csv-detective 0.7.4.dev969__py3-none-any.whl → 0.7.5.dev1009__py3-none-any.whl

csv_detective/__init__.py

@@ -1,3 +1,3 @@
 from .explore_csv import routine, routine_minio  # noqa
 
-__version__ = '0.7.4.dev'
+__version__ = '0.7.5.dev'

csv_detective/detect_fields/FR/geo/code_commune_insee/__init__.py

@@ -1,5 +1,9 @@
-from frformat import CodeCommuneInsee
+from frformat import CodeCommuneInsee, Millesime
 
 PROPORTION = 0.75
 
-_is = CodeCommuneInsee.is_valid
+_code_commune_insee = CodeCommuneInsee(Millesime.LATEST)
+
+
+def _is(val):
+    return _code_commune_insee.is_valid(val)

csv_detective/detect_fields/FR/geo/code_departement/__init__.py

@@ -1,7 +1,15 @@
-from frformat import NumeroDepartement
+from frformat import NumeroDepartement, Options, Millesime
 
 PROPORTION = 1
 
+_options = Options(
+    ignore_case=True,
+    ignore_accents=True,
+    replace_non_alphanumeric_with_space=True,
+    ignore_extra_whitespace=True
+)
+_numero_departement = NumeroDepartement(Millesime.LATEST, _options)
+
 
 def _is(val):
-    return isinstance(val, str) and NumeroDepartement.is_valid(val, strict=False)
+    return isinstance(val, str) and _numero_departement.is_valid(val)

csv_detective/detect_fields/FR/geo/code_fantoir/__init__.py

@@ -2,6 +2,8 @@ from frformat import CodeFantoir
 
 PROPORTION = 1
 
+_code_fantoir = CodeFantoir()
+
 
 def _is(val):
-    return isinstance(val, str) and CodeFantoir.is_valid(val)
+    return isinstance(val, str) and _code_fantoir.is_valid(val)

csv_detective/detect_fields/FR/geo/code_postal/__init__.py

@@ -2,4 +2,9 @@ from frformat import CodePostal
 
 PROPORTION = 0.9
 
-_is = CodePostal.is_valid
+_code_postal = CodePostal()
+
+
+def _is(val):
+
+    return _code_postal.is_valid(val)

csv_detective/detect_fields/FR/geo/code_region/__init__.py

@@ -1,8 +1,10 @@
-from frformat import CodeRegion
+from frformat import CodeRegion, Millesime
 
 PROPORTION = 1
 
+_code_region = CodeRegion(Millesime.LATEST)
+
 
 def _is(val):
     '''Renvoie True si val peut être un code_région, False sinon'''
-    return isinstance(val, str) and CodeRegion.is_valid(val)
+    return isinstance(val, str) and _code_region.is_valid(val)

csv_detective/detect_fields/FR/geo/commune/__init__.py

@@ -1,8 +1,16 @@
-from frformat import Commune
+from frformat import Commune, Options, Millesime
 
 PROPORTION = 0.9
 
+_options = Options(
+    ignore_case=True,
+    ignore_accents=True,
+    replace_non_alphanumeric_with_space=True,
+    ignore_extra_whitespace=True
+)
+_commune = Commune(Millesime.LATEST, _options)
+
 
 def _is(val):
     """Match avec le nom des communes"""
-    return isinstance(val, str) and Commune.is_valid(val, strict=False)
+    return isinstance(val, str) and _commune.is_valid(val)

csv_detective/detect_fields/FR/geo/departement/__init__.py

@@ -1,8 +1,16 @@
-from frformat import Departement
+from frformat import Departement, Options, Millesime
 
 PROPORTION = 0.9
 
+_options = Options(
+    ignore_case=True,
+    ignore_accents=True,
+    replace_non_alphanumeric_with_space=True,
+    ignore_extra_whitespace=True
+)
+_departement = Departement(Millesime.LATEST, _options)
+
 
 def _is(val):
     """Match avec le nom des departements"""
-    return isinstance(val, str) and Departement.is_valid(val, strict=False)
+    return isinstance(val, str) and _departement.is_valid(val)

csv_detective/detect_fields/FR/geo/insee_canton/__init__.py

@@ -1,8 +1,15 @@
-from frformat import Canton
+from frformat import Canton, Options, Millesime
 
