def evaluate_file(self, f):
"""
Evaluates the given file and returns the result.
"""
gt_path = f["gt_path"]
pdf_path = f["pdf_path"]
tool_path = f["tool_path"]
res = dict()
res["file"] = f
# Don't proceed, if the gt file doesn't exist.
if self.is_missing_or_empty(gt_path):
res["missing_gt_file"] = True
return res
# Don't proceed, if the pdf file doesn't exist.
if self.is_missing_or_empty(pdf_path):
res["missing_pdf_file"] = True
return res
# Read and format the groundtruth file.
gt = self.format_gt_file(gt_path)
# Read and format the tool's output.
tool_output = self.format_tool_file(tool_path)
# Don't proceed, if the gt is empty.
if not gt:
res["missing_gt_file"] = True
return res
# Don't proceed, if the tool output is empty.
if not tool_output:
res["missing_tool_file"] = True
return res
# Compute evaluation result.
res["evaluation_result"] = self.evaluate_strings(gt, tool_output)
res["num_ops"] = count_num_ops(res["evaluation_result"], self.args.junk)
res["prev_num_ops"] = self.deserialize(res)
# Trigger the event here to interact with user immediately, even in
# case of parallel processing.
self.handle_evaluation_result(res)
return res
def assert_equal(self, input1, input2, expected):
diff_phrases = doc_diff.doc_diff(input1, input2)
num_ops = doc_diff.count_num_ops(diff_phrases)
self.assertDictEqual(dict(num_ops), expected)
