Skip to content

P
pdf-miner
Commit 194f34af authored by quyuan's avatar quyuan
Browse files
Options

CI yaml

parent a6b22ae5
Show whitespace changes
Inline Side-by-side
Showing with 35 additions and 201 deletions
tools/base_data_ocr.json deleted 100644 → 0
View file @ a6b22ae5
{
"accuracy": 1.0,
"precision": 1.0,
"recall": 1.0,
"f1_score": 1.0,
"pdf间的平均编辑距离": 133.10256410256412,
"pdf间的平均bleu": 0.28838311595434046,
"分段准确率": 0.07220216606498195,
"行内公式准确率": {
"accuracy": 0.004835727492533068,
"precision": 0.008790072388831437,
"recall": 0.010634970284641852,
"f1_score": 0.009624911535739562
},
"行内公式编辑距离": 1.6176470588235294,
"行内公式bleu": 0.17154724654721457,
"行间公式准确率": {
"accuracy": 0.08490566037735849,
"precision": 0.1836734693877551,
"recall": 0.13636363636363635,
"f1_score": 0.1565217391304348
},
"行间公式编辑距离": 113.22222222222223,
"行间公式bleu": 0.2531053359913409,
"丢弃文本准确率": {
"accuracy": 0.00035398230088495576,
"precision": 0.0006389776357827476,
"recall": 0.0007930214115781126,
"f1_score": 0.0007077140835102619
},
"丢弃文本标签准确率": {
"color_background_header_txt_block": {
"precision": 0.0,
"recall": 0.0,
"f1-score": 0.0,
"support": 41.0
},
"header": {
"precision": 0.0,
"recall": 0.0,
"f1-score": 0.0,
"support": 4.0
},
"footnote": {
"precision": 1.0,
"recall": 0.009708737864077669,
"f1-score": 0.019230769230769232,
"support": 103.0
},
"on-table": {
"precision": 0.0,
"recall": 0.0,
"f1-score": 0.0,
"support": 665.0
},
"rotate": {
"precision": 0.0,
"recall": 0.0,
"f1-score": 0.0,
"support": 63.0
},
"on-image": {
"precision": 0.0,
"recall": 0.0,
"f1-score": 0.0,
"support": 380.0
},
"micro avg": {
"precision": 1.0,
"recall": 0.0007961783439490446,
"f1-score": 0.0015910898965791568,
"support": 1256.0
}
},
"丢弃图片准确率": {
"accuracy": 0.0,
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0
},
"丢弃表格准确率": {
"accuracy": 0.0,
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0
}
}
\ No newline at end of file
tools/base_data_text.json deleted 100644 → 0
View file @ a6b22ae5
{
"accuracy": 1.0,
"precision": 1.0,
"recall": 1.0,
"f1_score": 1.0,
"pdf间的平均编辑距离": 19.82051282051282,
"pdf间的平均bleu": 0.9002485609584511,
"阅读顺序编辑距离": 0.3176895306859206,
"分段准确率": 0.8989169675090253,
"行内公式准确率": {
"accuracy": 0.9782741738066095,
"precision": 0.9782741738066095,
"recall": 1.0,
"f1_score": 0.9890177880897139
},
"行内公式编辑距离": 0.0,
"行内公式bleu": 0.20340450120213166,
"行间公式准确率": {
"accuracy": 1.0,
"precision": 1.0,
"recall": 1.0,
"f1_score": 1.0
},
"行间公式编辑距离": 0.0,
"行间公式bleu": 0.3662262622386575,
"丢弃文本准确率": {
"accuracy": 0.867870036101083,
"precision": 0.9064856711915535,
"recall": 0.9532117367168914,
"f1_score": 0.9292616930807885
},
"丢弃文本标签准确率": {
"color_background_header_txt_block": {
"precision": 0.0,
"recall": 0.0,
"f1-score": 0.0,
"support": 41.0
},
"rotate": {
"precision": 1.0,
"recall": 0.9682539682539683,
"f1-score": 0.9838709677419355,
"support": 63.0
},
"footnote": {
"precision": 1.0,
"recall": 0.883495145631068,
"f1-score": 0.9381443298969072,
"support": 103.0
},
"header": {
"precision": 1.0,
"recall": 1.0,
"f1-score": 1.0,
"support": 4.0
},
"on-image": {
"precision": 0.9947643979057592,
"recall": 1.0,
"f1-score": 0.9973753280839895,
"support": 380.0
},
"on-table": {
"precision": 1.0,
"recall": 0.9443609022556391,
"f1-score": 0.97138437741686,
"support": 665.0
},
"micro avg": {
"precision": 0.9982847341337907,
"recall": 0.9267515923566879,
"f1-score": 0.9611890999174236,
"support": 1256.0
}
},
"丢弃图片准确率": {
"accuracy": 0.8666666666666667,
"precision": 0.9285714285714286,
"recall": 0.9285714285714286,
