merge license

@@ -33,6 +33,5 @@ data/*
 # 脚本
 src/*.sh
 
-test.py
-ocr_test.py
-ocr_test_2.py
\ No newline at end of file
+test*
+ocr_test.py
\ No newline at end of file

+class EDMSException(Exception):
+    pass

@@ -141,32 +141,22 @@ class BSWorkbook(Workbook):
                 # month_info process
                 month_info = month_mapping.setdefault('xxxx-xx', [])
                 month_info.append((ws.title, min_row, ws.max_row, 0))
-            elif len(month_list) == 1:
-                # reverse_trend_list process
-                reverse_trend = self.get_reverse_trend(dti.day, idx_list)
-                reverse_trend_list.append(reverse_trend)
-                # month_info process
-                month_info = month_mapping.setdefault(month_list[0], [])
-                day_mean = np.mean(dti.day.dropna())
-                if len(month_info) == 0:
-                    month_info.append((ws.title, min_row, ws.max_row, day_mean))
-                else:
-                    for i, item in enumerate(month_info):
-                        if day_mean <= item[-1]:
-                            month_info.insert(i, (ws.title, min_row, ws.max_row, day_mean))
-                            break
-                    else:
-                        month_info.append((ws.title, min_row, ws.max_row, day_mean))
             else:
                 # reverse_trend_list process
                 reverse_trend = self.get_reverse_trend(dti.day, idx_list)
                 reverse_trend_list.append(reverse_trend)
                 # month_info process
-                for i, item in enumerate(month_list[:-1]):
-                    month_mapping.setdefault(item, []).append(
-                        (ws.title, idx_list[i] + min_row, idx_list[i + 1] + min_row - 1, self.MAX_MEAN))
-                month_mapping.setdefault(month_list[-1], []).insert(
-                    0, (ws.title, idx_list[-1] + min_row, ws.max_row, 0))
+                day_idx = dti.day
+                idx_list_max_idx = len(idx_list) - 1
+                for i, item in enumerate(month_list):
+                    if i == idx_list_max_idx:
+                        day_mean = np.mean(day_idx[idx_list[i]:].dropna())
+                        month_mapping.setdefault(item, []).append(
+                            (ws.title, idx_list[i] + min_row, ws.max_row, day_mean))
+                    else:
+                        day_mean = np.mean(day_idx[idx_list[i]: idx_list[i + 1]].dropna())
+                        month_mapping.setdefault(item, []).append(
+                            (ws.title, idx_list[i] + min_row, idx_list[i + 1] + min_row - 1, day_mean))
 
     def build_metadata_rows(self, confidence, code, print_time, start_date, end_date):
         if start_date is None or end_date is None:
@@ -191,9 +181,9 @@ class BSWorkbook(Workbook):
     def create_meta_sheet(self, card):
         if self.worksheets[0].title == 'Sheet':
             ms = self.worksheets[0]
-            ms.title = '{0}({1})'.format(self.meta_sheet_title, card)
+            ms.title = '{0}({1})'.format(self.meta_sheet_title, card[-6:])
         else:
-            ms = self.create_sheet('{0}({1})'.format(self.meta_sheet_title, card))
+            ms = self.create_sheet('{0}({1})'.format(self.meta_sheet_title, card[-6:]))
         return ms
 
     def build_meta_sheet(self, card, confidence, code, print_time, start_date, end_date):
@@ -203,6 +193,26 @@ class BSWorkbook(Workbook):
             ms.append(row)
         return ms
 
+    @staticmethod
+    def amount_format(amount_str):
+        if not isinstance(amount_str, str) or amount_str == '':
+            return amount_str
+        # 1.替换
+        res_str = amount_str.translate(consts.TRANS)
+        # 2.删除多余的-
+        res_str = res_str[0] + res_str[1:].replace('-', '')
+        # 3.首字符处理
+        if res_str[0] in consts.ERROR_CHARS:
+            res_str = '-{0}'.format(res_str[1:])
+        # 4.逗号与句号处理
+        if len(res_str) >= 4:
+            period_idx = len(res_str) - 3
+            if res_str[period_idx] == '.' and res_str[period_idx - 1] == ',':
+                res_str = '{0}{1}'.format(res_str[:period_idx - 1], res_str[period_idx:])
+            elif res_str[period_idx] == ',':
+                res_str = '{0}.{1}'.format(res_str[:period_idx], res_str[period_idx + 1:])
+        return res_str
+
     def build_month_sheet(self, card, month_mapping, ms, is_reverse):
         tmp_ws = self.create_sheet('tmp_ws')
         for month in sorted(month_mapping.keys()):
@@ -235,29 +245,25 @@ class BSWorkbook(Workbook):
                 # 3.3.余额转数值
                 over_cell = rows[consts.OVER_IDX]
                 try:
-                    if isinstance(over_cell.value, str):
-                        over_cell.value = over_cell.value.translate(consts.TRANS)
-                    over_cell.value = locale.atof(over_cell.value)
+                    over_cell.value = locale.atof(self.amount_format(over_cell.value))
                 except Exception as e:
                     continue
                 else:
                     over_cell.number_format = numbers.FORMAT_NUMBER_COMMA_SEPARATED1
 
