From c2490d1ed5e8aa4ca360894865e7c923c30fed98 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Wed, 14 Feb 2018 23:16:59 +0100
Subject: [PATCH 01/28] more cleanup BP/CS/TSA/CIDEX/CEDEX

---
 addok_france/utils.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 779ab89..c3b01eb 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -51,8 +51,9 @@
 
 def clean_query(q):
     q = re.sub(r'([\d]{5})', r' \1 ', q, flags=re.IGNORECASE)
+    q = re.sub(r'(^| )(b\.?p\.?|cs|tsa|cidex) *(n *|no *|)[\d]* ?', '', q, flags=re.IGNORECASE)
+    q = re.sub(r'([\d]{2})[\d]{3}(.*)c(e|é)dex ?[\d]*', r'\1\2', q, flags=re.IGNORECASE)
     q = re.sub('c(e|é)dex ?[\d]*', '', q, flags=re.IGNORECASE)
-    q = re.sub(r'\b(bp|cs|tsa|cidex) *[\d]*', '', q, flags=re.IGNORECASE)
     q = re.sub('\d{,2}(e|[eè]me) ([eé]tage)', '', q, flags=re.IGNORECASE)
     q = re.sub(' {2,}', ' ', q, flags=re.IGNORECASE)
     q = re.sub('[ -]s/[ -]', ' sur ', q, flags=re.IGNORECASE)

From 253cb541fdd32defb4070ce496ecaf15d42ebf74 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Wed, 14 Feb 2018 23:35:42 +0100
Subject: [PATCH 02/28] =?UTF-8?q?cleanup=20BP/CS/TSA/CIDEX=20N=C2=B0=20+?=
 =?UTF-8?q?=20tests?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 addok_france/utils.py |  5 ++++-
 tests/test_utils.py   | 20 +++++++++++++++-----
 2 files changed, 19 insertions(+), 6 deletions(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index c3b01eb..093fbce 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -51,8 +51,11 @@
 
 def clean_query(q):
     q = re.sub(r'([\d]{5})', r' \1 ', q, flags=re.IGNORECASE)
-    q = re.sub(r'(^| )(b\.?p\.?|cs|tsa|cidex) *(n *|no *|)[\d]* ?', '', q, flags=re.IGNORECASE)
+    print(q)
+    q = re.sub(r'(^| )(b\.?p\.?|cs|tsa|cidex) *(n(o|°|) *|)[\d]+ *', r'\1', q, flags=re.IGNORECASE)
+    print(q)
     q = re.sub(r'([\d]{2})[\d]{3}(.*)c(e|é)dex ?[\d]*', r'\1\2', q, flags=re.IGNORECASE)
+    print(q)
     q = re.sub('c(e|é)dex ?[\d]*', '', q, flags=re.IGNORECASE)
     q = re.sub('\d{,2}(e|[eè]me) ([eé]tage)', '', q, flags=re.IGNORECASE)
     q = re.sub(' {2,}', ' ', q, flags=re.IGNORECASE)
diff --git a/tests/test_utils.py b/tests/test_utils.py
index 969ee91..9ea13cc 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -13,13 +13,17 @@
 
 @pytest.mark.parametrize("input,expected", [
     ("2 allée Jules Guesde 31068 TOULOUSE CEDEX 7",
-     "2 allée Jules Guesde 31068 TOULOUSE"),
+     "2 allée Jules Guesde 31 TOULOUSE"),
     ("7, avenue Léon-Blum 31507 Toulouse Cedex 5",
-     "7, avenue Léon-Blum 31507 Toulouse"),
+     "7, avenue Léon-Blum 31 Toulouse"),
     ("159, avenue Jacques-Douzans 31604 Muret Cedex",
-     "159, avenue Jacques-Douzans 31604 Muret"),
+     "159, avenue Jacques-Douzans 31 Muret"),
     ("2 allée Jules Guesde BP 7015 31068 TOULOUSE",
      "2 allée Jules Guesde 31068 TOULOUSE"),
+    ("2 allée Jules Guesde B.P. 7015 31068 TOULOUSE",
+     "2 allée Jules Guesde 31068 TOULOUSE"),
+    ("2 allée Jules Guesde B.P. N 7015 31068 TOULOUSE",
+     "2 allée Jules Guesde 31068 TOULOUSE"),
     ("BP 80111 159, avenue Jacques-Douzans 31604 Muret",
      "159, avenue Jacques-Douzans 31604 Muret"),
     ("12, place de l'Hôtel-de-Ville BP 46 02150 Sissonne",
@@ -27,7 +31,7 @@
     ("6, rue Winston-Churchill CS 40055 60321 Compiègne",
      "6, rue Winston-Churchill 60321 Compiègne"),
     ("BP 80111 159, avenue Jacques-Douzans 31604 Muret Cedex",
-     "159, avenue Jacques-Douzans 31604 Muret"),
+     "159, avenue Jacques-Douzans 31 Muret"),
     ("BP 20169 Cite administrative - 8e étage Rue Gustave-Delory 59017 Lille",
      "Cite administrative - Rue Gustave-Delory 59017 Lille"),
     ("12e étage Rue Gustave-Delory 59017 Lille",
@@ -52,9 +56,15 @@
     ("32bis Rue des Vosges93290",
      "32bis Rue des Vosges 93290"),
     ("20 avenue de Ségur TSA 30719 75334 Paris Cedex 07",
-     "20 avenue de Ségur 75334 Paris"),
+     "20 avenue de Ségur 75 Paris"),
+    ("20 avenue de Ségur TSA No30719 75334 Paris Cedex 07",
+     "20 avenue de Ségur 75 Paris"),
+    ("20 avenue de Ségur TSA N 30719 75334 Paris Cedex 07",
+     "20 avenue de Ségur 75 Paris"),
     ("20 rue saint germain CIDEX 304 89110 Poilly-sur-tholon",
      "20 rue saint germain 89110 Poilly-sur-tholon"),
+    ("20 rue saint germain CIDEX N°304 89110 Poilly-sur-tholon",
+     "20 rue saint germain 89110 Poilly-sur-tholon"),
 ])
 def test_clean_query(input, expected):
     assert clean_query(input) == expected

From 3f75ccfc8c15d27a70e602f60ef29d03fa154fa2 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Wed, 14 Feb 2018 23:36:22 +0100
Subject: [PATCH 03/28] do not break on queries like "12bis"

---
 addok_france/utils.py | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 093fbce..28efe54 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -120,14 +120,17 @@ def flag_housenumber(tokens):
 
 def fold_ordinal(s):
     """3bis => 3b."""
-    if s[0].isdigit() and not s.isdigit():
+    if s is not None and s !='' and s[0].isdigit() and not s.isdigit():
         try:
             number, ordinal = FOLD_PATTERN.findall(s)[0]
         except (IndexError, ValueError):
             pass
         else:
-            s = s.update('{}{}'.format(number,
+            try:
+                s = s.update('{}{}'.format(number,
                                        FOLD.get(ordinal.lower(), ordinal)))
+            except:
+                pass
     return s
 
