11.01.26

app.py

@@ -58,21 +58,30 @@ file_handler.setFormatter(logging.Formatter(
 class AHKParser:
     lab = "AHK"
 
-    def grab(self, text, labels):
+    def extract_table(self, text, headers):
         lines = [l.strip() for l in text.splitlines() if l.strip()]
-        idx = [i for i,l in enumerate(lines) if l in labels]
-        if not idx:
-            return {}
+        out = {}
+        for h in headers:
+            for i,l in enumerate(lines):
+                if l == h:
+                    for j in range(i+1, i+8):
+                        if j < len(lines) and lines[j].startswith(":"):
+                            out[h] = lines[j][1:].strip()
+                            break
+        return out
 
-        values = []
-        start = idx[-1] + 1
-        for l in lines[start:]:
-            if l.startswith(":"):
-                values.append(l[1:].strip())
-            if len(values) == len(labels):
-                break
-
-        return dict(zip(labels, values))
+    def extract_weights(self, text):
+        lines = [l.strip() for l in text.splitlines() if l.strip()]
+        res = {}
+        for i,l in enumerate(lines):
+            if l == "Bales Weighed":
+                headers = ["Bales","Gross","Tare","Net"]
+                for h in headers:
+                    for j in range(i, i+20):
+                        if j < len(lines) and lines[j].startswith(":"):
+                            res[h] = lines[j][1:].replace("kg","").strip()
+                            break
+        return res
 
     def parse(self, text):
         r = empty_weight_report("AHK")
@@ -85,14 +94,14 @@ class AHKParser:
         r["contract"]["invoice_no"] = safe_search(r"Client Reference:\s*([A-Z0-9\- /]+)", text)
         r["contract"]["commodity"]  = "Raw Cotton"
 
-        # parties
+        # buyer
         r["parties"]["buyer"] = safe_search(r"Buyer:\s*([A-Z0-9 ().,-]+)", text)
 
-        # shipment block
-        ship = self.grab(text, [
+        # shipment tables
+        ship = self.extract_table(text, [
             "Total Bales","Vessel","Voy. No.","B/L No.","B/L Date","Destination"
         ])
-        ship2 = self.grab(text, [
+        ship2 = self.extract_table(text, [
             "Growth","Arrival Date","First date of weighing",
             "Last Date of Weighing","Weighing method","Tare"
         ])
@@ -106,8 +115,8 @@ class AHKParser:
         r["contract"]["origin"]           = ship2.get("Growth")
 
         # weights
-        inv  = self.grab(text, ["Bales","Gross","Tare","Net"])
-        land = self.grab(section(text,"Bales Weighed","Outturn"),["Bales","Gross","Tare","Net"])
+        inv  = self.extract_table(text, ["Bales","Gross","Tare","Net"])
+        land = self.extract_weights(text)
 
         r["weights"]["invoice_net_kg"]  = to_float(inv.get("Net"))
         r["weights"]["gross_landed_kg"] = to_float(land.get("Gross"))