-                # 3.4.余额转数值
+                # 3.4.金额转数值
                 try:
                     try:
-                        if isinstance(amount_cell.value, str):  # TODO 可在转化数字失败后，再替换
-                            amount_cell.value = amount_cell.value.translate(consts.TRANS)
-                        amount_cell.value = locale.atof(amount_cell.value)
+                        amount_cell.value = locale.atof(self.amount_format(amount_cell.value))
                     except Exception as e:
                         try:
-                            if isinstance(rows[consts.INCOME_IDX].value, str):
-                                rows[consts.OUTLAY_IDX].value = rows[consts.INCOME_IDX].value.translate(consts.TRANS)
-                            amount_cell.value = locale.atof(rows[consts.OUTLAY_IDX].value)
+                            amount_cell.value = locale.atof(self.amount_format(rows[consts.INCOME_IDX].value))
+                            if amount_cell.value == 0:
+                                raise
+                            elif amount_cell.value < 0:
+                                amount_cell.value = -amount_cell.value
                         except Exception as e:
-                            if isinstance(rows[consts.OUTLAY_IDX].value, str):
-                                rows[consts.OUTLAY_IDX].value = rows[consts.OUTLAY_IDX].value.translate(consts.TRANS)
-                            amount_cell.value = locale.atof(rows[consts.OUTLAY_IDX].value)
+                            amount_cell.value = locale.atof(self.amount_format(rows[consts.OUTLAY_IDX].value))
                             if amount_cell.value > 0:
                                 amount_cell.value = -amount_cell.value
                 except Exception as e:
@@ -313,18 +319,18 @@ class BSWorkbook(Workbook):
         # }
         for card, summary in bs_summary.items():
             # 1.原表修剪、排列、按照月份分割
-            start_date = summary['start_date']
-            end_date = summary['end_date']
+            start_date = summary.get('start_date')
+            end_date = summary.get('end_date')
             date_statistics = False
             if start_date is None or end_date is None:
                 date_statistics = True
             date_list = []
             month_mapping = {}
             reverse_trend_list = []
-            for sheet in summary['sheet']:
+            for sheet in summary.get('sheet', []):
                 ws = self.get_sheet_by_name(sheet)
                 # 1.1.删除多余列、排列
-                min_row = self.sheet_prune(ws, summary['classify'])
+                min_row = self.sheet_prune(ws, summary.get('classify', 0))
                 # 1.2.按月份分割
                 self.sheet_split(ws, month_mapping, reverse_trend_list, min_row, date_list, date_statistics)
 
@@ -334,32 +340,43 @@ class BSWorkbook(Workbook):
 
             # 2.元信息提取表
             ms = self.build_meta_sheet(card,
-                                       summary['confidence'],
-                                       summary['code'],
-                                       summary['print_time'],
+                                       summary.get('confidence', 1),
+                                       summary.get('code'),
+                                       summary.get('print_time'),
                                        start_date,
                                        end_date)
 
             # 3.创建月份表、提取/高亮关键行
-            is_reverse = False
-            if sum(reverse_trend_list) > 0:  # 倒序处理
-                is_reverse = True
-                for month_list in month_mapping.values():
-                    month_list.sort(key=lambda x: x[-1], reverse=True)
+            # 倒序处理
+            is_reverse = True if sum(reverse_trend_list) > 0 else False
+            for month_list in month_mapping.values():
+                month_list.sort(key=lambda x: x[-1], reverse=is_reverse)
+
             self.build_month_sheet(card, month_mapping, ms, is_reverse)
 
             # 4.删除原表
-            for sheet in summary['sheet']:
+            for sheet in summary.get('sheet'):
                 self.remove(self.get_sheet_by_name(sheet))
 
     def license_rebuild(self, license_summary):
-        for en_key, cn_key in consts.LICENSE_ORDER:
-            ws = self.create_sheet(cn_key)
-            for bl in license_summary.get(en_key, []):
+        for classify, (_, name) in consts.LICENSE_ORDER:
+            res = license_summary.get(classify)
+            if res is None:
+                continue
+            ws = self.create_sheet(name)
+            for bl in res:
                 for bl_field in bl:
                     ws.append(bl_field)
                 ws.append((None, ))
 
-    def rebuild(self, bs_summary, license_summary):
+    def skip_img_sheet(self, skip_img):
+        if skip_img:
+            ws = self.create_sheet(consts.SKIP_IMG_SHEET_NAME)
+            ws.append(consts.SKIP_IMG_SHEET_HEADER)
+            for img_tuple in skip_img:
+                ws.append(img_tuple)
+
+    def rebuild(self, bs_summary, license_summary, skip_img):
         self.bs_rebuild(bs_summary)
-        # self.license_rebuild(license_summary)
+        self.license_rebuild(license_summary)
+        self.skip_img_sheet(skip_img)

@@ -25,7 +25,7 @@ class PDFHandler:
     def __init__(self, path, img_dir_path):
         self.path = path
         self.img_dir_path = img_dir_path
-        self.img_info_list = []
+        self.img_path_list = []
         self.xref_set = set()
 
     def get_img_save_path(self, pno, img_index=0, ext='png'):
@@ -38,7 +38,7 @@ class PDFHandler:
             pm = page.getPixmap(matrix=trans_2, alpha=False)
         img_save_path = self.get_img_save_path(page.number)
         pm.writePNG(img_save_path)
-        self.img_info_list.append((img_save_path, page.number, 0))
+        self.img_path_list.append(img_save_path)
 
     @staticmethod
     def getimage(pix):
@@ -88,7 +88,7 @@ class PDFHandler:
         with open(img_save_path, "wb") as f:
             f.write(img_data)
         self.xref_set.add(xref)
-        self.img_info_list.append((img_save_path, pno, img_index))
+        self.img_path_list.append(img_save_path)
 
     @staticmethod
     def split_il(il):
@@ -179,7 +179,7 @@ class PDFHandler:
                     img_save_path = self.get_img_save_path(pno, img_index, im_list[0][2])
                     new_img.save(img_save_path)
                     page_to_png = False
-                    self.img_info_list.append((img_save_path, pno, img_index))
+                    self.img_path_list.append(img_save_path)
 
         # 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: