show.py

# Copyright 2019 Google LLC
#
# Licensed under the Apache License, Version 2.0 (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.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Handler for showing the testcase detail page."""

from builtins import object
from builtins import range
import cgi
import datetime
import jinja2
import re

from base import utils
from build_management import revisions
from build_management import source_mapper
from config import db_config
from crash_analysis import severity_analyzer
from datastore import data_handler
from datastore import data_types
from fuzzing import leak_blacklist
from google_cloud_utils import blobs
from handlers import base_handler
from libs import access
from libs import auth
from libs import form
from libs import handler
from libs import helpers
from libs.issue_management import issue_tracker_utils
from metrics import crash_stats
from system import environment

FIND_SIMILAR_ISSUES_OPTIONS = [{
    'type': 'open',
    'label': 'Open'
}, {
    'type': 'all',
    'label': 'All'
}]

CRASH_STATE_REGEXES = [
    r'bad-cast to (.+) from.*', r'bad-cast to (.+)', r'(.+) +in +.+'
]

COMPILED_CRASH_STATE_REGEXES = [
    re.compile(r, flags=re.IGNORECASE) for r in CRASH_STATE_REGEXES
]

MAX_PREVIEW_STACKTRACE_LINE_COUNT = 25


def _parse_suspected_cls(predator_result):
  """Parse raw suspected_cls into dict."""
  if not predator_result:
    return None

  # The raw result contains some additional information that we don't need here.
  # Everything we're concerned with is a part of the "result" object included
  # with the response.
  predator_result = predator_result['result']
  return {
      'found': predator_result.get('found'),
      'suspected_project': predator_result.get('suspected_project'),
      'suspected_components': predator_result.get('suspected_components'),
      'changelists': predator_result.get('suspected_cls'),
      'feedback_url': predator_result.get('feedback_url'),
      'error_message': predator_result.get('error_message'),
  }


def highlight_common_stack_frames(crash_stacktrace):
  """Highlights common stack frames between first two stacks."""
  crash_stacks = [[]]
  highlighted_crash_stacktrace_lines = []
  old_frame_no = 0
  stack_index = 0
  stack_trace_line_format = '^ *#([0-9]+) *0x[0-9a-f]+ (.*)'

  for line in crash_stacktrace.splitlines():
    # Stacktrace separator prefix.
    if stack_index and line.startswith('+-'):
      break

    match = re.match(stack_trace_line_format, line)
    if match:
      frame_no = int(match.group(1))

      # This means we encountered another stack like free or alloc stack.
      if old_frame_no > frame_no:
        stack_index += 1
        crash_stacks.append([])

      crash_stacks[stack_index].append(match.group(2))
      old_frame_no = frame_no

  # If we have just one crash stack and no other stack,
  # then nothing to highlight.
  if stack_index == 0:
    return crash_stacktrace

  # Compare stack frames between first two stacks.
  match_index = -1
  start_index_crash_stack_1 = len(crash_stacks[0]) - 1
  start_index_crash_stack_2 = len(crash_stacks[1]) - 1
  while True:
    if (crash_stacks[0][start_index_crash_stack_1] !=
        crash_stacks[1][start_index_crash_stack_2]):
      break

    match_index = [start_index_crash_stack_1, start_index_crash_stack_2]

    if not start_index_crash_stack_1:
      break
    if not start_index_crash_stack_2:
      break

    start_index_crash_stack_1 -= 1
    start_index_crash_stack_2 -= 1

  # No match found, nothing to highlight.
  if match_index == -1:
    return crash_stacktrace

  old_frame_no = 0
  stack_index = 0
  frame_index = -1
  for line in crash_stacktrace.splitlines():
    match = re.match(stack_trace_line_format, line)
    if match:
      frame_no = int(match.group(1))