 

From 6d59df33d5dfb4b7fc09da4f4c2abf2844c46701 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Wed, 14 Feb 2018 23:38:36 +0100
Subject: [PATCH 04/28] print() removed

---
 addok_france/utils.py | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 28efe54..265cc01 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -51,11 +51,8 @@
 
 def clean_query(q):
     q = re.sub(r'([\d]{5})', r' \1 ', q, flags=re.IGNORECASE)
-    print(q)
     q = re.sub(r'(^| )(b\.?p\.?|cs|tsa|cidex) *(n(o|°|) *|)[\d]+ *', r'\1', q, flags=re.IGNORECASE)
-    print(q)
     q = re.sub(r'([\d]{2})[\d]{3}(.*)c(e|é)dex ?[\d]*', r'\1\2', q, flags=re.IGNORECASE)
-    print(q)
     q = re.sub('c(e|é)dex ?[\d]*', '', q, flags=re.IGNORECASE)
     q = re.sub('\d{,2}(e|[eè]me) ([eé]tage)', '', q, flags=re.IGNORECASE)
     q = re.sub(' {2,}', ' ', q, flags=re.IGNORECASE)

From 4e5be21774d2f9b437119ff15160f7652a9be5d0 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Wed, 14 Feb 2018 23:58:20 +0100
Subject: [PATCH 05/28] cleanup phone/fax numbers

---
 addok_france/utils.py |  3 ++-
 tests/test_utils.py   | 10 ++++++++++
 2 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 265cc01..205594a 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -50,11 +50,12 @@
 
 
 def clean_query(q):
-    q = re.sub(r'([\d]{5})', r' \1 ', q, flags=re.IGNORECASE)
     q = re.sub(r'(^| )(b\.?p\.?|cs|tsa|cidex) *(n(o|°|) *|)[\d]+ *', r'\1', q, flags=re.IGNORECASE)
     q = re.sub(r'([\d]{2})[\d]{3}(.*)c(e|é)dex ?[\d]*', r'\1\2', q, flags=re.IGNORECASE)
+    q = re.sub(r'([^\d ])([\d]{5})([^\d]|$)', r'\1 \2 ', q, flags=re.IGNORECASE)
     q = re.sub('c(e|é)dex ?[\d]*', '', q, flags=re.IGNORECASE)
     q = re.sub('\d{,2}(e|[eè]me) ([eé]tage)', '', q, flags=re.IGNORECASE)
+    q = re.sub(r'((fax|t[eé]l|t[eé]l[eé]copieur)[ :,\.]*|)(\d{10}|[0-9][0-9][ -\./]\d\d[-\./ ]\d\d[-\./ ]\d\d[-\./ ]\d\d)', '', q, flags=re.IGNORECASE)
     q = re.sub(' {2,}', ' ', q, flags=re.IGNORECASE)
     q = re.sub('[ -]s/[ -]', ' sur ', q, flags=re.IGNORECASE)
     q = re.sub('[ -]s/s[ -]', ' sous ', q, flags=re.IGNORECASE)
diff --git a/tests/test_utils.py b/tests/test_utils.py
index 9ea13cc..5e4ce00 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -65,6 +65,16 @@
      "20 rue saint germain 89110 Poilly-sur-tholon"),
     ("20 rue saint germain CIDEX N°304 89110 Poilly-sur-tholon",
      "20 rue saint germain 89110 Poilly-sur-tholon"),
+    ("20 rue saint germain 89110 Poilly-sur-tholon 01.23.45.67.89",
+     "20 rue saint germain 89110 Poilly-sur-tholon"),
+    ("32bis Rue des Vosges93290 fax: 0123456789",
+     "32bis Rue des Vosges 93290"),
+    ("32bis Rue des Vosges 93290 tel 01 23 45 67 89",
+     "32bis Rue des Vosges 93290"),
+    ("32bis Rue des Vosges 93290 telecopieur. 01/23/45/67/89",
+     "32bis Rue des Vosges 93290"),
+    ("32bis Rue des Vosges 93290 télécopieur, 01-23-45-67-89",
+     "32bis Rue des Vosges 93290"),
 ])
 def test_clean_query(input, expected):
     assert clean_query(input) == expected

From 454ca5006542d430677727e4474297a23e63fba2 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Thu, 15 Feb 2018 00:27:54 +0100
Subject: [PATCH 06/28] fold initiales: F F I > F F I FFI, etc

---
 addok_france/utils.py | 1 +
 tests/test_utils.py   | 2 ++
 2 files changed, 3 insertions(+)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 205594a..28100b8 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -60,6 +60,7 @@ def clean_query(q):
     q = re.sub('[ -]s/[ -]', ' sur ', q, flags=re.IGNORECASE)
     q = re.sub('[ -]s/s[ -]', ' sous ', q, flags=re.IGNORECASE)
     q = re.sub('^lieux?[ -]?dits?\\b(?=.)', '', q, flags=re.IGNORECASE)
+    q = re.sub(r'(^| )(([A-Z]) ([A-Z]) (([A-Z]) )?(([A-Z]) )?(([A-Z])( |$))?)', r'\1\2\3\4\6\8\10 ', q, flags=re.IGNORECASE)
     q = q.strip()
     return q
 
diff --git a/tests/test_utils.py b/tests/test_utils.py
index 5e4ce00..e5f128d 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -75,6 +75,8 @@
      "32bis Rue des Vosges 93290"),
     ("32bis Rue des Vosges 93290 télécopieur, 01-23-45-67-89",
      "32bis Rue des Vosges 93290"),
+    ("10 BLD DES F F I 85300 CHALLANS",
+     "10 BLD DES F F I FFI 85300 CHALLANS"),
 ])
 def test_clean_query(input, expected):
     assert clean_query(input) == expected

From 045c8f03a734ef823b41c399a411e6354b2049c5 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Thu, 15 Feb 2018 01:02:29 +0100
Subject: [PATCH 07/28] boite postale

---
 addok_france/utils.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 28100b8..f1640d2 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -50,7 +50,7 @@
 
 
 def clean_query(q):
-    q = re.sub(r'(^| )(b\.?p\.?|cs|tsa|cidex) *(n(o|°|) *|)[\d]+ *', r'\1', q, flags=re.IGNORECASE)
+    q = re.sub(r'(^| )(boite postale|b\.?p\.?|cs|tsa|cidex) *(n(o|°|) *|)[\d]+ *', r'\1', q, flags=re.IGNORECASE)
     q = re.sub(r'([\d]{2})[\d]{3}(.*)c(e|é)dex ?[\d]*', r'\1\2', q, flags=re.IGNORECASE)
     q = re.sub(r'([^\d ])([\d]{5})([^\d]|$)', r'\1 \2 ', q, flags=re.IGNORECASE)
     q = re.sub('c(e|é)dex ?[\d]*', '', q, flags=re.IGNORECASE)

From 2261c92108b33e999779ae6eca56ed851aa17b36 Mon Sep 17 00:00:00 2001
From: Christian Quest <cquest@openstreetmap.fr>
Date: Fri, 16 Feb 2018 18:56:47 +0100
Subject: [PATCH 08/28] WIP fold_initials "F F I" > "FFI"

---
 addok_france/__init__.py |  1 +
 addok_france/utils.py    | 15 +++++++++++++++
 tests/test_utils.py      | 30 +++++++++++++++++++++++++++++-
 3 files changed, 45 insertions(+), 1 deletion(-)

diff --git a/addok_france/__init__.py b/addok_france/__init__.py
index 2b20ccb..8f48d8c 100644
--- a/addok_france/__init__.py
+++ b/addok_france/__init__.py
@@ -14,6 +14,7 @@
 extract_address = yielder(utils.extract_address)
 glue_ordinal = utils.glue_ordinal
 fold_ordinal = yielder(utils.fold_ordinal)
+fold_initials = yielder(utils.fold_initials)
 flag_housenumber = utils.flag_housenumber
 make_labels = utils.make_labels
 remove_leading_zeros = yielder(utils.remove_leading_zeros)
diff --git a/addok_france/utils.py b/addok_france/utils.py
index f1640d2..4b4f94a 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -133,6 +133,21 @@ def fold_ordinal(s):
     return s
 
 
+def fold_initials(tokens):
+    initials = []
+    for _, token, next_ in neighborhood(tokens):
+        if len(token)==1:
+            initials.append(token)
+        else:
+            if len(initial)>2:
+                initials[0].update("".join(initials))
+                yield initials[0]
+            else:
+                for tk in initials:
+                    yield tk
+            yield token
+
+
 def remove_leading_zeros(s):
     """0003 => 3."""
     # Limit digits from 1 to 3 in order to avoid processing postcodes.
diff --git a/tests/test_utils.py b/tests/test_utils.py
index e5f128d..445ae1f 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -8,7 +8,7 @@
 from addok.helpers.text import Token
 from addok_france.utils import (clean_query, extract_address, flag_housenumber,
                                 fold_ordinal, glue_ordinal, make_labels,
-                                remove_leading_zeros)
+                                remove_leading_zeros, fold_initials)
 
 
 @pytest.mark.parametrize("input,expected", [
@@ -353,3 +353,31 @@ def test_make_municipality_labels(config):
         '59000 Lille',
         'Lille 59000',
     ]
+
+
+@pytest.mark.parametrize("inputs,expected", [
+    (['6', 'bis'], ['6bis']),
+    (['6'], ['6']),
+    (['6', 'avenue'], ['6', 'avenue']),
+    (['60', 'bis', 'avenue'], ['60bis', 'avenue']),
+    (['600', 'ter', 'avenue'], ['600ter', 'avenue']),
+    (['6', 'quinquies', 'avenue'], ['6quinquies', 'avenue']),
+    (['60', 'sexies', 'avenue'], ['60sexies', 'avenue']),
+    (['600', 'quater', 'avenue'], ['600quater', 'avenue']),
+    (['6', 's', 'avenue'], ['6s', 'avenue']),
+    (['60b', 'avenue'], ['60b', 'avenue']),
+    (['600', 'b', 'avenue'], ['600b', 'avenue']),
+    (['241', 'r', 'de'], ['241', 'r', 'de']),
+    (['120', 'r', 'renard'], ['120', 'r', 'renard']),
+    (['241', 'r', 'rue'], ['241r', 'rue']),
+    (['place', 'des', 'terreaux'], ['place', 'des', 'terreaux']),
+    (['rue', 'du', 'bis'], ['rue', 'du', 'bis']),
+])
+@pytest.mark.parametrize("input,expected", [
+    (['allee','a','b','c'], ['allee','abc']),
+    (['allee','a','b','c','toto'], ['allee','abc','toto']),
+    (['allee','a','b','c','d'], ['allee','abcd']),
+    (['allee','a','b','c','d','e'], ['allee','abcde']),
+])
+def test_fold_initials(input, expected):
+    assert fold_initials(Token(input)) == expected

From 5f11ac443104c8c90fe3c0825a5ffa8ad2409875 Mon Sep 17 00:00:00 2001
From: Christian Quest <cquest@openstreetmap.fr>
Date: Fri, 16 Feb 2018 19:48:27 +0100
Subject: [PATCH 09/28] fold_initials + tests

---
 addok_france/utils.py | 15 +++++++++------
 tests/test_utils.py   | 27 ++++++---------------------
 2 files changed, 15 insertions(+), 27 deletions(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 4b4f94a..6697a78 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -127,24 +127,27 @@ def fold_ordinal(s):
         else:
             try:
                 s = s.update('{}{}'.format(number,
-                                       FOLD.get(ordinal.lower(), ordinal)))
+                             FOLD.get(ordinal.lower(), ordinal)))
             except:
                 pass
     return s
 
 
 def fold_initials(tokens):
+    """ folds 'F F I' into 'FFI' """
     initials = []
     for _, token, next_ in neighborhood(tokens):
-        if len(token)==1:
+        isinitial = len(token) == 1 and token.isalpha()
+        if isinitial:
             initials.append(token)
-        else:
-            if len(initial)>2:
-                initials[0].update("".join(initials))
-                yield initials[0]
+        if not next_ or not isinitial:
+            if len(initials) > 2:
+                yield initials[0].update("".join(initials))
             else:
                 for tk in initials:
                     yield tk
+            initials = []
+        if not isinitial:
             yield token
 
 
diff --git a/tests/test_utils.py b/tests/test_utils.py
index 445ae1f..c08f55b 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -356,28 +356,13 @@ def test_make_municipality_labels(config):
 
 
 @pytest.mark.parametrize("inputs,expected", [
-    (['6', 'bis'], ['6bis']),
-    (['6'], ['6']),
-    (['6', 'avenue'], ['6', 'avenue']),
-    (['60', 'bis', 'avenue'], ['60bis', 'avenue']),
-    (['600', 'ter', 'avenue'], ['600ter', 'avenue']),
-    (['6', 'quinquies', 'avenue'], ['6quinquies', 'avenue']),
-    (['60', 'sexies', 'avenue'], ['60sexies', 'avenue']),
-    (['600', 'quater', 'avenue'], ['600quater', 'avenue']),
-    (['6', 's', 'avenue'], ['6s', 'avenue']),
-    (['60b', 'avenue'], ['60b', 'avenue']),
-    (['600', 'b', 'avenue'], ['600b', 'avenue']),
-    (['241', 'r', 'de'], ['241', 'r', 'de']),
-    (['120', 'r', 'renard'], ['120', 'r', 'renard']),
-    (['241', 'r', 'rue'], ['241r', 'rue']),
-    (['place', 'des', 'terreaux'], ['place', 'des', 'terreaux']),
-    (['rue', 'du', 'bis'], ['rue', 'du', 'bis']),
-])
-@pytest.mark.parametrize("input,expected", [
-    (['allee','a','b','c'], ['allee','abc']),
     (['allee','a','b','c','toto'], ['allee','abc','toto']),
+    (['allee','a','b','c','toto','d','e','f'], ['allee','abc','toto','def']),
+    (['allee','a','2','c','toto'], ['allee','a','2','c','toto']),
+    (['allee','a','b','c'], ['allee','abc']),
     (['allee','a','b','c','d'], ['allee','abcd']),
     (['allee','a','b','c','d','e'], ['allee','abcde']),
 ])
-def test_fold_initials(input, expected):
-    assert fold_initials(Token(input)) == expected
+def test_fold_initials(inputs, expected):
+    tokens = [Token(input_) for input_ in inputs]
+    assert list(fold_initials(tokens)) == expected

From 330169a89d923e9bb2d39339d2661db178711389 Mon Sep 17 00:00:00 2001
From: Christian Quest <cquest@openstreetmap.fr>
Date: Fri, 16 Feb 2018 19:48:58 +0100
Subject: [PATCH 10/28] pep8

---
 addok_france/utils.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 6697a78..ec2dd55 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -111,7 +111,7 @@ def flag_housenumber(tokens):
     found = False
     for previous, token, next_ in neighborhood(tokens):
         if ((token.is_first or (next_ and TYPES_PATTERN.match(next_)))
-                and NUMBER_PATTERN.match(token) and not found):
+            and NUMBER_PATTERN.match(token) and not found):
             token.kind = 'housenumber'
             found = True
         yield token
@@ -119,7 +119,7 @@ def flag_housenumber(tokens):
 
 def fold_ordinal(s):
     """3bis => 3b."""
-    if s is not None and s !='' and s[0].isdigit() and not s.isdigit():
+    if s is not None and s != '' and s[0].isdigit() and not s.isdigit():
         try:
             number, ordinal = FOLD_PATTERN.findall(s)[0]
         except (IndexError, ValueError):

From fd5d9ee94f81cb0f13c8f61aea8303ab089885b1 Mon Sep 17 00:00:00 2001
From: Christian Quest <cquest@openstreetmap.fr>
Date: Fri, 16 Feb 2018 20:25:13 +0100
Subject: [PATCH 11/28] pep8

---
 addok_france/utils.py | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index ec2dd55..ab6235a 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -50,9 +50,12 @@
 
 
 def clean_query(q):
-    q = re.sub(r'(^| )(boite postale|b\.?p\.?|cs|tsa|cidex) *(n(o|°|) *|)[\d]+ *', r'\1', q, flags=re.IGNORECASE)
-    q = re.sub(r'([\d]{2})[\d]{3}(.*)c(e|é)dex ?[\d]*', r'\1\2', q, flags=re.IGNORECASE)
-    q = re.sub(r'([^\d ])([\d]{5})([^\d]|$)', r'\1 \2 ', q, flags=re.IGNORECASE)
+    q = re.sub(r'(^| )(boite postale|b\.?p\.?|cs|tsa|cidex) *(n(o|°|) *|)[\d]+ *',
+               r'\1', q, flags=re.IGNORECASE)
+    q = re.sub(r'([\d]{2})[\d]{3}(.*)c(e|é)dex ?[\d]*', r'\1\2',
+               q, flags=re.IGNORECASE)
+    q = re.sub(r'([^\d ])([\d]{5})([^\d]|$)', r'\1 \2 ',
+               q, flags=re.IGNORECASE)
     q = re.sub('c(e|é)dex ?[\d]*', '', q, flags=re.IGNORECASE)
     q = re.sub('\d{,2}(e|[eè]me) ([eé]tage)', '', q, flags=re.IGNORECASE)
     q = re.sub(r'((fax|t[eé]l|t[eé]l[eé]copieur)[ :,\.]*|)(\d{10}|[0-9][0-9][ -\./]\d\d[-\./ ]\d\d[-\./ ]\d\d[-\./ ]\d\d)', '', q, flags=re.IGNORECASE)
@@ -60,7 +63,8 @@ def clean_query(q):
     q = re.sub('[ -]s/[ -]', ' sur ', q, flags=re.IGNORECASE)
     q = re.sub('[ -]s/s[ -]', ' sous ', q, flags=re.IGNORECASE)
     q = re.sub('^lieux?[ -]?dits?\\b(?=.)', '', q, flags=re.IGNORECASE)
-    q = re.sub(r'(^| )(([A-Z]) ([A-Z]) (([A-Z]) )?(([A-Z]) )?(([A-Z])( |$))?)', r'\1\2\3\4\6\8\10 ', q, flags=re.IGNORECASE)
+    q = re.sub(r'(^| )(([A-Z]) ([A-Z]) (([A-Z]) )?(([A-Z]) )?(([A-Z])( |$))?)',
+               r'\1\2\3\4\6\8\10 ', q, flags=re.IGNORECASE)
     q = q.strip()
     return q
 

From 3ccc267cf7d1630663a40936bdf7f479efc96676 Mon Sep 17 00:00:00 2001
From: Christian Quest <cquest@openstreetmap.fr>
Date: Fri, 16 Feb 2018 20:29:10 +0100
Subject: [PATCH 12/28] pep8

---
 tests/test_utils.py | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)

diff --git a/tests/test_utils.py b/tests/test_utils.py
index c08f55b..62962b5 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -356,12 +356,18 @@ def test_make_municipality_labels(config):
 
 
 @pytest.mark.parametrize("inputs,expected", [
-    (['allee','a','b','c','toto'], ['allee','abc','toto']),
-    (['allee','a','b','c','toto','d','e','f'], ['allee','abc','toto','def']),
-    (['allee','a','2','c','toto'], ['allee','a','2','c','toto']),
-    (['allee','a','b','c'], ['allee','abc']),
-    (['allee','a','b','c','d'], ['allee','abcd']),
-    (['allee','a','b','c','d','e'], ['allee','abcde']),
+    (['allee', 'a', 'b', 'c', 'toto'],
+     ['allee', 'abc', 'toto']),
+    (['allee', 'a', 'b', 'c', 'toto', 'd', 'e', 'f'],
+     ['allee', 'abc', 'toto', 'def']),
+    (['allee', 'a', '2', 'c', 'toto'],
+     ['allee', 'a', '2', 'c', 'toto']),
+    (['allee', 'a', 'b', 'c'],
+     ['allee', 'abc']),
+    (['allee', 'a', 'b', 'c', 'd'],
+     ['allee', 'abcd']),
+    (['allee', 'a', 'b', 'c', 'd', 'e'],
+     ['allee', 'abcde']),
 ])
 def test_fold_initials(inputs, expected):
     tokens = [Token(input_) for input_ in inputs]

From 9b90da3a0a597ec36ee84a5476a5c765f053ff96 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Fri, 16 Feb 2018 22:52:48 +0100
Subject: [PATCH 13/28] separate PR

---
 addok_france/utils.py | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index ab6235a..e4414ce 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -123,17 +123,14 @@ def flag_housenumber(tokens):
 
 def fold_ordinal(s):
     """3bis => 3b."""
-    if s is not None and s != '' and s[0].isdigit() and not s.isdigit():
+    if s[0].isdigit() and not s.isdigit():
         try:
             number, ordinal = FOLD_PATTERN.findall(s)[0]
         except (IndexError, ValueError):
             pass
         else:
-            try:
-                s = s.update('{}{}'.format(number,
-                             FOLD.get(ordinal.lower(), ordinal)))
-            except:
-                pass
+            s = s.update('{}{}'.format(number,
+                         FOLD.get(ordinal.lower(), ordinal)))
     return s
 
 

From a3017f85751fcc945221a12d320787a1320ef77f Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Sun, 18 Feb 2018 17:25:26 +0100
Subject: [PATCH 14/28] fold_initials is a yielder

---
 addok_france/__init__.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/addok_france/__init__.py b/addok_france/__init__.py
index 8f48d8c..4f55576 100644
--- a/addok_france/__init__.py
+++ b/addok_france/__init__.py
@@ -14,7 +14,7 @@
 extract_address = yielder(utils.extract_address)
 glue_ordinal = utils.glue_ordinal
 fold_ordinal = yielder(utils.fold_ordinal)
-fold_initials = yielder(utils.fold_initials)
+fold_initials = utils.fold_initials
 flag_housenumber = utils.flag_housenumber
 make_labels = utils.make_labels
 remove_leading_zeros = yielder(utils.remove_leading_zeros)

From bb92a1d35daf5e8c5cd5db222a2dda0162b9f1ed Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Wed, 21 Feb 2018 17:18:51 +0100
Subject: [PATCH 15/28] fold_initials > glue_initials

---
 addok_france/__init__.py | 2 +-
 addok_france/utils.py    | 6 ++----
 tests/test_utils.py      | 8 ++++----
 3 files changed, 7 insertions(+), 9 deletions(-)

diff --git a/addok_france/__init__.py b/addok_france/__init__.py
index 4f55576..c1ae1b7 100644
--- a/addok_france/__init__.py
+++ b/addok_france/__init__.py
@@ -14,7 +14,7 @@
 extract_address = yielder(utils.extract_address)
 glue_ordinal = utils.glue_ordinal
 fold_ordinal = yielder(utils.fold_ordinal)
-fold_initials = utils.fold_initials
+glue_initials = utils.glue_initials
 flag_housenumber = utils.flag_housenumber
 make_labels = utils.make_labels
 remove_leading_zeros = yielder(utils.remove_leading_zeros)
diff --git a/addok_france/utils.py b/addok_france/utils.py
index e4414ce..c7d001c 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -63,8 +63,6 @@ def clean_query(q):
     q = re.sub('[ -]s/[ -]', ' sur ', q, flags=re.IGNORECASE)
     q = re.sub('[ -]s/s[ -]', ' sous ', q, flags=re.IGNORECASE)
     q = re.sub('^lieux?[ -]?dits?\\b(?=.)', '', q, flags=re.IGNORECASE)
-    q = re.sub(r'(^| )(([A-Z]) ([A-Z]) (([A-Z]) )?(([A-Z]) )?(([A-Z])( |$))?)',
-               r'\1\2\3\4\6\8\10 ', q, flags=re.IGNORECASE)
     q = q.strip()
     return q
 
@@ -134,8 +132,8 @@ def fold_ordinal(s):
     return s
 
 
-def fold_initials(tokens):
-    """ folds 'F F I' into 'FFI' """
+def glue_initials(tokens):
+    """ glue 'F F I' into 'FFI' """
     initials = []
     for _, token, next_ in neighborhood(tokens):
         isinitial = len(token) == 1 and token.isalpha()
diff --git a/tests/test_utils.py b/tests/test_utils.py
index 62962b5..eea4772 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -8,7 +8,7 @@
 from addok.helpers.text import Token
 from addok_france.utils import (clean_query, extract_address, flag_housenumber,
                                 fold_ordinal, glue_ordinal, make_labels,
-                                remove_leading_zeros, fold_initials)
+                                remove_leading_zeros, glue_initials)
 
 
 @pytest.mark.parametrize("input,expected", [
@@ -76,7 +76,7 @@
     ("32bis Rue des Vosges 93290 télécopieur, 01-23-45-67-89",
      "32bis Rue des Vosges 93290"),
     ("10 BLD DES F F I 85300 CHALLANS",
-     "10 BLD DES F F I FFI 85300 CHALLANS"),
+     "10 BLD DES F F I 85300 CHALLANS"), # done by glue_initials
 ])
 def test_clean_query(input, expected):
     assert clean_query(input) == expected
@@ -369,6 +369,6 @@ def test_make_municipality_labels(config):
     (['allee', 'a', 'b', 'c', 'd', 'e'],
      ['allee', 'abcde']),
 ])
-def test_fold_initials(inputs, expected):
+def test_glue_initials(inputs, expected):
     tokens = [Token(input_) for input_ in inputs]
-    assert list(fold_initials(tokens)) == expected
+    assert list(glue_initials(tokens)) == expected

From 01cfa2a66360afdcab4814aafb040343c6db7778 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Sun, 18 Feb 2018 22:55:30 +0100
Subject: [PATCH 16/28] glue usual words like 'MONT' 'VAL' 'LE' 'LA' 'L' in an
 additionnal token

---
 addok_france/__init__.py |  1 +
 addok_france/utils.py    | 10 ++++++++++
 2 files changed, 11 insertions(+)

diff --git a/addok_france/__init__.py b/addok_france/__init__.py
index 2b20ccb..42ca4ab 100644
--- a/addok_france/__init__.py
+++ b/addok_france/__init__.py
@@ -14,6 +14,7 @@
 extract_address = yielder(utils.extract_address)
 glue_ordinal = utils.glue_ordinal
 fold_ordinal = yielder(utils.fold_ordinal)
+fold_words = utils.fold_words
 flag_housenumber = utils.flag_housenumber
 make_labels = utils.make_labels
 remove_leading_zeros = yielder(utils.remove_leading_zeros)
diff --git a/addok_france/utils.py b/addok_france/utils.py
index 779ab89..b1f4fba 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -127,6 +127,16 @@ def fold_ordinal(s):
     return s
 
 
+FOLD_WORDS = ["mont", "val", "le", "la", "l"]
+
+def fold_words(tokens):
+    """ folds 'MONT GRIFFON' into 'MONTGRIFFON' """
+    for _, token, next_ in neighborhood(tokens):
+        yield token
+        if token in FOLD_WORDS and next_ and next_.isalpha() and len(next_)>2:
+            yield token.update(token+next_)
+
+
 def remove_leading_zeros(s):
     """0003 => 3."""
     # Limit digits from 1 to 3 in order to avoid processing postcodes.

