Source code for galaxy.tool_util.verify

"""Module of utilities for verifying test results."""

import difflib
import filecmp
import hashlib
import json
import logging
import math
import os
import os.path
import re
import shutil
import tempfile
from typing import (
    Any,
    Callable,
    Dict,
    List,
    Optional,
    TYPE_CHECKING,
)

try:
    import numpy
except ImportError:
    pass
try:
    import pysam
except ImportError:
    pass
try:
    from PIL import Image
except ImportError:
    Image = None  # type: ignore[assignment, unused-ignore]
try:
    import tifffile
except ImportError:
    tifffile = None  # type: ignore[assignment, unused-ignore]


from galaxy.tool_util.parser.util import (
    DEFAULT_DELTA,
    DEFAULT_DELTA_FRAC,
    DEFAULT_EPS,
    DEFAULT_METRIC,
    DEFAULT_PIN_LABELS,
)
from galaxy.util import unicodify
from galaxy.util.compression_utils import get_fileobj
from .asserts import verify_assertions
from .test_data import TestDataResolver

if TYPE_CHECKING:
    import numpy.typing

log = logging.getLogger(__name__)

DEFAULT_TEST_DATA_RESOLVER = TestDataResolver()


[docs]def verify( item_label: str, output_content: bytes, attributes: Optional[Dict[str, Any]], filename: Optional[str] = None, get_filecontent: Optional[Callable[[str], bytes]] = None, get_filename: Optional[Callable[[str], str]] = None, keep_outputs_dir: Optional[str] = None, verify_extra_files: Optional[Callable] = None, mode="file", ): """Verify the content of a test output using test definitions described by attributes. Throw an informative assertion error if any of these tests fail. """ attributes = attributes or {} if get_filename is None: get_filecontent_: Callable[[str], bytes] if get_filecontent is None: get_filecontent_ = DEFAULT_TEST_DATA_RESOLVER.get_filecontent else: get_filecontent_ = get_filecontent def get_filename(filename: str) -> str: file_content = get_filecontent_(filename) local_name = make_temp_fname(fname=filename) with open(local_name, "wb") as f: f.write(file_content) return local_name # Check assertions... assertions = attributes.get("assert_list", None) if assertions is not None: try: verify_assertions(output_content, attributes["assert_list"], attributes.get("decompress", False)) except AssertionError as err: errmsg = f"{item_label} different than expected\n" errmsg += unicodify(err) raise AssertionError(errmsg) # Verify checksum attributes... # works with older Galaxy style md5=<expected_sum> or cwltest # style checksum=<hash_type>$<hash>. expected_checksum_type = None expected_checksum = None if attributes is not None and attributes.get("md5", None) is not None: expected_checksum_type = "md5" expected_checksum = attributes.get("md5") elif attributes is not None and attributes.get("checksum", None) is not None: checksum_value = attributes.get("checksum", None) expected_checksum_type, expected_checksum = checksum_value.split("$", 1) if expected_checksum_type: try: _verify_checksum(output_content, expected_checksum_type, expected_checksum) except AssertionError as err: errmsg = f"{item_label} different than expected\n" errmsg += unicodify(err) raise AssertionError(errmsg) # expected object might be None, so don't pull unless available has_expected_object = "object" in attributes if has_expected_object: assert filename is None expected_object = attributes.get("object") actual_object = json.loads(output_content) expected_object_type = type(expected_object) actual_object_type = type(actual_object) if expected_object_type != actual_object_type: message = f"Type mismatch between expected object ({expected_object_type}) and actual object ({actual_object_type})" raise AssertionError(message) if expected_object != actual_object: message = f"Expected object ({expected_object}) does not match actual object ({actual_object})" raise AssertionError(message) elif filename is not None: temp_name = make_temp_fname(fname=filename) with open(temp_name, "wb") as