edsl 0.1.39.dev3__py3-none-any.whl → 0.1.39.dev5__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 (341) hide show
  1. edsl/Base.py +413 -332
  2. edsl/BaseDiff.py +260 -260
  3. edsl/TemplateLoader.py +24 -24
  4. edsl/__init__.py +57 -49
  5. edsl/__version__.py +1 -1
  6. edsl/agents/Agent.py +1071 -867
  7. edsl/agents/AgentList.py +551 -413
  8. edsl/agents/Invigilator.py +284 -233
  9. edsl/agents/InvigilatorBase.py +257 -270
  10. edsl/agents/PromptConstructor.py +272 -354
  11. edsl/agents/QuestionInstructionPromptBuilder.py +128 -0
  12. edsl/agents/QuestionTemplateReplacementsBuilder.py +137 -0
  13. edsl/agents/__init__.py +2 -3
  14. edsl/agents/descriptors.py +99 -99
  15. edsl/agents/prompt_helpers.py +129 -129
  16. edsl/agents/question_option_processor.py +172 -0
  17. edsl/auto/AutoStudy.py +130 -117
  18. edsl/auto/StageBase.py +243 -230
  19. edsl/auto/StageGenerateSurvey.py +178 -178
  20. edsl/auto/StageLabelQuestions.py +125 -125
  21. edsl/auto/StagePersona.py +61 -61
  22. edsl/auto/StagePersonaDimensionValueRanges.py +88 -88
  23. edsl/auto/StagePersonaDimensionValues.py +74 -74
  24. edsl/auto/StagePersonaDimensions.py +69 -69
  25. edsl/auto/StageQuestions.py +74 -73
  26. edsl/auto/SurveyCreatorPipeline.py +21 -21
  27. edsl/auto/utilities.py +218 -224
  28. edsl/base/Base.py +279 -279
  29. edsl/config.py +177 -157
  30. edsl/conversation/Conversation.py +290 -290
  31. edsl/conversation/car_buying.py +59 -58
  32. edsl/conversation/chips.py +95 -95
  33. edsl/conversation/mug_negotiation.py +81 -81
  34. edsl/conversation/next_speaker_utilities.py +93 -93
  35. edsl/coop/CoopFunctionsMixin.py +15 -0
  36. edsl/coop/ExpectedParrotKeyHandler.py +125 -0
  37. edsl/coop/PriceFetcher.py +54 -54
  38. edsl/coop/__init__.py +2 -2
  39. edsl/coop/coop.py +1106 -1028
  40. edsl/coop/utils.py +131 -131
  41. edsl/data/Cache.py +573 -555
  42. edsl/data/CacheEntry.py +230 -233
  43. edsl/data/CacheHandler.py +168 -149
  44. edsl/data/RemoteCacheSync.py +186 -78
  45. edsl/data/SQLiteDict.py +292 -292
  46. edsl/data/__init__.py +5 -4
  47. edsl/data/orm.py +10 -10
  48. edsl/data_transfer_models.py +74 -73
  49. edsl/enums.py +202 -175
  50. edsl/exceptions/BaseException.py +21 -21
  51. edsl/exceptions/__init__.py +54 -54
  52. edsl/exceptions/agents.py +54 -42
  53. edsl/exceptions/cache.py +5 -5
  54. edsl/exceptions/configuration.py +16 -16
  55. edsl/exceptions/coop.py +10 -10
  56. edsl/exceptions/data.py +14 -14
  57. edsl/exceptions/general.py +34 -34
  58. edsl/exceptions/inference_services.py +5 -0
  59. edsl/exceptions/jobs.py +33 -33
  60. edsl/exceptions/language_models.py +63 -63
  61. edsl/exceptions/prompts.py +15 -15
  62. edsl/exceptions/questions.py +109 -91
  63. edsl/exceptions/results.py +29 -29
  64. edsl/exceptions/scenarios.py +29 -22
  65. edsl/exceptions/surveys.py +37 -37
  66. edsl/inference_services/AnthropicService.py +106 -87
  67. edsl/inference_services/AvailableModelCacheHandler.py +184 -0
  68. edsl/inference_services/AvailableModelFetcher.py +215 -0
  69. edsl/inference_services/AwsBedrock.py +118 -120
  70. edsl/inference_services/AzureAI.py +215 -217
  71. edsl/inference_services/DeepInfraService.py +18 -18
  72. edsl/inference_services/GoogleService.py +143 -148
  73. edsl/inference_services/GroqService.py +20 -20
  74. edsl/inference_services/InferenceServiceABC.py +80 -147
  75. edsl/inference_services/InferenceServicesCollection.py +138 -97
  76. edsl/inference_services/MistralAIService.py +120 -123
  77. edsl/inference_services/OllamaService.py +18 -18
  78. edsl/inference_services/OpenAIService.py +236 -224
  79. edsl/inference_services/PerplexityService.py +160 -163
  80. edsl/inference_services/ServiceAvailability.py +135 -0
  81. edsl/inference_services/TestService.py +90 -89
  82. edsl/inference_services/TogetherAIService.py +172 -170
  83. edsl/inference_services/data_structures.py +134 -0
  84. edsl/inference_services/models_available_cache.py +118 -118
  85. edsl/inference_services/rate_limits_cache.py +25 -25
  86. edsl/inference_services/registry.py +41 -41
  87. edsl/inference_services/write_available.py +10 -10
  88. edsl/jobs/AnswerQuestionFunctionConstructor.py +223 -0
  89. edsl/jobs/Answers.py +43 -56
  90. edsl/jobs/FetchInvigilator.py +47 -0
  91. edsl/jobs/InterviewTaskManager.py +98 -0
  92. edsl/jobs/InterviewsConstructor.py +50 -0
  93. edsl/jobs/Jobs.py +823 -898
  94. edsl/jobs/JobsChecks.py +172 -147
  95. edsl/jobs/JobsComponentConstructor.py +189 -0
  96. edsl/jobs/JobsPrompts.py +270 -268
  97. edsl/jobs/JobsRemoteInferenceHandler.py +311 -239
  98. edsl/jobs/JobsRemoteInferenceLogger.py +239 -0
  99. edsl/jobs/RequestTokenEstimator.py +30 -0
  100. edsl/jobs/__init__.py +1 -1
