Merge pull request #35 from myhloli/master

将ocr_parse逻辑切换到v2，并解决几个parse过程中的error

magic_pdf/pre_proc/ocr_dict_merge.py

@@ -252,7 +252,7 @@ def fix_image_block(block, img_blocks):
                     break
 
             # 根据list长度，判断img_block中是否有img_caption
-            if len(img_block['img_caption_bbox']) > 0:
+            if img_block['img_caption_bbox'] is not None:
                 img_caption_block, img_caption_spans = merge_spans_to_block(
                     block['spans'], img_block['img_caption_bbox'], BlockType.ImageCaption
                 )
@@ -280,7 +280,7 @@ def fix_table_block(block, table_blocks):
                     break
 
             # 根据list长度，判断table_block中是否有caption
-            if len(table_block['table_caption_bbox']) > 0:
+            if table_block['table_caption_bbox'] is not None:
                 table_caption_block, table_caption_spans = merge_spans_to_block(
                     block['spans'], table_block['table_caption_bbox'], BlockType.TableCaption
                 )
@@ -293,7 +293,7 @@ def fix_table_block(block, table_blocks):
                         block['spans'].remove(span)
 
             # 根据list长度，判断table_block中是否有table_note
-            if len(table_block['table_footnote_bbox']) > 0:
+            if table_block['table_footnote_bbox'] is not None:
                 table_footnote_block, table_footnote_spans = merge_spans_to_block(
                     block['spans'], table_block['table_footnote_bbox'], BlockType.TableFootnote
                 )

magic_pdf/pre_proc/ocr_span_list_modify.py

@@ -222,10 +222,10 @@ def get_qa_need_list_v2(blocks):
     interline_equations = []
 
     for block in blocks:
-        if block["type"] == BlockType.Image:
+        if block["block_type"] == BlockType.Image:
             images.append(block)
-        elif block["type"] == BlockType.Table:
+        elif block["block_type"] == BlockType.Table:
             tables.append(block)
-        elif block["type"] == BlockType.InterlineEquation:
+        elif block["block_type"] == BlockType.InterlineEquation:
             interline_equations.append(block)
     return images, tables, interline_equations

magic_pdf/user_api.py

@@ -15,7 +15,7 @@
 from loguru import logger
 
 from magic_pdf.rw import AbsReaderWriter
-from magic_pdf.pdf_parse_by_ocr import parse_pdf_by_ocr
+from magic_pdf.pdf_parse_by_ocr_v2 import parse_pdf_by_ocr
 from magic_pdf.pdf_parse_by_txt import parse_pdf_by_txt