      # This means we encountered another stack like free or alloc stack.
      if old_frame_no > frame_no:
        stack_index += 1
        frame_index = -1

      frame_index += 1
      old_frame_no = frame_no

      # We only care about highlighting the first two stacks.
      if stack_index <= 1 and frame_index >= match_index[stack_index]:
        line = '<b>%s</b>' % line

    highlighted_crash_stacktrace_lines.append(line)

  return '\n'.join(highlighted_crash_stacktrace_lines)


def filter_stacktrace(crash_stacktrace, crash_type, revisions_dict):
  """Clean up and format a stack trace for display."""
  if not crash_stacktrace:
    return ''

  filtered_crash_lines = []
  for line in crash_stacktrace.splitlines():
    # Html escape line content to prevent XSS.
    line = cgi.escape(line, quote=True)

    line = source_mapper.linkify_stack_frame(line, revisions_dict)

    filtered_crash_lines.append(line)

  filtered_crash_stacktrace = '\n'.join(filtered_crash_lines)

  if crash_type == leak_blacklist.DIRECT_LEAK_LABEL:
    return leak_blacklist.highlight_first_direct_leak(filtered_crash_stacktrace)

  return highlight_common_stack_frames(filtered_crash_stacktrace)


class Line(object):
  """Represent a stacktrace line."""

  def __init__(self, line_number, content, important):
    self.line_number = line_number
    self.content = content
    self.important = important

  def __str__(self):
    return 'Line(%d, "%s", %s)' % (self.line_number, self.content,
                                   self.important)

  def __eq__(self, other):
    return hash(self) == hash(other)

  def __hash__(self):
    return hash(self.__str__())

  def to_dict(self):
    return {
        'lineNumber': self.line_number,
        'content': self.content,
        'important': self.important,
        'type': 'Line'
    }


class Gap(object):
  """Represent a gap in a previewed stacktrace."""

  def __init__(self, size):
    self.size = size

  def __str__(self):
    return 'Gap(%d)' % self.size

  def __eq__(self, other):
    return hash(self) == hash(other)

  def __hash__(self):
    return hash(self.__str__())

  def to_dict(self):
    return {'type': 'Gap', 'size': self.size}


def _is_line_important(line_content, frames):
  """Check if the line contains a frame; it means the line is
     important."""
  for frame in frames:
    if frame in line_content:
      return True
  return False


def get_stack_frames(crash_state_lines):
  """Get the stack frames from the crash state. Sometimes the crash state
     contains a type of crash, e.g. 'Bad-cast to content::RenderWidget from
     content::RenderWidgetHostViewAura'. The stack frame is
     'content::RenderWidget'."""
  frames = []
  for line in crash_state_lines:
    added = False
    for regex in COMPILED_CRASH_STATE_REGEXES:
      matches = re.match(regex, line)
      if matches:
        frames.append(matches.group(1))
        added = True
        break

    if not added:
      frames.append(line)
  return frames


def convert_to_lines(raw_stacktrace, crash_state_lines, crash_type):
  """Convert an array of string to an array of Line."""
  if not raw_stacktrace or not raw_stacktrace.strip():
    return []

  raw_lines = raw_stacktrace.splitlines()

  frames = get_stack_frames(crash_state_lines)
  escaped_frames = [jinja2.escape(f) for f in frames]
  combined_frames = frames + escaped_frames

  # Certain crash types have their own customized frames that are not related to
  # the stacktrace. Therefore, we make our best effort to preview stacktrace
  # in a reasonable way; we preview around the the top of the stacktrace.
  for unique_type in data_types.CRASH_TYPES_WITH_UNIQUE_STATE:
    if crash_type.startswith(unique_type):
      combined_frames = ['ERROR']
      break

  lines = []
  for index, content in enumerate(raw_lines):
    important = _is_line_important(content, combined_frames)
    lines.append(Line(index + 1, content, important))
  return lines


def _preview_stacktrace(lines,
                        limit=MAX_PREVIEW_STACKTRACE_LINE_COUNT,
                        padding=7):
  """Preview stacktrace around crash_state."""
  preview_lines = set()
  length = len(lines)

  if length <= limit:
    return []

  for index, line in enumerate(lines):
    if line.important:
      this_preview_lines = set()

      for i in range(index - padding, index + padding + 1):
        if i < 0 or i >= length:
          continue
        this_preview_lines.add(lines[i])

