#!/usr/bin/env python
# Copyright (c) 2012 The Chromium Authors. All rights reserved.
# Use of this source code is governed by a BSD-style license that can be
# found in the LICENSE file.

"""Utility for checking and processing licensing information in third_party
directories.

Usage: licenses.py <command>

Commands:
  scan     scan third_party directories, verifying that we have licensing info
  credits  generate about:credits on stdout

(You can also import this as a module.)
"""

import argparse
import cgi
import os
import subprocess
import sys

# TODO(agrieve): Move build_utils.WriteDepFile into a non-android directory.
_REPOSITORY_ROOT = os.path.dirname(os.path.dirname(__file__))
sys.path.append(os.path.join(_REPOSITORY_ROOT, 'build/android/gyp/util'))
import build_utils


# Paths from the root of the tree to directories to skip.
PRUNE_PATHS = set([
    # Placeholder directory only, not third-party code.
    os.path.join('third_party','adobe'),

    # Already covered by //third_party/android_tools.
    os.path.join('third_party','android_tools_internal'),

    # Apache 2.0 license. See crbug.com/140478
    os.path.join('third_party','bidichecker'),

    # Build files only, not third-party code.
    os.path.join('third_party','widevine'),

    # Only binaries, used during development.
    os.path.join('third_party','valgrind'),

    # Used for development and test, not in the shipping product.
    os.path.join('build','secondary'),
    os.path.join('third_party','bison'),
    os.path.join('third_party','blanketjs'),
    os.path.join('third_party','chromite'),
    os.path.join('third_party','cygwin'),
    os.path.join('third_party','gles2_conform'),
    os.path.join('third_party','gnu_binutils'),
    os.path.join('third_party','gold'),
    os.path.join('third_party','gperf'),
    os.path.join('third_party','lighttpd'),
    os.path.join('third_party','llvm'),
    os.path.join('third_party','llvm-build'),
    os.path.join('third_party','mingw-w64'),
    os.path.join('third_party','nacl_sdk_binaries'),
    os.path.join('third_party','pefile'),
    os.path.join('third_party','perl'),
    os.path.join('third_party','psyco_win32'),
    os.path.join('third_party','pyelftools'),
    os.path.join('third_party','pylib'),
    os.path.join('third_party','pywebsocket'),
    os.path.join('third_party','syzygy'),

    # Chromium code.
    os.path.join('tools', 'swarming_client'),

    # Stuff pulled in from chrome-internal for official builds/tools.
    os.path.join('third_party', 'clear_cache'),
    os.path.join('third_party', 'gnu'),
    os.path.join('third_party', 'googlemac'),
    os.path.join('third_party', 'pcre'),
    os.path.join('third_party', 'psutils'),
    os.path.join('third_party', 'sawbuck'),
    # See crbug.com/350472
    os.path.join('chrome', 'browser', 'resources', 'chromeos', 'quickoffice'),
    # Chrome for Android proprietary code.
    os.path.join('clank'),

    # Redistribution does not require attribution in documentation.
    os.path.join('third_party','directxsdk'),

    # For testing only, presents on some bots.
    os.path.join('isolate_deps_dir'),

    # Overrides some WebRTC files, same license. Skip this one.
    os.path.join('third_party', 'webrtc_overrides'),
])

# Directories we don't scan through.
VCS_METADATA_DIRS = ('.svn', '.git')
PRUNE_DIRS = (VCS_METADATA_DIRS +
              ('out', 'Debug', 'Release',  # build files
               'layout_tests'))            # lots of subdirs

