Skip to content

P
pdf-miner
Commit f0c463ed authored by 许瑞's avatar 许瑞
Browse files
Options

Merge branch 'master' of https://github.com/myhloli/Magic-PDF

parents efed5faa 3d2fcc9d
Hide whitespace changes
Inline Side-by-side
Showing with 228 additions and 83 deletions
demo/ocr_demo.py
View file @ f0c463ed
......@@ -90,9 +90,10 @@ def ocr_parse_core(book_name, ocr_pdf_path, ocr_pdf_model_info, start_page_id=0,
if __name__ == '__main__':
# pdf_path = r"/home/cxu/workspace/Magic-PDF/ocr_demo/j.1540-627x.2006.00176.x.pdf"
# json_file_path = r"/home/cxu/workspace/Magic-PDF/ocr_demo/j.1540-627x.2006.00176.x.json"
pdf_path = r"/home/cxu/workspace/Magic-PDF/ocr_demo/j.1540-627x.2006.00176.x.pdf"
json_file_path = r"/home/cxu/workspace/Magic-PDF/ocr_demo/j.1540-627x.2006.00176.x.json"
# ocr_local_parse(pdf_path, json_file_path)
# book_name = "数学新星网/edu_00001236"
# ocr_online_parse(book_name)
book_name = "科数网/edu_00011318"
ocr_online_parse(book_name)
pass
magic_pdf/dict2md/ocr_mkcontent.py
View file @ f0c463ed
......@@ -72,18 +72,26 @@ def ocr_mk_mm_markdown_with_para(pdf_info_dict: dict):
markdown = []
for _, page_info in pdf_info_dict.items():
paras_of_layout = page_info.get("para_blocks")
page_markdown = ocr_mk_mm_markdown_with_para_core(paras_of_layout)
page_markdown = ocr_mk_mm_markdown_with_para_core(paras_of_layout, "mm")
markdown.extend(page_markdown)
return '\n\n'.join(markdown)
def ocr_mk_nlp_markdown_with_para(pdf_info_dict: dict):
markdown = []
for _, page_info in pdf_info_dict.items():
paras_of_layout = page_info.get("para_blocks")
page_markdown = ocr_mk_mm_markdown_with_para_core(paras_of_layout, "nlp")
markdown.extend(page_markdown)
return '\n\n'.join(markdown)
def ocr_mk_mm_markdown_with_para_and_pagination(pdf_info_dict: dict):
markdown_with_para_and_pagination = []
for page_no, page_info in pdf_info_dict.items():
paras_of_layout = page_info.get("para_blocks")
if not paras_of_layout:
continue
page_markdown = ocr_mk_mm_markdown_with_para_core(paras_of_layout)
page_markdown = ocr_mk_mm_markdown_with_para_core(paras_of_layout, "mm")
markdown_with_para_and_pagination.append({
'page_no': page_no,
'md_content': '\n\n'.join(page_markdown)
......@@ -91,7 +99,7 @@ def ocr_mk_mm_markdown_with_para_and_pagination(pdf_info_dict: dict):
return markdown_with_para_and_pagination
def ocr_mk_mm_markdown_with_para_core(paras_of_layout):
def ocr_mk_mm_markdown_with_para_core(paras_of_layout, mode):
page_markdown = []
for paras in paras_of_layout:
for para in paras:
......@@ -99,6 +107,7 @@ def ocr_mk_mm_markdown_with_para_core(paras_of_layout):
for line in para:
for span in line['spans']:
span_type = span.get('type')
content = ''
if span_type == ContentType.Text:
content = split_long_words(span['content'])
# content = span['content']
......@@ -107,9 +116,16 @@ def ocr_mk_mm_markdown_with_para_core(paras_of_layout):
elif span_type == ContentType.InterlineEquation:
content = f"\n$$\n{span['content']}\n$$\n"
elif span_type in [ContentType.Image, ContentType.Table]:
content = f"\n![]({join_path(s3_image_save_path, span['image_path'])})\n"
para_text += content + ' '
page_markdown.append(para_text.strip() + ' ')
if mode == 'mm':
content = f"\n![]({join_path(s3_image_save_path, span['image_path'])})\n"
elif mode == 'nlp':
pass
if content != '':
para_text += content + ' '
if para_text.strip() == '':
continue
else:
page_markdown.append(para_text.strip() + ' ')
return page_markdown
......