f: f.write(output_content) # If the server's env has GALAXY_TEST_SAVE, save the output file to that # directory. # This needs to be done before the call to `get_filename()` because that # may raise an exception if `filename` does not exist (e.g. when # generating a tool output file from scratch with # `planemo test --update_test_data`). if keep_outputs_dir: ofn = os.path.join(keep_outputs_dir, filename) out_dir = os.path.dirname(ofn) if not os.path.exists(out_dir): os.makedirs(out_dir) log.debug("keep_outputs_dir: %s, ofn: %s", keep_outputs_dir, ofn) try: shutil.copy(temp_name, ofn) except Exception: log.exception("Could not save output file %s to %s", temp_name, ofn) else: log.debug("## GALAXY_TEST_SAVE=%s. saved %s", keep_outputs_dir, ofn) if mode == "directory": # if verifying a file inside a extra_files_path directory # filename already point to a file that exists on disk local_name = filename else: filename_ = get_filename(filename) assert filename_, f"Failed to find output target for test {filename_}" local_name = filename_ compare = attributes.get("compare", "diff") try: if attributes.get("ftype", None) in [ "bam", "qname_sorted.bam", "qname_input_sorted.bam", "unsorted.bam", "cram", ]: try: local_fh, temp_name = _bam_to_sam(local_name, temp_name) local_name = local_fh.name except Exception as e: log.warning("%s. Will compare BAM files", unicodify(e)) if compare == "diff": files_diff(local_name, temp_name, attributes=attributes) elif compare == "re_match": files_re_match(local_name, temp_name, attributes=attributes) elif compare == "re_match_multiline": files_re_match_multiline(local_name, temp_name, attributes=attributes) elif compare == "sim_size": files_delta(local_name, temp_name, attributes=attributes) elif compare == "contains": files_contains(local_name, temp_name, attributes=attributes) elif compare == "image_diff": if Image and tifffile: files_image_diff(local_name, temp_name, attributes=attributes) else: raise Exception( "pillow and tifffile are not installed, but required to compare files using the 'image_diff' method" ) else: raise Exception(f"Unimplemented Compare type: {compare}") except AssertionError as err: errmsg = f"{item_label} different than expected, difference (using {compare}):\n" errmsg += f"( {local_name} v. {temp_name} )\n" errmsg += unicodify(err) raise AssertionError(errmsg) finally: if "GALAXY_TEST_NO_CLEANUP" not in os.environ: os.remove(temp_name) if verify_extra_files: extra_files = attributes.get("extra_files", None) if extra_files: verify_extra_files(extra_files)
[docs]def make_temp_fname(fname=None): """Safe temp name - preserve the file extension for tools that interpret it.""" suffix = os.path.split(fname)[-1] # ignore full path with tempfile.NamedTemporaryFile(prefix="tmp", suffix=suffix, delete=False) as temp: return temp.name
def _bam_to_sam(local_name, temp_name): temp_local = tempfile.NamedTemporaryFile(suffix=".sam", prefix="local_bam_converted_to_sam_") with tempfile.NamedTemporaryFile(suffix=".sam", prefix="history_bam_converted_to_sam_", delete=False) as temp: try: pysam.view("-h", "--no-PG", "-o", temp_local.name, local_name, catch_stdout=False) except Exception as e: msg = f"Converting local (test-data) BAM to SAM failed: {unicodify(e)}" raise Exception(msg) try: pysam.view("-h", "--no-PG", "-o", temp.name, temp_name, catch_stdout=False) except Exception as e: msg = f"Converting history BAM to SAM failed: {unicodify(e)}" raise Exception(msg) os.remove(temp_name) return temp_local, temp.name def _verify_checksum(data, checksum_type, expected_checksum_value): if checksum_type not in ["md5", "sha1", "sha256", "sha512"]: raise Exception(f"Unimplemented hash algorithm [{checksum_type}] encountered.") h = hashlib.new(checksum_type) h.update(data) actual_checksum_value = h.hexdigest() if expected_checksum_value != actual_checksum_value: template = "Output checksum [%s] does not match expected [%s] (using hash algorithm %s)." message = template % (actual_checksum_value, expected_checksum_value, checksum_type) raise AssertionError(message)