ADDITIONAL_PATHS = (
    os.path.join('breakpad'),
    os.path.join('chrome', 'common', 'extensions', 'docs', 'examples'),
    os.path.join('chrome', 'test', 'chromeos', 'autotest'),
    os.path.join('chrome', 'test', 'data'),
    os.path.join('native_client'),
    os.path.join('sdch', 'open-vcdiff'),
    os.path.join('testing', 'gmock'),
    os.path.join('testing', 'gtest'),
    os.path.join('tools', 'gyp'),
    os.path.join('tools', 'page_cycler', 'acid3'),
    os.path.join('url', 'third_party', 'mozilla'),
    os.path.join('v8'),
    # Fake directories to include the strongtalk and fdlibm licenses.
    os.path.join('v8', 'strongtalk'),
    os.path.join('v8', 'fdlibm'),
)


# Directories where we check out directly from upstream, and therefore
# can't provide a README.chromium.  Please prefer a README.chromium
# wherever possible.
SPECIAL_CASES = {
    os.path.join('native_client'): {
        "Name": "native client",
        "URL": "http://code.google.com/p/nativeclient",
        "License": "BSD",
    },
    os.path.join('sdch', 'open-vcdiff'): {
        "Name": "open-vcdiff",
        "URL": "https://github.com/google/open-vcdiff",
        "License": "Apache 2.0, MIT, GPL v2 and custom licenses",
        "License Android Compatible": "yes",
    },
    os.path.join('testing', 'gmock'): {
        "Name": "gmock",
        "URL": "http://code.google.com/p/googlemock",
        "License": "BSD",
        "License File": "NOT_SHIPPED",
    },
    os.path.join('testing', 'gtest'): {
        "Name": "gtest",
        "URL": "http://code.google.com/p/googletest",
        "License": "BSD",
        "License File": "NOT_SHIPPED",
    },
    os.path.join('third_party', 'angle'): {
        "Name": "Almost Native Graphics Layer Engine",
        "URL": "http://code.google.com/p/angleproject/",
        "License": "BSD",
    },
    os.path.join('third_party', 'cros_system_api'): {
        "Name": "Chromium OS system API",
        "URL": "http://www.chromium.org/chromium-os",
        "License": "BSD",
        # Absolute path here is resolved as relative to the source root.
        "License File": "/LICENSE.chromium_os",
    },
    os.path.join('third_party', 'lss'): {
        "Name": "linux-syscall-support",
        "URL": "http://code.google.com/p/linux-syscall-support/",
        "License": "BSD",
        "License File": "/LICENSE",
    },
    os.path.join('third_party', 'pdfium'): {
        "Name": "PDFium",
        "URL": "http://code.google.com/p/pdfium/",
        "License": "BSD",
    },
    os.path.join('third_party', 'ppapi'): {
        "Name": "ppapi",
        "URL": "http://code.google.com/p/ppapi/",
    },
    os.path.join('third_party', 'scons-2.0.1'): {
        "Name": "scons-2.0.1",
        "URL": "http://www.scons.org",
        "License": "MIT",
        "License File": "NOT_SHIPPED",
    },
    os.path.join('third_party', 'catapult'): {
        "Name": "catapult",
        "URL": "https://github.com/catapult-project/catapult",
        "License": "BSD",
        "License File": "NOT_SHIPPED",
    },
    os.path.join('third_party', 'v8-i18n'): {
        "Name": "Internationalization Library for v8",
        "URL": "http://code.google.com/p/v8-i18n/",
        "License": "Apache 2.0",
    },
    os.path.join('third_party', 'WebKit'): {
        "Name": "WebKit",
        "URL": "http://webkit.org/",
        "License": "BSD and LGPL v2 and LGPL v2.1",
        # Absolute path here is resolved as relative to the source root.
        "License File": "/third_party/WebKit/LICENSE_FOR_ABOUT_CREDITS",
    },
    os.path.join('third_party', 'webpagereplay'): {
        "Name": "webpagereplay",
        "URL": "http://code.google.com/p/web-page-replay",
        "License": "Apache 2.0",
        "License File": "NOT_SHIPPED",
    },
    os.path.join('tools', 'gyp'): {
        "Name": "gyp",
        "URL": "http://code.google.com/p/gyp",
        "License": "BSD",
        "License File": "NOT_SHIPPED",
    },
    os.path.join('v8'): {
        "Name": "V8 JavaScript Engine",
        "URL": "http://code.google.com/p/v8",
        "License": "BSD",
    },
    os.path.join('v8', 'strongtalk'): {
        "Name": "Strongtalk",
        "URL": "http://www.strongtalk.org/",
        "License": "BSD",
        # Absolute path here is resolved as relative to the source root.
        "License File": "/v8/LICENSE.strongtalk",
    },
    os.path.join('v8', 'fdlibm'): {
        "Name": "fdlibm",
        "URL": "http://www.netlib.org/fdlibm/",
        "License": "Freely Distributable",
        # Absolute path here is resolved as relative to the source root.
        "License File" : "/v8/LICENSE.fdlibm",
        "License Android Compatible" : "yes",
    },
    os.path.join('third_party', 'khronos_glcts'): {
        # These sources are not shipped, are not public, and it isn't
        # clear why they're tripping the license check.
        "Name": "khronos_glcts",
        "URL": "http://no-public-url",
        "License": "Khronos",
        "License File": "NOT_SHIPPED",
    },
    os.path.join('tools', 'telemetry', 'third_party', 'gsutil'): {
        "Name": "gsutil",
        "URL": "https://cloud.google.com/storage/docs/gsutil",
        "License": "Apache 2.0",
        "License File": "NOT_SHIPPED",
    },
    os.path.join('third_party', 'swiftshader'): {
        "Name": "SwiftShader",
        "URL": "https://swiftshader.googlesource.com/SwiftShader",
        "License": "Apache 2.0 and compatible licenses",
        "License Android Compatible": "yes",
        "License File": "/third_party/swiftshader/LICENSE.txt",
    },
}