magic_pdf/para/para_split.py
View file @ f0c463ed
......@@ -9,7 +9,7 @@ from magic_pdf.libs.ocr_content_type import ContentType
LINE_STOP_FLAG = ['.', '!', '?', '。', '！', '？',"：", ":", ")", "）", ";"]
INLINE_EQUATION = ContentType.InlineEquation
INTERLINE_EQUATION = ContentType.InterlineEquation
TEXT = "text"
TEXT = ContentType.Text
def __get_span_text(span):
......@@ -20,7 +20,7 @@ def __get_span_text(span):
return c
def __detect_list_lines(lines, new_layout_bboxes, lang='en'):
def __detect_list_lines(lines, new_layout_bboxes, lang):
"""
探测是否包含了列表，并且把列表的行分开.
这样的段落特点是，顶格字母大写/数字，紧跟着几行缩进的。缩进的行首字母含小写的。
......@@ -315,11 +315,14 @@ def __split_para_in_layoutbox(lines_group, new_layout_bbox, lang="en", char_avg_
return layout_paras, list_info
def __connect_list_inter_layout(layout_paras, new_layout_bbox, layout_list_info, page_num, lang="en"):
def __connect_list_inter_layout(layout_paras, new_layout_bbox, layout_list_info, page_num, lang):
"""
如果上个layout的最后一个段落是列表，下一个layout的第一个段落也是列表，那么将他们连接起来。
如果上个layout的最后一个段落是列表，下一个layout的第一个段落也是列表，那么将他们连接起来。 TODO 因为没有区分列表和段落，所以这个方法暂时不实现。
根据layout_list_info判断是不是列表。，下个layout的第一个段如果不是列表，那么看他们是否有几行都有相同的缩进。
"""
if len(layout_paras)==0 or len(layout_list_info)==0: # 0的时候最后的return 会出错
return layout_paras, [False, False]
for i in range(1, len(layout_paras)):
pre_layout_list_info = layout_list_info[i-1]
next_layout_list_info = layout_list_info[i]
......@@ -345,7 +348,37 @@ def __connect_list_inter_layout(layout_paras, new_layout_bbox, layout_list_info,
pre_last_para.extend(may_list_lines)
layout_paras[i] = layout_paras[i][len(may_list_lines):]
return layout_paras
return layout_paras, [layout_list_info[0][0], layout_list_info[-1][1]] # 同时还返回了这个页面级别的开头、结尾是不是列表的信息
def __connect_list_inter_page(pre_page_paras, next_page_paras, pre_page_layout_bbox, next_page_layout_bbox, pre_page_list_info, next_page_list_info, page_num, lang):
"""
如果上个layout的最后一个段落是列表，下一个layout的第一个段落也是列表，那么将他们连接起来。 TODO 因为没有区分列表和段落，所以这个方法暂时不实现。
根据layout_list_info判断是不是列表。，下个layout的第一个段如果不是列表，那么看他们是否有几行都有相同的缩进。
"""
if len(pre_page_paras)==0 or len(next_page_paras)==0: # 0的时候最后的return 会出错
return False
if pre_page_list_info[1] and not next_page_list_info[0]: # 前一个是列表结尾，后一个是非列表开头，此时检测是否有相同的缩进
logger.info(f"连接page {page_num} 内的list")
# 向layout_paras[i] 寻找开头具有相同缩进的连续的行
may_list_lines = []
for j in range(len(next_page_paras[0])):
line = next_page_paras[0][j]
if len(line)==1: # 只可能是一行，多行情况再需要分析了
if line[0]['bbox'][0] > __find_layout_bbox_by_line(line[0]['bbox'], next_page_layout_bbox)[0]:
may_list_lines.append(line[0])
else:
break
else:
break
# 如果这些行的缩进是相等的，那么连到上一个layout的最后一个段落上。
if len(may_list_lines)>0 and len(set([x['bbox'][0] for x in may_list_lines]))==1:
pre_page_paras[-1].append(may_list_lines)
next_page_paras[0] = next_page_paras[0][len(may_list_lines):]
return True
return False
def __find_layout_bbox_by_line(line_bbox, layout_bboxes):
......@@ -358,7 +391,7 @@ def __find_layout_bbox_by_line(line_bbox, layout_bboxes):
return None
def __connect_para_inter_layoutbox(layout_paras, new_layout_bbox, lang="en"):
def __connect_para_inter_layoutbox(layout_paras, new_layout_bbox, lang):
"""
layout之间进行分段。
主要是计算前一个layOut的最后一行和后一个layout的第一行是否可以连接。