[docs]def files_delta(file1, file2, attributes=None): """Check the contents of 2 files for size differences.""" if attributes is None: attributes = {} delta = attributes.get("delta", DEFAULT_DELTA) delta_frac = attributes.get("delta_frac", DEFAULT_DELTA_FRAC) s1 = os.path.getsize(file1) s2 = os.path.getsize(file2) if abs(s1 - s2) > delta: raise AssertionError( "Files %s=%db but %s=%db - compare by size (delta=%s) failed" % (file1, s1, file2, s2, delta) ) if delta_frac is not None and not (s1 - (s1 * delta_frac) <= s2 <= s1 + (s1 * delta_frac)): raise AssertionError( "Files %s=%db but %s=%db - compare by size (delta_frac=%s) failed" % (file1, s1, file2, s2, delta_frac) )
[docs]def get_compressed_formats(attributes): attributes = attributes or {} decompress = attributes.get("decompress") # None means all compressed formats are allowed return None if decompress else []
[docs]def files_diff(file1, file2, attributes=None): """Check the contents of 2 files for differences.""" attributes = attributes or {} def get_lines_diff(diff): count = 0 for line in diff: if (line.startswith("+") and not line.startswith("+++")) or ( line.startswith("-") and not line.startswith("---") ): count += 1 return count if not filecmp.cmp(file1, file2, shallow=False): compressed_formats = get_compressed_formats(attributes) is_pdf = False try: with get_fileobj(file2, compressed_formats=compressed_formats) as fh: history_data = fh.readlines() with get_fileobj(file1, compressed_formats=compressed_formats) as fh: local_file = fh.readlines() except UnicodeDecodeError: if file1.endswith(".pdf") or file2.endswith(".pdf"): is_pdf = True # Replace non-Unicode characters using unicodify(), # difflib.unified_diff doesn't work on list of bytes history_data = [ unicodify(line) for line in get_fileobj(file2, mode="rb", compressed_formats=compressed_formats) ] local_file = [ unicodify(line) for line in get_fileobj(file1, mode="rb", compressed_formats=compressed_formats) ] else: raise AssertionError("Binary data detected, not displaying diff") if attributes.get("sort", False): local_file.sort() history_data.sort() allowed_diff_count = int(attributes.get("lines_diff", 0)) diff = list(difflib.unified_diff(local_file, history_data, "local_file", "history_data")) diff_lines = get_lines_diff(diff) if diff_lines > allowed_diff_count: if "GALAXY_TEST_RAW_DIFF" in os.environ: diff_slice = diff else: if len(diff) < 60: diff_slice = diff[0:40] else: diff_slice = diff[:25] + ["********\n", "*SNIP *\n", "********\n"] + diff[-25:] # FIXME: This pdf stuff is rather special cased and has not been updated to consider lines_diff # due to unknown desired behavior when used in conjunction with a non-zero lines_diff # PDF forgiveness can probably be handled better by not special casing by __extension__ here # and instead using lines_diff or a regular expression matching # or by creating and using a specialized pdf comparison function if is_pdf: # PDF files contain creation dates, modification dates, ids and descriptions that change with each # new file, so we need to handle these differences. As long as the rest of the PDF file does # not differ we're ok. valid_diff_strs = ["description", "createdate", "creationdate", "moddate", "id", "producer", "creator"] valid_diff = False invalid_diff_lines = 0 for line in diff_slice: # Make sure to lower case strings before checking. line = line.lower() # Diff lines will always start with a + or - character, but handle special cases: '--- local_file \n', '+++ history_data \n' if ( (line.startswith("+") or line.startswith("-")) and line.find("local_file") < 0 and line.find("history_data") < 0 ): for vdf in valid_diff_strs: if line.find(vdf) < 0: valid_diff = False else: valid_diff = True # Stop checking as soon as we know we have a valid difference break if not valid_diff: invalid_diff_lines += 1 log.info( "## files diff on '%s' and '%s': lines_diff = %d, found diff = %d, found pdf invalid diff = %d" % (file1, file2, allowed_diff_count, diff_lines, invalid_diff_lines) ) if invalid_diff_lines > allowed_diff_count: # Print out diff_slice so we can see what failed log.info("###### diff_slice ######") raise AssertionError("".join(diff_slice)) else: log.info( "## files diff on '%s' and '%s': lines_diff = %d, found diff = %d" % (file1, file2, allowed_diff_count, diff_lines) ) raise AssertionError("".join(diff_slice))
