1 files changed, 92 insertions, 0 deletions
diff --git a/ishtar_common/utils.py b/ishtar_common/utils.py
index 09c470823..709f020a4 100644
--- a/ishtar_common/utils.py
+++ b/ishtar_common/utils.py
@@ -1362,6 +1362,98 @@ def create_slug(model, name, slug_attr="slug", max_length=100, pk=None):
     return final_slug
 
 
+def _get_parse_string(trunc_number=None):
+    def parse_string(value):
+        value = value.strip()
+        if value == "#EMPTY":
+            value = ""
+        value = value.replace("  ", " ")
+        if trunc_number:
+            value = value[:trunc_number]
+        return value
+
+    return parse_string
+
+
+parse_string = _get_parse_string()
+
+
+def parse_insee(value):
+    value = parse_string(value)
+    values = []
+    while len(value) > 4:
+        values.append(value[:5])
+        value = value[5:]
+    towns = []
+    Town = apps.get_model("ishtar_common", "Town")
+    for value in values:
+        try:
+            town = Town.objects.get(numero_insee=value)
+        except Town.DoesNotExist:
+            # sys.stderr.write('Numero INSEE : %s non existant en base'
+            # % value)
+            continue
+        towns.append(town)
+    return towns
+
+
+PARCEL_YEAR_REGEXP = re.compile(r"^([0-9]{4})[ :]+")
+PARCEL_SECTION_REGEXP = re.compile(
+    r"(?: )*(?:[Ss]ection(?:s)?)?(?: )*([A-Z][A-Z0-9]{0,3})[ :]*"
+    r"((?:(?: |;|,|[Pp]arcelle(?:s)?|n°|et|à|to)*[0-9]+[p]?)+)"
+)
+PARCEL_NB_RANGE_REGEXP = re.compile(r"([0-9]+[p]?) (?:à|to) ([0-9]+[p]?)")
+PARCEL_NB_REGEXP = re.compile(r"(?: |;|,|[Pp]arcelle(?:s)?|n°|et|à|to)*([0-9]+[p]?)")
+
+
+def parse_parcels(parcel_str, insee_code=None, owner=None):
+    parcels, town = [], None
+    if insee_code:
+        town = parse_insee(insee_code)
+        # manage only one town at a time
+        if len(town) >= 2 or not town:
+            return parcels
+        town = town[0]
+    parcel_str = parcel_str.strip().replace(
+        "\ufe50", ",").replace("\uff0c", ",").replace("\n", " ")
+    parcel_str = re.sub(r'\s+', ' ', parcel_str)
+    parcel_str = parcel_str.replace("à", "_aaaa_").replace("n°", "_nnnn_")
+    parcel_str = parcel_str.encode("ascii", "ignore").decode("utf-8")
+    parcel_str = parcel_str.replace("_aaaa_", "à").replace("_nnnn_", "n°")
+    m = PARCEL_YEAR_REGEXP.match(parcel_str)
+    year = None
+    if m:
+        year = m.groups()[0]
+        parcel_str = parcel_str[m.span()[1]:]
+    for parcel in PARCEL_SECTION_REGEXP.findall(parcel_str):
+        sector, nums = parcel[0], parcel[1]
+        for num in PARCEL_NB_REGEXP.findall(nums):
+            if len(str(num)) > 6:
+                continue
+            dct = {"year": year, "section": sector, "parcel_number": num}
+            if town:
+                dct["town"] = town
+            if owner:
+                dct["history_modifier"] = owner
+            parcels.append(dct)
+        for parcel_ranges in PARCEL_NB_RANGE_REGEXP.findall(nums):
+            lower_range, higher_range = parcel_ranges
+            try:
+                # the lower range itself has been already kept
+                lower_range = int(lower_range) + 1
+                higher_range = int(higher_range)
+            except ValueError:
+                continue
+            for num in range(lower_range, higher_range):
+                dct = {"year": year, "section": sector, "parcel_number": str(num)}
+                if town:
+                    dct["town"] = town
+                if owner:
+                    dct["history_modifier"] = owner
+                parcels.append(dct)
+    return parcels
+
+
 def get_all_field_names(model):
     return list(
         set(