......@@ -368,10 +401,19 @@ def __connect_para_inter_layoutbox(layout_paras, new_layout_bbox, lang="en"):
"""
connected_layout_paras = []
if len(layout_paras)==0:
return connected_layout_paras
connected_layout_paras.append(layout_paras[0])
for i in range(1, len(layout_paras)):
pre_last_line = layout_paras[i-1][-1][-1]
next_first_line = layout_paras[i][0][0]
try:
if len(layout_paras[i])==0 or len(layout_paras[i-1])==0: # TODO 考虑连接问题，
continue
pre_last_line = layout_paras[i-1][-1][-1]
next_first_line = layout_paras[i][0][0]
except Exception as e:
logger.error(f"page layout {i} has no line")
continue
pre_last_line_text = ''.join([__get_span_text(span) for span in pre_last_line['spans']])
pre_last_line_type = pre_last_line['spans'][-1]['type']
next_first_line_text = ''.join([__get_span_text(span) for span in next_first_line['spans']])
......@@ -400,7 +442,7 @@ def __connect_para_inter_layoutbox(layout_paras, new_layout_bbox, lang="en"):
return connected_layout_paras
def __connect_para_inter_page(pre_page_paras, next_page_paras, pre_page_layout_bbox, next_page_layout_bbox, lang):
def __connect_para_inter_page(pre_page_paras, next_page_paras, pre_page_layout_bbox, next_page_layout_bbox, page_num, lang):
"""
连接起来相邻两个页面的段落——前一个页面最后一个段落和后一个页面的第一个段落。
是否可以连接的条件：
......@@ -408,7 +450,7 @@ def __connect_para_inter_page(pre_page_paras, next_page_paras, pre_page_layout_b
2. 后一个页面的第一个段落第一行没有空白开头。
"""
# 有的页面可能压根没有文字
if len(pre_page_paras)==0 or len(next_page_paras)==0:
if len(pre_page_paras)==0 or len(next_page_paras)==0 or len(pre_page_paras[0])==0 or len(next_page_paras[0])==0: # TODO [[]]为什么出现在pre_page_paras里？
return False
pre_last_para = pre_page_paras[-1][-1]
next_first_para = next_page_paras[0][0]
......@@ -436,8 +478,85 @@ def __connect_para_inter_page(pre_page_paras, next_page_paras, pre_page_layout_b
else:
return False
def find_consecutive_true_regions(input_array):
start_index = None # 连续True区域的起始索引
regions = [] # 用于保存所有连续True区域的起始和结束索引
for i in range(len(input_array)):
# 如果我们找到了一个True值，并且当前并没有在连续True区域中
if input_array[i] and start_index is None:
start_index = i # 记录连续True区域的起始索引
def __do_split(blocks, layout_bboxes, new_layout_bbox, page_num, lang="en"):
# 如果我们找到了一个False值，并且当前在连续True区域中
elif not input_array[i] and start_index is not None:
# 如果连续True区域长度大于1，那么将其添加到结果列表中
if i - start_index > 1:
regions.append((start_index, i-1))
start_index = None # 重置起始索引
# 如果最后一个元素是True，那么需要将最后一个连续True区域加入到结果列表中
if start_index is not None and len(input_array) - start_index > 1:
regions.append((start_index, len(input_array)-1))
return regions
def __connect_middle_align_text(page_paras, new_layout_bbox, page_num, lang, debug_mode):
"""
找出来中间对齐的连续单行文本，如果连续行高度相同，那么合并为一个段落。
一个line居中的条件是：
1. 水平中心点跨越layout的中心点。
2. 左右两侧都有空白
"""
for layout_i, layout_para in enumerate(page_paras):
layout_box = new_layout_bbox[layout_i]
single_line_paras_tag = []
for i in range(len(layout_para)):
single_line_paras_tag.append(len(layout_para[i])==1 and layout_para[i][0]['spans'][0]['type']==TEXT)
"""找出来连续的单行文本，如果连续行高度相同，那么合并为一个段落。"""
consecutive_single_line_indices = find_consecutive_true_regions(single_line_paras_tag)
if len(consecutive_single_line_indices)>0:
index_offset = 0
"""检查这些行是否是高度相同的，居中的"""
for start, end in consecutive_single_line_indices:
start += index_offset
end += index_offset
line_hi = np.array([line[0]['bbox'][3]-line[0]['bbox'][1] for line in layout_para[start:end+1]])
first_line_text = ''.join([__get_span_text(span) for span in layout_para[start][0]['spans']])
if "Table" in first_line_text or "Figure" in first_line_text:
pass
if debug_mode:
logger.info(line_hi.std())
if line_hi.std()<2:
"""行高度相同，那么判断是否居中"""
all_left_x0 = [line[0]['bbox'][0] for line in layout_para[start:end+1]]
all_right_x1 = [line[0]['bbox'][2] for line in layout_para[start:end+1]]
layout_center = (layout_box[0] + layout_box[2]) / 2
if all([x0 < layout_center < x1 for x0, x1 in zip(all_left_x0, all_right_x1)]) \
and not all([x0==layout_box[0] for x0 in all_left_x0]) \
and not all([x1==layout_box[2] for x1 in all_right_x1]):
merge_para = [l[0] for l in layout_para[start:end+1]]
para_text = ''.join([__get_span_text(span) for line in merge_para for span in line['spans']])
if debug_mode:
logger.info(para_text)
layout_para[start:end+1] = [merge_para]
index_offset -= end-start
return
def __merge_signle_list_text(page_paras, new_layout_bbox, page_num, lang):
"""
找出来连续的单行文本，如果首行顶格，接下来的几个单行段落缩进对齐，那么合并为一个段落。
"""
pass
def __do_split_page(blocks, layout_bboxes, new_layout_bbox, page_num, lang):
"""
根据line和layout情况进行分段
先实现一个根据行末尾特征分段的简单方法。
......@@ -451,35 +570,54 @@ def __do_split(blocks, layout_bboxes, new_layout_bbox, page_num, lang="en"):
"""
lines_group = __group_line_by_layout(blocks, layout_bboxes, lang) # block内分段
layout_paras, layout_list_info = __split_para_in_layoutbox(lines_group, new_layout_bbox, lang) # layout内分段
layout_paras2 = __connect_list_inter_layout(layout_paras, new_layout_bbox, layout_list_info, page_num, lang) # layout之间连接列表段落
layout_paras2, page_list_info = __connect_list_inter_layout(layout_paras, new_layout_bbox, layout_list_info, page_num, lang) # layout之间连接列表段落
connected_layout_paras = __connect_para_inter_layoutbox(layout_paras2, new_layout_bbox, lang) # layout间链接段落
return connected_layout_paras
def para_split(pdf_info_dict, lang="en"):
return connected_layout_paras, page_list_info
def para_split(pdf_info_dict, debug_mode, lang="en"):
"""
根据line和layout情况进行分段
"""
new_layout_of_pages = [] # 数组的数组，每个元素是一个页面的layoutS
all_page_list_info = [] # 保存每个页面开头和结尾是否是列表
for page_num, page in pdf_info_dict.items():
blocks = page['preproc_blocks']
layout_bboxes = page['layout_bboxes']
new_layout_bbox = __common_pre_proc(blocks, layout_bboxes)
new_layout_of_pages.append(new_layout_bbox)
splited_blocks = __do_split(blocks, layout_bboxes, new_layout_bbox, page_num, lang)
splited_blocks, page_list_info = __do_split_page(blocks, layout_bboxes, new_layout_bbox, page_num, lang)
all_page_list_info.append(page_list_info)
page['para_blocks'] = splited_blocks
"""连接页面与页面之间的可能合并的段落"""
pdf_infos = list(pdf_info_dict.values())
for i, page in enumerate(pdf_info_dict.values()):
if i==0:
for page_num, page in enumerate(pdf_info_dict.values()):
if page_num==0:
continue
pre_page_paras = pdf_infos[i-1]['para_blocks']
next_page_paras = pdf_infos[i]['para_blocks']
pre_page_layout_bbox = new_layout_of_pages[i-1]
next_page_layout_bbox = new_layout_of_pages[i]
pre_page_paras = pdf_infos[page_num-1]['para_blocks']