      if (len(preview_lines | this_preview_lines)) <= limit:
        preview_lines |= this_preview_lines

  preview_lines = sorted(list(preview_lines), key=lambda l: l.line_number)
  preview_lines_with_gaps = []
  previous_line_number = 0

  for line in preview_lines:
    if (line.line_number - 1) > previous_line_number:
      preview_lines_with_gaps.append(
          Gap(line.line_number - previous_line_number - 1))

    preview_lines_with_gaps.append(line)
    previous_line_number = line.line_number

  if not preview_lines_with_gaps:
    return []

  if length > previous_line_number:
    preview_lines_with_gaps.append(Gap(length - previous_line_number))

  return preview_lines_with_gaps


def get_testcase_detail_by_id(testcase_id):
  """Get testcase detail for rendering the testcase detail page."""
  testcase = access.check_access_and_get_testcase(testcase_id)
  return get_testcase_detail(testcase)


def _get_revision_range_html_from_string(job_type, revision_range):
  """Return revision range html for a revision range and job type given a range
  string."""
  try:
    start_revision, end_revision = revision_range.split(':')
  except:
    return 'Bad revision range.'

  return _get_revision_range_html(job_type, start_revision, end_revision)


def _get_revision_range_html(job_type, start_revision, end_revision=None):
  """Return revision range html for a revision range and job type."""
  if end_revision is None:
    end_revision = start_revision

  component_rev_list = revisions.get_component_range_list(
      start_revision, end_revision, job_type)
  if not component_rev_list:
    return ('%s:%s (No component revisions found!)' % (start_revision,
                                                       end_revision))

  return revisions.format_revision_list(component_rev_list)


def _get_blob_size_string(blob_key):
  """Return blob size string."""
  blob_size = blobs.get_blob_size(blob_key)
  if blob_size is None:
    return None

  return utils.get_size_string(blob_size)


def _format_reproduction_help(reproduction_help):
  """Format a reproduction help string as HTML (linkified with break tags)."""
  if not reproduction_help:
    return ''

  return jinja2.utils.urlize(reproduction_help).replace('\n', '<br>')


def get_testcase_detail(testcase):
  """Get testcase detail for rendering the testcase detail page."""
  config = db_config.get()
  crash_address = testcase.crash_address
  crash_state = testcase.crash_state
  crash_state_lines = crash_state.strip().splitlines()
  crash_type = data_handler.get_crash_type_string(testcase)
  external_user = not access.has_access(job_type=testcase.job_type)
  issue_url = issue_tracker_utils.get_issue_url(testcase)
  metadata = testcase.get_metadata()
  original_testcase_size = _get_blob_size_string(testcase.fuzzed_keys)
  minimized_testcase_size = _get_blob_size_string(testcase.minimized_keys)
  has_issue_tracker = bool(data_handler.get_issue_tracker_name())

  fuzzer_display = data_handler.get_fuzzer_display(testcase)

  formatted_reproduction_help = _format_reproduction_help(
      data_handler.get_formatted_reproduction_help(testcase))
  # When we have a HELP_TEMPLATE, ignore any default values set for HELP_URL.
  if not formatted_reproduction_help:
    reproduction_help_url = data_handler.get_reproduction_help_url(
        testcase, config)
  else:
    reproduction_help_url = None

  if not testcase.regression:
    regression = 'Pending'
  elif testcase.regression == 'NA':
    regression = 'NA'
  else:
    regression = _get_revision_range_html_from_string(testcase.job_type,
                                                      testcase.regression)

  fixed_full = None
  if 'progression_pending' in metadata:
    fixed = 'Pending'
  elif not testcase.fixed:
    fixed = 'NO'
  elif testcase.fixed == 'NA':
    fixed = 'NA'
  elif testcase.fixed == 'Yes':
    fixed = 'YES'
  else:
    fixed = 'YES'
    fixed_full = _get_revision_range_html_from_string(testcase.job_type,
                                                      testcase.fixed)