 PROPORTION = 0.9
+_options = Options(
+    ignore_case=True,
+    ignore_accents=True,
+    replace_non_alphanumeric_with_space=True,
+    ignore_extra_whitespace=True
+)
+_canton = Canton(Millesime.LATEST, _options)
 
 
 def _is(val):
     """Match avec le nom des cantons"""
-    return isinstance(val, str) and Canton.is_valid(val, strict=False)
+    return isinstance(val, str) and _canton.is_valid(val)

csv_detective/detect_fields/FR/geo/latitude_l93/__init__.py

@@ -1,18 +1,18 @@
 from frformat import LatitudeL93
 from csv_detective.detect_fields.other.float import _is as is_float
+
 from csv_detective.detect_fields.other.float import float_casting
 
 
 PROPORTION = 0.9
 
+_latitudel93 = LatitudeL93()
+
 
 def _is(val):
     try:
-        if isinstance(val, (float, int)):
-            return LatitudeL93.is_valid(val)
-
-        elif isinstance(val, str) and is_float(val):
-            return LatitudeL93.is_valid(float_casting(val))
+        if isinstance(val, str) and is_float(val):
+            return _latitudel93.is_valid(float_casting(val))
 
         return False
 

csv_detective/detect_fields/FR/geo/longitude_l93/__init__.py

@@ -1,18 +1,18 @@
 from frformat import LongitudeL93
 from csv_detective.detect_fields.other.float import _is as is_float
+
 from csv_detective.detect_fields.other.float import float_casting
 
 
 PROPORTION = 0.9
 
+_longitudel93 = LongitudeL93()
+
 
 def _is(val):
     try:
-        if isinstance(val, (float, int)):
-            return LongitudeL93.is_valid(val)
-
-        elif isinstance(val, str) and is_float(val):
-            return LongitudeL93.is_valid(float_casting(val))
+        if isinstance(val, str) and is_float(val):
+            return _longitudel93.is_valid(float_casting(val))
 
         return False
 

csv_detective/detect_fields/FR/geo/pays/__init__.py

@@ -1,8 +1,16 @@
-from frformat import Pays
+from frformat import Pays, Options, Millesime
 
 PROPORTION = 0.6
 
+_options = Options(
+        ignore_case=True,
+        ignore_accents=True,
+        replace_non_alphanumeric_with_space=True,
+        ignore_extra_whitespace=True
+    )
+_pays = Pays(Millesime.LATEST, _options)
+
 
 def _is(val):
     """Match avec le nom des pays"""
-    return isinstance(val, str) and Pays.is_valid(val, strict=False)
+    return isinstance(val, str) and _pays.is_valid(val)

csv_detective/detect_fields/FR/geo/region/__init__.py

@@ -1,8 +1,48 @@
-from frformat import Region
+from frformat import Region, Options, Millesime
 
 PROPORTION = 1
 
+_extra_valid_values_set = frozenset({
+        "alsace",
+        "aquitaine",
+        "ara",
+        "aura",
+        "auvergne",
+        "auvergne et rhone alpes",
+        "basse normandie",
+        "bfc",
+        "bourgogne",
+        "bourgogne et franche comte",
+        "centre",
+        "champagne ardenne",
+        "franche comte",
+        "ge",
+        "haute normandie",
+        "hdf",
+        "languedoc roussillon",
+        "limousin",
+        "lorraine",
+        "midi pyrenees",
+        "nord pas de calais",
+        "npdc",
+        "paca",
+        "picardie",
+        "poitou charentes",
+        "reunion",
+        "rhone alpes",
+        })
+
+
+_options = Options(
+    ignore_case=True,
+    ignore_accents=True,
+    replace_non_alphanumeric_with_space=True,
+    ignore_extra_whitespace=True,
+    extra_valid_values=_extra_valid_values_set
+)
+_region = Region(Millesime.LATEST, _options)
+
 
