Merge branch 'master' of github.com:FAUSheppy/ths-blowerdoor-raven

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outdated = False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

@@ -7,31 +7,58 @@ import os.path
 BLOCK_TUP_TEXT = 4
 
 def load(filename):
+    print(filename)
     doc = fitz.open(filename)
     FIRST_P = True
 
     # 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]
 
+            kwErrorInfo = None
             if "Thermoscan" in text:
                 kwParts = text.split("\n")
                 kw = ""
@@ -42,7 +69,10 @@ def load(filename):
                     if not pClean:
                         continue
                     elif not kw:
-                        kw = int(pClean.split(". KW")[0])
+                        try:
+                            kw = int(pClean.split(". KW")[0])
+                        except ValueError:
+                            kwErrorInfo = "Kalenderwochen Info nicht gefunden."
                     elif not title:
                         title = pClean
                     elif not contractor:
@@ -50,10 +80,14 @@ def load(filename):
 
                 ISO_CAL_KW_LOC = 1
                 kwStartDate = startDateParsed.isocalendar()[ISO_CAL_KW_LOC]
-                if kw < kwStartDate:
+
-                    blowerdoorDate = "{} KW-{:02d}".format(startDateParsed.year +1, kw)
+                if kwErrorInfo:
+                    blowerdoorDate = None
                 else:
-                    blowerdoorDate = "{} KW-{}".format(startDateParsed.year, kw)
+                    if kw < kwStartDate:
+                        blowerdoorDate = "{} KW-{:02d}".format(startDateParsed.year +1, kw)
+                    else:
+                        blowerdoorDate = "{} KW-{}".format(startDateParsed.year, kw)
 
 
 
@@ -63,4 +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

@@ -2,6 +2,9 @@ import flask
 import argparse
 import glob
 import os
+from data import BlowerdoorData
+import datetime
+import os.path
 
 import eg_geiss_bauherren as parserBackend
 
@@ -10,9 +13,28 @@ app = flask.Flask("THS-Raven")
 @app.route("/")
 def root():
     allFiles = []
+    loaded = None
     for filename in glob.glob("static/files/*.pdf"):
-        allFiles.append(parserBackend.load(filename))
+        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

@@ -21,9 +21,15 @@
         <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;">(neueres Dokument verfügbar: {{ 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>
+               {% else %}
+               <td style="line-height: 45px;">0000 - Keine Informationen Gefunden</td>
+               {% endif %}
                <td style="line-height: 45px;">{{ bd.customer }}</td>
                <td style="line-height: 45px;">{{ bd.pdfDate.strftime("%Y/%m/%d") }}</td>
            </tr>