  last_tested = None
  last_tested_revision = (
      metadata.get('last_tested_revision') or testcase.crash_revision)
  if last_tested_revision:
    last_tested = _get_revision_range_html(testcase.job_type,
                                           last_tested_revision)

  crash_revision = testcase.crash_revision
  crash_revisions_dict = revisions.get_component_revisions_dict(
      crash_revision, testcase.job_type)
  crash_stacktrace = data_handler.get_stacktrace(testcase)
  crash_stacktrace = filter_stacktrace(crash_stacktrace, testcase.crash_type,
                                       crash_revisions_dict)
  crash_stacktrace = convert_to_lines(crash_stacktrace, crash_state_lines,
                                      crash_type)
  crash_stacktrace_preview_lines = _preview_stacktrace(crash_stacktrace)

  last_tested_crash_revision = metadata.get('last_tested_crash_revision')
  last_tested_crash_revisions_dict = revisions.get_component_revisions_dict(
      last_tested_crash_revision, testcase.job_type)
  last_tested_crash_stacktrace = data_handler.get_stacktrace(
      testcase, stack_attribute='last_tested_crash_stacktrace')
  last_tested_crash_stacktrace = filter_stacktrace(
      last_tested_crash_stacktrace, testcase.crash_type,
      last_tested_crash_revisions_dict)
  last_tested_crash_stacktrace = convert_to_lines(last_tested_crash_stacktrace,
                                                  crash_state_lines, crash_type)
  last_tested_crash_stacktrace_preview_lines = _preview_stacktrace(
      last_tested_crash_stacktrace)

  privileged_user = access.has_access(need_privileged_access=True)

  # Fix build url link. |storage.cloud.google.com| takes care of using the
  # right set of authentication credentials needed to access the link.
  if 'build_url' in metadata:
    metadata['build_url'] = metadata['build_url'].replace(
        'gs://', 'https://storage.cloud.google.com/')

  pending_blame_task = (
      testcase.has_blame() and 'blame_pending' in metadata and
      metadata['blame_pending'])
  pending_impact_task = (
      testcase.has_impacts() and not testcase.is_impact_set_flag)
  pending_minimize_task = not testcase.minimized_keys
  pending_progression_task = ('progression_pending' in metadata and
                              metadata['progression_pending'])
  pending_regression_task = not testcase.regression
  pending_stack_task = testcase.last_tested_crash_stacktrace == 'Pending'
  needs_refresh = (
      testcase.status == 'Pending' or
      ((testcase.status == 'Processed' or testcase.status == 'Duplicate') and
       (pending_blame_task or pending_impact_task or pending_minimize_task or
        pending_progression_task or pending_regression_task or
        pending_stack_task)))

  if data_types.SecuritySeverity.is_valid(testcase.security_severity):
    security_severity = severity_analyzer.severity_to_string(
        testcase.security_severity)
  else:
    security_severity = None

  auto_delete_timestamp = None
  auto_close_timestamp = None

  if testcase.one_time_crasher_flag:
    last_crash_time = (
        crash_stats.get_last_crash_time(testcase) or testcase.timestamp)

    # Set auto-delete timestamp for unreproducible testcases with
    # no associated bug.
    if not testcase.bug_information:
      auto_delete_timestamp = utils.utc_datetime_to_timestamp(
          last_crash_time + datetime.timedelta(
              days=data_types.UNREPRODUCIBLE_TESTCASE_NO_BUG_DEADLINE))

    # Set auto-close timestamp for unreproducible testcases with
    # an associated bug.
    if testcase.open and testcase.bug_information:
      auto_close_timestamp = utils.utc_datetime_to_timestamp(
          last_crash_time + datetime.timedelta(
              days=data_types.UNREPRODUCIBLE_TESTCASE_WITH_BUG_DEADLINE))

  memory_tool_display_string = environment.get_memory_tool_display_string(
      testcase.job_type)
  memory_tool_display_label = memory_tool_display_string.split(':')[0]
  memory_tool_display_value = memory_tool_display_string.split(':')[1].strip()

