|
69 | 69 |
|
70 | 70 | EXAMPLES_PATH = os.path.join(SPARK_HOME, "examples/src/main/python") |
71 | 71 | SCRIPTS_PATH = os.path.join(SPARK_HOME, "bin") |
| 72 | +DATA_PATH = os.path.join(SPARK_HOME, "data") |
72 | 73 | SCRIPTS_TARGET = os.path.join(TEMP_PATH, "bin") |
73 | 74 | JARS_TARGET = os.path.join(TEMP_PATH, "jars") |
74 | 75 | EXAMPLES_TARGET = os.path.join(TEMP_PATH, "examples") |
| 76 | +DATA_TARGET = os.path.join(TEMP_PATH, "data") |
75 | 77 |
|
76 | 78 |
|
77 | 79 | # Check and see if we are under the spark path in which case we need to build the symlink farm. |
@@ -114,11 +116,13 @@ def _supports_symlinks(): |
114 | 116 | os.symlink(JARS_PATH, JARS_TARGET) |
115 | 117 | os.symlink(SCRIPTS_PATH, SCRIPTS_TARGET) |
116 | 118 | os.symlink(EXAMPLES_PATH, EXAMPLES_TARGET) |
| 119 | + os.symlink(DATA_PATH, DATA_TARGET) |
117 | 120 | else: |
118 | 121 | # For windows fall back to the slower copytree |
119 | 122 | copytree(JARS_PATH, JARS_TARGET) |
120 | 123 | copytree(SCRIPTS_PATH, SCRIPTS_TARGET) |
121 | 124 | copytree(EXAMPLES_PATH, EXAMPLES_TARGET) |
| 125 | + copytree(DATA_PATH, DATA_TARGET) |
122 | 126 | else: |
123 | 127 | # If we are not inside of SPARK_HOME verify we have the required symlink farm |
124 | 128 | if not os.path.exists(JARS_TARGET): |
@@ -161,18 +165,21 @@ def _supports_symlinks(): |
161 | 165 | 'pyspark.jars', |
162 | 166 | 'pyspark.python.pyspark', |
163 | 167 | 'pyspark.python.lib', |
| 168 | + 'pyspark.data', |
164 | 169 | 'pyspark.examples.src.main.python'], |
165 | 170 | include_package_data=True, |
166 | 171 | package_dir={ |
167 | 172 | 'pyspark.jars': 'deps/jars', |
168 | 173 | 'pyspark.bin': 'deps/bin', |
169 | 174 | 'pyspark.python.lib': 'lib', |
| 175 | + 'pyspark.data': 'deps/data', |
170 | 176 | 'pyspark.examples.src.main.python': 'deps/examples', |
171 | 177 | }, |
172 | 178 | package_data={ |
173 | 179 | 'pyspark.jars': ['*.jar'], |
174 | 180 | 'pyspark.bin': ['*'], |
175 | 181 | 'pyspark.python.lib': ['*.zip'], |
| 182 | + 'pyspark.data': ['*'], |
176 | 183 | 'pyspark.examples.src.main.python': ['*.py', '*/*.py']}, |
177 | 184 | scripts=scripts, |
178 | 185 | license='http://www.apache.org/licenses/LICENSE-2.0', |
@@ -202,8 +209,10 @@ def _supports_symlinks(): |
202 | 209 | os.remove(os.path.join(TEMP_PATH, "jars")) |
203 | 210 | os.remove(os.path.join(TEMP_PATH, "bin")) |
204 | 211 | os.remove(os.path.join(TEMP_PATH, "examples")) |
| 212 | + os.remove(os.path.join(TEMP_PATH, "data")) |
205 | 213 | else: |
206 | 214 | rmtree(os.path.join(TEMP_PATH, "jars")) |
207 | 215 | rmtree(os.path.join(TEMP_PATH, "bin")) |
208 | 216 | rmtree(os.path.join(TEMP_PATH, "examples")) |
| 217 | + rmtree(os.path.join(TEMP_PATH, "data")) |
209 | 218 | os.rmdir(TEMP_PATH) |
0 commit comments