edsl 0.1.40.dev2__py3-none-any.whl → 0.1.42__py3-none-any.whl

edsl/results/Results.py

@@ -90,6 +90,7 @@ class Results(UserList, Mixins, Base):
         "comment",
         "generated_tokens",
         "cache_used",
+        "cache_keys",
     ]
 
     def __init__(
@@ -109,6 +110,7 @@ class Results(UserList, Mixins, Base):
         :param created_columns: A list of strings that are created columns.
         :param job_uuid: A string representing the job UUID.
         :param total_results: An integer representing the total number of results.
+        :cache: A Cache object.
         """
         super().__init__(data)
         from edsl.data.Cache import Cache
@@ -138,6 +140,16 @@ class Results(UserList, Mixins, Base):
         }
         return d
 
+    def _cache_keys(self):
+        cache_keys = []
+        for result in self:
+            cache_keys.extend(list(result["cache_keys"].values()))
+        return cache_keys
+
+    def relevant_cache(self, cache: Cache) -> Cache:
+        cache_keys = self._cache_keys()
+        return cache.subset(cache_keys)
+
     def insert(self, item):
         item_order = getattr(item, "order", None)
         if item_order is not None:
@@ -170,12 +182,12 @@ class Results(UserList, Mixins, Base):
         """
         total_cost = 0
         for result in self:
-            for key in result.raw_model_response:
+            for key in result["raw_model_response"]:
                 if key.endswith("_cost"):
-                    result_cost = result.raw_model_response[key]
+                    result_cost = result["raw_model_response"][key]
 
                     question_name = key.removesuffix("_cost")
-                    cache_used = result.cache_used_dict[question_name]
+                    cache_used = result["cache_used_dict"][question_name]
 
                     if isinstance(result_cost, (int, float)):
                         if include_cached_responses_in_cost:
@@ -349,7 +361,7 @@ class Results(UserList, Mixins, Base):
         self,
         sort: bool = False,
         add_edsl_version: bool = False,
-        include_cache: bool = False,
+        include_cache: bool = True,
         include_task_history: bool = False,
         include_cache_info: bool = True,
     ) -> dict[str, Any]:
@@ -635,7 +647,7 @@ class Results(UserList, Mixins, Base):
 
         >>> r = Results.example()
         >>> r.model_keys
-        ['frequency_penalty', 'logprobs', 'max_tokens', 'model', 'model_index', 'presence_penalty', 'temperature', 'top_logprobs', 'top_p']
+        ['frequency_penalty', 'inference_service', 'logprobs', 'max_tokens', 'model', 'model_index', 'presence_penalty', 'temperature', 'top_logprobs', 'top_p']
         """
         return sorted(self._data_type_to_keys["model"])
 

edsl/scenarios/FileStore.py

@@ -327,6 +327,38 @@ class FileStore(Scenario):
 
         return ConstructDownloadLink(self).create_link(custom_filename, style)
 
+    def to_pandas(self):
+        """
+        Convert the file content to a pandas DataFrame if supported by the file handler.
+
+        Returns:
+            pandas.DataFrame: The data from the file as a DataFrame
+
+        Raises:
+            AttributeError: If the file type's handler doesn't support pandas conversion
+        """
+        handler = FileMethods.get_handler(self.suffix)
+        if handler and hasattr(handler, "to_pandas"):
+            return handler(self.path).to_pandas()
+        raise AttributeError(
+            f"Converting {self.suffix} files to pandas DataFrame is not supported"
+        )
+
+    def __getattr__(self, name):
+        """
+        Delegate pandas DataFrame methods to the underlying DataFrame if this is a CSV file
+        """
+        if self.suffix == "csv":
+            # Get the pandas DataFrame
+            df = self.to_pandas()
+            # Check if the requested attribute exists in the DataFrame
+            if hasattr(df, name):
+                return getattr(df, name)
+        # If not a CSV or attribute doesn't exist in DataFrame, raise AttributeError
+        raise AttributeError(
+            f"'{self.__class__.__name__}' object has no attribute '{name}'"
+        )
+
 
 class CSVFileStore(FileStore):
     @classmethod

edsl/scenarios/PdfExtractor.py

@@ -2,14 +2,11 @@ import os
 
 
 class PdfExtractor:
-    def __init__(self, pdf_path: str, parent_object: object):
+    def __init__(self, pdf_path: str):
         self.pdf_path = pdf_path
-        self.constructor = parent_object.__class__
+        #self.constructor = parent_object.__class__
 
-    def get_object(self) -> object:
-        return self.constructor(self._get_pdf_dict())
-
-    def _get_pdf_dict(self) -> dict:
+    def get_pdf_dict(self) -> dict:
         # Ensure the file exists
         import fitz
 

edsl/scenarios/Scenario.py

@@ -358,7 +358,8 @@ class Scenario(Base, UserDict, ScenarioHtmlMixin):
     def from_pdf(cls, pdf_path: str):
         from edsl.scenarios.PdfExtractor import PdfExtractor
 
