tools/download-baselines.py

'''
Downloads the actual gm results most recently generated by the Skia buildbots,
and adds any new ones to SVN control.

This tool makes it much easier to check in new baselines, via the following
steps:

cd .../trunk
svn update
# make sure there are no files awaiting svn commit
tools/download-baselines.py gm/base-macmini-lion-fixed  # or other gm/ subdir
# upload CL for review
# validate that the new images look right
# commit CL


Copyright 2011 Google Inc.

Use of this source code is governed by a BSD-style license that can be
found in the LICENSE file.
'''

# common Python modules
import re
import sys
import urllib2

# modules declared within this same directory
import svn

# Where to download recently generated baseline images for each baseline type.
#
# For now this only works for our Mac buildbots; our other buildbots aren't
# uploading their results to a web server yet.
#
# Note also that these will currently work only within the Google corporate
# network; that will also change soon.
ACTUALS_BY_BASELINE_SUBDIR = {
    'gm/base-macmini':
        'http://172.29.92.185/b/build/slave/Skia_Mac_Float_NoDebug/gm/actual',
    'gm/base-macmini-fixed':
        'http://172.29.92.185/b/build/slave/Skia_Mac_Fixed_NoDebug/gm/actual',
    'gm/base-macmini-lion-fixed':
        'http://172.29.92.179/b/build/slave/Skia_MacMiniLion_Fixed_NoDebug/gm/actual',
    'gm/base-macmini-lion-float':
        'http://172.29.92.179/b/build/slave/Skia_MacMiniLion_Float_NoDebug/gm/actual',
}

IMAGE_REGEX = '.+\.png'
IMAGE_MIMETYPE = 'image/png'

def GetPlatformUrl(baseline_subdir):
    """Return URL within which the buildbots store generated baseline images,
    as of multiple svn revisions.

    Raises KeyError if we don't have a URL matching this baseline_subdir.

    @param baseline_subdir indicates which platform we want images for
    """
    try:
        return ACTUALS_BY_BASELINE_SUBDIR[baseline_subdir]
    except KeyError:
        raise KeyError(
            'unknown baseline_subdir "%s", try one of these instead: %s' % (
                baseline_subdir, ACTUALS_BY_BASELINE_SUBDIR.keys()))

def GetLatestResultsUrl(baseline_subdir):
    """Return URL from which we can download the MOST RECENTLY generated
    images for this baseline type.

    @param baseline_subdir indicates which platform we want images for
    """
    base_platform_url = GetPlatformUrl(baseline_subdir)
    print 'base_platform_url is %s' % base_platform_url

    # Find the most recently generated baseline images within base_platform_url
    response = urllib2.urlopen(base_platform_url)
    html = response.read()
    link_regex = re.compile('<a href="(.*)">')
    links = link_regex.findall(html)
    last_link = links[-1]
    most_recent_result_url = '%s/%s' % (base_platform_url, last_link)
    print 'most_recent_result_url is %s' % most_recent_result_url
    return most_recent_result_url

def DownloadMatchingFiles(source_url, filename_regex, dest_dir):
    """Download all files from source_url that match filename_regex, and save
    them (with their original filenames) in dest_dir.

    @param source_url
    @param filename_regex
    @param dest_dir
    """
    while source_url.endswith('/'):
        source_url = source_url[:-1]
    response = urllib2.urlopen(source_url)
    html = response.read()
    link_regex = re.compile('<a href="(%s)">' % filename_regex)
    links = link_regex.findall(html)
    for link in links:
        DownloadBinaryFile('%s/%s' % (source_url, link),
                           '%s/%s' % (dest_dir, link))

def DownloadBinaryFile(source_url, dest_path):
    """Download a single file from its source_url and save it to local disk
    at dest_path.

    @param source_url
    @param dest_path
    """
    print 'DownloadBinaryFile: %s -> %s' % (source_url, dest_path)
    url_fh = urllib2.urlopen(source_url)
    local_fh = open(dest_path, 'wb')
    local_fh.write(url_fh.read())
    local_fh.close()

def Main(arglist):
    """Download most recently generated baseline images for a given platform,
    and add any new ones to SVN control.

    @param arglist sys.argv or equivalent
    """
    num_args = len(arglist)
    if num_args != 2:
        raise Exception('usage: %s <baseline_subdir>' % __file__)

    baseline_subdir = arglist[1]
    while baseline_subdir.endswith('/'):
        baseline_subdir = baseline_subdir[:-1]

    results_url = GetLatestResultsUrl(baseline_subdir)
    DownloadMatchingFiles(results_url, IMAGE_REGEX, baseline_subdir)
    svn_handler = svn.Svn(baseline_subdir)
    new_files = svn_handler.GetNewFiles()
    if new_files:
        svn_handler.AddFiles(new_files)
        svn_handler.SetProperty(new_files, svn.PROPERTY_MIMETYPE,
                                IMAGE_MIMETYPE)

if __name__ == '__main__':
    Main(sys.argv)