From f6e2e0b2c6a2f039f0f738f2ff7b61b93fcb44d4 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Wed, 21 Feb 2018 17:13:52 +0100
Subject: [PATCH 17/28] flod_words > glue_words

---
 addok_france/__init__.py | 2 +-
 addok_france/utils.py    | 8 ++++----
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/addok_france/__init__.py b/addok_france/__init__.py
index 42ca4ab..10507f5 100644
--- a/addok_france/__init__.py
+++ b/addok_france/__init__.py
@@ -14,7 +14,7 @@
 extract_address = yielder(utils.extract_address)
 glue_ordinal = utils.glue_ordinal
 fold_ordinal = yielder(utils.fold_ordinal)
-fold_words = utils.fold_words
+glue_words = utils.glue_words
 flag_housenumber = utils.flag_housenumber
 make_labels = utils.make_labels
 remove_leading_zeros = yielder(utils.remove_leading_zeros)
diff --git a/addok_france/utils.py b/addok_france/utils.py
index b1f4fba..a5174e2 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -127,13 +127,13 @@ def fold_ordinal(s):
     return s
 
 
-FOLD_WORDS = ["mont", "val", "le", "la", "l"]
+GLUE_WORDS = ["mont", "val", "le", "la", "l", "champ"]
 
-def fold_words(tokens):
-    """ folds 'MONT GRIFFON' into 'MONTGRIFFON' """
+def glue_words(tokens):
+    """ glue 'MONT GRIFFON' into 'MONTGRIFFON' """
     for _, token, next_ in neighborhood(tokens):
         yield token
-        if token in FOLD_WORDS and next_ and next_.isalpha() and len(next_)>2:
+        if token in GLUE_WORDS and next_ and next_.isalpha() and len(next_)>2:
             yield token.update(token+next_)
 
 

From 221584ac16dcd3b2b63728b3c6dab4092f50fa80 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Wed, 21 Feb 2018 17:40:20 +0100
Subject: [PATCH 18/28] glue_words test

---
 tests/test_utils.py | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/tests/test_utils.py b/tests/test_utils.py
index 969ee91..efc4dc3 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -8,7 +8,7 @@
 from addok.helpers.text import Token
 from addok_france.utils import (clean_query, extract_address, flag_housenumber,
                                 fold_ordinal, glue_ordinal, make_labels,
-                                remove_leading_zeros)
+                                remove_leading_zeros, glue_words)
 
 
 @pytest.mark.parametrize("input,expected", [
@@ -331,3 +331,15 @@ def test_make_municipality_labels(config):
         '59000 Lille',
         'Lille 59000',
     ]
+
+
+@pytest.mark.parametrize("inputs,expected", [
+    (['mont', 'griffon'], ['mont', 'montgriffon', 'griffon']),
+    (['champ', 'vallon'], ['champ', 'champvallon', 'vallon']),
+    (['val', 'suzon'], ['val', 'valsuzon', 'suzon']),
+    (['l', 'a', 'peu', 'pres'], ['l', 'a', 'peu', 'pres']),
+    (['l', 'un', 'des'], ['l', 'un', 'des']),
+])
+def test_glue_ordinal(inputs, expected):
+    tokens = [Token(input_) for input_ in inputs]
+    assert list(glue_words(tokens)) == expected

From 60acc2337e0d61baacc5dbcaad712557e31868a6 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Tue, 29 Oct 2019 09:13:20 +0100
Subject: [PATCH 19/28] more tests

---
 tests/test_utils.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/tests/test_utils.py b/tests/test_utils.py
index 43f8967..8d0949e 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -62,6 +62,8 @@
      "20 avenue de Ségur 75 Paris"),
     ("20 avenue de Ségur TSA N 30719 75334 Paris Cedex 07",
      "20 avenue de Ségur 75 Paris"),
+    ("20 avenue de Ségur TSA N°30719 75334 Paris Cedex 07",
+     "20 avenue de Ségur 75 Paris"),
     ("20 rue saint germain CIDEX 304 89110 Poilly-sur-tholon",
      "20 rue saint germain 89110 Poilly-sur-tholon"),
     ("20 rue saint germain CIDEX N°304 89110 Poilly-sur-tholon",

From 5ecc80f0469f17d1501e03335f34e470486d4bac Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Tue, 29 Oct 2019 09:13:39 +0100
Subject: [PATCH 20/28] test 0 manquant sur postcode

---
 tests/test_utils.py | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/tests/test_utils.py b/tests/test_utils.py
index 8d0949e..ea0c608 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -80,6 +80,10 @@
      "32bis Rue des Vosges 93290"),
     ("10 BLD DES F F I 85300 CHALLANS",
      "10 BLD DES F F I 85300 CHALLANS"), # done by glue_initials
+    ("6 rue de suisse 6000 Nice",
+     "6 rue de suisse 06000 Nice"),
+    ("6000 rue de suisse 6000 Nice",
+     "6000 rue de suisse 06000 Nice"),
 ])
 def test_clean_query(input, expected):
     assert clean_query(input) == expected

From 3b2faa76151e8e27c62719253cb562855d3441ff Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Tue, 29 Oct 2019 09:15:19 +0100
Subject: [PATCH 21/28] =?UTF-8?q?prise=20en=20compte=20des=20abr=C3=A9viat?=
 =?UTF-8?q?ions=20dans=20l'extraction?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 addok_france/utils.py | 19 +++++++++++--------
 tests/test_utils.py   |  5 +++++
 2 files changed, 16 insertions(+), 8 deletions(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index be93f80..b240962 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -1,14 +1,17 @@
 import re
 
 TYPES = [
-    'av(enue)?', 'r(ue)?', 'b(oulevar)?d', 'all[ée]es?', 'impasse', 'place',
-    'chemin', 'rocade', 'route', 'l[ôo]tissement', 'mont[ée]e', 'c[ôo]te',
-    'clos', 'champ', 'bois', 'taillis', 'boucle', 'passage', 'domaine',
-    'étang', 'etang', 'quai', 'desserte', 'pré', 'porte', 'square', 'mont',
-    'r[ée]sidence', 'parc', 'cours?', 'promenade', 'hameau', 'faubourg',
-    'ilot', 'berges?', 'via', 'cit[ée]', 'sent(e|ier)', 'rond[- ][Pp]oint',
-    'pas(se)?', 'carrefour', 'traverse', 'giratoire', 'esplanade', 'voie',
-    'chauss[ée]e',
+    'av(enue)?', 'r(ue)?', 'b(oulevar|l?v?)?d', 'all([ée]es?)?', 'imp(asse)?', 'pl(ace)?',
+    'che?(m(in)?)?', 'rocade', 'r(ou)?te', 'l[ôo]t(issement)?', 'mont[ée]e', 'c[ôo]te',
+    'clos', 'ch(am)?p', 'bois', 'taillis', 'b(ou)?cle', 'pass(age)?', 'dom(aine)?',
+    '[ée]ta?ng', 'desserte', 'pré', 'porte', 'squ?(are)?', 'mont',
+    'r[ée]s(idence)?', 'parc', 'cours?', 'pro?m(enade)?', 'ham(eau)?', 'f(aubour|b|bour)?g',
+    'ilot', 'ber(ges?)?', 'via', 'cit[ée]', 'sent(e|ier)', 'rond[- ][Pp]oint', 'rd?pt',
+    'pas(se)?', 'carr?(efour)?', 'trav(erse)?', 'giratoire', 'espl?(anade)?', 'voie',
+    'chauss[ée]e', 'aer(odrome)?', 'gr(ande?)?', 'gr(e|es|s)?', 'anc(ien(ne)?)?', 'c(en)?tre',
+    'devi(ation)?', 'dig(ue)?', 'embr(anchement)?', 'jard(in)?', 'j(et)?te', 'p(asserel)?le',
+    'p(or)?te', 'p(lace)?tte', 'p(arvis|rv|vr)', 'q(ua|rt)(ier)?', 'qu?(ai)?',
+    'r(uel)?le','t(erra)?sse','tunn?(el)?', 'viad(uc)?', 'v(il)?la',
 ]
 TYPES_REGEX = '|'.join(
     map(lambda x: '[{}{}]{}'.format(x[0], x[0].upper(), x[1:]), TYPES)
diff --git a/tests/test_utils.py b/tests/test_utils.py
index ea0c608..0229d28 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -131,6 +131,11 @@ def test_clean_query(input, expected):
      "boulevard jean larrieu 44000 mont de marsan"),
     ("PARC D ACTIVITE DE SAUMATY 26 AV ANDRE ROUSSIN 13016 MARSEILLE 16",
      "26 AV ANDRE ROUSSIN 13016 MARSEILLE 16"),
+    # Abréviations
+    ("resid goelands 28  bis imp des petrels 76460 Saint-valery-en-caux",
+     "28  bis imp des petrels 76460 Saint-valery-en-caux"),
+    ("bla bla bl 28 r des moulins",
+     "28 r des moulins"),
     ("Non matching pattern",
      "Non matching pattern"),
 ])

From 25c34618ebe3ab6ab2731a2e8ee3a9f8d335ddaf Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Tue, 29 Oct 2019 09:15:58 +0100
Subject: [PATCH 22/28] suppression de bte/boite/case postale

---
 addok_france/utils.py | 2 +-
 tests/test_utils.py   | 6 ++++++
 2 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index b240962..8e70c28 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -53,7 +53,7 @@
 
 
 def clean_query(q):
-    q = re.sub(r'(^| )(boite postale|b\.?p\.?|cs|tsa|cidex) *(n(o|°|) *|)[\d]+ *',
+    q = re.sub(r'(^| )((b(oi)?te|case) postale|b\.?p\.?|cs|tsa|cidex) *(n(o|°|) *|)[\d]+ *',
                r'\1', q, flags=re.IGNORECASE)
     q = re.sub(r'([\d]{2})[\d]{3}(.*)c(e|é)dex ?[\d]*', r'\1\2',
                q, flags=re.IGNORECASE)
diff --git a/tests/test_utils.py b/tests/test_utils.py
index 0229d28..56af1c6 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -29,6 +29,12 @@
      "159, avenue Jacques-Douzans 31604 Muret"),
     ("12, place de l'Hôtel-de-Ville BP 46 02150 Sissonne",
      "12, place de l'Hôtel-de-Ville 02150 Sissonne"),
+    ("12, place de l'Hôtel-de-Ville boite postale 46 02150 Sissonne",
+     "12, place de l'Hôtel-de-Ville 02150 Sissonne"),
+    ("12, place de l'Hôtel-de-Ville case postale 46 02150 Sissonne",
+     "12, place de l'Hôtel-de-Ville 02150 Sissonne"),
+    ("12, place de l'Hôtel-de-Ville bte postale 46 02150 Sissonne",
+     "12, place de l'Hôtel-de-Ville 02150 Sissonne"),
     ("6, rue Winston-Churchill CS 40055 60321 Compiègne",
      "6, rue Winston-Churchill 60321 Compiègne"),
     ("BP 80111 159, avenue Jacques-Douzans 31604 Muret Cedex",

From 35709fe274f3b33a635450a613149c85addf1421 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Tue, 29 Oct 2019 09:16:46 +0100
Subject: [PATCH 23/28] =?UTF-8?q?prise=20en=20compte=20de=20"1er=20=C3=A9t?=
 =?UTF-8?q?age"?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 addok_france/utils.py | 2 +-
 tests/test_utils.py   | 2 ++
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 8e70c28..6e042bf 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -60,7 +60,7 @@ def clean_query(q):
     q = re.sub(r'([^\d ])([\d]{5})([^\d]|$)', r'\1 \2 ',
                q, flags=re.IGNORECASE)
     q = re.sub(r'c(e|é)dex ?[\d]*', '', q, flags=re.IGNORECASE)
-    q = re.sub(r'\d{,2}(e|[eè]me) ([eé]tage)', '', q, flags=re.IGNORECASE)
+    q = re.sub(r'\d{,2}(e|[eè]me|er) ([eé]tage)', '', q, flags=re.IGNORECASE)
     q = re.sub(r'((fax|t[eé]l|t[eé]l[eé]copieur)[ :,\.]*|)(\d{10}|[0-9][0-9][ -\./]\d\d[-\./ ]\d\d[-\./ ]\d\d[-\./ ]\d\d)', '', q, flags=re.IGNORECASE)
     q = re.sub(r' {2,}', ' ', q, flags=re.IGNORECASE)
     q = re.sub(r'[ -]s/[ -]', ' sur ', q, flags=re.IGNORECASE)
diff --git a/tests/test_utils.py b/tests/test_utils.py
index 56af1c6..a1f59d5 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -43,6 +43,8 @@
      "Cite administrative - Rue Gustave-Delory 59017 Lille"),
     ("12e étage Rue Gustave-Delory 59017 Lille",
      "Rue Gustave-Delory 59017 Lille"),
+    ("Rue Gustave-Delory 1er étage 59017 Lille",
+     "Rue Gustave-Delory 59017 Lille"),
     ("12eme étage Rue Gustave-Delory 59017 Lille",
      "Rue Gustave-Delory 59017 Lille"),
     ("12ème étage Rue Gustave-Delory 59017 Lille",

From aaf5f9592d29c342bd31a502f3a299f9e3ef6a96 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Tue, 29 Oct 2019 09:17:14 +0100
Subject: [PATCH 24/28] ss -> sous

---
 addok_france/utils.py | 2 +-
 tests/test_utils.py   | 1 +
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 6e042bf..90a5516 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -64,7 +64,7 @@ def clean_query(q):
     q = re.sub(r'((fax|t[eé]l|t[eé]l[eé]copieur)[ :,\.]*|)(\d{10}|[0-9][0-9][ -\./]\d\d[-\./ ]\d\d[-\./ ]\d\d[-\./ ]\d\d)', '', q, flags=re.IGNORECASE)
     q = re.sub(r' {2,}', ' ', q, flags=re.IGNORECASE)
     q = re.sub(r'[ -]s/[ -]', ' sur ', q, flags=re.IGNORECASE)
-    q = re.sub(r'[ -]s/s[ -]', ' sous ', q, flags=re.IGNORECASE)
+    q = re.sub(r'[ -]s/?s[ -]', ' sous ', q, flags=re.IGNORECASE)
     q = re.sub(r'^lieux?[ -]?dits?\b(?=.)', '', q, flags=re.IGNORECASE)
     q = re.sub(r' (\d{4}) ', r' 0\1 ', q, flags=re.IGNORECASE)
     q = q.strip()
diff --git a/tests/test_utils.py b/tests/test_utils.py
index a1f59d5..c8129ce 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -54,6 +54,7 @@
     ("air s/ l'adour", "air sur l'adour"),
     ("air-s/-l'adour", "air sur l'adour"),
     ("Saint Didier s/s Ecouves", "Saint Didier sous Ecouves"),
+    ("Saint Didier ss Ecouves", "Saint Didier sous Ecouves"),
     ("La Chapelle-aux-Brocs", "La Chapelle-aux-Brocs"),
     ("Lieu-Dit Les Chênes", "Les Chênes"),
     ("Lieu Dit Les Chênes", "Les Chênes"),

From 30e3f3b4c3ce05d9105d3c9f7a894de04376b925 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Sun, 8 Nov 2020 17:08:37 +0100
Subject: [PATCH 25/28] handle StopIteration exception (empty tokens)

---
 addok_france/utils.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 90a5516..37f98af 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -84,7 +84,11 @@ def neighborhood(iterable, first=None, last=None):
     """
     iterator = iter(iterable)
     previous = first
-    current = next(iterator)  # Throws StopIteration if empty.
+    try:
+        current = next(iterator)
+    except StopIteration:  # StopIteration if empty.
+        return
+
     for next_ in iterator:
         yield (previous, current, next_)
         previous = current

From 2a5e3b2cfb7ce0fc4c15a6ee6f67170febd918bb Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Sun, 8 Nov 2020 17:51:51 +0100
Subject: [PATCH 26/28] avoid bad transform in TYPES_REGEX due to leading [

---
 addok_france/utils.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/addok_france/utils.py b/addok_france/utils.py
index 37f98af..b9cc507 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -4,7 +4,7 @@
     'av(enue)?', 'r(ue)?', 'b(oulevar|l?v?)?d', 'all([ée]es?)?', 'imp(asse)?', 'pl(ace)?',
     'che?(m(in)?)?', 'rocade', 'r(ou)?te', 'l[ôo]t(issement)?', 'mont[ée]e', 'c[ôo]te',
     'clos', 'ch(am)?p', 'bois', 'taillis', 'b(ou)?cle', 'pass(age)?', 'dom(aine)?',
-    '[ée]ta?ng', 'desserte', 'pré', 'porte', 'squ?(are)?', 'mont',
+    'eta?ng', 'éta?ng', 'desserte', 'pré', 'porte', 'squ?(are)?', 'mont',
     'r[ée]s(idence)?', 'parc', 'cours?', 'pro?m(enade)?', 'ham(eau)?', 'f(aubour|b|bour)?g',
     'ilot', 'ber(ges?)?', 'via', 'cit[ée]', 'sent(e|ier)', 'rond[- ][Pp]oint', 'rd?pt',
     'pas(se)?', 'carr?(efour)?', 'trav(erse)?', 'giratoire', 'espl?(anade)?', 'voie',

From 1e1b9cadaaa85758ac0c881e0815fd6be7713297 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Sun, 22 Nov 2020 18:18:25 +0100
Subject: [PATCH 27/28] added: glue_refs, to glue D 412 into D412

---
 addok_france/__init__.py |  1 +
 addok_france/utils.py    | 19 +++++++++++++++++++
 2 files changed, 20 insertions(+)

diff --git a/addok_france/__init__.py b/addok_france/__init__.py
index 2aed9e7..8424187 100644
--- a/addok_france/__init__.py
+++ b/addok_france/__init__.py
@@ -19,3 +19,4 @@
 flag_housenumber = utils.flag_housenumber
 make_labels = utils.make_labels
 remove_leading_zeros = yielder(utils.remove_leading_zeros)
+glue_refs = utils.glue_refs
diff --git a/addok_france/utils.py b/addok_france/utils.py
index b9cc507..e4a74e7 100644
--- a/addok_france/utils.py
+++ b/addok_france/utils.py
@@ -169,6 +169,25 @@ def glue_initials(tokens):
             yield token
 
 
+GLUE_REFS = re.compile(r'^(a|n|rn|d|rd|m|rm)[0-9]+$', flags=re.IGNORECASE)
+
+def glue_refs(tokens):
+    ref = None
+    for _, token, next_ in neighborhood(tokens):
+        print(ref, token, next_)
+        if next_ and GLUE_REFS.match(token+next_):
+            ref = token+next_
+        elif next_ and ref and GLUE_REFS.match(ref+next_):
+            ref = ref+next_
+        elif ref:
+            yield token.update(re.sub(r'^r(n|d)', r'\1', ref))
+            ref  = None
+        elif GLUE_REFS.match(token):
+            yield token.update(re.sub(r'^r(n|d)', r'\1', token))
+        else:
+            yield token
+
+
 def remove_leading_zeros(s):
     """0003 => 3."""
     # Limit digits from 1 to 3 in order to avoid processing postcodes.

From 3000457f01fb4489944e926fbd87c7173e4bc8c2 Mon Sep 17 00:00:00 2001
From: cquest <cquest@openstreetmap.fr>
Date: Sun, 22 Nov 2020 18:27:59 +0100
Subject: [PATCH 28/28] glue_refs test

---
 tests/test_utils.py | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/tests/test_utils.py b/tests/test_utils.py
index c8129ce..c6bc65f 100644
--- a/tests/test_utils.py
+++ b/tests/test_utils.py
@@ -9,7 +9,7 @@
 from addok_france.utils import (clean_query, extract_address, flag_housenumber,
                                 fold_ordinal, glue_ordinal, make_labels,
                                 remove_leading_zeros, glue_words,
-                                glue_initials)
+                                glue_initials, glue_refs)
 
 
 @pytest.mark.parametrize("input,expected", [
@@ -175,6 +175,18 @@ def test_glue_ordinal(inputs, expected):
     assert list(glue_ordinal(tokens)) == expected
 
 
+@pytest.mark.parametrize("inputs,expected", [
+    (['d', '412'], ['d412']),
+    (['rd', '30'], ['d30']),
+    (['d', '30', 'a', '4'], ['d30', 'a4']),
+    (['route', 'd', '30', 'a', '4','b'], ['route', 'd30', 'a4', 'b']),
+    (['route', '30', 'a'], ['route', '30', 'a']),
+])
+def test_glue_refs(inputs, expected):
+    tokens = [Token(input_) for input_ in inputs]
+    assert list(glue_refs(tokens)) == expected
+
+
 @pytest.mark.parametrize("inputs,expected", [
     (['6b'], True),
     (['6'], True),