# Special value for 'License File' field used to indicate that the license file
# should not be used in about:credits.
NOT_SHIPPED = "NOT_SHIPPED"

# Paths for libraries that we have checked are not shipped on iOS. These are
# left out of the licenses file primarily because we don't want to cause a
# firedrill due to someone thinking that Chrome for iOS is using LGPL code
# when it isn't.
# This is a temporary hack; the real solution is crbug.com/178215
KNOWN_NON_IOS_LIBRARIES = set([
    os.path.join('base', 'third_party', 'symbolize'),
    os.path.join('base', 'third_party', 'xdg_mime'),
    os.path.join('base', 'third_party', 'xdg_user_dirs'),
    os.path.join('chrome', 'installer', 'mac', 'third_party', 'bsdiff'),
    os.path.join('chrome', 'installer', 'mac', 'third_party', 'xz'),
    os.path.join('chrome', 'test', 'data', 'third_party', 'kraken'),
    os.path.join('chrome', 'test', 'data', 'third_party', 'spaceport'),
    os.path.join('chrome', 'third_party', 'mock4js'),
    os.path.join('chrome', 'third_party', 'mozilla_security_manager'),
    os.path.join('third_party', 'WebKit'),
    os.path.join('third_party', 'angle'),
    os.path.join('third_party', 'apple_apsl'),
    os.path.join('third_party', 'apple_sample_code'),
    os.path.join('third_party', 'ashmem'),
    os.path.join('third_party', 'bspatch'),
    os.path.join('third_party', 'cacheinvalidation'),
    os.path.join('third_party', 'cld'),
    os.path.join('third_party', 'flot'),
    os.path.join('third_party', 'gtk+'),
    os.path.join('third_party', 'iaccessible2'),
    os.path.join('third_party', 'iccjpeg'),
    os.path.join('third_party', 'isimpledom'),
    os.path.join('third_party', 'jsoncpp'),
    os.path.join('third_party', 'khronos'),
    os.path.join('third_party', 'libXNVCtrl'),
    os.path.join('third_party', 'libevent'),
    os.path.join('third_party', 'libjpeg'),
    os.path.join('third_party', 'libusb'),
    os.path.join('third_party', 'libva'),
    os.path.join('third_party', 'libxslt'),
    os.path.join('third_party', 'lss'),
    os.path.join('third_party', 'lzma_sdk'),
    os.path.join('third_party', 'mesa'),
    os.path.join('third_party', 'molokocacao'),
    os.path.join('third_party', 'motemplate'),
    os.path.join('third_party', 'mozc'),
    os.path.join('third_party', 'mozilla'),
    os.path.join('third_party', 'npapi'),
    os.path.join('third_party', 'ots'),
    os.path.join('third_party', 'ppapi'),
    os.path.join('third_party', 'qcms'),
    os.path.join('third_party', 're2'),
    os.path.join('third_party', 'safe_browsing'),
    os.path.join('third_party', 'sfntly'),
    os.path.join('third_party', 'smhasher'),
    os.path.join('third_party', 'sudden_motion_sensor'),
    os.path.join('third_party', 'swiftshader'),
    os.path.join('third_party', 'swig'),
    os.path.join('third_party', 'talloc'),
    os.path.join('third_party', 'tcmalloc'),
    os.path.join('third_party', 'usb_ids'),
    os.path.join('third_party', 'v8-i18n'),
    os.path.join('third_party', 'wtl'),
    os.path.join('third_party', 'yasm'),
    os.path.join('v8', 'strongtalk'),
])