"f1_score": 0.9285714285714286
},
"丢弃表格准确率": {
"accuracy": 0,
"precision": 0,
"recall": 0,
"f1_score": 0
}
}
\ No newline at end of file
tools/ocr_badcase.py
View file @ 194f34af
......@@ -756,7 +756,7 @@ def merge_json_data(json_test_df, json_standard_df):
return inner_merge, standard_exist, test_exist
def save_results(result_dict,overall_report_dict,badcase_path,overall_path,):
def save_results(result_dict,overall_report_dict,badcase_path,overall_path, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url):
"""
将结果字典保存为JSON文件至指定路径。
......@@ -764,19 +764,20 @@ def save_results(result_dict,overall_report_dict,badcase_path,overall_path,):
- result_dict: 包含计算结果的字典。
- overall_path: 结果文件的保存路径，包括文件名。
"""
with open(overall_path, 'w', encoding='utf-8') as f:
# 将结果字典转换为JSON格式并写入文件
json.dump(overall_report_dict, f, ensure_ascii=False, indent=4)
final_overall_path = upload_to_s3(overall_path, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url)
overall_path_res = "OCR抽取方案整体评测指标结果请查看：" + final_overall_path
print(f'\033[31m{overall_path_res}\033[0m')
# 打开指定的文件以写入
with open(badcase_path, 'w', encoding='utf-8') as f:
# 将结果字典转换为JSON格式并写入文件
json.dump(result_dict, f, ensure_ascii=False, indent=4)
badcase_path_res = "OCR抽取方案评测badcase输出报告查看：" + badcase_path
final_badcase_path = upload_to_s3(badcase_path, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url)
badcase_path_res = "OCR抽取方案评测badcase输出报告查看：" + final_badcase_path
print(f'\033[31m{badcase_path_res}\033[0m')
with open(overall_path, 'w', encoding='utf-8') as f:
# 将结果字典转换为JSON格式并写入文件
json.dump(overall_report_dict, f, ensure_ascii=False, indent=4)
overall_path_res = "OCR抽取方案整体评测指标结果请查看：" + overall_path
print(f'\033[31m{overall_path_res}\033[0m')
def upload_to_s3(file_path, bucket_name, s3_directory, AWS_ACCESS_KEY, AWS_SECRET_KEY, END_POINT_URL):
"""
上传文件到Amazon S3
......@@ -792,8 +793,9 @@ def upload_to_s3(file_path, bucket_name, s3_directory, AWS_ACCESS_KEY, AWS_SECRE
# 上传文件到S3
s3.upload_file(file_path, bucket_name, s3_object_key)
print(f"文件 {file_path} 成功上传到S3存储桶 {bucket_name} 中的目录 {s3_directory}，文件名为 {file_name}")
s3_path = f"http://st.bigdata.shlab.tech/S3_Browser?output_path=s3://{bucket_name}/{s3_directory}/{file_name}"
return s3_path
#print(f"文件 {file_path} 成功上传到S3存储桶 {bucket_name} 中的目录 {s3_directory}，文件名为 {file_name}")
except FileNotFoundError:
print(f"文件 {file_path} 未找到，请检查文件路径是否正确。")
except NoCredentialsError:
......@@ -801,6 +803,7 @@ def upload_to_s3(file_path, bucket_name, s3_directory, AWS_ACCESS_KEY, AWS_SECRE
except ClientError as e:
print(f"上传文件时发生错误：{e}")
def generate_filename(badcase_path,overall_path):
"""
生成带有当前时间戳的输出文件名。
......@@ -864,17 +867,19 @@ def main(standard_file, test_file, zip_file, badcase_path, overall_path,base_dat
badcase_file,overall_file = generate_filename(badcase_path,overall_path)
# 保存结果到JSON文件
save_results(result_dict, overall_report_dict,badcase_file,overall_file)
#save_results(result_dict, overall_report_dict,badcase_file,overall_file)
save_results(result_dict, overall_report_dict,badcase_file,overall_file, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url)
result=compare_edit_distance(base_data_path, overall_report_dict)
"""
if all([s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url]):
try:
upload_to_s3(badcase_file, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url)
upload_to_s3(overall_file, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url)
except Exception as e:
print(f"上传到S3时发生错误: {e}")
print(result)
"""
#print(result)
assert result == 1
if __name__ == "__main__":
......
