ebank part 2

@@ -39,6 +39,14 @@ class PDFHandler:
         self.page_text_list = []
         self.page_text_list = []
         self.pdf_info = {}
         self.pdf_info = {}
         self.img_path_pno_list = []
         self.img_path_pno_list = []
+        self.ebank_title_list = [
+            '微信支付交易明细证明',
+            '支付宝收支明细证明',
+            '招商银行交易流水',
+            '中国工商银行借记账户历史明细',
+            '中国建设银行个人活期账户全部交易明细',
+            '平安银行个人账户交易明细清单',
+        ]
 
 
     def get_suffix(self, file_name):
     def get_suffix(self, file_name):
         if file_name is None:
         if file_name is None:
@@ -260,20 +268,27 @@ class PDFHandler:
             page = pdf.loadPage(pno)
             page = pdf.loadPage(pno)
             self.page_to_png(page)
             self.page_to_png(page)
 
 
+    def title_is_ebank(self, char):
+        for title in self.ebank_title_list:
+            if title.find(char) != -1 or char.find(title) != -1:
+                return True
+        return False
+
     def check_ebank(self, pdf):
     def check_ebank(self, pdf):
-        # page_text_list = []
+        page_text_list = []
         text_item_sum = 0
         text_item_sum = 0
+        in_ebank_set = False
         for pno in range(pdf.pageCount):
         for pno in range(pdf.pageCount):
             page = pdf.loadPage(pno)
             page = pdf.loadPage(pno)
-            # if page.rotation is None:
-            #     rotation = 0
-            # elif isinstance(page.rotation, int):
-            #     divisor, remainder = divmod(page.rotation, 90)
-            #     if remainder != 0:
-            #         return
-            #     rotation = divmod(divisor, 4)[1]
-            # else:
-            #     return
+            if page.rotation is None:
+                rotation = 0
+            elif isinstance(page.rotation, int):
+                divisor, remainder = divmod(page.rotation, 90)
+                if remainder != 0:
+                    return
+                rotation = divmod(divisor, 4)[1]
+            else:
+                return
             textpage = page.getTextPage()
             textpage = page.getTextPage()
             text = textpage.extractDICT()
             text = textpage.extractDICT()
             text_list = []
             text_list = []
@@ -284,22 +299,24 @@ class PDFHandler:
                         bbox = span.get('bbox')
                         bbox = span.get('bbox')
                         if char.strip() == '':
                         if char.strip() == '':
                             continue
                             continue
+                        if pno == 0 and self.title_is_ebank(char):
+                            in_ebank_set = True
                         text_list.append((bbox, char))
                         text_list.append((bbox, char))
             text_item_sum += len(text_list)
             text_item_sum += len(text_list)
             if text_item_sum < (pno + 1) * 5:
             if text_item_sum < (pno + 1) * 5:
                 return
                 return
-            # else:
-            #     page_text_list.append(
-            #         {
-            #             'width': text.get('width'),
-            #             'height': text.get('height'),
-            #             'rotation': rotation,
-            #             'text': text_list
-            #         }
-            #     )
-        # self.is_ebank = True
+            else:
+                page_text_list.append(
+                    {
+                        'width': text.get('width'),
+                        'height': text.get('height'),
+                        'rotation': rotation,
+                        'text': text_list
+                    }
+                )
+        self.is_ebank = in_ebank_set
         self.is_e_pdf = True
         self.is_e_pdf = True
-        # self.page_text_list = page_text_list
+        self.page_text_list = page_text_list
 
 
     def e_contract_process(self):
     def e_contract_process(self):
         os.makedirs(self.img_dir_path, exist_ok=True)
         os.makedirs(self.img_dir_path, exist_ok=True)