  101. edsl/jobs/async_interview_runner.py +138 -0
  102. edsl/jobs/buckets/BucketCollection.py +104 -63
  103. edsl/jobs/buckets/ModelBuckets.py +65 -65
  104. edsl/jobs/buckets/TokenBucket.py +283 -251
  105. edsl/jobs/buckets/TokenBucketAPI.py +211 -0
  106. edsl/jobs/buckets/TokenBucketClient.py +191 -0
  107. edsl/jobs/check_survey_scenario_compatibility.py +85 -0
  108. edsl/jobs/data_structures.py +120 -0
  109. edsl/jobs/decorators.py +35 -0
  110. edsl/jobs/interviews/Interview.py +396 -661
  111. edsl/jobs/interviews/InterviewExceptionCollection.py +99 -99
  112. edsl/jobs/interviews/InterviewExceptionEntry.py +186 -186
  113. edsl/jobs/interviews/InterviewStatistic.py +63 -63
  114. edsl/jobs/interviews/InterviewStatisticsCollection.py +25 -25
  115. edsl/jobs/interviews/InterviewStatusDictionary.py +78 -78
  116. edsl/jobs/interviews/InterviewStatusLog.py +92 -92
  117. edsl/jobs/interviews/ReportErrors.py +66 -66
  118. edsl/jobs/interviews/interview_status_enum.py +9 -9
  119. edsl/jobs/jobs_status_enums.py +9 -0
  120. edsl/jobs/loggers/HTMLTableJobLogger.py +304 -0
  121. edsl/jobs/results_exceptions_handler.py +98 -0
  122. edsl/jobs/runners/JobsRunnerAsyncio.py +151 -466
  123. edsl/jobs/runners/JobsRunnerStatus.py +297 -330
  124. edsl/jobs/tasks/QuestionTaskCreator.py +244 -242
  125. edsl/jobs/tasks/TaskCreators.py +64 -64
  126. edsl/jobs/tasks/TaskHistory.py +470 -450
  127. edsl/jobs/tasks/TaskStatusLog.py +23 -23
  128. edsl/jobs/tasks/task_status_enum.py +161 -163
  129. edsl/jobs/tokens/InterviewTokenUsage.py +27 -27
  130. edsl/jobs/tokens/TokenUsage.py +34 -34
  131. edsl/language_models/ComputeCost.py +63 -0
  132. edsl/language_models/LanguageModel.py +626 -668
  133. edsl/language_models/ModelList.py +164 -155
  134. edsl/language_models/PriceManager.py +127 -0
  135. edsl/language_models/RawResponseHandler.py +106 -0
  136. edsl/language_models/RegisterLanguageModelsMeta.py +184 -184
  137. edsl/language_models/ServiceDataSources.py +0 -0
  138. edsl/language_models/__init__.py +2 -3
  139. edsl/language_models/fake_openai_call.py +15 -15
  140. edsl/language_models/fake_openai_service.py +61 -61
  141. edsl/language_models/key_management/KeyLookup.py +63 -0
  142. edsl/language_models/key_management/KeyLookupBuilder.py +273 -0
  143. edsl/language_models/key_management/KeyLookupCollection.py +38 -0
  144. edsl/language_models/key_management/__init__.py +0 -0
  145. edsl/language_models/key_management/models.py +131 -0
  146. edsl/language_models/model.py +256 -0
  147. edsl/language_models/repair.py +156 -156
  148. edsl/language_models/utilities.py +65 -64
  149. edsl/notebooks/Notebook.py +263 -258
  150. edsl/notebooks/NotebookToLaTeX.py +142 -0
  151. edsl/notebooks/__init__.py +1 -1
  152. edsl/prompts/Prompt.py +352 -362
  153. edsl/prompts/__init__.py +2 -2
  154. edsl/questions/ExceptionExplainer.py +77 -0
  155. edsl/questions/HTMLQuestion.py +103 -0
  156. edsl/questions/QuestionBase.py +518 -664
  157. edsl/questions/QuestionBasePromptsMixin.py +221 -217
  158. edsl/questions/QuestionBudget.py +227 -227
  159. edsl/questions/QuestionCheckBox.py +359 -359
  160. edsl/questions/QuestionExtract.py +180 -182
  161. edsl/questions/QuestionFreeText.py +113 -114
  162. edsl/questions/QuestionFunctional.py +166 -166
  163. edsl/questions/QuestionList.py +223 -231
  164. edsl/questions/QuestionMatrix.py +265 -0
  165. edsl/questions/QuestionMultipleChoice.py +330 -286
  166. edsl/questions/QuestionNumerical.py +151 -153
  167. edsl/questions/QuestionRank.py +314 -324
  168. edsl/questions/Quick.py +41 -41
  169. edsl/questions/SimpleAskMixin.py +74 -73
  170. edsl/questions/__init__.py +27 -26
  171. edsl/questions/{AnswerValidatorMixin.py → answer_validator_mixin.py} +334 -289
  172. edsl/questions/compose_questions.py +98 -98
  173. edsl/questions/data_structures.py +20 -0
  174. edsl/questions/decorators.py +21 -21
  175. edsl/questions/derived/QuestionLikertFive.py +76 -76
  176. edsl/questions/derived/QuestionLinearScale.py +90 -87
  177. edsl/questions/derived/QuestionTopK.py +93 -93
  178. edsl/questions/derived/QuestionYesNo.py +82 -82
  179. edsl/questions/descriptors.py +427 -413
  180. edsl/questions/loop_processor.py +149 -0
  181. edsl/questions/prompt_templates/question_budget.jinja +13 -13
  182. edsl/questions/prompt_templates/question_checkbox.jinja +32 -32
  183. edsl/questions/prompt_templates/question_extract.jinja +11 -11
  184. edsl/questions/prompt_templates/question_free_text.jinja +3 -3
  185. edsl/questions/prompt_templates/question_linear_scale.jinja +11 -11
  186. edsl/questions/prompt_templates/question_list.jinja +17 -17
  187. edsl/questions/prompt_templates/question_multiple_choice.jinja +33 -33
