
                os.remove(original_path)
            os.rename(backup_path, original_path)
        else:
            print "... error relicensing, file may be corrupted"
        # fallback to type_ for string exceptions
        type_, value, tb = sys.exc_info()
        return _relicensing_error(value or type_,
                                  original_path, _errors)


def _add_license_to_file(original_path,
                         (initial_copyright_date,
                          initial_developer,
                          original_code_is,
                          original_code_date,
                          backup,
                          results,
                          _errors)):
    """Relicense the given file.

        "original_path" is the file to relicense
        "initial_copyright_date"
        "initial_developer"
        "original_code_is"
        "original_code_date"
            User-specified values to use for the new license. All but
            "original_code_date" are required.
        "backup" (optional, default false) is a boolean indicating if
            backups should be made
        "results" is a dictionary in which to store statistics and errors.
            See relicense() for schema.
        "_errors" is a dictionary on which errors are reported
            (keyed by file path) when the force option is in effect.

    The function does not return anything.
    """
    log.debug("_add_license_to_file(original_path='%s')", original_path)
    print original_path

    # Ensure can access file.
    if not os.access(original_path, os.R_OK|os.W_OK):
        return _relicensing_error("cannot access", original_path, _errors)
    else:
        log.info("have read/write access")

    # Skip binary files.
    try:
        if _is_binary(original_path):
            print "... binary, skipping this file"
            results["binary"] += 1
            return
    except Exception, ex:
        return _relicensing_error(
            "error determining if file is binary: %s" % ex,
            original_path, _errors)

    try:
        lic_info = _get_license_info(original_path, 0)
    except RelicError, ex:
        return _relicensing_error(ex, original_path, _errors)

    # Return/abort if cannot or do not need to re-license.
    parts = lic_info["parts"]
    if lic_info["parts"]: # has a license
        results["license"] += 1
        print "... license found, skipping this file"
        return
    #... else we need to add a license to this file.
    print "... no license found, need to add one"

    # Load license info.
    lic_info["initial_developer"] = initial_developer
    print "... initial developer: %(initial_developer)s" % lic_info
    lic_info["initial_copyright_date"] = initial_copyright_date
    print "... initial copyright date: %(initial_copyright_date)s" % lic_info
    lic_info["original_code_is"] = original_code_is
    print "... original code is: %(original_code_is)s" % lic_info
    if original_code_date:
        lic_info["original_code_date"] = original_code_date
        print "... original code date: %(original_code_date)s" % lic_info
    else:
        lic_info["original_code_date"] = None

    # Determine what line we can start putting the license block on.
    # Typically this would be line 0, but for the following exceptions:
    # - Shebang (#!) lines
    # - Emacs local variables line:
    #     /* -*- Mode: C++; ... -*- */
    #   This line does not HAVE to be first, but that seems to be a
    #   trend, so might as well honour it.
    # - XML magic "number":  <?xml version="2.0" ... ?>
    #   where "..." might include newlines
    startline = 0
    try:
        comment_delim_sets = _get_comment_delim_sets(original_path)
    except RelicError, ex:
        return _relicensing_error(ex, original_path, _errors, 1)
    fin = open(original_path, 'r')
    try:
        lines = fin.readlines()
    finally:
        fin.close()
    # If this is an XML file, advance past the magic number tag.
    if lines and lines[0].find("<?xml") != -1:
        line = lines[0]
        if (line.find('encoding="utf-8"') != -1
            and line.startswith("\xef\xbb\xbf")):
            # remove UTF-8 BOM
            # Note: this is hardly robust Unicode XML handling :)
            line = line[3:]
        if line.startswith("<?xml"):
            end_index = lines[startline].find("?>")
            while startline < len(lines):
                startline += 1
                if end_index != -1: # found end of tag
                    break
            # Note: this does not catch something like this:
            #   <?xml version="2.0"?> <?stylesheet ...
            #   ...?>
            # but that is just crazy.
    # else, advance past a possible shebang line.
    else:
        for comment_delims in comment_delim_sets:
            if (len(comment_delims) == 1 and comment_delims[0] == "#"
                and lines[0].startswith("#!")):
                startline += 1
    # Advance past an Emacs local variable line.
    comment_delims = None
    if lines[startline].find("-*-") != -1:
        for comment_delims in comment_delim_sets:
            if lines[startline].find(comment_delims[0]) != -1:
                break
        else:
            # We were hoping to be able to determine which of the set of
            # possible commenting styles was in use by finding the
            # comment start token on the same line as the -*-
            # Emacs-modeline signifier, but could not. This likely means
            # that this file uses a block-style comment but the block
            # doesn't start on the same line. Fallback to the
            # block-style comment delimiter set, if there is one.
            for comment_delims in comment_delim_sets:
                if len(comment_delims) == 3:
                    break
            else:
                comment_delims = comment_delim_sets[0]

