Compare commits
1 Commits
boyin_rag
...
frontier_2
| Author | SHA1 | Date | |
|---|---|---|---|
|
7243724300 |
@@ -697,15 +697,6 @@ def _merge_pdfs_ng(pdf1_path, pdf2_path, output_path):
|
|||||||
),
|
),
|
||||||
0,
|
0,
|
||||||
)
|
)
|
||||||
if "/Annots" in page1:
|
|
||||||
page1_annot_id = [annot.idnum for annot in page1["/Annots"]]
|
|
||||||
else:
|
|
||||||
page1_annot_id = []
|
|
||||||
|
|
||||||
if "/Annots" in page2:
|
|
||||||
page2_annot_id = [annot.idnum for annot in page2["/Annots"]]
|
|
||||||
else:
|
|
||||||
page2_annot_id = []
|
|
||||||
if "/Annots" in new_page:
|
if "/Annots" in new_page:
|
||||||
annotations = new_page["/Annots"]
|
annotations = new_page["/Annots"]
|
||||||
for i, annot in enumerate(annotations):
|
for i, annot in enumerate(annotations):
|
||||||
@@ -720,7 +711,8 @@ def _merge_pdfs_ng(pdf1_path, pdf2_path, output_path):
|
|||||||
if "/S" in action and action["/S"] == "/GoTo":
|
if "/S" in action and action["/S"] == "/GoTo":
|
||||||
# 内部链接：跳转到文档中的某个页面
|
# 内部链接：跳转到文档中的某个页面
|
||||||
dest = action.get("/D") # 目标页或目标位置
|
dest = action.get("/D") # 目标页或目标位置
|
||||||
if dest and annot.idnum in page2_annot_id:
|
# if dest and annot.idnum in page2_annot_id:
|
||||||
|
if dest in pdf2_reader.named_destinations:
|
||||||
# 获取原始文件中跳转信息，包括跳转页面
|
# 获取原始文件中跳转信息，包括跳转页面
|
||||||
destination = pdf2_reader.named_destinations[
|
destination = pdf2_reader.named_destinations[
|
||||||
dest
|
dest
|
||||||
@@ -732,24 +724,39 @@ def _merge_pdfs_ng(pdf1_path, pdf2_path, output_path):
|
|||||||
)
|
)
|
||||||
# 更新跳转信息，跳转到对应的页面和，指定坐标 (100, 150)，缩放比例为 100%
|
# 更新跳转信息，跳转到对应的页面和，指定坐标 (100, 150)，缩放比例为 100%
|
||||||
# “/D”:[10,'/XYZ',100,100,0]
|
# “/D”:[10,'/XYZ',100,100,0]
|
||||||
annot_obj["/A"].update(
|
if destination.dest_array[1] == "/XYZ":
|
||||||
{
|
annot_obj["/A"].update(
|
||||||
NameObject("/D"): ArrayObject(
|
{
|
||||||
[
|
NameObject("/D"): ArrayObject(
|
||||||
NumberObject(page_number),
|
[
|
||||||
destination.dest_array[1],
|
NumberObject(page_number),
|
||||||
FloatObject(
|
destination.dest_array[1],
|
||||||
destination.dest_array[2]
|
FloatObject(
|
||||||
+ int(
|
destination.dest_array[
|
||||||
page1.mediaBox.getWidth()
|
2
|
||||||
)
|
]
|
||||||
),
|
+ int(
|
||||||
destination.dest_array[3],
|
page1.mediaBox.getWidth()
|
||||||
destination.dest_array[4],
|
)
|
||||||
]
|
),
|
||||||
) # 确保键和值是 PdfObject
|
destination.dest_array[3],
|
||||||
}
|
destination.dest_array[4],
|
||||||
)
|
]
|
||||||
|
) # 确保键和值是 PdfObject
|
||||||
|
}
|
||||||
|
)
|
||||||
|
else:
|
||||||
|
annot_obj["/A"].update(
|
||||||
|
{
|
||||||
|
NameObject("/D"): ArrayObject(
|
||||||
|
[
|
||||||
|
NumberObject(page_number),
|
||||||
|
destination.dest_array[1],
|
||||||
|
]
|
||||||
|
) # 确保键和值是 PdfObject
|
||||||
|
}
|
||||||
|
)
|
||||||
|
|
||||||
rect = annot_obj.get("/Rect")
|
rect = annot_obj.get("/Rect")
|
||||||
# 更新点击坐标
|
# 更新点击坐标
|
||||||
rect = ArrayObject(
|
rect = ArrayObject(
|
||||||
@@ -773,7 +780,9 @@ def _merge_pdfs_ng(pdf1_path, pdf2_path, output_path):
|
|||||||
): rect # 确保键和值是 PdfObject
|
): rect # 确保键和值是 PdfObject
|
||||||
}
|
}
|
||||||
)
|
)
|
||||||
if dest and annot.idnum in page1_annot_id:
|
# if dest and annot.idnum in page1_annot_id:
|
||||||
|
if dest in pdf1_reader.named_destinations:
|
||||||
|
|
||||||
# 获取原始文件中跳转信息，包括跳转页面
|
# 获取原始文件中跳转信息，包括跳转页面
|
||||||
destination = pdf1_reader.named_destinations[
|
destination = pdf1_reader.named_destinations[
|
||||||
dest
|
dest
|
||||||
@@ -785,21 +794,36 @@ def _merge_pdfs_ng(pdf1_path, pdf2_path, output_path):
|
|||||||
)
|
)
|
||||||
# 更新跳转信息，跳转到对应的页面和，指定坐标 (100, 150)，缩放比例为 100%
|
# 更新跳转信息，跳转到对应的页面和，指定坐标 (100, 150)，缩放比例为 100%
|
||||||
# “/D”:[10,'/XYZ',100,100,0]
|
# “/D”:[10,'/XYZ',100,100,0]
|
||||||
annot_obj["/A"].update(
|
if destination.dest_array[1] == "/XYZ":
|
||||||
{
|
annot_obj["/A"].update(
|
||||||
NameObject("/D"): ArrayObject(
|
{
|
||||||
[
|
NameObject("/D"): ArrayObject(
|
||||||
NumberObject(page_number),
|
[
|
||||||
destination.dest_array[1],
|
NumberObject(page_number),
|
||||||
FloatObject(
|
destination.dest_array[1],
|
||||||
destination.dest_array[2]
|
FloatObject(
|
||||||
),
|
destination.dest_array[
|
||||||
destination.dest_array[3],
|
2
|
||||||
destination.dest_array[4],
|
]
|
||||||
]
|
),
|
||||||
) # 确保键和值是 PdfObject
|
destination.dest_array[3],
|
||||||
}
|
destination.dest_array[4],
|
||||||
)
|
]
|
||||||
|
) # 确保键和值是 PdfObject
|
||||||
|
}
|
||||||
|
)
|
||||||
|
else:
|
||||||
|
annot_obj["/A"].update(
|
||||||
|
{
|
||||||
|
NameObject("/D"): ArrayObject(
|
||||||
|
[
|
||||||
|
NumberObject(page_number),
|
||||||
|
destination.dest_array[1],
|
||||||
|
]
|
||||||
|
) # 确保键和值是 PdfObject
|
||||||
|
}
|
||||||
|
)
|
||||||
|
|
||||||
rect = annot_obj.get("/Rect")
|
rect = annot_obj.get("/Rect")
|
||||||
rect = ArrayObject(
|
rect = ArrayObject(
|
||||||
[
|
[
|
||||||
@@ -820,14 +844,12 @@ def _merge_pdfs_ng(pdf1_path, pdf2_path, output_path):
|
|||||||
elif "/S" in action and action["/S"] == "/URI":
|
elif "/S" in action and action["/S"] == "/URI":
|
||||||
# 外部链接：跳转到某个URI
|
# 外部链接：跳转到某个URI
|
||||||
uri = action.get("/URI")
|
uri = action.get("/URI")
|
||||||
|
|
||||||
output_writer.addPage(new_page)
|
output_writer.addPage(new_page)
|
||||||
# Save the merged PDF file
|
# Save the merged PDF file
|
||||||
with open(output_path, "wb") as output_file:
|
with open(output_path, "wb") as output_file:
|
||||||
output_writer.write(output_file)
|
output_writer.write(output_file)
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
def _merge_pdfs_legacy(pdf1_path, pdf2_path, output_path):
|
def _merge_pdfs_legacy(pdf1_path, pdf2_path, output_path):
|
||||||
import PyPDF2 # PyPDF2这个库有严重的内存泄露问题，把它放到子进程中运行，从而方便内存的释放
|
import PyPDF2 # PyPDF2这个库有严重的内存泄露问题，把它放到子进程中运行，从而方便内存的释放
|
||||||
|
|
||||||
|
|||||||
Reference in New Issue
Block a user