  188. edsl/questions/prompt_templates/question_numerical.jinja +36 -36
  189. edsl/questions/{QuestionBaseGenMixin.py → question_base_gen_mixin.py} +168 -161
  190. edsl/questions/question_registry.py +177 -177
  191. edsl/questions/{RegisterQuestionsMeta.py → register_questions_meta.py} +71 -71
  192. edsl/questions/{ResponseValidatorABC.py → response_validator_abc.py} +188 -174
  193. edsl/questions/response_validator_factory.py +34 -0
  194. edsl/questions/settings.py +12 -12
  195. edsl/questions/templates/budget/answering_instructions.jinja +7 -7
  196. edsl/questions/templates/budget/question_presentation.jinja +7 -7
  197. edsl/questions/templates/checkbox/answering_instructions.jinja +10 -10
  198. edsl/questions/templates/checkbox/question_presentation.jinja +22 -22
  199. edsl/questions/templates/extract/answering_instructions.jinja +7 -7
  200. edsl/questions/templates/likert_five/answering_instructions.jinja +10 -10
  201. edsl/questions/templates/likert_five/question_presentation.jinja +11 -11
  202. edsl/questions/templates/linear_scale/answering_instructions.jinja +5 -5
  203. edsl/questions/templates/linear_scale/question_presentation.jinja +5 -5
  204. edsl/questions/templates/list/answering_instructions.jinja +3 -3
  205. edsl/questions/templates/list/question_presentation.jinja +5 -5
  206. edsl/questions/templates/matrix/__init__.py +1 -0
  207. edsl/questions/templates/matrix/answering_instructions.jinja +5 -0
  208. edsl/questions/templates/matrix/question_presentation.jinja +20 -0
  209. edsl/questions/templates/multiple_choice/answering_instructions.jinja +9 -9
  210. edsl/questions/templates/multiple_choice/question_presentation.jinja +11 -11
  211. edsl/questions/templates/numerical/answering_instructions.jinja +6 -6
  212. edsl/questions/templates/numerical/question_presentation.jinja +6 -6
  213. edsl/questions/templates/rank/answering_instructions.jinja +11 -11
  214. edsl/questions/templates/rank/question_presentation.jinja +15 -15
  215. edsl/questions/templates/top_k/answering_instructions.jinja +8 -8
  216. edsl/questions/templates/top_k/question_presentation.jinja +22 -22
  217. edsl/questions/templates/yes_no/answering_instructions.jinja +6 -6
  218. edsl/questions/templates/yes_no/question_presentation.jinja +11 -11
  219. edsl/results/CSSParameterizer.py +108 -108
  220. edsl/results/Dataset.py +587 -424
  221. edsl/results/DatasetExportMixin.py +594 -731
  222. edsl/results/DatasetTree.py +295 -275
  223. edsl/results/MarkdownToDocx.py +122 -0
  224. edsl/results/MarkdownToPDF.py +111 -0
  225. edsl/results/Result.py +557 -465
  226. edsl/results/Results.py +1183 -1165
  227. edsl/results/ResultsExportMixin.py +45 -43
  228. edsl/results/ResultsGGMixin.py +121 -121
  229. edsl/results/TableDisplay.py +125 -198
  230. edsl/results/TextEditor.py +50 -0
  231. edsl/results/__init__.py +2 -2
  232. edsl/results/file_exports.py +252 -0
  233. edsl/results/{ResultsFetchMixin.py → results_fetch_mixin.py} +33 -33
  234. edsl/results/{Selector.py → results_selector.py} +145 -135
  235. edsl/results/{ResultsToolsMixin.py → results_tools_mixin.py} +98 -98
  236. edsl/results/smart_objects.py +96 -0
  237. edsl/results/table_data_class.py +12 -0
  238. edsl/results/table_display.css +77 -77
  239. edsl/results/table_renderers.py +118 -0
  240. edsl/results/tree_explore.py +115 -115
  241. edsl/scenarios/ConstructDownloadLink.py +109 -0
  242. edsl/scenarios/DocumentChunker.py +102 -0
  243. edsl/scenarios/DocxScenario.py +16 -0
  244. edsl/scenarios/FileStore.py +511 -632
  245. edsl/scenarios/PdfExtractor.py +40 -0
  246. edsl/scenarios/Scenario.py +498 -601
  247. edsl/scenarios/ScenarioHtmlMixin.py +65 -64
  248. edsl/scenarios/ScenarioList.py +1458 -1287
  249. edsl/scenarios/ScenarioListExportMixin.py +45 -52
  250. edsl/scenarios/ScenarioListPdfMixin.py +239 -261
  251. edsl/scenarios/__init__.py +3 -4
  252. edsl/scenarios/directory_scanner.py +96 -0
  253. edsl/scenarios/file_methods.py +85 -0
  254. edsl/scenarios/handlers/__init__.py +13 -0
  255. edsl/scenarios/handlers/csv.py +38 -0
  256. edsl/scenarios/handlers/docx.py +76 -0
  257. edsl/scenarios/handlers/html.py +37 -0
  258. edsl/scenarios/handlers/json.py +111 -0
  259. edsl/scenarios/handlers/latex.py +5 -0
  260. edsl/scenarios/handlers/md.py +51 -0
  261. edsl/scenarios/handlers/pdf.py +68 -0
  262. edsl/scenarios/handlers/png.py +39 -0
  263. edsl/scenarios/handlers/pptx.py +105 -0
  264. edsl/scenarios/handlers/py.py +294 -0
  265. edsl/scenarios/handlers/sql.py +313 -0
  266. edsl/scenarios/handlers/sqlite.py +149 -0
  267. edsl/scenarios/handlers/txt.py +33 -0
  268. edsl/scenarios/{ScenarioJoin.py → scenario_join.py} +131 -127
  269. edsl/scenarios/scenario_selector.py +156 -0
