airbyte-cdk 0.50.0__py3-none-any.whl → 0.50.2__py3-none-any.whl
Sign up to get free protection for your applications and to get access to all the features.
- airbyte_cdk/entrypoint.py +7 -0
- airbyte_cdk/sources/declarative/declarative_component_schema.yaml +3 -3
- airbyte_cdk/sources/declarative/models/declarative_component_schema.py +3 -3
- airbyte_cdk/sources/file_based/availability_strategy/default_file_based_availability_strategy.py +9 -9
- airbyte_cdk/sources/file_based/config/csv_format.py +42 -6
- airbyte_cdk/sources/file_based/file_based_source.py +4 -5
- airbyte_cdk/sources/file_based/file_types/csv_parser.py +114 -59
- airbyte_cdk/sources/file_based/stream/cursor/__init__.py +2 -2
- airbyte_cdk/sources/file_based/stream/cursor/{file_based_cursor.py → abstract_file_based_cursor.py} +9 -1
- airbyte_cdk/sources/file_based/stream/cursor/default_file_based_cursor.py +10 -10
- airbyte_cdk/sources/file_based/stream/default_file_based_stream.py +15 -2
- {airbyte_cdk-0.50.0.dist-info → airbyte_cdk-0.50.2.dist-info}/METADATA +1 -1
- {airbyte_cdk-0.50.0.dist-info → airbyte_cdk-0.50.2.dist-info}/RECORD +25 -24
- unit_tests/sources/file_based/config/test_csv_format.py +23 -0
- unit_tests/sources/file_based/file_types/test_csv_parser.py +50 -18
- unit_tests/sources/file_based/helpers.py +5 -0
- unit_tests/sources/file_based/in_memory_files_source.py +11 -3
- unit_tests/sources/file_based/scenarios/csv_scenarios.py +1254 -47
- unit_tests/sources/file_based/scenarios/incremental_scenarios.py +6 -5
- unit_tests/sources/file_based/scenarios/scenario_builder.py +8 -7
- unit_tests/sources/file_based/stream/test_default_file_based_cursor.py +13 -12
- unit_tests/sources/file_based/test_scenarios.py +30 -0
- {airbyte_cdk-0.50.0.dist-info → airbyte_cdk-0.50.2.dist-info}/LICENSE.txt +0 -0
- {airbyte_cdk-0.50.0.dist-info → airbyte_cdk-0.50.2.dist-info}/WHEEL +0 -0
- {airbyte_cdk-0.50.0.dist-info → airbyte_cdk-0.50.2.dist-info}/top_level.txt +0 -0
|
@@ -45,45 +45,68 @@ single_csv_scenario = (
|
|
|
45
45
|
"properties": {
|
|
46
46
|
"streams": {
|
|
47
47
|
"title": "The list of streams to sync",
|
|
48
|
-
"description":
|
|
48
|
+
"description": "Each instance of this configuration defines a <a href=\"https://docs.airbyte.com/cloud/core-concepts#stream\">stream</a>. Use this to define which files belong in the stream, their format, and how they should be parsed and validated. When sending data to warehouse destination such as Snowflake or BigQuery, each stream is a separate table.",
|
|
49
49
|
"order": 10,
|
|
50
50
|
"type": "array",
|
|
51
51
|
"items": {
|
|
52
52
|
"title": "FileBasedStreamConfig",
|
|
53
53
|
"type": "object",
|
|
54
54
|
"properties": {
|
|
55
|
-
"name": {
|
|
55
|
+
"name": {
|
|
56
|
+
"title": "Name",
|
|
57
|
+
"description": "The name of the stream.",
|
|
58
|
+
"type": "string"
|
|
59
|
+
},
|
|
56
60
|
"file_type": {
|
|
57
61
|
"title": "File Type",
|
|
58
62
|
"description": "The data file type that is being extracted for a stream.",
|
|
59
|
-
"type": "string"
|
|
63
|
+
"type": "string"
|
|
60
64
|
},
|
|
61
65
|
"globs": {
|
|
62
66
|
"title": "Globs",
|
|
63
|
-
"description":
|
|
67
|
+
"description": "The pattern used to specify which files should be selected from the file system. For more information on glob pattern matching look <a href=\"https://en.wikipedia.org/wiki/Glob_(programming)\">here</a>.",
|
|
64
68
|
"type": "array",
|
|
65
|
-
"items": {
|
|
69
|
+
"items": {
|
|
70
|
+
"type": "string"
|
|
71
|
+
}
|
|
66
72
|
},
|
|
67
73
|
"validation_policy": {
|
|
68
74
|
"title": "Validation Policy",
|
|
69
75
|
"description": "The name of the validation policy that dictates sync behavior when a record does not adhere to the stream schema.",
|
|
70
|
-
"type": "string"
|
|
76
|
+
"type": "string"
|
|
71
77
|
},
|
|
72
78
|
"input_schema": {
|
|
73
79
|
"title": "Input Schema",
|
|
74
80
|
"description": "The schema that will be used to validate records extracted from the file. This will override the stream schema that is auto-detected from incoming files.",
|
|
75
|
-
"oneOf": [
|
|
81
|
+
"oneOf": [
|
|
82
|
+
{
|
|
83
|
+
"type": "object"
|
|
84
|
+
},
|
|
85
|
+
{
|
|
86
|
+
"type": "string"
|
|
87
|
+
}
|
|
88
|
+
]
|
|
76
89
|
},
|
|
77
90
|
"primary_key": {
|
|
78
91
|
"title": "Primary Key",
|
|
79
92
|
"description": "The column or columns (for a composite key) that serves as the unique identifier of a record.",
|
|
80
|
-
"oneOf": [
|
|
93
|
+
"oneOf": [
|
|
94
|
+
{
|
|
95
|
+
"type": "string"
|
|
96
|
+
},
|
|
97
|
+
{
|
|
98
|
+
"type": "array",
|
|
99
|
+
"items": {
|
|
100
|
+
"type": "string"
|
|
101
|
+
}
|
|
102
|
+
}
|
|
103
|
+
]
|
|
81
104
|
},
|
|
82
105
|
"days_to_sync_if_history_is_full": {
|
|
83
106
|
"title": "Days To Sync If History Is Full",
|
|
84
107
|
"description": "When the state history of the file store is full, syncs will only read files that were last modified in the provided day range.",
|
|
85
108
|
"default": 3,
|
|
86
|
-
"type": "integer"
|
|
109
|
+
"type": "integer"
|
|
87
110
|
},
|
|
88
111
|
"format": {
|
|
89
112
|
"oneOf": [
|
|
@@ -100,16 +123,18 @@ single_csv_scenario = (
|
|
|
100
123
|
"filetype": {
|
|
101
124
|
"title": "Filetype",
|
|
102
125
|
"default": "avro",
|
|
103
|
-
"enum": [
|
|
104
|
-
|
|
126
|
+
"enum": [
|
|
127
|
+
"avro"
|
|
128
|
+
],
|
|
129
|
+
"type": "string"
|
|
105
130
|
},
|
|
106
131
|
"decimal_as_float": {
|
|
107
132
|
"title": "Convert Decimal Fields to Floats",
|
|
108
133
|
"description": "Whether to convert decimal fields to floats. There is a loss of precision when converting decimals to floats, so this is not recommended.",
|
|
109
134
|
"default": False,
|
|
110
|
-
"type": "boolean"
|
|
111
|
-
}
|
|
112
|
-
}
|
|
135
|
+