        if len(comment_delims) == 1: # line-style comments
            startline += 1
        else: # block-style comments
            in_comment = 0
            while startline < len(lines):
                line = lines[startline]
                linepos = 0
                while linepos < len(line):
                    if not in_comment:
                        i = line.find(comment_delims[0], linepos)
                        if i == -1:
                            break
                        else:
                            in_comment = 1
                            linepos = i+1
                    else:
                        i = line.find(comment_delims[-1], linepos)
                        if i == -1:
                            break
                        else:
                            in_comment = 0
                            linepos = i+1
                startline += 1
                if not in_comment:
                    break
    
    # Put the license block together.
    # - build up the license block from the appropriate parts
    trilicense_name = "MPL/GPL/LGPL"
    print "... adding %s tri-license starting at line %s (zero-based)"\
          % (trilicense_name, startline)
    trilicense =  _g_trilicense_parts["mpl"]
    if lic_info["original_code_date"] is not None:
        trilicense += _g_trilicense_parts["original_code_is_with_date"] % lic_info
    else:
        trilicense += _g_trilicense_parts["original_code_is"] % lic_info
    trilicense += _g_trilicense_parts["initial_developer"] % lic_info
    if lic_info.get("contributors"):
        contributors = "  " + "\n  ".join(lic_info["contributors"]) + "\n"
    else:
        contributors = ""
    trilicense += _g_trilicense_parts["contributors"] % contributors
    trilicense += _g_trilicense_parts["gpl/lgpl for mpl"]
    # - add the comment delimiters
    if comment_delims is None:
        for comment_delims in comment_delim_sets:
            if lines[startline].find(comment_delims[0]) != -1:
                break
            elif len(comment_delims) == 3 and lines[startline].find(comment_delims[1]) != -1:
                break
        else:
            # We were hoping to be able to determine which of the set of
            # possible commenting styles was in use by finding the
            # comment start token on the same line as the -*-
            # Emacs-modeline signifier, but could not. This likely means
            # that this file uses a block-style comment but the block
            # doesn't start on the same line. Fallback to the
            # block-style comment delimiter set, if there is one.
            for comment_delims in comment_delim_sets:
                if len(comment_delims) == 3:
                    break
            else:
                comment_delims = comment_delim_sets[0]
        print "comment delims were none: %r" % comment_delims
    t_lines = trilicense.splitlines()
    if len(comment_delims) == 1: # line-style comments
        for i in range(len(t_lines)):
            if t_lines[i]:
                t_lines[i] = comment_delims[0] + ' ' + t_lines[i]
            else: # don't add trailing whitespace
                t_lines[i] = comment_delims[0]
    else: # block-style comments
        if t_lines[0]:
            t_lines[0] = comment_delims[0] + ' ' + t_lines[0]
        else: # don't add trailing whitespace
            t_lines[0] = comment_delims[0]
        for i in range(1, len(t_lines)):
            if t_lines[i]:
                t_lines[i] = comment_delims[1] + ' ' + t_lines[i]
            else: # don't add trailing whitespace
                t_lines[i] = comment_delims[1]
        t_lines[-1] += ' ' + comment_delims[-1]
    for i in range(len(t_lines)): t_lines[i] += '\n'
    t_lines[-1] += '\n' # add a blank line at end of lic block
    trilicense_lines = t_lines
    #pprint.pprint(t_lines)