tools/text_badcase.py
View file @ 194f34af
......@@ -768,7 +768,7 @@ def merge_json_data(json_test_df, json_standard_df):
return inner_merge, standard_exist, test_exist
def save_results(result_dict,overall_report_dict,badcase_path,overall_path,):
def save_results(result_dict,overall_report_dict,badcase_path,overall_path, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url):
"""
将结果字典保存为JSON文件至指定路径。
......@@ -776,18 +776,21 @@ def save_results(result_dict,overall_report_dict,badcase_path,overall_path,):
- result_dict: 包含计算结果的字典。
- overall_path: 结果文件的保存路径，包括文件名。
"""
with open(overall_path, 'w', encoding='utf-8') as f:
# 将结果字典转换为JSON格式并写入文件
json.dump(overall_report_dict, f, ensure_ascii=False, indent=4)
final_overall_path = upload_to_s3(overall_path, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url)
overall_path_res = "文本型PDF抽取方案整体评测指标结果请查看：" + final_overall_path
print(f'\033[31m{overall_path_res}\033[0m')
# 打开指定的文件以写入
with open(badcase_path, 'w', encoding='utf-8') as f:
# 将结果字典转换为JSON格式并写入文件
json.dump(result_dict, f, ensure_ascii=False, indent=4)
badcase_path_res = "文本型PDF抽取方案评测badcase输出报告查看：" + badcase_path
final_badcase_path = upload_to_s3(badcase_path, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url)
badcase_path_res = "文本型PDF抽取方案评测badcase输出报告查看：" + final_badcase_path
print(f'\033[31m{badcase_path_res}\033[0m')
with open(overall_path, 'w', encoding='utf-8') as f:
# 将结果字典转换为JSON格式并写入文件
json.dump(overall_report_dict, f, ensure_ascii=False, indent=4)
overall_path_res = "文本型PDF抽取方案整体评测指标结果请查看：" + overall_path
print(f'\033[31m{overall_path_res}\033[0m')
def upload_to_s3(file_path, bucket_name, s3_directory, AWS_ACCESS_KEY, AWS_SECRET_KEY, END_POINT_URL):
......@@ -805,8 +808,9 @@ def upload_to_s3(file_path, bucket_name, s3_directory, AWS_ACCESS_KEY, AWS_SECRE
# 上传文件到S3
s3.upload_file(file_path, bucket_name, s3_object_key)
print(f"文件 {file_path} 成功上传到S3存储桶 {bucket_name} 中的目录 {s3_directory}，文件名为 {file_name}")
s3_path = f"http://st.bigdata.shlab.tech/S3_Browser?output_path=s3://{bucket_name}/{s3_directory}/{file_name}"
return s3_path
#print(f"文件 {file_path} 成功上传到S3存储桶 {bucket_name} 中的目录 {s3_directory}，文件名为 {file_name}")
except FileNotFoundError:
print(f"文件 {file_path} 未找到，请检查文件路径是否正确。")
except NoCredentialsError:
......@@ -875,17 +879,17 @@ def main(standard_file, test_file, zip_file, badcase_path, overall_path,base_dat
badcase_file,overall_file = generate_filename(badcase_path,overall_path)
# 保存结果到JSON文件
save_results(result_dict, overall_report_dict,badcase_file,overall_file)
save_results(result_dict, overall_report_dict,badcase_file,overall_file, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url)
result=compare_edit_distance(base_data_path, overall_report_dict)
"""
if all([s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url]):
try:
upload_to_s3(badcase_file, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url)
upload_to_s3(overall_file, s3_bucket_name, s3_file_directory, aws_access_key, aws_secret_key, end_point_url)
except Exception as e:
print(f"上传到S3时发生错误: {e}")
print(result)
"""
assert result == 1
if __name__ == "__main__":
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment