From 4144d6a21b1364b7150d29859477a5807b6b48df Mon Sep 17 00:00:00 2001 From: chenyao <chenyao@situdata.com> Date: Fri, 17 Jan 2025 16:29:04 +0800 Subject: [PATCH] 将切图恢复为小图 --- src/common/tools/pdf_to_img.py | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/src/common/tools/pdf_to_img.py b/src/common/tools/pdf_to_img.py index e463334..4c68c22 100644 --- a/src/common/tools/pdf_to_img.py +++ b/src/common/tools/pdf_to_img.py @@ -345,7 +345,7 @@ class PDFHandler: # 3.3 碎图分组大于2、全过滤、含特殊filter,特殊处理:整个页面保存为png图片 if page_to_png: page = pdf.loadPage(pno) - self.page_to_png(page, is_big_img=True) + self.page_to_png(page) def title_is_ebank(self, char): new_char = normalize('NFKC', char) @@ -450,7 +450,7 @@ class PDFHandler: # 1.页面图片对象数目为0时,保存整个页面为png图片 if self.is_e_pdf or self.is_ebank or len(il) == 0: page = pdf.loadPage(pno) - self.page_to_png(page, is_big_img=True) + self.page_to_png(page) # 2.页面图片对象数目为1时: # 小图(如电子账单的盖章):保存整个页面为png图片 # 大图:提取图片对象 @@ -459,13 +459,13 @@ class PDFHandler: # 小图 if width < WH_COUPLE_1[0] and height < WH_COUPLE_1[1]: page = pdf.loadPage(pno) - self.page_to_png(page, is_big_img=True) + self.page_to_png(page) # 大图 elif width >= WH_COUPLE_6[0] or height >= WH_COUPLE_6[1]: self.is_new_modify = 1 is_big_img = (width < WH_COUPLE_7[0] and height < WH_COUPLE_7[1]) # 防止图片过大 page = pdf.loadPage(pno) - self.page_to_png(page, is_big_img=True) + self.page_to_png(page, is_big_img=is_big_img) elif xref not in self.xref_set: self.extract_single_image(pdf, xref, smask, colorspace, pno) # 3.页面图片对象数目大于1时,特殊处理 @@ -480,7 +480,7 @@ class PDFHandler: with fitz.Document(self.path) as pdf: for pno in range(pdf.pageCount): page = pdf.loadPage(pno) - self.page_to_png(page, is_big_img=True) + self.page_to_png(page) self.img_count = len(self.img_path_list) def ebank_draw(self): -- libgit2 0.24.0