    # Skip out now if doing a dry-run.
    if _g_dry_run:
        results["added"] += 1
        return

    # Make a backup.
    if backup:
        backup_path = _make_backup_path(original_path)
        print "... backing up to '%s'" % backup_path
        try:
            shutil.copy(original_path, backup_path)
        except EnvironmentError, ex:
            return _relicensing_error(ex, original_path, _errors)

    # Add the license to the file.
    try:
        lines[startline:startline] = trilicense_lines

        fout = open(original_path, "w")
        try:
            fout.write(''.join(lines))
        finally:
            fout.close()
        results["added"] += 1
        print "... done adding license to '%s'" % original_path
    except:
        if backup:
            print "... error adding license, restoring original"
            if os.path.exists(original_path):
                os.remove(original_path)
            os.rename(backup_path, original_path)
        else:
            print "... error adding license, file may be corrupted"
        # fallback to type_ for string exceptions
        type_, value, tb = sys.exc_info()
        return _relicensing_error(value or type_,
                                  original_path, _errors)


def _traverse_dir((file_handler, results), dirname, names):
    """os.path.walk target to traverse the give dir"""
    log.debug("_traverse_dir((file_handler, results), dirname='%s', "
              "names=%s)", dirname, names)
    # Iterate over names backwards because may modify it in-place.
    # Modifying it in-place ensures that removed entries are not
    # traversed by os.path.walk.
    for i in range(len(names)-1, -1, -1):
        path = os.path.join(dirname, names[i])
        if os.path.isdir(path):
            if _should_skip_dir(path):
                del names[i]
                continue
        if os.path.isfile(path):
            if _should_skip_file(path):
                del names[i]
                continue
            if file_handler is not None:
                file_handler(path, results)

def _traverse(paths, file_handler, arg):
    """Traverse the given path(s) and call the given callback for each.
    
        "paths" is either a list of files or directories, or it is an
            input stream with a path on each line.
        "file_handler" is a callable to be called on each file traversed.
            It is called with the following signature:
                file_handler(path, arg)
        "arg" is some object passed to each callback. This is useful for
            recording results.

    This method takes care of skipping files and directories that should
    be skipped according to .cvsignore files and the configured skip
    paths.  This method does not return anything.
    """
    log.debug("_traverse(paths=%s, file_handler=%s, arg=%s)",
              paths, file_handler, arg)

    for path in paths:
        if path[-1] == "\n": path = path[:-1] # chomp if 'paths' is a stream
        if not os.path.exists(path):
            log.warn("'%s' does not exist, skipping", path)
        elif os.path.isfile(path):
            if _should_skip_file(path):
                continue
            if file_handler is not None:
                file_handler(path, arg)
        elif os.path.isdir(path):
            if _should_skip_dir(path):
                continue
            os.path.walk(path, _traverse_dir, (file_handler, arg))
        else:
            raise RelicError("unexpected path type '%s'" % path)



#---- public routines

def relicense(paths,
              fallback_initial_copyright_date=None,
              fallback_initial_developer=None,
              fallback_original_code_is=None,
              fallback_original_code_date=None,
              switch_to_mpl=0,
              backup=0, 
              force_relicensing=0,
              _errors=None):
    """Relicense the given file(s) (or files in the given dir).
    
        "paths" is either a list of files or directories, or it is an
            input stream with a path on each line.
        "fallback_initial_copyright_date"
        "fallback_initial_developer"
        "fallback_original_code_is"
        "fallback_original_code_date"
            User-specified fallback values to use for the new license
            block if they cannot be found in the original.
        "switch_to_mpl" (optional, default false) is a boolean
            indicating if an NPL-based license should be converted to
            MPL.
        "backup" (optional, default false)is a boolean indicating if
            backups should be made
        "force_relicensing" (option, default false) is a boolean
            indicating if relicensing should happen even if the license
            block looks complete
        "_errors" (optional) is a dictionary on which errors are reported
            (keyed by file path) when the force option is in effect.