class LicenseError(Exception):
    """We raise this exception when a directory's licensing info isn't
    fully filled out."""
    pass

def AbsolutePath(path, filename, root):
    """Convert a path in README.chromium to be absolute based on the source
    root."""
    if filename.startswith('/'):
        # Absolute-looking paths are relative to the source root
        # (which is the directory we're run from).
        absolute_path = os.path.join(root, filename[1:])
    else:
        absolute_path = os.path.join(root, path, filename)
    if os.path.exists(absolute_path):
        return absolute_path
    return None

def ParseDir(path, root, require_license_file=True, optional_keys=None):
    """Examine a third_party/foo component and extract its metadata."""

    # Parse metadata fields out of README.chromium.
    # We examine "LICENSE" for the license file by default.
    metadata = {
        "License File": "LICENSE",  # Relative path to license text.
        "Name": None,               # Short name (for header on about:credits).
        "URL": None,                # Project home page.
        "License": None,            # Software license.
        }

    if optional_keys is None:
        optional_keys = []

    if path in SPECIAL_CASES:
        metadata.update(SPECIAL_CASES[path])
    else:
        # Try to find README.chromium.
        readme_path = os.path.join(root, path, 'README.chromium')
        if not os.path.exists(readme_path):
            raise LicenseError("missing README.chromium or licenses.py "
                               "SPECIAL_CASES entry")

        for line in open(readme_path):
            line = line.strip()
            if not line:
                break
            for key in metadata.keys() + optional_keys:
                field = key + ": "
                if line.startswith(field):
                    metadata[key] = line[len(field):]

    # Check that all expected metadata is present.
    errors = []
    for key, value in metadata.iteritems():
        if not value:
            errors.append("couldn't find '" + key + "' line "
                          "in README.chromium or licences.py "
                          "SPECIAL_CASES")

    # Special-case modules that aren't in the shipping product, so don't need
    # their license in about:credits.
    if metadata["License File"] != NOT_SHIPPED:
        # Check that the license file exists.
        for filename in (metadata["License File"], "COPYING"):
            license_path = AbsolutePath(path, filename, root)
            if license_path is not None:
                break

        if require_license_file and not license_path:
            errors.append("License file not found. "
                          "Either add a file named LICENSE, "
                          "import upstream's COPYING if available, "
                          "or add a 'License File:' line to "
                          "README.chromium with the appropriate path.")
        metadata["License File"] = license_path