[docs]def files_re_match(file1, file2, attributes=None): """Check the contents of 2 files for differences using re.match.""" attributes = attributes or {} join_char = "" to_strip = os.linesep compressed_formats = get_compressed_formats(attributes) try: with get_fileobj(file2, compressed_formats=compressed_formats) as fh: history_data = fh.readlines() with get_fileobj(file1, compressed_formats=compressed_formats) as fh: local_file = fh.readlines() except UnicodeDecodeError: join_char = b"" to_strip = os.linesep.encode("utf-8") with open(file2, "rb") as fh: history_data = fh.readlines() with open(file1, "rb") as fh: local_file = fh.readlines() assert len(local_file) == len(history_data), ( "Data File and Regular Expression File contain a different number of lines (%d != %d)\nHistory Data (first 40 lines):\n%s" % (len(local_file), len(history_data), join_char.join(history_data[:40])) ) if attributes.get("sort", False): history_data.sort() local_file.sort() lines_diff = int(attributes.get("lines_diff", 0)) line_diff_count = 0 diffs = [] for regex_line, data_line in zip(local_file, history_data): regex_line = regex_line.rstrip(to_strip) data_line = data_line.rstrip(to_strip) if not re.match(regex_line, data_line): line_diff_count += 1 diffs.append(f"Regular Expression: {regex_line}, Data file: {data_line}\n") if line_diff_count > lines_diff: raise AssertionError( "Regular expression did not match data file (allowed variants=%i):\n%s" % (lines_diff, "".join(diffs)) )
[docs]def files_re_match_multiline(file1, file2, attributes=None): """Check the contents of 2 files for differences using re.match in multiline mode.""" attributes = attributes or {} join_char = "" compressed_formats = get_compressed_formats(attributes) try: with get_fileobj(file2, compressed_formats=compressed_formats) as fh: history_data = fh.readlines() with get_fileobj(file1, compressed_formats=compressed_formats) as fh: local_file = fh.read() except UnicodeDecodeError: join_char = b"" with open(file2, "rb") as fh: history_data = fh.readlines() with open(file1, "rb") as fh: local_file = fh.read() if attributes.get("sort", False): history_data.sort() history_data = join_char.join(history_data) # lines_diff not applicable to multiline matching assert re.match(local_file, history_data, re.MULTILINE), "Multiline Regular expression did not match data file"
[docs]def files_contains(file1, file2, attributes=None): """Check the contents of file2 for substrings found in file1, on a per-line basis.""" # TODO: allow forcing ordering of contains attributes = attributes or {} to_strip = os.linesep compressed_formats = get_compressed_formats(attributes) try: with get_fileobj(file2, compressed_formats=compressed_formats) as fh: history_data = fh.read() with get_fileobj(file1, compressed_formats=compressed_formats) as fh: local_file = fh.readlines() except UnicodeDecodeError: to_strip = os.linesep.encode("utf-8") with open(file2, "rb") as fh: history_data = fh.read() with open(file1, "rb") as fh: local_file = fh.readlines() lines_diff = int(attributes.get("lines_diff", 0)) line_diff_count = 0 for contains in local_file: contains = contains.rstrip(to_strip) if contains not in history_data: line_diff_count += 1 if line_diff_count > lines_diff: raise AssertionError(f"Failed to find '{contains}' in history data. (lines_diff={lines_diff}).")