 def _is(val):
     """Match avec le nom des regions"""
-    return isinstance(val, str) and Region.is_valid(val, strict=False)
+    return isinstance(val, str) and _region.is_valid(val)

csv_detective/detect_fields/FR/other/code_rna/__init__.py

@@ -2,6 +2,8 @@ from frformat import CodeRNA
 
 PROPORTION = 0.9
 
+_code_rna = CodeRNA()
+
 
 def _is(val):
-    return isinstance(val, str) and CodeRNA.is_valid(val)
+    return isinstance(val, str) and _code_rna.is_valid(val)

{csv_detective-0.7.4.dev969.data → csv_detective-0.7.5.dev1009.data}/data/share/csv_detective/CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## Current (in progress)
 
+- Nothing yet
+
+## 0.7.4 (2024-11-15)
+
 - Enable calling main functions from base [#97](https://github.com/datagouv/csv-detective/pull/97)
 - Better detection of ints and floats [#94](https://github.com/datagouv/csv-detective/pull/94)
 - Better handle NaN values [#96](https://github.com/datagouv/csv-detective/pull/96)

{csv_detective-0.7.4.dev969.dist-info → csv_detective-0.7.5.dev1009.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: csv_detective
-Version: 0.7.4.dev969
+Version: 0.7.5.dev1009
 Summary: Detect CSV column content
 Home-page: https://github.com/etalab/csv_detective
 Author: Etalab
@@ -26,5 +26,5 @@ Requires-Dist: odfpy==1.4.1
 Requires-Dist: requests==2.32.3
 Requires-Dist: responses==0.25.0
 Requires-Dist: python-magic==0.4.27
-Requires-Dist: frformat==0.3.0
+Requires-Dist: frformat==0.4.0
 

{csv_detective-0.7.4.dev969.dist-info → csv_detective-0.7.5.dev1009.dist-info}/RECORD