    if errors:
        raise LicenseError(";\n".join(errors))
    return metadata


def ContainsFiles(path, root):
    """Determines whether any files exist in a directory or in any of its
    subdirectories."""
    for _, dirs, files in os.walk(os.path.join(root, path)):
        if files:
            return True
        for vcs_metadata in VCS_METADATA_DIRS:
            if vcs_metadata in dirs:
                dirs.remove(vcs_metadata)
    return False


def FilterDirsWithFiles(dirs_list, root):
    # If a directory contains no files, assume it's a DEPS directory for a
    # project not used by our current configuration and skip it.
    return [x for x in dirs_list if ContainsFiles(x, root)]


def FindThirdPartyDirs(prune_paths, root):
    """Find all third_party directories underneath the source root."""
    third_party_dirs = set()
    for path, dirs, files in os.walk(root):
        path = path[len(root)+1:]  # Pretty up the path.

        if path in prune_paths:
            dirs[:] = []
            continue

        # Prune out directories we want to skip.
        # (Note that we loop over PRUNE_DIRS so we're not iterating over a
        # list that we're simultaneously mutating.)
        for skip in PRUNE_DIRS:
            if skip in dirs:
                dirs.remove(skip)

        if os.path.basename(path) == 'third_party':
            # Add all subdirectories that are not marked for skipping.
            for dir in dirs:
                dirpath = os.path.join(path, dir)
                if dirpath not in prune_paths:
                    third_party_dirs.add(dirpath)

            # Don't recurse into any subdirs from here.
            dirs[:] = []
            continue

        # Don't recurse into paths in ADDITIONAL_PATHS, like we do with regular
        # third_party/foo paths.
        if path in ADDITIONAL_PATHS:
            dirs[:] = []

    for dir in ADDITIONAL_PATHS:
        if dir not in prune_paths:
            third_party_dirs.add(dir)

    return third_party_dirs


def FindThirdPartyDirsWithFiles(root):
    third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, root)
    return FilterDirsWithFiles(third_party_dirs, root)


# Many builders do not contain 'gn' in their PATH, so use the GN binary from
# //buildtools.
def _GnBinary():
    exe = 'gn'
    if sys.platform == 'linux2':
        subdir = 'linux64'
    elif sys.platform == 'darwin':
        subdir = 'mac'
    elif sys.platform == 'win32':
        subdir, exe = 'win', 'gn.exe'
    else:
        raise RuntimeError("Unsupported platform '%s'." % sys.platform)

    return os.path.join(_REPOSITORY_ROOT, 'buildtools', subdir, exe)


def FindThirdPartyDeps(gn_out_dir, gn_target):
    if not gn_out_dir:
        raise RuntimeError("--gn-out-dir is required if --gn-target is used.")

    gn_deps = subprocess.check_output([_GnBinary(), "desc", gn_out_dir,
                                       gn_target,
                                       "deps", "--as=buildfile", "--all"])
    third_party_deps = set()
    for build_dep in gn_deps.split():
        if ("third_party" in build_dep and
                os.path.basename(build_dep) == "BUILD.gn"):
            third_party_deps.add(os.path.dirname(build_dep))
    return third_party_deps


def ScanThirdPartyDirs(root=None):
    """Scan a list of directories and report on any problems we find."""
    if root is None:
      root = os.getcwd()
    third_party_dirs = FindThirdPartyDirsWithFiles(root)

    errors = []
    for path in sorted(third_party_dirs):
        try:
            metadata = ParseDir(path, root)
        except LicenseError, e:
            errors.append((path, e.args[0]))
            continue

    for path, error in sorted(errors):
        print path + ": " + error

    return len(errors) == 0


def GenerateCredits(
        file_template_file, entry_template_file, output_file, target_os,
        gn_out_dir, gn_target, depfile=None):
    """Generate about:credits."""

    def EvaluateTemplate(template, env, escape=True):
        """Expand a template with variables like {{foo}} using a
        dictionary of expansions."""
        for key, val in env.items():
            if escape:
                val = cgi.escape(val)
            template = template.replace('{{%s}}' % key, val)
        return template

    if gn_target:
        third_party_dirs = FindThirdPartyDeps(gn_out_dir, gn_target)

