outdated

eg_geiss_bauherren.py

@@ -14,21 +14,46 @@ def load(filename):
     # pop vars
     customer = "NOT_FOUND"
     location = ""
+    inDocumentDate = None
     startDate = doc.metadata["creationDate"].split("D:")[1].split("+")[0]
     startDateParsed = dateutil.parser.parse(startDate)
     blowerdoorDate = "NOT_FOUND"
 
+    datumNext = False
+    page = -1
     for p in doc:
+        page += 1
         blocks = p.get_text("blocks")
         for i in range(0, len(blocks)):
         
+
             text = blocks[i][BLOCK_TUP_TEXT]
         
             textNoSpaceNewline = text.replace("\n", "")
             textNoSpaceNewline = textNoSpaceNewline.replace(" ", "")
+
+            
+            if datumNext and page == 0:
+                try:
+                    #if "Bauablaufplan11.pdf" in filename:
+                    #    print(textNoSpaceNewline)
+                    inDocumentDate= dateutil.parser.parse(textNoSpaceNewline)
+                    datumNext = False
+                except ValueError:
+                    try:
+                        split = textNoSpaceNewline.split(".de")[1]
+                        inDocumentDate = dateutil.parser.parse(split)
+                    except ValueError:
+                        pass
+                    except IndexError:
+                        pass
+
             if FIRST_P and i < 3 and textNoSpaceNewline:
                 FIRST_P = False
                 customer = text
+
+            if "Datum:" in text:
+                datumNext = True
             
             if "Bauort:" in text:
                 location += text.split("Bauort:")[1]
@@ -72,4 +97,4 @@ def load(filename):
 
     filename = filename.replace("\\","/")
     return data.BlowerdoorData(filename, os.path.basename(filename), location, 
-                                    customer, startDateParsed, blowerdoorDate)
+                                    customer, startDateParsed, blowerdoorDate, inDocumentDate)