@@ -1,4 +1,4 @@
-csv_detective/__init__.py,sha256=Wxshf_Kon4kxH2VS-pa1Tp8tIPzzs7rpe3ruh4oCZqU,83
+csv_detective/__init__.py,sha256=giVhs0g13y4U2H0WiVBLcrvytcMxQ1LiCd2i03XITwQ,83
 csv_detective/cli.py,sha256=Ua7SE1wMH2uFUsTmfumh4nJk7O06okpMd2gvjUDO1II,1048
 csv_detective/detection.py,sha256=AuXlPOZfzqznZY2ybAAgaXIq6qVITYd3MXf2CoigI3I,22097
 csv_detective/explore_csv.py,sha256=X5yZS3WCUsafUMcs5tOnDTeMGzMnfr0iB9vEDx7xiqg,16977
@@ -10,24 +10,24 @@ csv_detective/detect_fields/__init__.py,sha256=CchNbi1vrgIGh_uBexXZTzfjBETDY0kQL
 csv_detective/detect_fields/FR/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 csv_detective/detect_fields/FR/geo/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 csv_detective/detect_fields/FR/geo/adresse/__init__.py,sha256=e5JqMNOPxx0Ivju3zAHCGMopZroCpR4vr3DJKlQhMz4,1675
-csv_detective/detect_fields/FR/geo/code_commune_insee/__init__.py,sha256=l3-4WlLacNVngWWcgNhxwYMACFEKQRky_KJo_M7g5fc,90
-csv_detective/detect_fields/FR/geo/code_departement/__init__.py,sha256=cv53Vw0uqsXu1zl47JR7WPGP0PKjWgNJ_2ibrKAc3tU,153
-csv_detective/detect_fields/FR/geo/code_fantoir/__init__.py,sha256=rRnOsK5Ax8Dy1MyGUq-o7Kcs89okCXpwllJcOF79Ns0,127
-csv_detective/detect_fields/FR/geo/code_postal/__init__.py,sha256=DHFC0I05Iywt9eVgflLtsmK78PYomI2KNiAuGQRm4CM,77
-csv_detective/detect_fields/FR/geo/code_region/__init__.py,sha256=AEirwDhkjCpymsaB09Nc8OmcexN2eaLNyipizYe3m4Q,195
-csv_detective/detect_fields/FR/geo/commune/__init__.py,sha256=mfLJu1elZiNGF5Uh565HvOTUjRrAKqj45QjYQ41uw0w,176
-csv_detective/detect_fields/FR/geo/departement/__init__.py,sha256=9TUR7YnYhkJDxpUjlK2BRudDCsvHuH57sXpne7Kjb1g,188
-csv_detective/detect_fields/FR/geo/insee_canton/__init__.py,sha256=k09WqKkB-RgR1Dr0nvO8iaxyvROj2wcV3t8Vc4JJSdQ,173
-csv_detective/detect_fields/FR/geo/latitude_l93/__init__.py,sha256=15g9DJBvJUXQwOS3vz9l-77as3e1AC7sTaMQVyj5xHg,496
+csv_detective/detect_fields/FR/geo/code_commune_insee/__init__.py,sha256=tfHdqUnCQ0cv-fBo3Cy--8UNXzgjld4kseI5eQ_sR4E,187
+csv_detective/detect_fields/FR/geo/code_departement/__init__.py,sha256=unr-Y4zquKSM5PVUiQGnOm-zQvaN8qd3v_XHf0W2VH8,378
+csv_detective/detect_fields/FR/geo/code_fantoir/__init__.py,sha256=27bCkZP5w7tpsKUdOIXuiAG90DTdw066CWg3G5HtsKE,160
+csv_detective/detect_fields/FR/geo/code_postal/__init__.py,sha256=e1SdnW8zVSxrRMm-CeK9tlkLzORP2C6KOInTWnB7h3o,134
+csv_detective/detect_fields/FR/geo/code_region/__init__.py,sha256=y-TPljkf-W209tp7V0RnJ34936XxB6FA2-XPYK3DV8I,253
+csv_detective/detect_fields/FR/geo/commune/__init__.py,sha256=tZ4d1BQd9Xow0SWBcmuGlnX-RKHDzCstdY9AsXM6-Nk,379
+csv_detective/detect_fields/FR/geo/departement/__init__.py,sha256=je2zLsPlK_X189bbmKzf4BJSEoFShxMz2eQNXB7hsh0,399
+csv_detective/detect_fields/FR/geo/insee_canton/__init__.py,sha256=3uNN_Iha6dFfm24CluUmkHFg6nj7kRQaXrHDEcLfyjY,373