    This method does not return anything. It will raise RelicError if
    there is a problem. Note that OSError/IOError may also be raised.
    """
    log.debug("relicense(paths=%s, backup=%r)", paths, backup)
    results = {
        "relicensed": 0,
        "no license": 0,
        "good": 0,
        "binary": 0,
    }
    _traverse(paths, _relicense_file,
              (fallback_initial_copyright_date,
               fallback_initial_developer,
               fallback_original_code_is,
               fallback_original_code_date,
               switch_to_mpl,
               backup,
               results,
               force_relicensing,
               _errors))
    print
    print "--------------------- Summary of Results ------------------------"
    print "Files skipped b/c they are binary:", results["binary"]
    print "Files skipped b/c they already had proper license:", results["good"]
    print "Files skipped b/c they had no license:", results["no license"]
    if _g_dry_run:
        print "Files re-licensed: %d (dry-run)" % results["relicensed"]
    else:
        print "Files re-licensed:", results["relicensed"]
    print "-----------------------------------------------------------------"


def addlicense(paths,
               initial_copyright_date,
               initial_developer,
               original_code_is,
               original_code_date=None,
               backup=0, 
               _errors=None):
    """Add a license to those of the given file(s) that do not appear to
    have one.
    
        "paths" is either a list of files or directories, or it is an
            input stream with a path on each line.
        "initial_copyright_date"
        "initial_developer"
        "original_code_is"
        "original_code_date"
            User-specified values to use for the new license block. All
            but "original_code_date" are required.
        "backup" (optional, default false) is a boolean indicating if
            backups should be made
        "_errors" (optional) is a dictionary on which errors are reported
            (keyed by file path) when the force option is in effect.

    This method does not return anything. It will raise RelicError if
    there is a problem. Note that OSError/IOError may also be raised.
    """
    log.debug("addlicense(paths=%s, backup=%r)", paths, backup)
    if not initial_copyright_date:
        raise RelicError("no Initial Developer copyright year was "
                         "specified -- use the -y option")
    if not initial_developer:
        raise RelicError("no 'Initial Developer' section was specified "
                         "-- use the -i option")
    if not original_code_is:
        raise RelicError("no 'Original Code is' section was specified "
                         "-- use the -o option")

    results = {
        "added": 0,
        "license": 0,
        "binary": 0,
    }
    _traverse(paths, _add_license_to_file,
              (initial_copyright_date,
               initial_developer,
               original_code_is,
               original_code_date,
               backup,
               results,
               _errors))
    print
    print "--------------------- Summary of Results ------------------------"
    print "Files skipped b/c they are binary:", results["binary"]
    print "Files skipped b/c they already had a license:", results["license"]
    if _g_dry_run:
        print "Files to which a license was added: %d (dry-run)" % results["added"]
    else:
        print "Files to which a license was added: %d" % results["added"]
    print "-----------------------------------------------------------------"


def report(paths, switch_to_mpl=0, show_initial=1, quick=0, _errors=None):
    """Report on the existing licenses in the given file(s).
    
        "paths" is either a list of files or directories, or it is an
            input stream with a path on each line.
        "switch_to_mpl" (optional, default false) is a boolean
            indicating if an NPL-based license should be converted to
            MPL.
        "show_initial" (optional, default true) is a boolean indicating
            if the initial developer should be displayed for each file.
        "quick" (optional, default false) is a boolean indicating if only
            basic license checking should be applied.
        "_errors" (optional) is a dictionary on which errors are reported
            (keyed by file path) when the force option is in effect.

    This method does not return anything. It will raise RelicError if
    there is a problem.
    """
    log.debug("report(paths=%s)", paths)
    results = {}
    _traverse(paths,\
	      _report_on_file,\
	      (results, switch_to_mpl, show_initial, quick, _errors))


def statistics(paths, extended=0, quick=0, _errors=None):
    """Show a summary table of licenses in files in the given path(s).

        "paths" is either a list of files or directories, or it is an
            input stream with a path on each line.
        "extended" (optional) is a boolean indicating if extended
            statistics should be shown
        "quick" (optional) is a boolean indicating if quick scan mode should
            be enabled.
        "_errors" (optional) is a dictionary on which errors are reported
            (keyed by file path) when the force option is in effect.