next_page_paras = pdf_infos[page_num]['para_blocks']
pre_page_layout_bbox = new_layout_of_pages[page_num-1]
next_page_layout_bbox = new_layout_of_pages[page_num]
is_conn= __connect_para_inter_page(pre_page_paras, next_page_paras, pre_page_layout_bbox, next_page_layout_bbox, lang)
if is_conn:
logger.info(f"连接了第{i-1}页和第{i}页的段落")
is_conn = __connect_para_inter_page(pre_page_paras, next_page_paras, pre_page_layout_bbox, next_page_layout_bbox, page_num, lang)
if debug_mode:
if is_conn:
logger.info(f"连接了第{page_num-1}页和第{page_num}页的段落")
is_list_conn = __connect_list_inter_page(pre_page_paras, next_page_paras, pre_page_layout_bbox, next_page_layout_bbox, all_page_list_info[page_num-1], all_page_list_info[page_num], page_num, lang)
if debug_mode:
if is_list_conn:
logger.info(f"连接了第{page_num-1}页和第{page_num}页的列表段落")
"""接下来可能会漏掉一些特别的一些可以合并的内容，对他们进行段落连接
1. 正文中有时出现一个行顶格，接下来几行缩进的情况。
2. 居中的一些连续单行，如果高度相同，那么可能是一个段落。
"""
for page_num, page in enumerate(pdf_info_dict.values()):
page_paras = page['para_blocks']
new_layout_bbox = new_layout_of_pages[page_num]
__connect_middle_align_text(page_paras, new_layout_bbox, page_num, lang, debug_mode=debug_mode)
__merge_signle_list_text(page_paras, new_layout_bbox, page_num, lang)
magic_pdf/pdf_parse_by_ocr.py
View file @ f0c463ed
......@@ -57,16 +57,16 @@ def construct_page_component(blocks, layout_bboxes, page_id, page_w, page_h, lay
def parse_pdf_by_ocr(
pdf_path,
s3_pdf_profile,
pdf_model_output,
save_path,
book_name,
pdf_model_profile=None,
image_s3_config=None,
start_page_id=0,
end_page_id=None,
debug_mode=False,
pdf_path,
s3_pdf_profile,
pdf_model_output,
save_path,
book_name,
pdf_model_profile=None,
image_s3_config=None,
start_page_id=0,
end_page_id=None,
debug_mode=False,
):
pdf_bytes = read_file(pdf_path, s3_pdf_profile)
save_tmp_path = os.path.join(os.path.dirname(__file__), "../..", "tmp", "unittest")
......@@ -95,7 +95,6 @@ def parse_pdf_by_ocr(
start_time = time.time()
end_page_id = end_page_id if end_page_id else len(pdf_docs) - 1
for page_id in range(start_page_id, end_page_id + 1):
......@@ -125,13 +124,6 @@ def parse_pdf_by_ocr(
page_id, page, ocr_page_info, md_bookname_save_path, debug_mode=debug_mode
)
# 构建需要remove的bbox列表
# need_remove_spans_bboxes = []
# need_remove_spans_bboxes.extend(page_no_bboxes)
# need_remove_spans_bboxes.extend(header_bboxes)
# need_remove_spans_bboxes.extend(footer_bboxes)
# need_remove_spans_bboxes.extend(footnote_bboxes)
# 构建需要remove的bbox字典
need_remove_spans_bboxes_dict = {
DropTag.PAGE_NUMBER: page_no_bboxes,
......@@ -199,50 +191,48 @@ def parse_pdf_by_ocr(
else:
continue
# 删除重叠spans中较小的那些
'''删除重叠spans中较小的那些'''
spans, dropped_spans_by_span_overlap = remove_overlaps_min_spans(spans)
# 删除remove_span_block_bboxes中的bbox
# spans = remove_spans_by_bboxes(spans, need_remove_spans_bboxes)
# 按qa要求，增加drop相关数据
'''
删除remove_span_block_bboxes中的bbox
并增加drop相关数据
'''
spans, dropped_spans_by_removed_bboxes = remove_spans_by_bboxes_dict(spans, need_remove_spans_bboxes_dict)
# 对image和table截图
'''对image和table截图'''
spans = cut_image_and_table(spans, page, page_id, book_name, save_path, img_s3_client)
# 行内公式调整, 高度调整至与同行文字高度一致(优先左侧, 其次右侧)
'''行内公式调整, 高度调整至与同行文字高度一致(优先左侧, 其次右侧)'''
displayed_list = []
text_inline_lines = []