  helpers.log('Testcase %s' % testcase.key.id(), helpers.VIEW_OPERATION)
  return {
      'id':
          testcase.key.id(),
      'crash_type':
          crash_type,
      'crash_address':
          crash_address,
      'crash_state':
          crash_state,  # Used by reproduce tool.
      'crash_state_lines':
          crash_state_lines,
      'crash_revision':
          testcase.crash_revision,
      'csrf_token':
          form.generate_csrf_token(),
      'external_user':
          external_user,
      'footer':
          testcase.comments,
      'formatted_reproduction_help':
          formatted_reproduction_help,
      'fixed':
          fixed,
      'fixed_full':
          fixed_full,
      'issue_url':
          issue_url,
      'is_admin':
          auth.is_current_user_admin(),
      'metadata':
          metadata,
      'minimized_testcase_size':
          minimized_testcase_size,
      'needs_refresh':
          needs_refresh,
      'original_testcase_size':
          original_testcase_size,
      'privileged_user':
          privileged_user,
      'regression':
          regression,
      'crash_stacktrace': {
          'lines':
              crash_stacktrace,
          'preview_lines':
              crash_stacktrace_preview_lines,
          'revision':
              revisions.get_real_revision(
                  crash_revision, testcase.job_type, display=True)
      },
      'last_tested_crash_stacktrace': {
          'lines':
              last_tested_crash_stacktrace,
          'preview_lines':
              last_tested_crash_stacktrace_preview_lines,
          'revision':
              revisions.get_real_revision(
                  last_tested_crash_revision, testcase.job_type, display=True)
      },
      'security_severity':
          security_severity,
      'security_severities':
          data_types.SecuritySeverity.list(),
      'stats': {
          'min_hour': crash_stats.get_min_hour(),
          'max_hour': crash_stats.get_max_hour(),
      },
      'suspected_cls':
          _parse_suspected_cls(metadata.get('predator_result')),
      'testcase':
          testcase,
      'timestamp':
          utils.utc_datetime_to_timestamp(testcase.timestamp),
      'show_blame':
          testcase.has_blame(),
      'show_impact':
          testcase.has_impacts(),
      'impacts_production':
          testcase.impacts_production(),
      'find_similar_issues_options':
          FIND_SIMILAR_ISSUES_OPTIONS,
      'auto_delete_timestamp':
          auto_delete_timestamp,
      'auto_close_timestamp':
          auto_close_timestamp,
      'memory_tool_display_label':
          memory_tool_display_label,
      'memory_tool_display_value':
          memory_tool_display_value,
      'last_tested':
          last_tested,
      'is_admin_or_not_oss_fuzz':
          is_admin_or_not_oss_fuzz(),
      'has_issue_tracker':
          has_issue_tracker,
      'reproduction_help_url':
          reproduction_help_url,
      'is_local_development':
          environment.is_running_on_app_engine_development(),
      'fuzzer_display':
          vars(fuzzer_display),
  }


def is_admin_or_not_oss_fuzz():
  """Return True if the current user is an admin or if this is not OSS-Fuzz."""
  return not utils.is_oss_fuzz() or auth.is_current_user_admin()


class Handler(base_handler.Handler):
  """Handler that shows a testcase in detail."""

  @handler.get(handler.HTML)
  def get(self, testcase_id):
    """Serve the testcase detail HTML page."""
    values = {'info': get_testcase_detail_by_id(testcase_id)}
    self.render('testcase-detail.html', values)


class DeprecatedHandler(base_handler.Handler):
  """Deprecated handler to show old style testcase link with key."""

  def get(self):
    """Serve the redirect to the current test case detail page."""
    testcase_id = self.request.get('key')
    if not testcase_id:
      raise helpers.EarlyExitException('No testcase key provided.', 400)

    self.redirect('/testcase-detail/%s' % testcase_id)


class RefreshHandler(base_handler.Handler):
  """Handler that shows a testcase in detail through JSON."""

  @handler.post(handler.JSON, handler.JSON)
  @handler.oauth
  def post(self):
    """Serve the testcase detail JSON."""
    testcase_id = self.request.get('testcaseId')
    self.render_json(get_testcase_detail_by_id(testcase_id))