  270. edsl/shared.py +1 -1
  271. edsl/study/ObjectEntry.py +173 -173
  272. edsl/study/ProofOfWork.py +113 -113
  273. edsl/study/SnapShot.py +80 -80
  274. edsl/study/Study.py +521 -528
  275. edsl/study/__init__.py +4 -4
  276. edsl/surveys/ConstructDAG.py +92 -0
  277. edsl/surveys/DAG.py +148 -148
  278. edsl/surveys/EditSurvey.py +221 -0
  279. edsl/surveys/InstructionHandler.py +100 -0
  280. edsl/surveys/Memory.py +31 -31
  281. edsl/surveys/MemoryManagement.py +72 -0
  282. edsl/surveys/MemoryPlan.py +244 -244
  283. edsl/surveys/Rule.py +327 -326
  284. edsl/surveys/RuleCollection.py +385 -387
  285. edsl/surveys/RuleManager.py +172 -0
  286. edsl/surveys/Simulator.py +75 -0
  287. edsl/surveys/Survey.py +1280 -1801
  288. edsl/surveys/SurveyCSS.py +273 -261
  289. edsl/surveys/SurveyExportMixin.py +259 -259
  290. edsl/surveys/{SurveyFlowVisualizationMixin.py → SurveyFlowVisualization.py} +181 -179
  291. edsl/surveys/SurveyQualtricsImport.py +284 -284
  292. edsl/surveys/SurveyToApp.py +141 -0
  293. edsl/surveys/__init__.py +5 -3
  294. edsl/surveys/base.py +53 -53
  295. edsl/surveys/descriptors.py +60 -56
  296. edsl/surveys/instructions/ChangeInstruction.py +48 -49
  297. edsl/surveys/instructions/Instruction.py +56 -65
  298. edsl/surveys/instructions/InstructionCollection.py +82 -77
  299. edsl/templates/error_reporting/base.html +23 -23
  300. edsl/templates/error_reporting/exceptions_by_model.html +34 -34
  301. edsl/templates/error_reporting/exceptions_by_question_name.html +16 -16
  302. edsl/templates/error_reporting/exceptions_by_type.html +16 -16
  303. edsl/templates/error_reporting/interview_details.html +115 -115
  304. edsl/templates/error_reporting/interviews.html +19 -19
  305. edsl/templates/error_reporting/overview.html +4 -4
  306. edsl/templates/error_reporting/performance_plot.html +1 -1
  307. edsl/templates/error_reporting/report.css +73 -73
  308. edsl/templates/error_reporting/report.html +117 -117
  309. edsl/templates/error_reporting/report.js +25 -25
  310. edsl/tools/__init__.py +1 -1
  311. edsl/tools/clusters.py +192 -192
  312. edsl/tools/embeddings.py +27 -27
  313. edsl/tools/embeddings_plotting.py +118 -118
  314. edsl/tools/plotting.py +112 -112
  315. edsl/tools/summarize.py +18 -18
  316. edsl/utilities/PrettyList.py +56 -0
  317. edsl/utilities/SystemInfo.py +28 -28
  318. edsl/utilities/__init__.py +22 -22
  319. edsl/utilities/ast_utilities.py +25 -25
  320. edsl/utilities/data/Registry.py +6 -6
  321. edsl/utilities/data/__init__.py +1 -1
  322. edsl/utilities/data/scooter_results.json +1 -1
  323. edsl/utilities/decorators.py +77 -77
  324. edsl/utilities/gcp_bucket/cloud_storage.py +96 -96
  325. edsl/utilities/interface.py +627 -627
  326. edsl/utilities/is_notebook.py +18 -0
  327. edsl/utilities/is_valid_variable_name.py +11 -0
  328. edsl/utilities/naming_utilities.py +263 -263
  329. edsl/utilities/remove_edsl_version.py +24 -0
  330. edsl/utilities/repair_functions.py +28 -28
  331. edsl/utilities/restricted_python.py +70 -70
  332. edsl/utilities/utilities.py +436 -424
  333. {edsl-0.1.39.dev3.dist-info → edsl-0.1.39.dev5.dist-info}/LICENSE +21 -21
  334. {edsl-0.1.39.dev3.dist-info → edsl-0.1.39.dev5.dist-info}/METADATA +13 -11
  335. edsl-0.1.39.dev5.dist-info/RECORD +358 -0
  336. {edsl-0.1.39.dev3.dist-info → edsl-0.1.39.dev5.dist-info}/WHEEL +1 -1
  337. edsl/language_models/KeyLookup.py +0 -30
  338. edsl/language_models/registry.py +0 -190
  339. edsl/language_models/unused/ReplicateBase.py +0 -83
  340. edsl/results/ResultsDBMixin.py +0 -238
  341. edsl-0.1.39.dev3.dist-info/RECORD +0 -277
@@ -1,289 +1,334 @@
1
- """Mixin with validators for LLM answers to questions."""
2
-
3
- import re
4
- from typing import Any, Type, Union
5
- from edsl.exceptions import (
6
- QuestionAnswerValidationError,
7
- )
8
-
9
-
10
- class AnswerValidatorMixin:
11
- """
12
- Mixin with validators for LLM answers to questions.
13
-
14
- - Template validation: validators for the entire answer object format
15
- - Value validation: validators for specific values
16
- - Question specific validation: validators for specific question types
17
- """
18
-
19
- def failing_job(self):
20
- from edsl import Agent
21
-
22
- a = Agent()
23
-
24
- def f(self, question, scenario):
25
- return []
26
-
27
- a.add_direct_question_answering_method(f, validate_response=True)
28
- from edsl import QuestionNumerical
29
-
30
- q = QuestionNumerical.example()
31
- results = q.by(a).run()
32
- return results
33
-
34
- #####################
35
- # TEMPLATE VALIDATION
36
- #####################
37
- def _validate_answer_template_basic(self, answer: Any) -> None:
38
- """Check that the answer (i) is a dictionary (ii) has an 'answer' key.