+csv_detective/detect_fields/FR/geo/latitude_l93/__init__.py,sha256=5v07RJbi12eoPa-e_-q8xlWBew80FPMxsggcMgZQiI8,438
 csv_detective/detect_fields/FR/geo/latitude_wgs_fr_metropole/__init__.py,sha256=WjPHg8s0ND6bOwS-yo6FP1dnwD-6SWg9oH1K0avHsbI,344
-csv_detective/detect_fields/FR/geo/longitude_l93/__init__.py,sha256=4ZJZRIyr4RSaLzMoxoVNME-HrA2_mF1V1CluxgTGp_0,499
+csv_detective/detect_fields/FR/geo/longitude_l93/__init__.py,sha256=ZwThPSfbRwNHA_anuplxTPYHK-WMduc_np2Xw9XsApM,442
 csv_detective/detect_fields/FR/geo/longitude_wgs_fr_metropole/__init__.py,sha256=d4fLK4IndwllDhsddyTbyRiPfc8O9wT0pLIRI_C3QvQ,344
-csv_detective/detect_fields/FR/geo/pays/__init__.py,sha256=kFXGruWjn5EfKVQIfjiLEhNc73p_N2VgZCl-l8DIHqs,166
-csv_detective/detect_fields/FR/geo/region/__init__.py,sha256=pajL5nr2zWnzCiVsC9SZcfMfjXLUJXm0QKLcyMecVYg,171
+csv_detective/detect_fields/FR/geo/pays/__init__.py,sha256=2q5T4SmCK6ZFF1mrv7d-q9tOIQKBcROI24y_UYIuvz0,383
+csv_detective/detect_fields/FR/geo/region/__init__.py,sha256=JbFKDd4jAnd9yb7YqP36MoLdO1JFPm1cg60fGXt6ZvI,1074
 csv_detective/detect_fields/FR/other/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 csv_detective/detect_fields/FR/other/code_csp_insee/__init__.py,sha256=X0NT6YbBg9PrxIcBwzUCQuBiv_QdDdqb3CJnrlent28,566
 csv_detective/detect_fields/FR/other/code_csp_insee/code_csp_insee.txt,sha256=rbcjtMP6qTZ7BTU6ZegkiXKCruqY_m9Ep6ZgRabFS_E,2486
-csv_detective/detect_fields/FR/other/code_rna/__init__.py,sha256=7bQiT-Mx7e7lW2MSydKXCIk_D8xjWLdWhQIxT7q4fG4,121
+csv_detective/detect_fields/FR/other/code_rna/__init__.py,sha256=Z0RjMBt1--ZL7Jd1RsHAQCCbTAQk_BnlnTq8VF1o_VA,146
 csv_detective/detect_fields/FR/other/code_waldec/__init__.py,sha256=g9n5sOjRlk4I9YFZjdaTYrXf8ftXRDunGZOUpYhN4fA,295
 csv_detective/detect_fields/FR/other/csp_insee/__init__.py,sha256=XacU_3rwXqtdbw_ULTSnu0OOtx0w_rKlviCrLmNdHjc,496
 csv_detective/detect_fields/FR/other/csp_insee/csp_insee.txt,sha256=kgKaKc-5PHu5U4--ugLjpFyMNtTU9CGdZ9ANU3YAsM4,32879
@@ -125,17 +125,17 @@ csv_detective/detect_labels/temp/date/__init__.py,sha256=GrIbo64WVM3hi7ShBRKKyKU
 csv_detective/detect_labels/temp/datetime_iso/__init__.py,sha256=Ih9l56nBcdmGLyWDavVUWuUUuVZBz9QUDE1hHzADvVg,1157
 csv_detective/detect_labels/temp/datetime_rfc822/__init__.py,sha256=DQ_h4uDW1e6qu2rATEhgGKw6O-vVi7HbDhbEDDCT9uY,1175
 csv_detective/detect_labels/temp/year/__init__.py,sha256=zPF_mvhzhXMAlHPAskS8mhuxjLj2AlKpV4ss8Q4tDms,1150
-csv_detective-0.7.4.dev969.data/data/share/csv_detective/CHANGELOG.md,sha256=Cf1Jgo4-ZbpMdF78GZW6CqVBbjDwA2m0O9Js0Fv9M7w,6547
-csv_detective-0.7.4.dev969.data/data/share/csv_detective/LICENSE.AGPL.txt,sha256=2N5ReRelkdqkR9a-KP-y-shmcD5P62XoYiG-miLTAzo,34519
-csv_detective-0.7.4.dev969.data/data/share/csv_detective/README.md,sha256=Qr8xRXc-dxQ-tdXCpCTCKp1Uliqq84r0UOlPRNuGCpI,9506
+csv_detective-0.7.5.dev1009.data/data/share/csv_detective/CHANGELOG.md,sha256=4ABp5UF2L6tPg-eK7Dj6NWgnFnkU74BwhrMzrRGJ2Lw,6585