    This method does not return anything.
    """
    #XXX Info gathering returns a lot more info now. We might be able to
    #    output more interesting stats.
    log.debug("statistics(paths=%s, extended=%s)",
              paths, extended)
    results = {}
    _traverse(paths, _gather_info_on_file, (results, _errors))

    # Process results and print out some stats.
    stats = {
        # <lic type>: [<number of hits>, [<files>...]]
    }
    for file, info in results.items():
        lic_types = [p for p in info["parts"]
                     if p not in ["block_begin", "block_end"]]
        if not lic_types:
            name = "<none found>"
        elif "unknown" in lic_types:
            name = "<unknown license>"
        # Distinguish between complete mpl/gpl/lgpl (i.e. with the block
        # begin and end tokens) and incomplete mpl/gpl/lgpl. Likewise
        # NPL.
        elif (info["parts"] == ["block_begin", "mpl", "gpl", "lgpl", "block_end"]):
            name = "mpl/gpl/lgpl (complete)"
        elif (info["parts"] == ["block_begin", "npl", "gpl", "lgpl", "block_end"]):
            name = "npl/gpl/lgpl (complete)"
        else:
            name = "/".join(lic_types)
        if name not in stats: stats[name] = [0, []]
        stats[name][0] += 1
        stats[name][1].append(file)

    statslist = [(hits, name, files) for name, (hits, files) in stats.items()]
    statslist.sort()    # sort by number of hits
    statslist.reverse() # most common first
    print "Summary of Licenses in Files"
    print "============================"
    print " Number  Percent License"
    print "------- -------- -----------"
          #    115   55.55% npl/gpl
    for hits, name, files in statslist:
        print "%7d %7.2f%% %s"\
              % (hits, (hits*100.0/len(results)), name)
        #XXX Removed for now because I am not clear if this is at all
        #    useful.
        #if extended:
        #    hits_per_ext = {}
        #    for file in files:
        #        ext = os.path.splitext(file)[1]
        #        if ext not in hits_per_ext: hits_per_ext[ext] = 0
        #        hits_per_ext[ext] += 1
        #    hits_per_ext_list = [(h, e) for e, h in hits_per_ext.items()]
        #    hits_per_ext_list.sort()
        #    hits_per_ext_list.reverse()
        #    for ext_hits, ext in hits_per_ext_list:
        #        if not ext: ext = "<no extension>"
        #        print "         %7d %s" % (ext_hits, ext)
    print "----------------------------"
    print "%7d files processed" % len(results)

    # Print some other interesting statistics.
    no_original_code_is = []
    no_initial_developer = []
    unindented_contributor_lines = []
    for file, info in results.items():
        if "original_code_is" in info and info["original_code_is"] is None:
            no_original_code_is.append(file)
        if "initial_developer" in info and info["initial_developer"] is None:
            no_initial_developer.append(file)
        if info.get("unindented_contributor_lines"):
            unindented_contributor_lines.append(file)
    print
    print "Licensed files with no 'Initial Developer...' info: %d" % len(no_initial_developer)
    if extended:
        for f in no_initial_developer:
            print "    %s" % f
    print "Licensed files with no 'Original Code is...' info: %d" % len(no_original_code_is)
    if extended:
        for f in no_original_code_is:
            print "    %s" % f
    print "Licensed files with improperly indented 'Contributor(s):' line(s): %d" % len(unindented_contributor_lines)
    if extended:
        for f in unindented_contributor_lines:
            print "    %s" % f
    if extended:
        for hits, name, files in statslist:
            print "Files in license category '%s'" % name
            sortedFiles = files[:]
            sortedFiles.sort()
            for file in sortedFiles:
                print "    %s" % file
    print



