outdated

data.py

@@ -1,11 +1,14 @@
 class BlowerdoorData:
-    def __init__(self, path, docName, location, customer, pdfDate, blowerdoorDate):
+    def __init__(self, path, docName, location, customer, pdfDate, blowerdoorDate, inDocumentDate=None):
         self.path = path
         self.docName = docName
         self.location = location
         self.customer = customer
         self.blowerdoorDate = blowerdoorDate
         self.pdfDate = pdfDate
+        self.inDocumentDate = inDocumentDate
+
+        self.outdated = False
 
     
 #print("Bauort: " + location)

eg_geiss_bauherren.py

@@ -14,22 +14,47 @@ 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)

server.py

@@ -15,12 +15,26 @@ def root():
     allFiles = []
     loaded = None
     for filename in glob.glob("static/files/*.pdf"):
+        loaded = parserBackend.load(filename)
         try:
             loaded = parserBackend.load(filename)
         except Exception:
             loaded = BlowerdoorData(os.path.basename(filename), os.path.basename(filename), "", "", datetime.datetime.now(), datetime.datetime.now())
         allFiles.append(loaded)
 
+    # check duplicates
+    duplicateCheckMap = dict()
+    for f in allFiles:
+        if f.inDocumentDate:
+            duplicateCheckMap.update({ f.customer + f.location : f })
+    
+    for f in allFiles:
+        key = f.customer + f.location
+        if key in duplicateCheckMap and not f is duplicateCheckMap[key]:
+            if f.inDocumentDate <= duplicateCheckMap[key].inDocumentDate:
+                f.outdated = True
+
+
     return flask.render_template("index.html", listContent=allFiles)
 
 @app.route("/get-file")

templates/index.html

@@ -21,7 +21,9 @@
         <tbody>
             {% for bd in listContent %}
             <tr>
-               <td style="line-height: 45px;"><a target="_blank" href="/get-file?basename={{ bd.docName }}">{{ bd.docName }}</a></td>
+               <td style="line-height: 45px;"><a target="_blank" href="/get-file?basename={{ bd.docName }}">{{ bd.docName }}</a>
+                {% if bd.outdated %}<p style="color: red;">(älter: {{ bd.inDocumentDate.strftime("%d.%m.%Y") }})</p>{% endif %}
+               </td>
                <td style="line-height: 45px;">{{ bd.location }}</td>
                {% if bd.blowerdoorDate %}
                <td style="line-height: 45px;">{{ bd.blowerdoorDate }}</td>