+csv_detective-0.7.5.dev1009.data/data/share/csv_detective/LICENSE.AGPL.txt,sha256=2N5ReRelkdqkR9a-KP-y-shmcD5P62XoYiG-miLTAzo,34519
+csv_detective-0.7.5.dev1009.data/data/share/csv_detective/README.md,sha256=Qr8xRXc-dxQ-tdXCpCTCKp1Uliqq84r0UOlPRNuGCpI,9506
 tests/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-tests/test_fields.py,sha256=Uq9eJaK3D8b_lDd_4Q3aMGUHP4NkrpY6g07LUnJcDDc,10587
+tests/test_fields.py,sha256=kXel-hiyQYrJ3OLmwUMg1K3DKbbwBLvUplxZWxpp18I,10605
 tests/test_file.py,sha256=1fEOu3bArGBaarRKAoTXAF3cSIGJfFN3UIwOW6esWRs,6399
 tests/test_labels.py,sha256=6MOKrGznkwU5fjZ_3oiB6Scmb480Eu-9geBJs0UDLds,159
 tests/test_structure.py,sha256=SVsnluVoIIprYw_67I1_gB3cp9m1wlO8C7SpdsLW8cM,1161
-csv_detective-0.7.4.dev969.dist-info/LICENSE.AGPL.txt,sha256=2N5ReRelkdqkR9a-KP-y-shmcD5P62XoYiG-miLTAzo,34519
-csv_detective-0.7.4.dev969.dist-info/METADATA,sha256=m6_AMJ_ku1rE4sFMdSG-y5A3ML2Z5excqpR-a9a7VkM,1089
-csv_detective-0.7.4.dev969.dist-info/WHEEL,sha256=R06PA3UVYHThwHvxuRWMqaGcr-PuniXahwjmQRFMEkY,91
-csv_detective-0.7.4.dev969.dist-info/entry_points.txt,sha256=JjweTReFqKJmuvkegzlew2j3D5pZzfxvbEGOtGVGmaY,56
-csv_detective-0.7.4.dev969.dist-info/top_level.txt,sha256=M0Nv646VHo-49zWjPkwo2C48UmtfddV8_9mEZeIxy8Q,20
-csv_detective-0.7.4.dev969.dist-info/RECORD,,
+csv_detective-0.7.5.dev1009.dist-info/LICENSE.AGPL.txt,sha256=2N5ReRelkdqkR9a-KP-y-shmcD5P62XoYiG-miLTAzo,34519
+csv_detective-0.7.5.dev1009.dist-info/METADATA,sha256=dd1MAjbcCr_c82zoDPcO5zbfB8KLLDd8WdDCMWtchzA,1090
+csv_detective-0.7.5.dev1009.dist-info/WHEEL,sha256=R06PA3UVYHThwHvxuRWMqaGcr-PuniXahwjmQRFMEkY,91
+csv_detective-0.7.5.dev1009.dist-info/entry_points.txt,sha256=JjweTReFqKJmuvkegzlew2j3D5pZzfxvbEGOtGVGmaY,56
+csv_detective-0.7.5.dev1009.dist-info/top_level.txt,sha256=M0Nv646VHo-49zWjPkwo2C48UmtfddV8_9mEZeIxy8Q,20
+csv_detective-0.7.5.dev1009.dist-info/RECORD,,

tests/test_fields.py

@@ -282,26 +282,26 @@ def test_do_not_match_canton():
 
 # latitude_l93
 def test_match_latitude_l93():
-    vals = [6037008, 7123528.5, "7124528,5"]
+    vals = ["6037008", "7123528.5", "7124528,5"]
     for val in vals:
         assert latitude_l93._is(val)
 
 
 def test_do_not_match_latitude_93():
-    vals = [0, -6734529.6, 7245669.8, "3422674,78", "32_34"]
+    vals = ["0", "-6734529.6", "7245669.8", "3422674,78", "32_34"]
     for val in vals:
         assert not latitude_l93._is(val)
 
 
 # longitude_l93
 def test_match_longitude_l93():
-    vals = [0, -154, "1265783,45", 34723.4]
+    vals = ["0", "-154", "1265783,45", "34723.4"]
     for val in vals:
         assert longitude_l93._is(val)
 
 
 def test_do_not_match_longitude_93():
-    vals = [1456669.8, "-776225", "346_3214"]
+    vals = ["1456669.8", "-776225", "346_3214"]
     for val in vals:
         assert not longitude_l93._is(val)