fg-data-profiling 4.19.0__py2.py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (238) hide show
  1. data_profiling/__init__.py +34 -0
  2. data_profiling/compare_reports.py +359 -0
  3. data_profiling/config.py +496 -0
  4. data_profiling/config_default.yaml +223 -0
  5. data_profiling/config_minimal.yaml +222 -0
  6. data_profiling/controller/__init__.py +1 -0
  7. data_profiling/controller/console.py +125 -0
  8. data_profiling/controller/pandas_decorator.py +21 -0
  9. data_profiling/expectations_report.py +117 -0
  10. data_profiling/model/__init__.py +4 -0
  11. data_profiling/model/alerts.py +780 -0
  12. data_profiling/model/correlations.py +163 -0
  13. data_profiling/model/dataframe.py +35 -0
  14. data_profiling/model/describe.py +210 -0
  15. data_profiling/model/description.py +108 -0
  16. data_profiling/model/duplicates.py +14 -0
  17. data_profiling/model/expectation_algorithms.py +112 -0
  18. data_profiling/model/handler.py +81 -0
  19. data_profiling/model/missing.py +146 -0
  20. data_profiling/model/pairwise.py +33 -0
  21. data_profiling/model/pandas/__init__.py +55 -0
  22. data_profiling/model/pandas/correlations_pandas.py +207 -0
  23. data_profiling/model/pandas/dataframe_pandas.py +26 -0
  24. data_profiling/model/pandas/describe_boolean_pandas.py +43 -0
  25. data_profiling/model/pandas/describe_categorical_pandas.py +274 -0
  26. data_profiling/model/pandas/describe_counts_pandas.py +63 -0
  27. data_profiling/model/pandas/describe_date_pandas.py +77 -0
  28. data_profiling/model/pandas/describe_file_pandas.py +56 -0
  29. data_profiling/model/pandas/describe_generic_pandas.py +36 -0
  30. data_profiling/model/pandas/describe_image_pandas.py +255 -0
  31. data_profiling/model/pandas/describe_numeric_pandas.py +175 -0
  32. data_profiling/model/pandas/describe_path_pandas.py +63 -0
  33. data_profiling/model/pandas/describe_supported_pandas.py +41 -0
  34. data_profiling/model/pandas/describe_text_pandas.py +62 -0
  35. data_profiling/model/pandas/describe_timeseries_pandas.py +222 -0
  36. data_profiling/model/pandas/describe_url_pandas.py +57 -0
  37. data_profiling/model/pandas/discretize_pandas.py +81 -0
  38. data_profiling/model/pandas/duplicates_pandas.py +56 -0
  39. data_profiling/model/pandas/imbalance_pandas.py +35 -0
  40. data_profiling/model/pandas/missing_pandas.py +42 -0
  41. data_profiling/model/pandas/sample_pandas.py +38 -0
  42. data_profiling/model/pandas/summary_pandas.py +101 -0
  43. data_profiling/model/pandas/table_pandas.py +56 -0
  44. data_profiling/model/pandas/timeseries_index_pandas.py +33 -0
  45. data_profiling/model/pandas/utils_pandas.py +27 -0
  46. data_profiling/model/sample.py +37 -0
  47. data_profiling/model/spark/__init__.py +48 -0
  48. data_profiling/model/spark/correlations_spark.py +152 -0
  49. data_profiling/model/spark/dataframe_spark.py +34 -0
  50. data_profiling/model/spark/describe_boolean_spark.py +27 -0
  51. data_profiling/model/spark/describe_categorical_spark.py +28 -0
  52. data_profiling/model/spark/describe_counts_spark.py +105 -0
  53. data_profiling/model/spark/describe_date_spark.py +51 -0
  54. data_profiling/model/spark/describe_generic_spark.py +30 -0
  55. data_profiling/model/spark/describe_numeric_spark.py +155 -0
  56. data_profiling/model/spark/describe_supported_spark.py +33 -0
  57. data_profiling/model/spark/describe_text_spark.py +25 -0
  58. data_profiling/model/spark/duplicates_spark.py +54 -0
  59. data_profiling/model/spark/missing_spark.py +96 -0
  60. data_profiling/model/spark/sample_spark.py +43 -0
  61. data_profiling/model/spark/summary_spark.py +95 -0
  62. data_profiling/model/spark/table_spark.py +58 -0
  63. data_profiling/model/spark/timeseries_index_spark.py +12 -0
  64. data_profiling/model/summarizer.py +207 -0
  65. data_profiling/model/summary.py +66 -0
  66. data_profiling/model/summary_algorithms.py +276 -0
  67. data_profiling/model/table.py +10 -0
  68. data_profiling/model/timeseries_index.py +16 -0
  69. data_profiling/model/typeset.py +365 -0
  70. data_profiling/model/typeset_relations.py +143 -0
  71. data_profiling/profile_report.py +573 -0
  72. data_profiling/report/__init__.py +4 -0
  73. data_profiling/report/formatters.py +346 -0
  74. data_profiling/report/presentation/__init__.py +1 -0
  75. data_profiling/report/presentation/core/__init__.py +39 -0
  76. data_profiling/report/presentation/core/alerts.py +18 -0
  77. data_profiling/report/presentation/core/collapse.py +24 -0
  78. data_profiling/report/presentation/core/container.py +50 -0
  79. data_profiling/report/presentation/core/correlation_table.py +21 -0
  80. data_profiling/report/presentation/core/dropdown.py +44 -0
  81. data_profiling/report/presentation/core/duplicate.py +16 -0
  82. data_profiling/report/presentation/core/frequency_table.py +14 -0
  83. data_profiling/report/presentation/core/frequency_table_small.py +16 -0
  84. data_profiling/report/presentation/core/html.py +14 -0
  85. data_profiling/report/presentation/core/image.py +34 -0
  86. data_profiling/report/presentation/core/item_renderer.py +17 -0
  87. data_profiling/report/presentation/core/renderable.py +42 -0
  88. data_profiling/report/presentation/core/root.py +35 -0
  89. data_profiling/report/presentation/core/sample.py +20 -0
  90. data_profiling/report/presentation/core/scores.py +32 -0
  91. data_profiling/report/presentation/core/table.py +26 -0
  92. data_profiling/report/presentation/core/toggle_button.py +14 -0
  93. data_profiling/report/presentation/core/variable.py +40 -0
  94. data_profiling/report/presentation/core/variable_info.py +36 -0
  95. data_profiling/report/presentation/flavours/__init__.py +9 -0
  96. data_profiling/report/presentation/flavours/flavour_html.py +64 -0
  97. data_profiling/report/presentation/flavours/flavour_widget.py +61 -0
  98. data_profiling/report/presentation/flavours/flavours.py +43 -0
  99. data_profiling/report/presentation/flavours/html/__init__.py +47 -0
  100. data_profiling/report/presentation/flavours/html/alerts.py +10 -0
  101. data_profiling/report/presentation/flavours/html/collapse.py +7 -0
  102. data_profiling/report/presentation/flavours/html/container.py +58 -0
  103. data_profiling/report/presentation/flavours/html/correlation_table.py +13 -0
  104. data_profiling/report/presentation/flavours/html/dropdown.py +7 -0
  105. data_profiling/report/presentation/flavours/html/duplicate.py +24 -0
  106. data_profiling/report/presentation/flavours/html/frequency_table.py +20 -0
  107. data_profiling/report/presentation/flavours/html/frequency_table_small.py +15 -0
  108. data_profiling/report/presentation/flavours/html/html.py +6 -0
  109. data_profiling/report/presentation/flavours/html/image.py +7 -0
  110. data_profiling/report/presentation/flavours/html/root.py +14 -0
  111. data_profiling/report/presentation/flavours/html/sample.py +12 -0
  112. data_profiling/report/presentation/flavours/html/scores.py +11 -0
  113. data_profiling/report/presentation/flavours/html/table.py +7 -0
  114. data_profiling/report/presentation/flavours/html/templates/alerts/alert_constant.html +1 -0
  115. data_profiling/report/presentation/flavours/html/templates/alerts/alert_constant_length.html +1 -0
  116. data_profiling/report/presentation/flavours/html/templates/alerts/alert_dirty_category.html +1 -0
  117. data_profiling/report/presentation/flavours/html/templates/alerts/alert_duplicates.html +1 -0
  118. data_profiling/report/presentation/flavours/html/templates/alerts/alert_empty.html +1 -0
  119. data_profiling/report/presentation/flavours/html/templates/alerts/alert_high_cardinality.html +1 -0
  120. data_profiling/report/presentation/flavours/html/templates/alerts/alert_high_correlation.html +4 -0
  121. data_profiling/report/presentation/flavours/html/templates/alerts/alert_imbalance.html +1 -0
  122. data_profiling/report/presentation/flavours/html/templates/alerts/alert_infinite.html +1 -0
  123. data_profiling/report/presentation/flavours/html/templates/alerts/alert_missing.html +1 -0
  124. data_profiling/report/presentation/flavours/html/templates/alerts/alert_near_duplicates.html +1 -0
  125. data_profiling/report/presentation/flavours/html/templates/alerts/alert_non_stationary.html +1 -0
  126. data_profiling/report/presentation/flavours/html/templates/alerts/alert_seasonal.html +1 -0
  127. data_profiling/report/presentation/flavours/html/templates/alerts/alert_skewed.html +1 -0
  128. data_profiling/report/presentation/flavours/html/templates/alerts/alert_truncated.html +1 -0
  129. data_profiling/report/presentation/flavours/html/templates/alerts/alert_type_date.html +1 -0
  130. data_profiling/report/presentation/flavours/html/templates/alerts/alert_uniform.html +1 -0
  131. data_profiling/report/presentation/flavours/html/templates/alerts/alert_unique.html +1 -0
  132. data_profiling/report/presentation/flavours/html/templates/alerts/alert_unsupported.html +1 -0
  133. data_profiling/report/presentation/flavours/html/templates/alerts/alert_zeros.html +1 -0
  134. data_profiling/report/presentation/flavours/html/templates/alerts.html +47 -0
  135. data_profiling/report/presentation/flavours/html/templates/collapse.html +11 -0
  136. data_profiling/report/presentation/flavours/html/templates/correlation_table.html +5 -0
  137. data_profiling/report/presentation/flavours/html/templates/diagram.html +11 -0
  138. data_profiling/report/presentation/flavours/html/templates/dropdown.html +16 -0
  139. data_profiling/report/presentation/flavours/html/templates/duplicate.html +5 -0
  140. data_profiling/report/presentation/flavours/html/templates/frequency_table.html +45 -0
  141. data_profiling/report/presentation/flavours/html/templates/frequency_table_small.html +34 -0
  142. data_profiling/report/presentation/flavours/html/templates/report.html +26 -0
  143. data_profiling/report/presentation/flavours/html/templates/sample.html +10 -0
  144. data_profiling/report/presentation/flavours/html/templates/scores.html +78 -0
  145. data_profiling/report/presentation/flavours/html/templates/sequence/batch_grid.html +16 -0
  146. data_profiling/report/presentation/flavours/html/templates/sequence/grid.html +18 -0
  147. data_profiling/report/presentation/flavours/html/templates/sequence/list.html +7 -0
  148. data_profiling/report/presentation/flavours/html/templates/sequence/named_list.html +8 -0
  149. data_profiling/report/presentation/flavours/html/templates/sequence/overview_tabs.html +30 -0
  150. data_profiling/report/presentation/flavours/html/templates/sequence/scores.html +3 -0
  151. data_profiling/report/presentation/flavours/html/templates/sequence/sections.html +13 -0
  152. data_profiling/report/presentation/flavours/html/templates/sequence/select.html +40 -0
  153. data_profiling/report/presentation/flavours/html/templates/sequence/tabs.html +30 -0
  154. data_profiling/report/presentation/flavours/html/templates/table.html +38 -0
  155. data_profiling/report/presentation/flavours/html/templates/toggle_button.html +18 -0
  156. data_profiling/report/presentation/flavours/html/templates/variable.html +7 -0
  157. data_profiling/report/presentation/flavours/html/templates/variable_info.html +49 -0
  158. data_profiling/report/presentation/flavours/html/templates/wrapper/assets/bootstrap.bundle.min.js +7 -0
  159. data_profiling/report/presentation/flavours/html/templates/wrapper/assets/bootstrap.min.css +6 -0
  160. data_profiling/report/presentation/flavours/html/templates/wrapper/assets/cosmo.bootstrap.min.css +12 -0
  161. data_profiling/report/presentation/flavours/html/templates/wrapper/assets/flatly.bootstrap.min.css +12 -0
  162. data_profiling/report/presentation/flavours/html/templates/wrapper/assets/script.js +52 -0
  163. data_profiling/report/presentation/flavours/html/templates/wrapper/assets/simplex.bootstrap.min.css +12 -0
  164. data_profiling/report/presentation/flavours/html/templates/wrapper/assets/style.css +253 -0
  165. data_profiling/report/presentation/flavours/html/templates/wrapper/assets/united.bootstrap.min.css +12 -0
  166. data_profiling/report/presentation/flavours/html/templates/wrapper/footer.html +7 -0
  167. data_profiling/report/presentation/flavours/html/templates/wrapper/javascript.html +18 -0
  168. data_profiling/report/presentation/flavours/html/templates/wrapper/navigation.html +36 -0
  169. data_profiling/report/presentation/flavours/html/templates/wrapper/style.html +53 -0
  170. data_profiling/report/presentation/flavours/html/templates.py +76 -0
  171. data_profiling/report/presentation/flavours/html/toggle_button.py +7 -0
  172. data_profiling/report/presentation/flavours/html/variable.py +7 -0
  173. data_profiling/report/presentation/flavours/html/variable_info.py +7 -0
  174. data_profiling/report/presentation/flavours/widget/__init__.py +49 -0
  175. data_profiling/report/presentation/flavours/widget/alerts.py +45 -0
  176. data_profiling/report/presentation/flavours/widget/collapse.py +43 -0
  177. data_profiling/report/presentation/flavours/widget/container.py +121 -0
  178. data_profiling/report/presentation/flavours/widget/correlation_table.py +14 -0
  179. data_profiling/report/presentation/flavours/widget/dropdown.py +31 -0
  180. data_profiling/report/presentation/flavours/widget/duplicate.py +14 -0
  181. data_profiling/report/presentation/flavours/widget/frequency_table.py +57 -0
  182. data_profiling/report/presentation/flavours/widget/frequency_table_small.py +66 -0
  183. data_profiling/report/presentation/flavours/widget/html.py +11 -0
  184. data_profiling/report/presentation/flavours/widget/image.py +26 -0
  185. data_profiling/report/presentation/flavours/widget/notebook.py +81 -0
  186. data_profiling/report/presentation/flavours/widget/root.py +10 -0
  187. data_profiling/report/presentation/flavours/widget/sample.py +14 -0
  188. data_profiling/report/presentation/flavours/widget/table.py +30 -0
  189. data_profiling/report/presentation/flavours/widget/toggle_button.py +17 -0
  190. data_profiling/report/presentation/flavours/widget/variable.py +12 -0
  191. data_profiling/report/presentation/flavours/widget/variable_info.py +11 -0
  192. data_profiling/report/presentation/frequency_table_utils.py +141 -0
  193. data_profiling/report/structure/__init__.py +1 -0
  194. data_profiling/report/structure/correlations.py +123 -0
  195. data_profiling/report/structure/overview.py +376 -0
  196. data_profiling/report/structure/report.py +457 -0
  197. data_profiling/report/structure/variables/__init__.py +35 -0
  198. data_profiling/report/structure/variables/render_boolean.py +132 -0
  199. data_profiling/report/structure/variables/render_categorical.py +566 -0
  200. data_profiling/report/structure/variables/render_common.py +31 -0
  201. data_profiling/report/structure/variables/render_complex.py +102 -0
  202. data_profiling/report/structure/variables/render_count.py +172 -0
  203. data_profiling/report/structure/variables/render_date.py +143 -0
  204. data_profiling/report/structure/variables/render_file.py +70 -0
  205. data_profiling/report/structure/variables/render_generic.py +45 -0
  206. data_profiling/report/structure/variables/render_image.py +204 -0
  207. data_profiling/report/structure/variables/render_path.py +134 -0
  208. data_profiling/report/structure/variables/render_real.py +314 -0
  209. data_profiling/report/structure/variables/render_text.py +189 -0
  210. data_profiling/report/structure/variables/render_timeseries.py +371 -0
  211. data_profiling/report/structure/variables/render_url.py +132 -0
  212. data_profiling/report/utils.py +34 -0
  213. data_profiling/serialize_report.py +143 -0
  214. data_profiling/utils/__init__.py +1 -0
  215. data_profiling/utils/backend.py +9 -0
  216. data_profiling/utils/cache.py +59 -0
  217. data_profiling/utils/common.py +142 -0
  218. data_profiling/utils/compat.py +31 -0
  219. data_profiling/utils/dataframe.py +238 -0
  220. data_profiling/utils/logger.py +53 -0
  221. data_profiling/utils/notebook.py +8 -0
  222. data_profiling/utils/paths.py +45 -0
  223. data_profiling/utils/progress_bar.py +15 -0
  224. data_profiling/utils/styles.py +22 -0
  225. data_profiling/utils/versions.py +19 -0
  226. data_profiling/version.py +1 -0
  227. data_profiling/visualisation/__init__.py +1 -0
  228. data_profiling/visualisation/context.py +87 -0
  229. data_profiling/visualisation/missing.py +138 -0
  230. data_profiling/visualisation/plot.py +1158 -0
  231. data_profiling/visualisation/utils.py +113 -0
  232. fg_data_profiling-4.19.0.dist-info/METADATA +362 -0
  233. fg_data_profiling-4.19.0.dist-info/RECORD +238 -0
  234. fg_data_profiling-4.19.0.dist-info/WHEEL +6 -0
  235. fg_data_profiling-4.19.0.dist-info/entry_points.txt +3 -0
  236. fg_data_profiling-4.19.0.dist-info/licenses/LICENSE +21 -0
  237. fg_data_profiling-4.19.0.dist-info/top_level.txt +2 -0
  238. ydata_profiling/__init__.py +43 -0
@@ -0,0 +1,52 @@
1
+ document.addEventListener("DOMContentLoaded", () => {
2
+ /* Tooltip */
3
+ document.querySelectorAll('[data-bs-toggle="tooltip"]').forEach(el => new bootstrap.Tooltip(el));
4
+
5
+ /* Link */
6
+ document.querySelectorAll("a:not([href^='#'])").forEach(el => el.setAttribute("target", "_blank"));
7
+
8
+ /* Variables Dropdown */
9
+ const variablesDropdown = document.querySelector("select#variables-dropdown");
10
+ variablesDropdown.addEventListener("change", () => {
11
+ const searchText = variablesDropdown.value.toLowerCase();
12
+
13
+ document.querySelectorAll(".variable").forEach(el => {
14
+ const title = el.firstElementChild.firstElementChild.firstElementChild.getAttribute("title").toLowerCase();
15
+ let isMatch = title === (searchText);
16
+
17
+ if (searchText === "") {
18
+ isMatch = true
19
+ }
20
+
21
+ if (isMatch) {
22
+ el.parentElement.classList.remove("d-none");
23
+ } else {
24
+ el.parentElement.classList.add("d-none");
25
+ }
26
+ });
27
+ });
28
+
29
+ /* Correlation */
30
+ document.querySelector("#toggle-correlation-description")?.click(() => {
31
+ document.querySelectorAll(".correlation-description").forEach(el => el.classList.toggle("d-none"));
32
+
33
+ document.querySelectorAll(".tab-pane[id^=correlations_tab-]").forEach(el => {
34
+ const diagramLength = el.querySelectorAll(".tab-pane[id^=correlations_tab-]").length;
35
+
36
+ let classes = [];
37
+ if (diagramLength === 1) {
38
+ classes = ["col-sm-6", "col-sm-12"];
39
+ } else if (diagramLength === 2) {
40
+ classes = ["col-sm-4", "col-sm-6"];
41
+ } else if (diagramLength === 3) {
42
+ classes = ["col-sm-3", "col-sm-4"];
43
+ } else {
44
+ console.log("unsupported number of reports");
45
+ }
46
+
47
+ document.querySelectorAll(".correlation-diagram").forEach(el => {
48
+ classes.forEach(clz => el.classList.toggle(clz));
49
+ });
50
+ });
51
+ });
52
+ });