match therion files to wallets, scans names therein

2 files changed, 40 insertions, 5 deletions

diff --git a/parsers/drawings.py b/parsers/drawings.py
index 4b3e44d..1c5a1e8 100644
--- a/parsers/drawings.py
+++ b/parsers/drawings.py
@@ -29,7 +29,7 @@ def find_dwg_file(dwgfile, path):
     mscansdir = re.search(r"(\d\d\d\d#X?\d+\w?|1995-96kh|92-94Surveybookkh|1991surveybook|smkhs)/(.*?(?:png|jpg|pdf|jpeg|gif|txt))$", path)
     if mscansdir:
         scanswalletl = Wallet.objects.filter(walletname=mscansdir.group(1))
-        # This should be chnaged to properly detect if a list of folders is returned and do something sensible, not just pick the first.
+        # This should be changed to properly detect if a list of folders is returned and do something sensible, not just pick the first.
         if len(scanswalletl):
             wallet = scanswalletl[0]
             if len(scanswalletl) > 1:
@@ -70,10 +70,43 @@ def find_dwg_file(dwgfile, path):
 
     dwgfile.save()
 
-def findimageinsert(therionfile, xth_me):
+def findwalletimage(therionfile, foundpath):
     '''Tries to link the drawing file (Therion format) to the referenced image (scan) file
     '''
-    pass
+    foundpath = foundpath.strip("{}")
+    mscansdir = re.search(r"(\d\d\d\d#\d+\w?|1995-96kh|92-94Surveybookkh|1991surveybook|smkhs)", foundpath)
+    if mscansdir:
+        scanswalletl = Wallet.objects.filter(walletname=mscansdir.group(1))
+        # This should be changed to properly detect if a list of folders is returned and do something sensible, not just pick the first.
+        if len(scanswalletl):
+            wallet = scanswalletl[0]
+            if len(scanswalletl) > 1:
+                message = "! More than one scan FOLDER matches filter query. [{}]: {} {} {}".format(therionfile, mscansdir.group(1), foundpath)
+                print(message)
+                DataIssue.objects.create(parser='Therion', message=message)
+        if wallet:
+            therionfile.dwgwallets.add(wallet)
+            
+            scanfilename = Path(foundpath).name
+            scansfilel = wallet.singlescan_set.filter(name=scanfilename, wallet=wallet)
+            if len(scansfilel):
+                # message = f'! {len(scansfilel)} {scansfilel} = {scanfilename} found in the wallet specified {wallet.walletname}'
+                # print(message)
+                if len(scansfilel) > 1:
+                    plist =[]
+                    for sf in scansfilel:
+                        plist.append(sf.ffile)
+                    message = "! More than one image FILENAME matches filter query. [{}]: {} {} {} {} {}".format(scansfilel[0], mscansdir.group(1), mscansdir.group(2), dwgfile.dwgpath, path, plist)
+                    print(message)
+                    DataIssue.objects.create(parser='Therion', message=message)
+                scansfile = scansfilel[0]
+                therionfile.scans.add(scansfile)
+            else:
+                message = f'! Scanned file {scanfilename} mentioned in "{therionfile.dwgpath}" is not actually found in {wallet.walletname}'
+                wurl = f'/survey_scans/{wallet.walletname}/'.replace("#",":")
+                print(message)
+                DataIssue.objects.create(parser='Therion', message=message, url = wurl)
+
 
 def findimportinsert(therionfile, imp):
     '''Tries to link the scrap (Therion format) to the referenced therion scrap
@@ -120,10 +153,12 @@ def settherionfileinfo(filetuple):
         message = f'! Un-parsed image filename: {therionfile.dwgname} : {xth_me.split()[-3]} - {therionfile.dwgpath}'
         # print(message)
         # DataIssue.objects.create(parser='xTherion', message=message, url=f'/dwgdataraw/{therionfile.dwgpath}')
+        # ! Un-parsed image filename: 107coldest : ../../../expofiles/surveyscans/2015/2015#20/notes.jpg - therion/plan/107coldest.th2
+
         with open('therionrefs.log', 'a') as lg:
             lg.write(message + '\n')
 
-        findimageinsert(therionfile, xth_me)
+        findwalletimage(therionfile, xth_me.split()[-3])
         
     for inp in rx_input.findall(ttext):
         # if this 'input' is a .th2 file we have already seen, then we can assign this as a sub-file
diff --git a/parsers/scans.py b/parsers/scans.py
index 28d0325..6439549 100644
--- a/parsers/scans.py
+++ b/parsers/scans.py
@@ -93,7 +93,7 @@ def LoadListScansFile(wallet):
     
     c=0
     for (fyf, ffyf, fisdiryf) in gld:
-        if re.search(r"\.(?:png|jpg|jpeg|pdf|svg|gif)(?i)$", fyf):
+        if re.search(r"\.(?:png|jpg|jpeg|pdf|svg|gif|xvi)(?i)$", fyf):
             singlescan = SingleScan(ffile=ffyf, name=fyf, wallet=wallet)
             singlescan.save()
             c+=1