        # Sanity-check to raise a build error if invalid gn_... settings are
        # somehow passed to this script.
        if not third_party_dirs:
            raise RuntimeError("No deps found.")
    else:
        third_party_dirs = FindThirdPartyDirs(PRUNE_PATHS, _REPOSITORY_ROOT)

    if not file_template_file:
        file_template_file = os.path.join(_REPOSITORY_ROOT, 'components',
                                          'about_ui', 'resources',
                                          'about_credits.tmpl')
    if not entry_template_file:
        entry_template_file = os.path.join(_REPOSITORY_ROOT, 'components',
                                           'about_ui', 'resources',
                                           'about_credits_entry.tmpl')

    entry_template = open(entry_template_file).read()
    entries = []
    for path in third_party_dirs:
        try:
            metadata = ParseDir(path, _REPOSITORY_ROOT)
        except LicenseError:
            # TODO(phajdan.jr): Convert to fatal error (http://crbug.com/39240).
            continue
        if metadata['License File'] == NOT_SHIPPED:
            continue
        if target_os == 'ios' and not gn_target:
            # Skip over files that are known not to be used on iOS. But
            # skipping is unnecessary if GN was used to query the actual
            # dependencies.
            # TODO(lambroslambrou): Remove this step once the iOS build is
            # updated to provide --gn-target to this script.
            if path in KNOWN_NON_IOS_LIBRARIES:
                continue
        env = {
            'name': metadata['Name'],
            'url': metadata['URL'],
            'license': open(metadata['License File'], 'rb').read(),
        }
        entry = {
            'name': metadata['Name'],
            'content': EvaluateTemplate(entry_template, env),
            'license_file': metadata['License File'],
        }
        entries.append(entry)

    entries.sort(key=lambda entry: (entry['name'], entry['content']))
    entries_contents = '\n'.join([entry['content'] for entry in entries])
    file_template = open(file_template_file).read()
    template_contents = "<!-- Generated by licenses.py; do not edit. -->"
    template_contents += EvaluateTemplate(file_template,
                                          {'entries': entries_contents},
                                          escape=False)

    if output_file:
      with open(output_file, 'w') as output:
        output.write(template_contents)
    else:
      print template_contents

    if depfile:
      assert output_file
      # Add in build.ninja so that the target will be considered dirty whenever
      # gn gen is run. Otherwise, it will fail to notice new files being added.
      # This is still no perfect, as it will fail if no build files are changed,
      # but a new README.chromium / LICENSE is added. This shouldn't happen in
      # practice however.
      license_file_list = (entry['license_file'] for entry in entries)
      license_file_list = (os.path.relpath(p) for p in license_file_list)
      license_file_list = sorted(set(license_file_list))
      build_utils.WriteDepfile(depfile, output_file,
                               license_file_list + ['build.ninja'])

    return True


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument('--file-template',
                        help='Template HTML to use for the license page.')
    parser.add_argument('--entry-template',
                        help='Template HTML to use for each license.')
    parser.add_argument('--target-os',
                        help='OS that this build is targeting.')
    parser.add_argument('--gn-out-dir',
                        help='GN output directory for scanning dependencies.')
    parser.add_argument('--gn-target',
                        help='GN target to scan for dependencies.')
    parser.add_argument('command', choices=['help', 'scan', 'credits'])
    parser.add_argument('output_file', nargs='?')
    build_utils.AddDepfileOption(parser)
    args = parser.parse_args()

    if args.command == 'scan':
        if not ScanThirdPartyDirs():
            return 1
    elif args.command == 'credits':
        if not GenerateCredits(args.file_template, args.entry_template,
                               args.output_file, args.target_os,
                               args.gn_out_dir, args.gn_target, args.depfile):
            return 1
    else:
        print __doc__
        return 1


if __name__ == '__main__':
  sys.exit(main())