"type": "boolean"
|
|
136
|
+
}
|
|
137
|
+
}
|
|
113
138
|
},
|
|
114
139
|
{
|
|
115
140
|
"title": "CsvFormat",
|
|
@@ -118,37 +143,39 @@ single_csv_scenario = (
|
|
|
118
143
|
"filetype": {
|
|
119
144
|
"title": "Filetype",
|
|
120
145
|
"default": "csv",
|
|
121
|
-
"enum": [
|
|
122
|
-
|
|
146
|
+
"enum": [
|
|
147
|
+
"csv"
|
|
148
|
+
],
|
|
149
|
+
"type": "string"
|
|
123
150
|
},
|
|
124
151
|
"delimiter": {
|
|
125
152
|
"title": "Delimiter",
|
|
126
153
|
"description": "The character delimiting individual cells in the CSV data. This may only be a 1-character string. For tab-delimited data enter '\\t'.",
|
|
127
154
|
"default": ",",
|
|
128
|
-
"type": "string"
|
|
155
|
+
"type": "string"
|
|
129
156
|
},
|
|
130
157
|
"quote_char": {
|
|
131
158
|
"title": "Quote Character",
|
|
132
159
|
"description": "The character used for quoting CSV values. To disallow quoting, make this field blank.",
|
|
133
|
-
"default":
|
|
134
|
-
"type": "string"
|
|
160
|
+
"default": "\"",
|
|
161
|
+
"type": "string"
|
|
135
162
|
},
|
|
136
163
|
"escape_char": {
|
|
137
164
|
"title": "Escape Character",
|
|
138
165
|
"description": "The character used for escaping special characters. To disallow escaping, leave this field blank.",
|
|
139
|
-
"type": "string"
|
|
166
|
+
"type": "string"
|
|
140
167
|
},
|
|
141
168
|
"encoding": {
|
|
142
169
|
"title": "Encoding",
|
|
143
|
-
"description":
|
|
170
|
+
"description": "The character encoding of the CSV data. Leave blank to default to <strong>UTF8</strong>. See <a href=\"https://docs.python.org/3/library/codecs.html#standard-encodings\" target=\"_blank\">list of python encodings</a> for allowable options.",
|
|
144
171
|
"default": "utf8",
|
|
145
|
-
"type": "string"
|
|
172
|
+
"type": "string"
|
|
146
173
|
},
|
|
147
174
|
"double_quote": {
|
|
148
175
|
"title": "Double Quote",
|
|
149
176
|
"description": "Whether two quotes in a quoted CSV value denote a single quote in the data.",
|
|
150
177
|
"default": True,
|
|
151
|
-
"type": "boolean"
|
|
178
|
+
"type": "boolean"
|
|
152
179
|
},
|
|
153
180
|
"quoting_behavior": {
|
|
154
181
|
"title": "Quoting Behavior",
|
|
@@ -158,10 +185,72 @@ single_csv_scenario = (
|
|
|
158
185
|
"Quote All",
|
|
159
186
|
"Quote Special Characters",
|
|
160
187
|
"Quote Non-numeric",
|
|
161
|
-
"Quote None"
|
|
188
|
+
"Quote None"
|
|
189
|
+
]
|
|
190
|
+
},
|
|
191
|
+
"null_values": {
|
|
192
|
+
"title": "Null Values",
|
|
193
|
+
"description": "A set of case-sensitive strings that should be interpreted as null values. For example, if the value 'NA' should be interpreted as null, enter 'NA' in this field.",
|
|
194
|
+
"default": [],
|
|
195
|
+
"type": "array",
|
|
196
|
+
"items": {
|
|
197
|
+
"type": "string"
|
|
198
|
+
},
|
|
199
|
+
"uniqueItems": True
|
|
200
|
+
},
|
|
201
|
+
"skip_rows_before_header": {
|
|
202
|
+
"title": "Skip Rows Before Header",
|
|
203
|
+
"description": "The number of rows to skip before the header row. For example, if the header row is on the 3rd row, enter 2 in this field.",
|
|
204
|
+
"default": 0,
|
|
205
|
+
"type": "integer"
|
|
206
|
+
},
|
|
207
|
+
"skip_rows_after_header": {
|
|
208
|
+
"title": "Skip Rows After Header",
|
|
209
|
+
"description": "The number of rows to skip after the header row.",
|
|
210
|
+
"default": 0,
|
|
211
|
+
"type": "integer"
|
|
212
|
+
},
|
|
213
|
+
"autogenerate_column_names": {
|
|
214
|
+
"title": "Autogenerate Column Names",
|
|
215
|
+
"description": "Whether to autogenerate column names if column_names is empty. If true, column names will be of the form \u201cf0\u201d, \u201cf1\u201d\u2026 If false, column names will be read from the first CSV row after skip_rows_before_header.",
|
|
216
|
+
"default": False,
|
|
217
|
+
"type": "boolean"
|
|
218
|
+
},
|
|
219
|
+
"true_values": {
|
|
220
|
+
"title": "True Values",
|
|
221
|
+
"description": "A set of case-sensitive strings that should be interpreted as true values.",
|
|
222
|
+
"default": [
|
|
223
|
+
"y",
|
|
224
|
+
"yes",
|
|
225
|
+
"t",
|
|
226
|
+
"true",
|
|
227
|
+
"on",
|
|
228
|
+
"1"
|
|
162
229
|
],
|
|
230
|
+
"type": "array",
|
|
231
|
+
"items": {
|
|
232
|
+
"type": "string"
|
|
233
|
+
},
|
|
234
|
+
"uniqueItems": True
|
|
163
235
|
},
|
|
164
|
-
|
|
236
|
+
"false_values": {
|
|
237
|
+
"title": "False Values",
|
|
238
|
+
"description": "A set of case-sensitive strings that should be interpreted as false values.",
|
|
239
|
+
"default": [
|
|
240
|
+
"n",
|
|
241
|
+
"no",
|
|
242
|
+
"f",
|
|
243
|
+
"false",
|
|
244
|
+
"off",
|
|
245
|
+
"0"
|
|
246
|
+
],
|
|
247
|
+
"type": "array",
|
|
248
|
+
"items": {
|
|
249
|
+
"type": "string"
|
|
250
|
+
},
|
|
251
|
+
"uniqueItems": True
|
|
252
|
+
}
|
|
253
|
+
}
|
|
165
254
|
},
|
|
166
255
|
{
|
|
167
256
|
"title": "JsonlFormat",
|
|
@@ -170,10 +259,12 @@ single_csv_scenario = (
|
|
|
170
259
|
"filetype": {
|
|
171
260
|
"title": "Filetype",
|
|
172
261
|
"default": "jsonl",
|
|
173
|
-
"enum": [
|
|
174
|
-
|
|
262
|
+
"enum": [
|
|
263
|
+
"jsonl"
|
|
264
|
+
],
|
|
265
|
+
"type": "string"
|
|
175
266
|
}
|
|
176
|
-
}
|
|
267
|
+
}
|
|
177
268
|
},
|
|
178
269
|
{
|
|
179
270
|
"title": "ParquetFormat",
|
|
@@ -182,50 +273,67 @@ single_csv_scenario = (
|
|
|
182
273
|
"filetype": {
|
|
183
274
|
"title": "Filetype",
|
|
184
275
|
"default": "parquet",
|
|
185
|
-
"enum": [
|
|
186
|
-
|
|
276
|
+
"enum": [
|
|
277
|
+
"parquet"
|
|
278
|
+
],
|
|
279
|
+
"type": "string"
|
|
187
280
|
},
|
|
188
281
|
"decimal_as_float": {
|
|
189
282
|
"title": "Convert Decimal Fields to Floats",
|
|
190
283
|
"description": "Whether to convert decimal fields to floats. There is a loss of precision when converting decimals to floats, so this is not recommended.",
|
|
191
284
|
"default": False,
|
|
192
|
-
"type": "boolean"
|
|
193
|
-
}
|
|
194
|
-
}
|
|
195
|
-
}
|
|
285
|
+
"type": "boolean"
|
|
286
|
+
}
|
|
287
|
+
}
|
|
288
|
+
}
|
|
196
289
|
]
|
|
197
|
-
}
|
|
290
|
+
}
|
|
198
291
|
},
|
|
199
292
|
{
|
|
200
293
|
"title": "Legacy Format",
|
|
201
|
-
"required": [
|
|
294
|
+
"required": [
|
|
295
|
+
"filetype"
|
|
296
|
+
],
|
|
202
297
|
"type": "object",
|
|
203
|
-
"properties": {
|
|
204
|
-
|
|
298
|
+
"properties": {
|
|
299
|
+
"filetype": {
|
|
300
|
+
"title": "Filetype",
|
|
301
|
+
"type": "string"
|
|
302
|
+
}
|
|
303
|
+
}
|
|
304
|
+
}
|
|
205
305
|
]
|
|
206
306
|
},
|
|
207
307
|
"schemaless": {
|
|
208
308
|
"title": "Schemaless",
|
|
209
309
|
"description": "When enabled, syncs will not validate or structure records against the stream's schema.",
|
|
210
310
|
"default": False,
|
|
211
|
-
"type": "boolean"
|
|
212
|
-
}
|
|
311
|
+
"type": "boolean"
|
|
312
|
+
}
|
|
213
313
|
},
|
|
214
|
-
"required": [
|
|
215
|
-
|
|
314
|
+
"required": [
|
|
315
|
+
"name",
|
|
316
|
+
"file_type",
|
|
317
|
+
"validation_policy"
|
|
318
|
+
]
|
|
319
|
+
}
|
|
216
320
|
},
|
|
217
321
|
"start_date": {
|
|
218
322
|
"title": "Start Date",
|
|
219
323
|
"description": "UTC date and time in the format 2017-01-25T00:00:00Z. Any file modified before this date will not be replicated.",
|
|
220
|
-
"examples": [
|
|
324
|
+
"examples": [
|
|
325
|
+
"2021-01-01T00:00:00Z"
|
|
326
|
+
],
|
|
221
327
|
"format": "date-time",
|
|
222
328
|
"pattern": "^[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2}Z$",
|
|
223
329
|
"order": 1,
|
|
224
|
-
"type": "string"
|
|
225
|
-
}
|
|
330
|
+
"type": "string"
|
|
331
|
+
}
|
|
226
332
|
},
|
|
227
|
-
"required": [
|
|
228
|
-
|
|
333
|
+
"required": [
|
|
334
|
+
"streams"
|
|
335
|
+
]
|
|
336
|
+
}
|
|
229
337
|
}
|
|
230
338
|
)
|
|
231
339
|
.set_expected_catalog(
|
|
@@ -1531,3 +1639,1102 @@ schemaless_with_user_input_schema_fails_connection_check_multi_stream_scenario =
|
|
|
1531
1639
|
.set_expected_discover_error(ConfigValidationError, FileBasedSourceError.CONFIG_VALIDATION_ERROR.value)
|
|
1532
1640
|
.set_expected_read_error(ConfigValidationError, FileBasedSourceError.CONFIG_VALIDATION_ERROR.value)
|
|
1533
1641
|
).build()
|
|
1642
|
+
|
|
1643
|
+
csv_string_can_be_null_with_input_schemas_scenario = (
|
|
1644
|
+
TestScenarioBuilder()
|
|
1645
|
+
.set_name("csv_string_can_be_null_with_input_schema")
|
|
1646
|
+
.set_config(
|
|
1647
|
+
{
|
|
1648
|
+
"streams": [
|
|
1649
|
+
{
|
|
1650
|
+
"name": "stream1",
|
|
1651
|
+
"file_type": "csv",
|
|
1652
|
+
"globs": ["*"],
|
|
1653
|
+
"validation_policy": "emit_record",
|
|
1654
|
+
"input_schema": {"col1": "string", "col2": "string"},
|
|
1655
|
+
"format": {
|
|
1656
|
+
"csv": {
|
|
1657
|
+
"filetype": "csv",
|
|
1658
|
+
"null_values": ["null"],
|
|
1659
|
+
}
|
|
1660
|
+
}
|
|
1661
|
+
}
|
|
1662
|
+
],
|
|
1663
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
1664
|
+
}
|
|
1665
|
+
)
|
|
1666
|
+
.set_files(
|
|
1667
|
+
{
|
|
1668
|
+
"a.csv": {
|
|
1669
|
+
"contents": [
|
|
1670
|
+
("col1", "col2"),
|
|
1671
|
+
("2", "null"),
|
|
1672
|
+
],
|
|
1673
|
+
"last_modified": "2023-06-05T03:54:07.000000Z",
|
|
1674
|
+
}
|
|
1675
|
+
}
|
|
1676
|
+
)
|
|
1677
|
+
.set_file_type("csv")
|
|
1678
|
+
.set_expected_catalog(
|
|
1679
|
+
{
|
|
1680
|
+
"streams": [
|
|
1681
|
+
{
|
|
1682
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
1683
|
+
"json_schema": {
|
|
1684
|
+
"type": "object",
|
|
1685
|
+
"properties": {
|
|
1686
|
+
"col1": {
|
|
1687
|
+
"type": "string"
|
|
1688
|
+
},
|
|
1689
|
+
"col2": {
|
|
1690
|
+
"type": "string"
|
|
1691
|
+
},
|
|
1692
|
+
"_ab_source_file_last_modified": {
|
|
1693
|
+
"type": "string"
|
|
1694
|
+
},
|
|
1695
|
+
"_ab_source_file_url": {
|
|
1696
|
+
"type": "string"
|
|
1697
|
+
},
|
|
1698
|
+
},
|
|
1699
|
+
},
|
|
1700
|
+
"name": "stream1",
|
|
1701
|
+
"source_defined_cursor": True,
|
|
1702
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
1703
|
+
}
|
|
1704
|
+
]
|
|
1705
|
+
}
|
|
1706
|
+
)
|
|
1707
|
+
.set_expected_records(
|
|
1708
|
+
[
|
|
1709
|
+
{"data": {"col1": "2", "col2": None, "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
1710
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
1711
|
+
]
|
|
1712
|
+
)
|
|
1713
|
+
).build()
|
|
1714
|
+
|
|
1715
|
+
csv_string_not_null_if_no_null_values_scenario = (
|
|
1716
|
+
TestScenarioBuilder()
|
|
1717
|
+
.set_name("csv_string_not_null_if_no_null_values")
|
|
1718
|
+
.set_config(
|
|
1719
|
+
{
|
|
1720
|
+
"streams": [
|
|
1721
|
+
{
|
|
1722
|
+
"name": "stream1",
|
|
1723
|
+
"file_type": "csv",
|
|
1724
|
+
"globs": ["*"],
|
|
1725
|
+
"validation_policy": "emit_record",
|
|
1726
|
+
"format": {
|
|
1727
|
+
"csv": {
|
|
1728
|
+
"filetype": "csv",
|
|
1729
|
+
}
|
|
1730
|
+
}
|
|
1731
|
+
}
|
|
1732
|
+
],
|
|
1733
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
1734
|
+
}
|
|
1735
|
+
)
|
|
1736
|
+
.set_files(
|
|
1737
|
+
{
|
|
1738
|
+
"a.csv": {
|
|
1739
|
+
"contents": [
|
|
1740
|
+
("col1", "col2"),
|
|
1741
|
+
("2", "null"),
|
|
1742
|
+
],
|
|
1743
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
1744
|
+
}
|
|
1745
|
+
}
|
|
1746
|
+
)
|
|
1747
|
+
.set_file_type("csv")
|
|
1748
|
+
.set_expected_catalog(
|
|
1749
|
+
{
|
|
1750
|
+
"streams": [
|
|
1751
|
+
{
|
|
1752
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
1753
|
+
"json_schema": {
|
|
1754
|
+
"type": "object",
|
|
1755
|
+
"properties": {
|
|
1756
|
+
"col1": {
|
|
1757
|
+
"type": ["null", "string"]
|
|
1758
|
+
},
|
|
1759
|
+
"col2": {
|
|
1760
|
+
"type": ["null", "string"]
|
|
1761
|
+
},
|
|
1762
|
+
"_ab_source_file_last_modified": {
|
|
1763
|
+
"type": "string"
|
|
1764
|
+
},
|
|
1765
|
+
"_ab_source_file_url": {
|
|
1766
|
+
"type": "string"
|
|
1767
|
+
},
|
|
1768
|
+
},
|
|
1769
|
+
},
|
|
1770
|
+
"name": "stream1",
|
|
1771
|
+
"source_defined_cursor": True,
|
|
1772
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
1773
|
+
}
|
|
1774
|
+
]
|
|
1775
|
+
}
|
|
1776
|
+
)
|
|
1777
|
+
.set_expected_records(
|
|
1778
|
+
[
|
|
1779
|
+
{"data": {"col1": "2", "col2": "null", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
1780
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
1781
|
+
]
|
|
1782
|
+
)
|
|
1783
|
+
).build()
|
|
1784
|
+
|
|
1785
|
+
csv_strings_can_be_null_not_quoted_scenario = (
|
|
1786
|
+
TestScenarioBuilder()
|
|
1787
|
+
.set_name("csv_strings_can_be_null_no_input_schema")
|
|
1788
|
+
.set_config(
|
|
1789
|
+
{
|
|
1790
|
+
"streams": [
|
|
1791
|
+
{
|
|
1792
|
+
"name": "stream1",
|
|
1793
|
+
"file_type": "csv",
|
|
1794
|
+
"globs": ["*"],
|
|
1795
|
+
"validation_policy": "emit_record",
|
|
1796
|
+
"format": {
|
|
1797
|
+
"csv": {
|
|
1798
|
+
"filetype": "csv",
|
|
1799
|
+
"null_values": ["null"]
|
|
1800
|
+
}
|
|
1801
|
+
}
|
|
1802
|
+
}
|
|
1803
|
+
],
|
|
1804
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
1805
|
+
}
|
|
1806
|
+
)
|
|
1807
|
+
.set_files(
|
|
1808
|
+
{
|
|
1809
|
+
"a.csv": {
|
|
1810
|
+
"contents": [
|
|
1811
|
+
("col1", "col2"),
|
|
1812
|
+
("2", "null"),
|
|
1813
|
+
],
|
|
1814
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
1815
|
+
}
|
|
1816
|
+
}
|
|
1817
|
+
)
|
|
1818
|
+
.set_file_type("csv")
|
|
1819
|
+
.set_expected_catalog(
|
|
1820
|
+
{
|
|
1821
|
+
"streams": [
|
|
1822
|
+
{
|
|
1823
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
1824
|
+
"json_schema": {
|
|
1825
|
+
"type": "object",
|
|
1826
|
+
"properties": {
|
|
1827
|
+
"col1": {
|
|
1828
|
+
"type": ["null", "string"]
|
|
1829
|
+
},
|
|
1830
|
+
"col2": {
|
|
1831
|
+
"type": ["null", "string"]
|
|
1832
|
+
},
|
|
1833
|
+
"_ab_source_file_last_modified": {
|
|
1834
|
+
"type": "string"
|
|
1835
|
+
},
|
|
1836
|
+
"_ab_source_file_url": {
|
|
1837
|
+
"type": "string"
|
|
1838
|
+
},
|
|
1839
|
+
},
|
|
1840
|
+
},
|
|
1841
|
+
"name": "stream1",
|
|
1842
|
+
"source_defined_cursor": True,
|
|
1843
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
1844
|
+
}
|
|
1845
|
+
]
|
|
1846
|
+
}
|
|
1847
|
+
)
|
|
1848
|
+
.set_expected_records(
|
|
1849
|
+
[
|
|
1850
|
+
{"data": {"col1": "2", "col2": None, "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
1851
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
1852
|
+
]
|
|
1853
|
+
)
|
|
1854
|
+
).build()
|
|
1855
|
+
|
|
1856
|
+
csv_newline_in_values_quoted_value_scenario = (
|
|
1857
|
+
TestScenarioBuilder()
|
|
1858
|
+
.set_name("csv_newline_in_values_quoted_value")
|
|
1859
|
+
.set_config(
|
|
1860
|
+
{
|
|
1861
|
+
"streams": [
|
|
1862
|
+
{
|
|
1863
|
+
"name": "stream1",
|
|
1864
|
+
"file_type": "csv",
|
|
1865
|
+
"globs": ["*"],
|
|
1866
|
+
"validation_policy": "emit_record",
|
|
1867
|
+
"format": {
|
|
1868
|
+
"csv": {
|
|
1869
|
+
"filetype": "csv",
|
|
1870
|
+
"quoting_behavior": "Quote All"
|
|
1871
|
+
}
|
|
1872
|
+
}
|
|
1873
|
+
}
|
|
1874
|
+
],
|
|
1875
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
1876
|
+
}
|
|
1877
|
+
)
|
|
1878
|
+
.set_files(
|
|
1879
|
+
{
|
|
1880
|
+
"a.csv": {
|
|
1881
|
+
"contents": [
|
|
1882
|
+
'''"col1","col2"''',
|
|
1883
|
+
'''"2","val\n2"''',
|
|
1884
|
+
],
|
|
1885
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
1886
|
+
}
|
|
1887
|
+
}
|
|
1888
|
+
)
|
|
1889
|
+
.set_file_type("csv")
|
|
1890
|
+
.set_expected_catalog(
|
|
1891
|
+
{
|
|
1892
|
+
"streams": [
|
|
1893
|
+
{
|
|
1894
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
1895
|
+
"json_schema": {
|
|
1896
|
+
"type": "object",
|
|
1897
|
+
"properties": {
|
|
1898
|
+
"col1": {
|
|
1899
|
+
"type": ["null", "string"]
|
|
1900
|
+
},
|
|
1901
|
+
"col2": {
|
|
1902
|
+
"type": ["null", "string"]
|
|
1903
|
+
},
|
|
1904
|
+
"_ab_source_file_last_modified": {
|
|
1905
|
+
"type": "string"
|
|
1906
|
+
},
|
|
1907
|
+
"_ab_source_file_url": {
|
|
1908
|
+
"type": "string"
|
|
1909
|
+
},
|
|
1910
|
+
},
|
|
1911
|
+
},
|
|
1912
|
+
"name": "stream1",
|
|
1913
|
+
"source_defined_cursor": True,
|
|
1914
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
1915
|
+
}
|
|
1916
|
+
]
|
|
1917
|
+
}
|
|
1918
|
+
)
|
|
1919
|
+
.set_expected_records(
|
|
1920
|
+
[
|
|
1921
|
+
{"data": {"col1": "2", "col2": 'val\n2', "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
1922
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
1923
|
+
]
|
|
1924
|
+
)
|
|
1925
|
+
).build()
|
|
1926
|
+
|
|
1927
|
+
csv_newline_in_values_not_quoted_scenario = (
|
|
1928
|
+
TestScenarioBuilder()
|
|
1929
|
+
.set_name("csv_newline_in_values_not_quoted")
|
|
1930
|
+
.set_config(
|
|
1931
|
+
{
|
|
1932
|
+
"streams": [
|
|
1933
|
+
{
|
|
1934
|
+
"name": "stream1",
|
|
1935
|
+
"file_type": "csv",
|
|
1936
|
+
"globs": ["*"],
|
|
1937
|
+
"validation_policy": "emit_record",
|
|
1938
|
+
"format": {
|
|
1939
|
+
"csv": {
|
|
1940
|
+
"filetype": "csv",
|
|
1941
|
+
}
|
|
1942
|
+
}
|
|
1943
|
+
}
|
|
1944
|
+
],
|
|
1945
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
1946
|
+
}
|
|
1947
|
+
)
|
|
1948
|
+
.set_files(
|
|
1949
|
+
{
|
|
1950
|
+
"a.csv": {
|
|
1951
|
+
"contents": [
|
|
1952
|
+
'''col1,col2''',
|
|
1953
|
+
'''2,val\n2''',
|
|
1954
|
+
],
|
|
1955
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
1956
|
+
}
|
|
1957
|
+
}
|
|
1958
|
+
)
|
|
1959
|
+
.set_file_type("csv")
|
|
1960
|
+
.set_expected_catalog(
|
|
1961
|
+
{
|
|
1962
|
+
"streams": [
|
|
1963
|
+
{
|
|
1964
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
1965
|
+
"json_schema": {
|
|
1966
|
+
"type": "object",
|
|
1967
|
+
"properties": {
|
|
1968
|
+
"col1": {
|
|
1969
|
+
"type": ["null", "string"]
|
|
1970
|
+
},
|
|
1971
|
+
"col2": {
|
|
1972
|
+
"type": ["null", "string"]
|
|
1973
|
+
},
|
|
1974
|
+
"_ab_source_file_last_modified": {
|
|
1975
|
+
"type": "string"
|
|
1976
|
+
},
|
|
1977
|
+
"_ab_source_file_url": {
|
|
1978
|
+
"type": "string"
|
|
1979
|
+
},
|
|
1980
|
+
},
|
|
1981
|
+
},
|
|
1982
|
+
"name": "stream1",
|
|
1983
|
+
"source_defined_cursor": True,
|
|
1984
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
1985
|
+
}
|
|
1986
|
+
]
|
|
1987
|
+
}
|
|
1988
|
+
)
|
|
1989
|
+
.set_expected_records(
|
|
1990
|
+
[
|
|
1991
|
+
# Note that the value for col2 is truncated to "val" because the newline is not escaped
|
|
1992
|
+
{"data": {"col1": "2", "col2": 'val', "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
1993
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
1994
|
+
]
|
|
1995
|
+
)
|
|
1996
|
+
.set_expected_logs({"read": [
|
|
1997
|
+
{
|
|
1998
|
+
"level": "ERROR",
|
|
1999
|
+
"message": "Error parsing record. This could be due to a mismatch between the config's file type and the actual file type, or because the file or record is not parseable. stream=stream1 file=a.csv line_no=2 n_skipped=0",
|
|
2000
|
+
}
|
|
2001
|
+
]})
|
|
2002
|
+
).build()
|
|
2003
|
+
|
|
2004
|
+
csv_escape_char_is_set_scenario = (
|
|
2005
|
+
TestScenarioBuilder()
|
|
2006
|
+
.set_name("csv_escape_char_is_set")
|
|
2007
|
+
.set_config(
|
|
2008
|
+
{
|
|
2009
|
+
"streams": [
|
|
2010
|
+
{
|
|
2011
|
+
"name": "stream1",
|
|
2012
|
+
"file_type": "csv",
|
|
2013
|
+
"globs": ["*"],
|
|
2014
|
+
"validation_policy": "emit_record",
|
|
2015
|
+
"format": {
|
|
2016
|
+
"csv": {
|
|
2017
|
+
"filetype": "csv",
|
|
2018
|
+
"double_quotes": False,
|
|
2019
|
+
"quote_char": '"',
|
|
2020
|
+
"delimiter": ",",
|
|
2021
|
+
"escape_char": "\\",
|
|
2022
|
+
"quoting_behavior": "Quote All",
|
|
2023
|
+
|
|
2024
|
+
}
|
|
2025
|
+
}
|
|
2026
|
+
}
|
|
2027
|
+
],
|
|
2028
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
2029
|
+
}
|
|
2030
|
+
)
|
|
2031
|
+
.set_files(
|
|
2032
|
+
{
|
|
2033
|
+
"a.csv": {
|
|
2034
|
+
"contents": [
|
|
2035
|
+
'''col1,col2''',
|
|
2036
|
+
'''val11,"val\\"2"''',
|
|
2037
|
+
],
|
|
2038
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
2039
|
+
}
|
|
2040
|
+
}
|
|
2041
|
+
)
|
|
2042
|
+
.set_file_type("csv")
|
|
2043
|
+
.set_expected_catalog(
|
|
2044
|
+
{
|
|
2045
|
+
"streams": [
|
|
2046
|
+
{
|
|
2047
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
2048
|
+
"json_schema": {
|
|
2049
|
+
"type": "object",
|
|
2050
|
+
"properties": {
|
|
2051
|
+
"col1": {
|
|
2052
|
+
"type": ["null", "string"]
|
|
2053
|
+
},
|
|
2054
|
+
"col2": {
|
|
2055
|
+
"type": ["null", "string"]
|
|
2056
|
+
},
|
|
2057
|
+
"_ab_source_file_last_modified": {
|
|
2058
|
+
"type": "string"
|
|
2059
|
+
},
|
|
2060
|
+
"_ab_source_file_url": {
|
|
2061
|
+
"type": "string"
|
|
2062
|
+
},
|
|
2063
|
+
},
|
|
2064
|
+
},
|
|
2065
|
+
"name": "stream1",
|
|
2066
|
+
"source_defined_cursor": True,
|
|
2067
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
2068
|
+
}
|
|
2069
|
+
]
|
|
2070
|
+
}
|
|
2071
|
+
)
|
|
2072
|
+
.set_expected_records(
|
|
2073
|
+
[
|
|
2074
|
+
{"data": {"col1": 'val11', "col2": 'val"2', "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
2075
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
2076
|
+
]
|
|
2077
|
+
)
|
|
2078
|
+
).build()
|
|
2079
|
+
|
|
2080
|
+
csv_double_quote_is_set_scenario = (
|
|
2081
|
+
TestScenarioBuilder()
|
|
2082
|
+
.set_name("csv_doublequote_is_set")
|
|
2083
|
+
# This scenario tests that quotes are properly escaped when double_quotes is True
|
|
2084
|
+
.set_config(
|
|
2085
|
+
{
|
|
2086
|
+
"streams": [
|
|
2087
|
+
{
|
|
2088
|
+
"name": "stream1",
|
|
2089
|
+
"file_type": "csv",
|
|
2090
|
+
"globs": ["*"],
|
|
2091
|
+
"validation_policy": "emit_record",
|
|
2092
|
+
"format": {
|
|
2093
|
+
"csv": {
|
|
2094
|
+
"filetype": "csv",
|
|
2095
|
+
"double_quotes": True,
|
|
2096
|
+
"quote_char": '"',
|
|
2097
|
+
"delimiter": ",",
|
|
2098
|
+
"quoting_behavior": "Quote All",
|
|
2099
|
+
|
|
2100
|
+
}
|
|
2101
|
+
}
|
|
2102
|
+
}
|
|
2103
|
+
],
|
|
2104
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
2105
|
+
}
|
|
2106
|
+
)
|
|
2107
|
+
.set_files(
|
|
2108
|
+
{
|
|
2109
|
+
"a.csv": {
|
|
2110
|
+
"contents": [
|
|
2111
|
+
'''col1,col2''',
|
|
2112
|
+
'''val11,"val""2"''',
|
|
2113
|
+
],
|
|
2114
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
2115
|
+
}
|
|
2116
|
+
}
|
|
2117
|
+
)
|
|
2118
|
+
.set_file_type("csv")
|
|
2119
|
+
.set_expected_catalog(
|
|
2120
|
+
{
|
|
2121
|
+
"streams": [
|
|
2122
|
+
{
|
|
2123
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
2124
|
+
"json_schema": {
|
|
2125
|
+
"type": "object",
|
|
2126
|
+
"properties": {
|
|
2127
|
+
"col1": {
|
|
2128
|
+
"type": ["null", "string"]
|
|
2129
|
+
},
|
|
2130
|
+
"col2": {
|
|
2131
|
+
"type": ["null", "string"]
|
|
2132
|
+
},
|
|
2133
|
+
"_ab_source_file_last_modified": {
|
|
2134
|
+
"type": "string"
|
|
2135
|
+
},
|
|
2136
|
+
"_ab_source_file_url": {
|
|
2137
|
+
"type": "string"
|
|
2138
|
+
},
|
|
2139
|
+
},
|
|
2140
|
+
},
|
|
2141
|
+
"name": "stream1",
|
|
2142
|
+
"source_defined_cursor": True,
|
|
2143
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
2144
|
+
}
|
|
2145
|
+
]
|
|
2146
|
+
}
|
|
2147
|
+
)
|
|
2148
|
+
.set_expected_records(
|
|
2149
|
+
[
|
|
2150
|
+
{"data": {"col1": 'val11', "col2": 'val"2', "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
2151
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
2152
|
+
]
|
|
2153
|
+
)
|
|
2154
|
+
).build()
|
|
2155
|
+
|
|
2156
|
+
csv_custom_delimiter_with_escape_char_scenario = (
|
|
2157
|
+
TestScenarioBuilder()
|
|
2158
|
+
.set_name("csv_custom_delimiter_with_escape_char")
|
|
2159
|
+
# This scenario tests that a value can contain the delimiter if it is wrapped in the quote_char
|
|
2160
|
+
.set_config(
|
|
2161
|
+
{
|
|
2162
|
+
"streams": [
|
|
2163
|
+
{
|
|
2164
|
+
"name": "stream1",
|
|
2165
|
+
"file_type": "csv",
|
|
2166
|
+
"globs": ["*"],
|
|
2167
|
+
"validation_policy": "emit_record",
|
|
2168
|
+
"format": {
|
|
2169
|
+
"csv": {
|
|
2170
|
+
"filetype": "csv",
|
|
2171
|
+
"double_quotes": True,
|
|
2172
|
+
"quote_char": '@',
|
|
2173
|
+
"delimiter": "|",
|
|
2174
|
+
"escape_char": "+"
|
|
2175
|
+
}
|
|
2176
|
+
}
|
|
2177
|
+
}
|
|
2178
|
+
],
|
|
2179
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
2180
|
+
}
|
|
2181
|
+
)
|
|
2182
|
+
.set_files(
|
|
2183
|
+
{
|
|
2184
|
+
"a.csv": {
|
|
2185
|
+
"contents": [
|
|
2186
|
+
'''col1|col2''',
|
|
2187
|
+
'''val"1,1|val+|2''',
|
|
2188
|
+
],
|
|
2189
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
2190
|
+
}
|
|
2191
|
+
}
|
|
2192
|
+
)
|
|
2193
|
+
.set_file_type("csv")
|
|
2194
|
+
.set_expected_catalog(
|
|
2195
|
+
{
|
|
2196
|
+
"streams": [
|
|
2197
|
+
{
|
|
2198
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
2199
|
+
"json_schema": {
|
|
2200
|
+
"type": "object",
|
|
2201
|
+
"properties": {
|
|
2202
|
+
"col1": {
|
|
2203
|
+
"type": ["null", "string"]
|
|
2204
|
+
},
|
|
2205
|
+
"col2": {
|
|
2206
|
+
"type": ["null", "string"]
|
|
2207
|
+
},
|
|
2208
|
+
"_ab_source_file_last_modified": {
|
|
2209
|
+
"type": "string"
|
|
2210
|
+
},
|
|
2211
|
+
"_ab_source_file_url": {
|
|
2212
|
+
"type": "string"
|
|
2213
|
+
},
|
|
2214
|
+
},
|
|
2215
|
+
},
|
|
2216
|
+
"name": "stream1",
|
|
2217
|
+
"source_defined_cursor": True,
|
|
2218
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
2219
|
+
}
|
|
2220
|
+
]
|
|
2221
|
+
}
|
|
2222
|
+
)
|
|
2223
|
+
.set_expected_records(
|
|
2224
|
+
[
|
|
2225
|
+
{"data": {"col1": 'val"1,1', "col2": 'val|2', "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
2226
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
2227
|
+
]
|
|
2228
|
+
)
|
|
2229
|
+
).build()
|
|
2230
|
+
|
|
2231
|
+
csv_custom_delimiter_in_double_quotes_scenario = (
|
|
2232
|
+
TestScenarioBuilder()
|
|
2233
|
+
.set_name("csv_custom_delimiter_in_double_quotes")
|
|
2234
|
+
# This scenario tests that a value can contain the delimiter if it is wrapped in the quote_char
|
|
2235
|
+
.set_config(
|
|
2236
|
+
{
|
|
2237
|
+
"streams": [
|
|
2238
|
+
{
|
|
2239
|
+
"name": "stream1",
|
|
2240
|
+
"file_type": "csv",
|
|
2241
|
+
"globs": ["*"],
|
|
2242
|
+
"validation_policy": "emit_record",
|
|
2243
|
+
"format": {
|
|
2244
|
+
"csv": {
|
|
2245
|
+
"filetype": "csv",
|
|
2246
|
+
"double_quotes": True,
|
|
2247
|
+
"quote_char": '@',
|
|
2248
|
+
"delimiter": "|",
|
|
2249
|
+
}
|
|
2250
|
+
}
|
|
2251
|
+
}
|
|
2252
|
+
],
|
|
2253
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
2254
|
+
}
|
|
2255
|
+
)
|
|
2256
|
+
.set_files(
|
|
2257
|
+
{
|
|
2258
|
+
"a.csv": {
|
|
2259
|
+
"contents": [
|
|
2260
|
+
'''col1|col2''',
|
|
2261
|
+
'''val"1,1|@val|2@''',
|
|
2262
|
+
],
|
|
2263
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
2264
|
+
}
|
|
2265
|
+
}
|
|
2266
|
+
)
|
|
2267
|
+
.set_file_type("csv")
|
|
2268
|
+
.set_expected_catalog(
|
|
2269
|
+
{
|
|
2270
|
+
"streams": [
|
|
2271
|
+
{
|
|
2272
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
2273
|
+
"json_schema": {
|
|
2274
|
+
"type": "object",
|
|
2275
|
+
"properties": {
|
|
2276
|
+
"col1": {
|
|
2277
|
+
"type": ["null", "string"]
|
|
2278
|
+
},
|
|
2279
|
+
"col2": {
|
|
2280
|
+
"type": ["null", "string"]
|
|
2281
|
+
},
|
|
2282
|
+
"_ab_source_file_last_modified": {
|
|
2283
|
+
"type": "string"
|
|
2284
|
+
},
|
|
2285
|
+
"_ab_source_file_url": {
|
|
2286
|
+
"type": "string"
|
|
2287
|
+
},
|
|
2288
|
+
},
|
|
2289
|
+
},
|
|
2290
|
+
"name": "stream1",
|
|
2291
|
+
"source_defined_cursor": True,
|
|
2292
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
2293
|
+
}
|
|
2294
|
+
]
|
|
2295
|
+
}
|
|
2296
|
+
)
|
|
2297
|
+
.set_expected_records(
|
|
2298
|
+
[
|
|
2299
|
+
{"data": {"col1": 'val"1,1', "col2": 'val|2', "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
2300
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
2301
|
+
]
|
|
2302
|
+
)
|
|
2303
|
+
).build()
|
|
2304
|
+
|
|
2305
|
+
|
|
2306
|
+
csv_skip_before_header_scenario = (
|
|
2307
|
+
TestScenarioBuilder()
|
|
2308
|
+
.set_name("csv_skip_before_header")
|
|
2309
|
+
.set_config(
|
|
2310
|
+
{
|
|
2311
|
+
"streams": [
|
|
2312
|
+
{
|
|
2313
|
+
"name": "stream1",
|
|
2314
|
+
"file_type": "csv",
|
|
2315
|
+
"globs": ["*"],
|
|
2316
|
+
"validation_policy": "emit_record",
|
|
2317
|
+
"format": {
|
|
2318
|
+
"csv": {
|
|
2319
|
+
"filetype": "csv",
|
|
2320
|
+
"skip_rows_before_header": 2
|
|
2321
|
+
}
|
|
2322
|
+
}
|
|
2323
|
+
}
|
|
2324
|
+
],
|
|
2325
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
2326
|
+
}
|
|
2327
|
+
)
|
|
2328
|
+
.set_files(
|
|
2329
|
+
{
|
|
2330
|
+
"a.csv": {
|
|
2331
|
+
"contents": [
|
|
2332
|
+
("skip_this", "skip_this"),
|
|
2333
|
+
("skip_this_too", "skip_this_too"),
|
|
2334
|
+
("col1", "col2"),
|
|
2335
|
+
("val11", "val12"),
|
|
2336
|
+
],
|
|
2337
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
2338
|
+
}
|
|
2339
|
+
}
|
|
2340
|
+
)
|
|
2341
|
+
.set_file_type("csv")
|
|
2342
|
+
.set_expected_catalog(
|
|
2343
|
+
{
|
|
2344
|
+
"streams": [
|
|
2345
|
+
{
|
|
2346
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
2347
|
+
"json_schema": {
|
|
2348
|
+
"type": "object",
|
|
2349
|
+
"properties": {
|
|
2350
|
+
"col1": {
|
|
2351
|
+
"type": ["null", "string"]
|
|
2352
|
+
},
|
|
2353
|
+
"col2": {
|
|
2354
|
+
"type": ["null", "string"]
|
|
2355
|
+
},
|
|
2356
|
+
"_ab_source_file_last_modified": {
|
|
2357
|
+
"type": "string"
|
|
2358
|
+
},
|
|
2359
|
+
"_ab_source_file_url": {
|
|
2360
|
+
"type": "string"
|
|
2361
|
+
},
|
|
2362
|
+
},
|
|
2363
|
+
},
|
|
2364
|
+
"name": "stream1",
|
|
2365
|
+
"source_defined_cursor": True,
|
|
2366
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
2367
|
+
}
|
|
2368
|
+
]
|
|
2369
|
+
}
|
|
2370
|
+
)
|
|
2371
|
+
.set_expected_records(
|
|
2372
|
+
[
|
|
2373
|
+
{"data": {"col1": "val11", "col2": "val12", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
2374
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
2375
|
+
]
|
|
2376
|
+
)
|
|
2377
|
+
).build()
|
|
2378
|
+
|
|
2379
|
+
csv_skip_after_header_scenario = (
|
|
2380
|
+
TestScenarioBuilder()
|
|
2381
|
+
.set_name("csv_skip_after_header")
|
|
2382
|
+
.set_config(
|
|
2383
|
+
{
|
|
2384
|
+
"streams": [
|
|
2385
|
+
{
|
|
2386
|
+
"name": "stream1",
|
|
2387
|
+
"file_type": "csv",
|
|
2388
|
+
"globs": ["*"],
|
|
2389
|
+
"validation_policy": "emit_record",
|
|
2390
|
+
"format": {
|
|
2391
|
+
"csv": {
|
|
2392
|
+
"filetype": "csv",
|
|
2393
|
+
"skip_rows_after_header": 2
|
|
2394
|
+
}
|
|
2395
|
+
}
|
|
2396
|
+
}
|
|
2397
|
+
],
|
|
2398
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
2399
|
+
}
|
|
2400
|
+
)
|
|
2401
|
+
.set_files(
|
|
2402
|
+
{
|
|
2403
|
+
"a.csv": {
|
|
2404
|
+
"contents": [
|
|
2405
|
+
("col1", "col2"),
|
|
2406
|
+
("skip_this", "skip_this"),
|
|
2407
|
+
("skip_this_too", "skip_this_too"),
|
|
2408
|
+
("val11", "val12"),
|
|
2409
|
+
],
|
|
2410
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
2411
|
+
}
|
|
2412
|
+
}
|
|
2413
|
+
)
|
|
2414
|
+
.set_file_type("csv")
|
|
2415
|
+
.set_expected_catalog(
|
|
2416
|
+
{
|
|
2417
|
+
"streams": [
|
|
2418
|
+
{
|
|
2419
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
2420
|
+
"json_schema": {
|
|
2421
|
+
"type": "object",
|
|
2422
|
+
"properties": {
|
|
2423
|
+
"col1": {
|
|
2424
|
+
"type": ["null", "string"]
|
|
2425
|
+
},
|
|
2426
|
+
"col2": {
|
|
2427
|
+
"type": ["null", "string"]
|
|
2428
|
+
},
|
|
2429
|
+
"_ab_source_file_last_modified": {
|
|
2430
|
+
"type": "string"
|
|
2431
|
+
},
|
|
2432
|
+
"_ab_source_file_url": {
|
|
2433
|
+
"type": "string"
|
|
2434
|
+
},
|
|
2435
|
+
},
|
|
2436
|
+
},
|
|
2437
|
+
"name": "stream1",
|
|
2438
|
+
"source_defined_cursor": True,
|
|
2439
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
2440
|
+
}
|
|
2441
|
+
]
|
|
2442
|
+
}
|
|
2443
|
+
)
|
|
2444
|
+
.set_expected_records(
|
|
2445
|
+
[
|
|
2446
|
+
{"data": {"col1": "val11", "col2": "val12", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
2447
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
2448
|
+
]
|
|
2449
|
+
)
|
|
2450
|
+
).build()
|
|
2451
|
+
|
|
2452
|
+
|
|
2453
|
+
csv_skip_before_and_after_header_scenario = (
|
|
2454
|
+
TestScenarioBuilder()
|
|
2455
|
+
.set_name("csv_skip_before_after_header")
|
|
2456
|
+
.set_config(
|
|
2457
|
+
{
|
|
2458
|
+
"streams": [
|
|
2459
|
+
{
|
|
2460
|
+
"name": "stream1",
|
|
2461
|
+
"file_type": "csv",
|
|
2462
|
+
"globs": ["*"],
|
|
2463
|
+
"validation_policy": "emit_record",
|
|
2464
|
+
"format": {
|
|
2465
|
+
"csv": {
|
|
2466
|
+
"filetype": "csv",
|
|
2467
|
+
"skip_rows_before_header": 1,
|
|
2468
|
+
"skip_rows_after_header": 1,
|
|
2469
|
+
}
|
|
2470
|
+
}
|
|
2471
|
+
}
|
|
2472
|
+
],
|
|
2473
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
2474
|
+
}
|
|
2475
|
+
)
|
|
2476
|
+
.set_files(
|
|
2477
|
+
{
|
|
2478
|
+
"a.csv": {
|
|
2479
|
+
"contents": [
|
|
2480
|
+
("skip_this", "skip_this"),
|
|
2481
|
+
("col1", "col2"),
|
|
2482
|
+
("skip_this_too", "skip_this_too"),
|
|
2483
|
+
("val11", "val12"),
|
|
2484
|
+
],
|
|
2485
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
2486
|
+
}
|
|
2487
|
+
}
|
|
2488
|
+
)
|
|
2489
|
+
.set_file_type("csv")
|
|
2490
|
+
.set_expected_catalog(
|
|
2491
|
+
{
|
|
2492
|
+
"streams": [
|
|
2493
|
+
{
|
|
2494
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
2495
|
+
"json_schema": {
|
|
2496
|
+
"type": "object",
|
|
2497
|
+
"properties": {
|
|
2498
|
+
"col1": {
|
|
2499
|
+
"type": ["null", "string"]
|
|
2500
|
+
},
|
|
2501
|
+
"col2": {
|
|
2502
|
+
"type": ["null", "string"]
|
|
2503
|
+
},
|
|
2504
|
+
"_ab_source_file_last_modified": {
|
|
2505
|
+
"type": "string"
|
|
2506
|
+
},
|
|
2507
|
+
"_ab_source_file_url": {
|
|
2508
|
+
"type": "string"
|
|
2509
|
+
},
|
|
2510
|
+
},
|
|
2511
|
+
},
|
|
2512
|
+
"name": "stream1",
|
|
2513
|
+
"source_defined_cursor": True,
|
|
2514
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
2515
|
+
}
|
|
2516
|
+
]
|
|
2517
|
+
}
|
|
2518
|
+
)
|
|
2519
|
+
.set_expected_records(
|
|
2520
|
+
[
|
|
2521
|
+
{"data": {"col1": "val11", "col2": "val12", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
2522
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
2523
|
+
]
|
|
2524
|
+
)
|
|
2525
|
+
).build()
|
|
2526
|
+
|
|
2527
|
+
csv_autogenerate_column_names_scenario = (
|
|
2528
|
+
TestScenarioBuilder()
|
|
2529
|
+
.set_name("csv_autogenerate_column_names")
|
|
2530
|
+
.set_config(
|
|
2531
|
+
{
|
|
2532
|
+
"streams": [
|
|
2533
|
+
{
|
|
2534
|
+
"name": "stream1",
|
|
2535
|
+
"file_type": "csv",
|
|
2536
|
+
"globs": ["*"],
|
|
2537
|
+
"validation_policy": "emit_record",
|
|
2538
|
+
"format": {
|
|
2539
|
+
"csv": {
|
|
2540
|
+
"filetype": "csv",
|
|
2541
|
+
"autogenerate_column_names": True,
|
|
2542
|
+
}
|
|
2543
|
+
}
|
|
2544
|
+
}
|
|
2545
|
+
],
|
|
2546
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
2547
|
+
}
|
|
2548
|
+
)
|
|
2549
|
+
.set_files(
|
|
2550
|
+
{
|
|
2551
|
+
"a.csv": {
|
|
2552
|
+
"contents": [
|
|
2553
|
+
("val11", "val12"),
|
|
2554
|
+
],
|
|
2555
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
2556
|
+
}
|
|
2557
|
+
}
|
|
2558
|
+
)
|
|
2559
|
+
.set_file_type("csv")
|
|
2560
|
+
.set_expected_catalog(
|
|
2561
|
+
{
|
|
2562
|
+
"streams": [
|
|
2563
|
+
{
|
|
2564
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
2565
|
+
"json_schema": {
|
|
2566
|
+
"type": "object",
|
|
2567
|
+
"properties": {
|
|
2568
|
+
"f0": {
|
|
2569
|
+
"type": ["null", "string"]
|
|
2570
|
+
},
|
|
2571
|
+
"f1": {
|
|
2572
|
+
"type": ["null", "string"]
|
|
2573
|
+
},
|
|
2574
|
+
"_ab_source_file_last_modified": {
|
|
2575
|
+
"type": "string"
|
|
2576
|
+
},
|
|
2577
|
+
"_ab_source_file_url": {
|
|
2578
|
+
"type": "string"
|
|
2579
|
+
},
|
|
2580
|
+
},
|
|
2581
|
+
},
|
|
2582
|
+
"name": "stream1",
|
|
2583
|
+
"source_defined_cursor": True,
|
|
2584
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
2585
|
+
}
|
|
2586
|
+
]
|
|
2587
|
+
}
|
|
2588
|
+
)
|
|
2589
|
+
.set_expected_records(
|
|
2590
|
+
[
|
|
2591
|
+
{"data": {"f0": "val11", "f1": "val12", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
2592
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
2593
|
+
]
|
|
2594
|
+
)
|
|
2595
|
+
).build()
|
|
2596
|
+
|
|
2597
|
+
csv_custom_bool_values_scenario = (
|
|
2598
|
+
TestScenarioBuilder()
|
|
2599
|
+
.set_name("csv_custom_bool_values")
|
|
2600
|
+
.set_config(
|
|
2601
|
+
{
|
|
2602
|
+
"streams": [
|
|
2603
|
+
{
|
|
2604
|
+
"name": "stream1",
|
|
2605
|
+
"file_type": "csv",
|
|
2606
|
+
"globs": ["*"],
|
|
2607
|
+
"validation_policy": "emit_record",
|
|
2608
|
+
"input_schema": {"col1": "boolean", "col2": "boolean"},
|
|
2609
|
+
"format": {
|
|
2610
|
+
"csv": {
|
|
2611
|
+
"filetype": "csv",
|
|
2612
|
+
"true_values": ["this_is_true"],
|
|
2613
|
+
"false_values": ["this_is_false"],
|
|
2614
|
+
}
|
|
2615
|
+
}
|
|
2616
|
+
}
|
|
2617
|
+
],
|
|
2618
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
2619
|
+
}
|
|
2620
|
+
)
|
|
2621
|
+
.set_files(
|
|
2622
|
+
{
|
|
2623
|
+
"a.csv": {
|
|
2624
|
+
"contents": [
|
|
2625
|
+
("col1", "col2"),
|
|
2626
|
+
("this_is_true", "this_is_false"),
|
|
2627
|
+
],
|
|
2628
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
2629
|
+
}
|
|
2630
|
+
}
|
|
2631
|
+
)
|
|
2632
|
+
.set_file_type("csv")
|
|
2633
|
+
.set_expected_catalog(
|
|
2634
|
+
{
|
|
2635
|
+
"streams": [
|
|
2636
|
+
{
|
|
2637
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
2638
|
+
"json_schema": {
|
|
2639
|
+
"type": "object",
|
|
2640
|
+
"properties": {
|
|
2641
|
+
"col1": {
|
|
2642
|
+
"type": "boolean"
|
|
2643
|
+
},
|
|
2644
|
+
"col2": {
|
|
2645
|
+
"type": "boolean"
|
|
2646
|
+
},
|
|
2647
|
+
"_ab_source_file_last_modified": {
|
|
2648
|
+
"type": "string"
|
|
2649
|
+
},
|
|
2650
|
+
"_ab_source_file_url": {
|
|
2651
|
+
"type": "string"
|
|
2652
|
+
},
|
|
2653
|
+
},
|
|
2654
|
+
},
|
|
2655
|
+
"name": "stream1",
|
|
2656
|
+
"source_defined_cursor": True,
|
|
2657
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
2658
|
+
}
|
|
2659
|
+
]
|
|
2660
|
+
}
|
|
2661
|
+
)
|
|
2662
|
+
.set_expected_records(
|
|
2663
|
+
[
|
|
2664
|
+
{"data": {"col1": True, "col2": False, "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
2665
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
2666
|
+
]
|
|
2667
|
+
)
|
|
2668
|
+
).build()
|
|
2669
|
+
|
|
2670
|
+
csv_custom_null_values_scenario = (
|
|
2671
|
+
TestScenarioBuilder()
|
|
2672
|
+
.set_name("csv_custom_null_values")
|
|
2673
|
+
.set_config(
|
|
2674
|
+
{
|
|
2675
|
+
"streams": [
|
|
2676
|
+
{
|
|
2677
|
+
"name": "stream1",
|
|
2678
|
+
"file_type": "csv",
|
|
2679
|
+
"globs": ["*"],
|
|
2680
|
+
"validation_policy": "emit_record",
|
|
2681
|
+
"input_schema": {"col1": "boolean", "col2": "string"},
|
|
2682
|
+
"format": {
|
|
2683
|
+
"csv": {
|
|
2684
|
+
"filetype": "csv",
|
|
2685
|
+
"null_values": ["null"],
|
|
2686
|
+
}
|
|
2687
|
+
}
|
|
2688
|
+
}
|
|
2689
|
+
],
|
|
2690
|
+
"start_date": "2023-06-04T03:54:07Z"
|
|
2691
|
+
}
|
|
2692
|
+
)
|
|
2693
|
+
.set_files(
|
|
2694
|
+
{
|
|
2695
|
+
"a.csv": {
|
|
2696
|
+
"contents": [
|
|
2697
|
+
("col1", "col2"),
|
|
2698
|
+
("null", "na"),
|
|
2699
|
+
],
|
|
2700
|
+
"last_modified": "2023-06-05T03:54:07.000Z",
|
|
2701
|
+
}
|
|
2702
|
+
}
|
|
2703
|
+
)
|
|
2704
|
+
.set_file_type("csv")
|
|
2705
|
+
.set_expected_catalog(
|
|
2706
|
+
{
|
|
2707
|
+
"streams": [
|
|
2708
|
+
{
|
|
2709
|
+
"default_cursor_field": ["_ab_source_file_last_modified"],
|
|
2710
|
+
"json_schema": {
|
|
2711
|
+
"type": "object",
|
|
2712
|
+
"properties": {
|
|
2713
|
+
"col1": {
|
|
2714
|
+
"type": "boolean"
|
|
2715
|
+
},
|
|
2716
|
+
"col2": {
|
|
2717
|
+
"type": "string"
|
|
2718
|
+
},
|
|
2719
|
+
"_ab_source_file_last_modified": {
|
|
2720
|
+
"type": "string"
|
|
2721
|
+
},
|
|
2722
|
+
"_ab_source_file_url": {
|
|
2723
|
+
"type": "string"
|
|
2724
|
+
},
|
|
2725
|
+
},
|
|
2726
|
+
},
|
|
2727
|
+
"name": "stream1",
|
|
2728
|
+
"source_defined_cursor": True,
|
|
2729
|
+
"supported_sync_modes": ["full_refresh", "incremental"],
|
|
2730
|
+
}
|
|
2731
|
+
]
|
|
2732
|
+
}
|
|
2733
|
+
)
|
|
2734
|
+
.set_expected_records(
|
|
2735
|
+
[
|
|
2736
|
+
{"data": {"col1": None, "col2": "na", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
|
|
2737
|
+
"_ab_source_file_url": "a.csv"}, "stream": "stream1"},
|
|
2738
|
+
]
|
|
2739
|
+
)
|
|
2740
|
+
).build()
|