modify_y_axis(spans, displayed_list, text_inline_lines)
# 模型识别错误的行间公式, type类型转换成行内公式
'''模型识别错误的行间公式, type类型转换成行内公式'''
spans = modify_inline_equation(spans, displayed_list, text_inline_lines)
# bbox去除粘连
'''bbox去除粘连'''
spans = remove_overlap_between_bbox(spans)
# 对tpye=["interline_equation", "image", "table"]进行额外处理,如果左边有字的话,将该span的bbox中y0调整至不高于文字的y0
'''
对tpye=["interline_equation", "image", "table"]进行额外处理,
如果左边有字的话,将该span的bbox中y0调整至不高于文字的y0
'''
spans = adjust_bbox_for_standalone_block(spans)
# 从ocr_page_info中解析layout信息(按自然阅读方向排序,并修复重叠和交错的bad case)
'''从ocr_page_info中解析layout信息(按自然阅读方向排序,并修复重叠和交错的bad case)'''
layout_bboxes, layout_tree = layout_detect(ocr_page_info['subfield_dets'], page, ocr_page_info)
# 将spans合并成line(在layout内,从上到下,从左到右)
'''将spans合并成line(在layout内,从上到下,从左到右)'''
lines, dropped_spans_by_layout = merge_spans_to_line_by_layout(spans, layout_bboxes)
# 将lines合并成block
'''将lines合并成block'''
blocks = merge_lines_to_block(lines)
# 根据block合并段落
#para_blocks = para_split(blocks, layout_bboxes)
# 获取QA需要外置的list
'''获取QA需要外置的list'''
images, tables, interline_equations, inline_equations = get_qa_need_list(blocks)
# drop的span_list合并
'''drop的span_list合并'''
dropped_spans = []
dropped_spans.extend(dropped_spans_by_span_overlap)
dropped_spans.extend(dropped_spans_by_removed_bboxes)
......@@ -263,19 +253,18 @@ def parse_pdf_by_ocr(
elif span['type'] in [ContentType.InlineEquation, ContentType.InterlineEquation]:
dropped_equation_block.append(span)
# 构造pdf_info_dict
'''构造pdf_info_dict'''
page_info = construct_page_component(blocks, layout_bboxes, page_id, page_w, page_h, layout_tree,
images, tables, interline_equations, inline_equations,
dropped_text_block, dropped_image_block, dropped_table_block, dropped_equation_block,
dropped_text_block, dropped_image_block, dropped_table_block,
dropped_equation_block,
need_remove_spans_bboxes_dict)
pdf_info_dict[f"page_{page_id}"] = page_info
"""分段"""
para_split(pdf_info_dict)
# 在测试时,保存调试信息
para_split(pdf_info_dict, debug_mode=debug_mode)
'''在测试时,保存调试信息'''
if debug_mode:
params_file_save_path = join_path(
save_tmp_path, "md", book_name, "preproc_out.json"
......
magic_pdf/pipeline.py
View file @ f0c463ed
......@@ -7,7 +7,7 @@ from magic_pdf.dict2md.ocr_mkcontent import (
ocr_mk_nlp_markdown,
ocr_mk_mm_markdown,
ocr_mk_mm_standard_format,
ocr_mk_mm_markdown_with_para, ocr_mk_mm_markdown_with_para_and_pagination,
ocr_mk_mm_markdown_with_para, ocr_mk_mm_markdown_with_para_and_pagination, ocr_mk_nlp_markdown_with_para,
)
from magic_pdf.libs.commons import (
read_file,
......@@ -510,7 +510,8 @@ def ocr_pdf_intermediate_dict_to_markdown_with_para(jso: dict, debug_mode=False)
pdf_intermediate_dict = jso["pdf_intermediate_dict"]
# 将 pdf_intermediate_dict 解压
pdf_intermediate_dict = JsonCompressor.decompress_json(pdf_intermediate_dict)
markdown_content = ocr_mk_mm_markdown_with_para(pdf_intermediate_dict)
# markdown_content = ocr_mk_mm_markdown_with_para(pdf_intermediate_dict)
markdown_content = ocr_mk_nlp_markdown_with_para(pdf_intermediate_dict)
jso["content"] = markdown_content
logger.info(
f"book_name is:{get_data_source(jso)}/{jso['file_id']},markdown content length is {len(markdown_content)}",
......
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