763_references_in_table

Author: vovanbravin

app/main/checks/report_checks/literature_references.py

@@ -87,27 +87,44 @@ def check(self):
                     '''
         return answer(False, result_str)
 
+    def search_references_in_table(self, index_table: int, prev_ref, array_of_references, ref_sequence) -> int:
+        """Функция поиска ссылок в таблицах (ищет в каждой клетке)"""
+        for cell in self.file.file.tables[index_table]._cells:
+            prev_ref =  self.search_references_in_text(cell.text, prev_ref, array_of_references, ref_sequence)
+        return prev_ref
+
+    def search_references_in_text(self, text: str, prev_ref, array_of_references, ref_sequence) -> int:
+        """Функция поиска ссылок в переданном тексте"""
+        reg_exp = r'\[[\^]{0,1}[\d \-,]+\]'
+        detected_references = re.findall(reg_exp, text)
+        if detected_references:
+            for reference_raw in detected_references:
+                reference = reference_raw.replace('^', '')  # TODO: kostyl'...
+                for one_part in re.split(r'[\[\],]', reference):
+                    if re.match(r'\d+[ \-]+\d+', one_part):
+                        start, end = re.split(r'[ -]+', one_part)
+                        for k in range(int(start), int(end) + 1):
+                            prev_ref = self.add_references(k, prev_ref, array_of_references, ref_sequence)
+                    elif one_part != '':
+                        prev_ref = self.add_references(int(one_part), prev_ref, array_of_references, ref_sequence)
+        return prev_ref
+
     def search_references(self, start_par):
+        """Функция поиска ссылок в документе"""
         prev_ref = 0
         ref_sequence = []
         array_of_references = set()
-        reg_exp = r'\[[\^]{0,1}[\d \-,]+\]'  # md can use [^5] format for hyperlink
         for i in range(0, start_par):
-            if isinstance(self.file.paragraphs[i], str):
-                detected_references = re.findall(reg_exp, self.file.paragraphs[i])
-            else:
-                detected_references = re.findall(reg_exp, self.file.paragraphs[i].paragraph_text)
-
-            if detected_references:
-                for reference_raw in detected_references:
-                    reference = reference_raw.replace('^', '')      # TODO: kostyl'...
-                    for one_part in re.split(r'[\[\],]', reference):
-                        if re.match(r'\d+[ \-]+\d+', one_part):
-                            start, end = re.split(r'[ -]+', one_part)
-                            for k in range(int(start), int(end) + 1):
-                                prev_ref = self.add_references(k, prev_ref, array_of_references, ref_sequence)
-                        elif one_part != '':
-                            prev_ref = self.add_references(int(one_part), prev_ref, array_of_references, ref_sequence)
+            text = self.file.paragraphs[i] if isinstance(self.file.paragraphs[i], str) else self.file.paragraphs[i].paragraph_text
+            match = re.search(r'Таблица ([.\d]+)', text)
+            if match:
+                index_table = int(match.group(1)) - 1
+                prev_ref = self.search_references_in_table(index_table, prev_ref, array_of_references, ref_sequence)
+                text = None
+
+            if text is not None:
+                prev_ref  = self.search_references_in_text(text, prev_ref, array_of_references, ref_sequence)
+
         if ref_sequence:
             if ref_sequence[0][1] == '0':
                 ref_sequence[0] = ref_sequence[0].replace('[0],', '')