def _singleobject_intersection_over_union( mask1: "numpy.typing.NDArray", mask2: "numpy.typing.NDArray", ) -> "numpy.floating": return numpy.logical_and(mask1, mask2).sum() / numpy.logical_or(mask1, mask2).sum() def _multiobject_intersection_over_union( mask1: "numpy.typing.NDArray", mask2: "numpy.typing.NDArray", pin_labels: Optional[List[int]] = None, repeat_reverse: bool = True, ) -> List["numpy.floating"]: iou_list: List[numpy.floating] = [] for label1 in numpy.unique(mask1): cc1 = mask1 == label1 # If the label is in `pin_labels`, then use the same label value to find the corresponding object in the second mask. if pin_labels is not None and label1 in pin_labels: cc2 = mask2 == label1 iou_list.append(_singleobject_intersection_over_union(cc1, cc2)) # Otherwise, use the object with the largest IoU value, excluding the pinned labels. else: cc1_iou_list: List[numpy.floating] = [] for label2 in numpy.unique(mask2[cc1]): if pin_labels is not None and label2 in pin_labels: continue cc2 = mask2 == label2 cc1_iou_list.append(_singleobject_intersection_over_union(cc1, cc2)) iou_list.append(max(cc1_iou_list)) # type: ignore[type-var, unused-ignore] # https://github.com/python/typeshed/issues/12562 if repeat_reverse: iou_list.extend(_multiobject_intersection_over_union(mask2, mask1, pin_labels, repeat_reverse=False)) return iou_list
[docs]def intersection_over_union( mask1: "numpy.typing.NDArray", mask2: "numpy.typing.NDArray", pin_labels: Optional[List[int]] = None ) -> "numpy.floating": """Compute the intersection over union (IoU) for the objects in two masks containing lables. The IoU is computed for each uniquely labeled image region (object), and the overall minimum value is returned (i.e. the worst value). To compute the IoU for each object, the corresponding object in the other mask needs to be determined. The object correspondences are not necessarily symmetric. By default, the corresponding object in the other mask is determined as the one with the largest IoU value. If the label of an object is listed in `pin_labels`, then the corresponding object in the other mask is determined as the object with the same label value. Objects with labels listed in `pin_labels` also cannot correspond to objects with different labels. This is particularly useful when specific image regions must always be labeled with a designated label value (e.g., the image background is often labeled with 0 or -1). """ assert mask1.dtype == mask2.dtype assert mask1.ndim == mask2.ndim == 2 assert mask1.shape == mask2.shape for label in pin_labels or []: count = sum(label in mask for mask in (mask1, mask2)) count_str = {1: "one", 2: "both"} assert count == 2, f"Label {label} is pinned but missing in {count_str[2 - count]} of the images." return min(_multiobject_intersection_over_union(mask1, mask2, pin_labels)) # type: ignore[type-var, unused-ignore] # https://github.com/python/typeshed/issues/12562
def _parse_label_list(label_list_str: Optional[str]) -> List[int]: if label_list_str is None: return [] else: return [int(label.strip()) for label in label_list_str.split(",") if len(label_list_str) > 0]
[docs]def get_image_metric( attributes: Dict[str, Any] ) -> Callable[["numpy.typing.NDArray", "numpy.typing.NDArray"], "numpy.floating"]: metric_name = attributes.get("metric", DEFAULT_METRIC) pin_labels = _parse_label_list(attributes.get("pin_labels", DEFAULT_PIN_LABELS)) metrics = { "mae": lambda arr1, arr2: numpy.abs(arr1 - arr2).mean(), # Convert to float before squaring to prevent overflows "mse": lambda arr1, arr2: numpy.square((arr1 - arr2).astype(float)).mean(), "rms": lambda arr1, arr2: math.sqrt(numpy.square((arr1 - arr2).astype(float)).mean()), "fro": lambda arr1, arr2: numpy.linalg.norm((arr1 - arr2).reshape(1, -1), "fro"), "iou": lambda arr1, arr2: 1 - intersection_over_union(arr1, arr2, pin_labels), } try: return metrics[metric_name] except KeyError: raise ValueError(f'No such metric: "{metric_name}"')
def _load_image(filepath: str) -> "numpy.typing.NDArray": """ Reads the given image, trying tifffile and Pillow for reading. """ # Try reading with tifffile first. It fails if the file is not a TIFF. try: arr = tifffile.imread(filepath) # If tifffile failed, then the file is not a tifffile. In that case, try with Pillow. except tifffile.TiffFileError: with Image.open(filepath) as im: arr = numpy.array(im) # Return loaded image return arr
[docs]def files_image_diff(file1: str, file2: str, attributes: Optional[Dict[str, Any]] = None) -> None: """Check the pixel data of 2 image files for differences.""" attributes = attributes or {} arr1 = _load_image(file1) arr2 = _load_image(file2) if arr1.dtype != arr2.dtype: raise AssertionError(f"Image data types did not match ({arr1.dtype}, {arr2.dtype}).") if arr1.shape != arr2.shape: raise AssertionError(f"Image dimensions did not match ({arr1.shape}, {arr2.shape}).") distance = get_image_metric(attributes)(arr1, arr2) distance_eps = attributes.get("eps", DEFAULT_EPS) if distance > distance_eps: raise AssertionError(f"Image difference {distance} exceeds eps={distance_eps}.")