39
-
40
- - E.g., both {'answer': 1} and {'answer': {'a': 1}, 'other_key'=[1,2,3]} are valid
41
- """
42
- if not isinstance(answer, dict):
43
- raise QuestionAnswerValidationError(
44
- f"Answer must be a dictionary (got {answer})."
45
- )
46
- if not "answer" in answer:
47
- raise QuestionAnswerValidationError(
48
- f"Answer must have an 'answer' key (got {answer})."
49
- )
50
-
51
- #####################
52
- # VALUE VALIDATION
53
- #####################
54
- def _validate_answer_key_value(
55
- self, answer: dict[str, Any], key: str, of_type: Type
56
- ) -> None:
57
- """Check that the value of a key is of the specified type."""
58
- if not isinstance(answer.get(key), of_type):
59
- raise QuestionAnswerValidationError(
60
- f"""Answer key '{key}' must be of type {of_type.__name__};
61
- (got {answer.get(key)}) which is of type {type(answer.get(key))}."""
62
- )
63
-
64
- def _validate_answer_key_value_numeric(
65
- self, answer: dict[str, Any], key: str
66
- ) -> None:
67
- """Check that the value is numeric (int or float).
68
- Can also deal with strings that contain commas and other characters.
69
-
70
- """
71
- value = answer.get(key)
72
- initial_value = value
73
- if type(value) == str:
74
- value = value.replace(",", "")
75
- value = "".join(re.findall(r"[-+]?\d*\.\d+|\d+", value))
76
- if value.isdigit():
77
- value = int(value)
78
- else:
79
- try:
80
- float(value)
81
- value = float(value)
82
- except ValueError:
83
- raise QuestionAnswerValidationError(
84
- f"Answer should be numerical (int or float). Got '{initial_value}'"
85
- )
86
- return None
87
- elif type(value) == int or type(value) == float:
88
- return None
89
- else:
90
- raise QuestionAnswerValidationError(
91
- f"Answer should be numerical (int or float)."
92
- )
93
-
94
- #####################
95
- # QUESTION SPECIFIC VALIDATION
96
- #####################
97
- def _validate_answer_budget(self, answer: dict[str, Any]) -> None:
98
- """Validate QuestionBudget-specific answer.
99
-
100
- Check that answer["answer"]:
101
- - has keys that are in the range of the number of options
102
- - has values that are non-negative integers
103
- - has values that sum to `budget_sum`
104
- - contains all keys in the range of the number of options
105
- """
106
- answer = answer.get("answer")
107
- budget_sum = self.budget_sum
108
- acceptable_answer_keys = set(range(len(self.question_options)))
109
- answer_keys = set([int(k) for k in answer.keys()])
110
- current_sum = sum(answer.values())
111
- if not current_sum == budget_sum:
112
- raise QuestionAnswerValidationError(
113
- f"Budget sum must be {budget_sum}, but got {current_sum}."
114
- )
115
- if any(v < 0 for v in answer.values()):
116
- raise QuestionAnswerValidationError(
117
- f"Budget values must be positive, but got {answer_keys}."
118
- )
119
- if any([int(key) not in acceptable_answer_keys for key in answer.keys()]):
120
- raise QuestionAnswerValidationError(
121
- f"Budget keys must be in {acceptable_answer_keys}, but got {answer_keys}."
122
- )
123
- if acceptable_answer_keys != answer_keys:
124
- missing_keys = acceptable_answer_keys - answer_keys
125
- raise QuestionAnswerValidationError(
126
- f"All but keys must be represented in the answer. Missing: {missing_keys}."
127
- )
128
-
129
- def _validate_answer_checkbox(self, answer: dict[str, Union[str, int]]) -> None:
130
- """Validate QuestionCheckbox-specific answer.
131
-
132
- :param answer: Answer to validate
133
-
134
-
135
- Check that answer["answer"]:
136
- - has elements that are strings, bytes-like objects or real numbers evaluating to integers
137
- - has elements that are in the range of the number of options
138
- - has at least `min_selections` elements, if provided
139
- - has at most `max_selections` elements, if provided
140
- """
141
- answer_codes = answer["answer"]
142
- try:
143
- answer_codes = [int(k) for k in answer["answer"]]
144
- except:
145
- raise QuestionAnswerValidationError(
146
- f"Answer codes must be a list of strings, bytes-like objects or real numbers (got {answer['answer']})."
147
- )
148
- acceptable_values = list(range(len(self.question_options)))
149
- for answer_code in answer_codes:
150
- if answer_code not in acceptable_values:
151
- raise QuestionAnswerValidationError(
152
- f"Answer code {answer_code} has elements not in {acceptable_values}."
153
- )
154
- if self.min_selections is not None and len(answer_codes) < self.min_selections:
155
- raise QuestionAnswerValidationError(
156
- f"Answer codes, {answer_codes}, has fewer than {self.min_selections} options selected."
157
- )
158
- if self.max_selections is not None and len(answer_codes) > self.max_selections:
159
- raise QuestionAnswerValidationError(
160
- f"Answer codes, {answer_codes}, has more than {self.max_selections} options selected."
161
- )
162
-
163
- def _validate_answer_extract(self, answer: dict[str, Any]) -> None:
164
- """Validate QuestionExtract-specific answer.
165
-
166
- Check that answer["answer"]:
167
- - does not have keys that are not in the answer template
168
- - has all keys that are in the answer template
169
- """
170
- value = answer.get("answer")
171
- acceptable_answer_keys = set(self.answer_template.keys())
172
- if any([key not in acceptable_answer_keys for key in value.keys()]):
173
- raise QuestionAnswerValidationError(
174
- f"Answer keys must be in {acceptable_answer_keys}, but got {value.keys()}."
175
- )
176
- if any([key not in value.keys() for key in acceptable_answer_keys]):
177
- raise QuestionAnswerValidationError(
178
- f"Answer must have all keys in {acceptable_answer_keys}, but got {value.keys()}."
179
- )
180
-
181
- def _validate_answer_list(self, answer: dict[str, Union[list, str]]) -> None:
182
- """Validate QuestionList-specific answer.
183
-
184
- Check that answer["answer"]:
185
- - is not empty, if `allow_nonresponse` is False
186
- - has no more than `max_list_items` elements
187
- - has no empty strings
188
- """
189
- value = answer.get("answer")
190
- if (
191
- hasattr(self, "allow_nonresponse")
192
- and self.allow_nonresponse == False
193
- and (value == [] or value is None)
194
- ):
195
- raise QuestionAnswerValidationError("You must provide a response.")
196
-
197
- if (
198
- hasattr(self, "max_list_items")
199
- and self.max_list_items is not None
200
- and (len(value) > self.max_list_items)
201
- ):
202
- raise QuestionAnswerValidationError("Response has too many items.")
203
-
204
- if any([item == "" for item in value]):
205
- raise QuestionAnswerValidationError(
206
- f"Answer cannot contain empty strings, but got {value}."
207
- )
208
-
209
- def _validate_answer_numerical(self, answer: dict) -> None:
210
- """Validate QuestionNumerical-specific answer.
211
-
212
- Check that answer["answer"]:
213
- - is not less than `min_value`
214
- - is not greater than `max_value`
215
- """
216
- value = float(answer.get("answer"))
217
- if self.min_value is not None and value < self.min_value:
218
- raise QuestionAnswerValidationError(
219
- f"Value {value} is less than {self.min_value}"
220
- )
221
- if self.max_value is not None and value > self.max_value:
222
- raise QuestionAnswerValidationError(
223
- f"Value {value} is greater than {self.max_value}"
224
- )
225
- return value
226
-
227
- def _validate_answer_multiple_choice(
228
- self, answer: dict[str, Union[str, int]]
229
- ) -> None:
230
- """Validate QuestionMultipleChoice-specific answer.
231
-
232
- Check that answer["answer"]:
233
- - is a string, bytes-like object or real number
234
- - is a non-negative integer
235
- - is in the range of the number of options
236
- """
237
- try:
238
- value = int(answer.get("answer"))
239
- except:
240
- raise QuestionAnswerValidationError(
241
- f"Answer code must be a string, a bytes-like object or a real number (got {answer.get('answer')})."
242
- )
243
- if not value >= 0:
244
- raise QuestionAnswerValidationError(
245
- f"Answer code must be a non-negative integer (got {value})."
246
- )
247
- if int(value) not in range(len(self.question_options)):
248
- raise QuestionAnswerValidationError(
249
- f"Answer code {value} must be in {list(range(len(self.question_options)))}."
250
- )
251
-
252
- def _validate_answer_rank(self, answer: dict[str, Union[str, int]]) -> None:
253
- """Validate QuestionRank-specific answer.
254
-
255
- Check that answer["answer"]:
256
- - contains only integers
257
- - contains only integers in the range of the number of options
258
- - has the correct number of elements
259
- """
260
- value = answer.get("answer")
261
- acceptable_values = list(range(len(self.question_options)))
262
- for v in value:
263
- try:
264
- answer_code = int(v)
265
- except ValueError:
266
- raise QuestionAnswerValidationError(
267
- f"Rank answer {value} has elements that are not integers, namely {v}."
268
- )
269
- except TypeError:
270
- raise QuestionAnswerValidationError(
271
- f"Rank answer {value} has elements that are not integers, namely {v}."
272
- )
273
- if answer_code not in acceptable_values:
274
- raise QuestionAnswerValidationError(
275
- f"Answer {value} has elements not in {acceptable_values}, namely {v}."
276
- )
277
- if len(value) != self.num_selections:
278
- raise QuestionAnswerValidationError(
279
- f"Rank answer {value}, but exactly {self.num_selections} selections required."
280
- )
281
-
282
-
283
- if __name__ == "__main__":
284
- pass
285
- # import doctest
286
-
287
- # doctest.testmod(optionflags=doctest.ELLIPSIS)
288
-
289
- # results = AnswerValidatorMixin().failing_job()
1
+ """Mixin with validators for LLM answers to questions."""
2
+
3
+ import re
4
+ from typing import Any, Type, Union
5
+ from edsl.exceptions.questions import (
6
+ QuestionAnswerValidationError,
7
+ )
8
+
9
+
10
+ class AnswerValidatorMixin:
11
+ """
12
+ Mixin with validators for LLM answers to questions.
13
+
14
+ - Template validation: validators for the entire answer object format
15
+ - Value validation: validators for specific values
16
+ - Question specific validation: validators for specific question types
17
+ """
18
+
19
+ def failing_job(self):
20
+ from edsl import Agent
21
+
22
+ a = Agent()
23
+
24
+ def f(self, question, scenario):
25
+ return []
26
+
27
+ a.add_direct_question_answering_method(f, validate_response=True)
28
+ from edsl import QuestionNumerical
29
+
30
+ q = QuestionNumerical.example()
31
+ results = q.by(a).run()
32
+ return results
33
+
34
+ #####################
35
+ # TEMPLATE VALIDATION
36
+ #####################
37
+ def _validate_answer_template_basic(self, answer: Any) -> None:
38
+ """Check that the answer (i) is a dictionary (ii) has an 'answer' key.
39
+
40
+ - E.g., both {'answer': 1} and {'answer': {'a': 1}, 'other_key'=[1,2,3]} are valid
41
+ """
42
+ if not isinstance(answer, dict):
43
+ raise QuestionAnswerValidationError(
44
+ f"Answer must be a dictionary (got {answer})."
45
+ )
46
+ if not "answer" in answer:
47
+ raise QuestionAnswerValidationError(
48
+ f"Answer must have an 'answer' key (got {answer})."
49
+ )
50
+
51
+ #####################
52
+ # VALUE VALIDATION
53
+ #####################
54
+ def _validate_answer_key_value(
55
+ self, answer: dict[str, Any], key: str, of_type: Type
56
+ ) -> None:
57
+ """Check that the value of a key is of the specified type."""
58
+ if not isinstance(answer.get(key), of_type):
59
+ raise QuestionAnswerValidationError(
60
+ f"""Answer key '{key}' must be of type {of_type.__name__};
61
+ (got {answer.get(key)}) which is of type {type(answer.get(key))}."""
62
+ )
63
+
64
+ def _validate_answer_key_value_numeric(
65
+ self, answer: dict[str, Any], key: str
66
+ ) -> None:
67
+ """Check that the value is numeric (int or float).
68
+ Can also deal with strings that contain commas and other characters.
69
+
70
+ """
71
+ value = answer.get(key)
72
+ initial_value = value
73
+ if type(value) == str:
74
+ value = value.replace(",", "")
75
+ value = "".join(re.findall(r"[-+]?\d*\.\d+|\d+", value))
76
+ if value.isdigit():
77
+ value = int(value)
78
+ else:
79
+ try:
80
+ float(value)
81
+ value = float(value)
82
+ except ValueError:
83
+ raise QuestionAnswerValidationError(
84
+ f"Answer should be numerical (int or float). Got '{initial_value}'"
85
+ )
86
+ return None
87
+ elif type(value) == int or type(value) == float:
88
+ return None
89
+ else:
90
+ raise QuestionAnswerValidationError(
91
+ f"Answer should be numerical (int or float)."
92
+ )
93
+
94
+ #####################
95
+ # QUESTION SPECIFIC VALIDATION
96
+ #####################
97
+ def _validate_answer_budget(self, answer: dict[str, Any]) -> None:
98
+ """Validate QuestionBudget-specific answer.
99
+
100
+ Check that answer["answer"]:
101
+ - has keys that are in the range of the number of options
102
+ - has values that are non-negative integers
103
+ - has values that sum to `budget_sum`
104
+ - contains all keys in the range of the number of options
105
+ """
106
+ answer = answer.get("answer")
107
+ budget_sum = self.budget_sum
108
+ acceptable_answer_keys = set(range(len(self.question_options)))
109
+ answer_keys = set([int(k) for k in answer.keys()])
110
+ current_sum = sum(answer.values())
111
+ if not current_sum == budget_sum:
112
+ raise QuestionAnswerValidationError(
113
+ f"Budget sum must be {budget_sum}, but got {current_sum}."
114
+ )
115
+ if any(v < 0 for v in answer.values()):
116
+ raise QuestionAnswerValidationError(
117
+ f"Budget values must be positive, but got {answer_keys}."
118
+ )
119
+ if any([int(key) not in acceptable_answer_keys for key in answer.keys()]):
120
+ raise QuestionAnswerValidationError(
121
+ f"Budget keys must be in {acceptable_answer_keys}, but got {answer_keys}."
122
+ )
123
+ if acceptable_answer_keys != answer_keys:
124
+ missing_keys = acceptable_answer_keys - answer_keys
125
+ raise QuestionAnswerValidationError(
126
+ f"All but keys must be represented in the answer. Missing: {missing_keys}."
127
+ )
128
+
129
+ def _validate_answer_checkbox(self, answer: dict[str, Union[str, int]]) -> None:
130
+ """Validate QuestionCheckbox-specific answer.
131
+
132
+ :param answer: Answer to validate
133
+
134
+
135
+ Check that answer["answer"]:
136
+ - has elements that are strings, bytes-like objects or real numbers evaluating to integers
137
+ - has elements that are in the range of the number of options
138
+ - has at least `min_selections` elements, if provided
139
+ - has at most `max_selections` elements, if provided
140
+ """
141
+ answer_codes = answer["answer"]
142
+ try:
143
+ answer_codes = [int(k) for k in answer["answer"]]
144
+ except:
145
+ raise QuestionAnswerValidationError(
146
+ f"Answer codes must be a list of strings, bytes-like objects or real numbers (got {answer['answer']})."
147
+ )
148
+ acceptable_values = list(range(len(self.question_options)))
149
+ for answer_code in answer_codes:
150
+ if answer_code not in acceptable_values:
151
+ raise QuestionAnswerValidationError(
152
+ f"Answer code {answer_code} has elements not in {acceptable_values}."
153
+ )
154
+ if self.min_selections is not None and len(answer_codes) < self.min_selections:
155
+ raise QuestionAnswerValidationError(
156
+ f"Answer codes, {answer_codes}, has fewer than {self.min_selections} options selected."
157
+ )
158
+ if self.max_selections is not None and len(answer_codes) > self.max_selections:
159
+ raise QuestionAnswerValidationError(
160
+ f"Answer codes, {answer_codes}, has more than {self.max_selections} options selected."
161
+ )
162
+
163
+ def _validate_answer_extract(self, answer: dict[str, Any]) -> None:
164
+ """Validate QuestionExtract-specific answer.
165
+
166
+ Check that answer["answer"]:
167
+ - does not have keys that are not in the answer template
168
+ - has all keys that are in the answer template
169
+ """
170
+ value = answer.get("answer")
171
+ acceptable_answer_keys = set(self.answer_template.keys())
172
+ if any([key not in acceptable_answer_keys for key in value.keys()]):
173
+ raise QuestionAnswerValidationError(
174
+ f"Answer keys must be in {acceptable_answer_keys}, but got {value.keys()}."
175
+ )
176
+ if any([key not in value.keys() for key in acceptable_answer_keys]):
177
+ raise QuestionAnswerValidationError(
178
+ f"Answer must have all keys in {acceptable_answer_keys}, but got {value.keys()}."
179
+ )
180
+
181
+ def _validate_answer_list(self, answer: dict[str, Union[list, str]]) -> None:
182
+ """Validate QuestionList-specific answer.
183
+
184
+ Check that answer["answer"]:
185
+ - is not empty, if `allow_nonresponse` is False
186
+ - has no more than `max_list_items` elements
187
+ - has no empty strings
188
+ """
189
+ value = answer.get("answer")
190
+ if (
191
+ hasattr(self, "allow_nonresponse")
192
+ and self.allow_nonresponse == False
193
+ and (value == [] or value is None)
194
+ ):
195
+ raise QuestionAnswerValidationError("You must provide a response.")
196
+
197
+ if (
198
+ hasattr(self, "max_list_items")
199
+ and self.max_list_items is not None
200
+ and (len(value) > self.max_list_items)
201
+ ):
202
+ raise QuestionAnswerValidationError("Response has too many items.")
203
+
204
+ if any([item == "" for item in value]):
205
+ raise QuestionAnswerValidationError(
206
+ f"Answer cannot contain empty strings, but got {value}."
207
+ )
208
+
209
+ def _validate_answer_numerical(self, answer: dict) -> None:
210
+ """Validate QuestionNumerical-specific answer.
211
+
212
+ Check that answer["answer"]:
213
+ - is not less than `min_value`
214
+ - is not greater than `max_value`
215
+ """
216
+ try:
217
+ value = float(answer.get("answer"))
218
+ except ValueError:
219
+ raise QuestionAnswerValidationError(
220
+ f"Answer must real number or convertible to a real number (got {answer.get('answer')})."
221
+ )
222
+ if self.min_value is not None and value < self.min_value:
223
+ raise QuestionAnswerValidationError(
224
+ f"Value {value} is less than {self.min_value}"
225
+ )
226
+ if self.max_value is not None and value > self.max_value:
227
+ raise QuestionAnswerValidationError(
228
+ f"Value {value} is greater than {self.max_value}"
229
+ )
230
+ return value
231
+
232
+ def _validate_answer_multiple_choice(
233
+ self, answer: dict[str, Union[str, int]]
234
+ ) -> None:
235
+ """Validate QuestionMultipleChoice-specific answer.
236
+
237
+ Check that answer["answer"]:
238
+ - is a string, bytes-like object or real number
239
+ - is a non-negative integer
240
+ - is in the range of the number of options
241
+ """
242
+ try:
243
+ value = int(answer.get("answer"))
244
+ except:
245
+ raise QuestionAnswerValidationError(
246
+ f"Answer code must be a string, a bytes-like object or a real number (got {answer.get('answer')})."
247
+ )
248
+ if not value >= 0:
249
+ raise QuestionAnswerValidationError(
250
+ f"Answer code must be a non-negative integer (got {value})."
251
+ )
252
+ if int(value) not in range(len(self.question_options)):
253
+ raise QuestionAnswerValidationError(
254
+ f"Answer code {value} must be in {list(range(len(self.question_options)))}."
255
+ )
256
+
257
+ def _validate_answer_rank(self, answer: dict[str, Union[str, int]]) -> None:
258
+ """Validate QuestionRank-specific answer.
259
+
260
+ Check that answer["answer"]:
261
+ - contains only integers
262
+ - contains only integers in the range of the number of options
263
+ - has the correct number of elements
264
+ """
265
+ value = answer.get("answer")
266
+ acceptable_values = list(range(len(self.question_options)))
267
+ for v in value:
268
+ try:
269
+ answer_code = int(v)
270
+ except ValueError:
271
+ raise QuestionAnswerValidationError(
272
+ f"Rank answer {value} has elements that are not integers, namely {v}."
273
+ )
274
+ except TypeError:
275
+ raise QuestionAnswerValidationError(
276
+ f"Rank answer {value} has elements that are not integers, namely {v}."
277
+ )
278
+ if answer_code not in acceptable_values:
279
+ raise QuestionAnswerValidationError(
280
+ f"Answer {value} has elements not in {acceptable_values}, namely {v}."
281
+ )
282
+ if len(value) != self.num_selections:
283
+ raise QuestionAnswerValidationError(
284
+ f"Rank answer {value}, but exactly {self.num_selections} selections required."
285
+ )
286
+
287
+ def _validate_answer_matrix(self, answer: dict[str, Any]) -> None:
288
+ """Validate QuestionMatrix-specific answer.
289
+
290
+ Check that answer["answer"]:
291
+ - is a dictionary
292
+ - has all required question_items as keys
293
+ - has values that are valid options from question_options
294
+ - has the correct number of responses (one per item)
295
+ """
296
+ value = answer.get("answer")
297
+
298
+ # Check that answer is a dictionary
299
+ if not isinstance(value, dict):
300
+ raise QuestionAnswerValidationError(
301
+ f"Matrix answer must be a dictionary mapping items to responses (got {value})"
302
+ )
303
+
304
+ # Check that all required items are present
305
+ required_items = set(self.question_items)
306
+ provided_items = set(value.keys())
307
+
308
+ if missing_items := (required_items - provided_items):
309
+ raise QuestionAnswerValidationError(
310
+ f"Missing responses for items: {missing_items}"
311
+ )
312
+
313
+ if extra_items := (provided_items - required_items):
314
+ raise QuestionAnswerValidationError(
315
+ f"Unexpected responses for items: {extra_items}"
316
+ )
317
+
318
+ # Check that all responses are valid options
319
+ valid_options = set(self.question_options)
320
+ for item, response in value.items():
321
+ if response not in valid_options:
322
+ raise QuestionAnswerValidationError(
323
+ f"Invalid response '{response}' for item '{item}'. "
324
+ f"Must be one of: {valid_options}"
325
+ )
326
+
327
+
328
+ if __name__ == "__main__":
329
+ pass
330
+ # import doctest
331
+
332
+ # doctest.testmod(optionflags=doctest.ELLIPSIS)
333
+
334
+ # results = AnswerValidatorMixin().failing_job()