#---- mainline

def main(argv):
    try:
        opts, args = getopt.getopt(argv[1:], "VvadhqfML:sxry:i:o:D:ARI",
            ["version", "verbose", "all", "help", "debug",
             "dry-run", "force", "MPL", "license=",
             "statistics", "relicense", "backup", "add", "defaults",
             "force-relicense", "initial-developers", "quick"])
    except getopt.GetoptError, ex:
        log.error(str(ex))
        log.error("Try `%s --help'.", argv[0])
        return 2

    debug = 0
    mode = "report"
    extended = 0
    backup = 0
    quick = 0
    force_relicensing = 0
    fallback_initial_copyright_date = None
    fallback_initial_developer = None
    fallback_original_code_is = None
    fallback_original_code_date = None
    switch_to_mpl = 0
    show_initial = 0
    for opt, optarg in opts:
        if opt in ("-h", "--help"):
            sys.stdout.write(__doc__)
            return
        elif opt in ("-V", "--version"):
            ver = '.'.join([str(part) for part in _version_])
            print "relic %s" % ver
            return
        elif opt in ("-v", "--verbose"):
            log.setLevel(logging.INFO)
        elif opt in ("-a", "--all"):
            global _g_check_all
            _g_check_all = 1
        elif opt in ("-M", "--MPL"):
            switch_to_mpl = 1
        elif opt in ("-d", "--debug"):
            log.setLevel(logging.DEBUG)
            debug = 1
        elif opt in ("--dry-run",):
            global _g_dry_run
            _g_dry_run = 1
        elif opt in ("-f", "--force"):
            global _g_force
            _g_force = 1
        elif opt in ("-s", "--statistics"):
            mode = "statistics"
        elif opt in ("-x",):
            extended = 1
        elif opt in ("-r", "--relicense"):
            mode = "relicense"
        elif opt in ("-R", "--force-relicense"):
            mode = "relicense"
            force_relicensing = 1
        elif opt in ("-A", "--add"):
            mode = "add"
        elif opt == "--backup":
            backup = 1
        elif opt == "-y":
            fallback_initial_copyright_date = optarg
        elif opt == "-i":
            fallback_initial_developer = optarg
        elif opt == "-o":
            fallback_original_code_is = optarg
        elif opt == "-D":
            fallback_original_code_date = optarg
        elif opt in ("-I", "--initial-developers"):
            show_initial = 1
        elif opt == "--defaults":
            fallback_original_code_is = "mozilla.org Code"
            fallback_initial_copyright_date = "2001"
            fallback_initial_developer = "Netscape Communications Corporation"
        elif opt in ("-q", "--quick"):
            quick = 1

    try:
        # Prepare the input.
        if not args:
            log.debug("no given files, trying stdin")
            paths = sys.stdin
        else:
            paths = args

        # Invoke the requested action.
        _errors = {}
        if mode == "relicense":
            relicense(paths,
                      fallback_initial_copyright_date,
                      fallback_initial_developer,
                      fallback_original_code_is,
                      fallback_original_code_date,
                      switch_to_mpl,
                      backup,
                      force_relicensing,
                      _errors=_errors)
        elif mode == "statistics":
            statistics(paths, extended, quick, _errors=_errors)
        elif mode == "report":
            report(paths, switch_to_mpl, show_initial, quick, _errors=_errors)
        elif mode == "add":
            addlicense(paths,
                       fallback_initial_copyright_date,
                       fallback_initial_developer,
                       fallback_original_code_is,
                       fallback_original_code_date,
                       backup,
                       _errors=_errors)
        else:
            raise RelicError("unexpected mode: '%s'" % mode)

        # Report any delayed errors.
        if _errors:
            print
            print "=================== Summary of Errors ==========================="
            print "Files with processing errors:", len(_errors)
            print "================================================================="
            for file, error in _errors.items():
                print "%s: %s" % (file, error)
                print
            print "================================================================="
    except RelicError, ex:
        log.error(str(ex) +
                  " (the --force option can be used to skip problematic "
                  "files and continue processing rather than aborting)")
        if debug:
            print
            import traceback
            traceback.print_exception(*sys.exc_info())
        return 1
    except KeyboardInterrupt:
        pass


if __name__ == "__main__":
    sys.exit( main(sys.argv) )