-        return PdfExtractor(pdf_path, cls).get_object()
+        extractor = PdfExtractor(pdf_path)
+        return Scenario(extractor.get_pdf_dict())
 
     @classmethod
     def from_docx(cls, docx_path: str) -> "Scenario":

edsl/scenarios/handlers/csv.py

@@ -36,3 +36,14 @@ class CsvMethods(FileMethods):
         with tempfile.NamedTemporaryFile(delete=False, suffix=".csv") as f:
             df.to_csv(f.name, index=False)
         return f.name
+
+    def to_pandas(self):
+        """
+        Convert the CSV file to a pandas DataFrame.
+
+        Returns:
+            pandas.DataFrame: The data from the CSV as a DataFrame
+        """
+        import pandas as pd
+
+        return pd.read_csv(self.path)

edsl/surveys/Survey.py

@@ -391,6 +391,10 @@ class Survey(SurveyExportMixin, Base):
 
             if (class_name := pass_dict.get("edsl_class_name")) == "QuestionBase":
                 return QuestionBase
+            elif pass_dict.get("edsl_class_name") == "QuestionDict":
+                from edsl.questions.QuestionDict import QuestionDict
+
+                return QuestionDict
             elif class_name == "Instruction":
                 from edsl.surveys.instructions.Instruction import Instruction
 
@@ -1277,4 +1281,4 @@ if __name__ == "__main__":
     import doctest
 
     # doctest.testmod(optionflags=doctest.ELLIPSIS | doctest.SKIP)
-    doctest.testmod(optionflags=doctest.ELLIPSIS)
+    doctest.testmod(optionflags=doctest.ELLIPSIS)

edsl/templates/error_reporting/base.html

@@ -3,7 +3,7 @@
 <head>
     <meta charset="UTF-8">
     <meta name="viewport" content="width=device-width, initial-scale=1.0">
-    <title>Exception Details</title>
+    <title>Exceptions Report</title>
     <style>
     {{ css }}
     </style>
@@ -15,9 +15,7 @@
 </head>
 <body>
     {% include 'overview.html' %}
-    {% include 'exceptions_by_type.html' %}
-    {% include 'exceptions_by_model.html' %}
-    {% include 'exceptions_by_question_name.html' %}
+    {% include 'exceptions_table.html' %}
     {% include 'interviews.html' %}
     {% include 'performance_plot.html' %}
 </body>

edsl/templates/error_reporting/exceptions_table.html

@@ -0,0 +1,35 @@
+<style>
+    th, td {
+        padding: 0 10px; /* This applies the padding uniformly to all td elements */
+    }
+</style>
+
+<table border="1">
+    <thead>
+        <tr>
+            <th>Exception Type</th>
+            <th>Service</th>
+            <th>Model</th>
+            <th>Question Name</th>
+            <th>Total</th>
+        </tr>
+    </thead>
+    <tbody>
+        {% for (exception_type, service, model, question_name), count in exceptions_table.items() %}
+            <tr>
+                <td>{{ exception_type }}</td>
+                <td>{{ service }}</td>
+                <td>{{ model }}</td>
+                <td>{{ question_name }}</td>
+                <td>{{ count }}</td>
+            </tr>
+        {% endfor %}
+    </tbody>
+</table>
+<p>
+    <i>Note:</i> You may encounter repeated exceptions where retries were attempted. 
+    You can modify the maximum number of attempts for failed API calls in `edsl/config.py`. 
+</p>
+<p>
+    Click to expand the details below for information about each exception, including code for reproducing it.
+</p>

edsl/templates/error_reporting/interview_details.html

@@ -1,43 +1,67 @@
-<div class="question">question_name: {{ question }}</div>
-
+<style>
+    td {
+        padding: 0 10px; /* This applies the padding uniformly to all td elements */
+    }
+    .toggle-btn {
+        background-color: #4CAF50;
+        color: white;
+        border: none;
+        padding: 10px 20px;
+        text-align: center;
+        text-decoration: none;
+        display: inline-block;
+        font-size: 16px;
+        margin: 4px 2px;
+        cursor: pointer;
+        border-radius: 8px;
+        white-space: nowrap;
+    }
+    .toggle-btn span.collapse {
+        display: none;
+    }
+    .exception-content {
+        max-width: 100%; /* Adjust this value based on your layout */
+        overflow-x: auto; /* Enables horizontal scrolling */
+    }
+</style>
 
-<h2>Exception details</h2>
+<div class="question">question_name: {{ question }}</div>
 
 {% for exception_message in exceptions %}
 <div class="exception-detail">
-        <div class="exception-header">
+    <div class="exception-header">
         <span class="exception-exception">Exception: {{ exception_message.name }}</span>
-        <button class="toggle-btn">▼</button>
-        </div>
-        <div class="exception-content">
+        <button id="toggleBtn" class="toggle-btn" onclick="toggleButton(this)" aria-expanded="false">
+            <span class="expand"> ▼ </span>
+        </button>             
+    </div>
+    <div class="exception-content">
         <table border="1">
-            <tr>
-                <th>Key</th>
-                <th>Value</th>
-            </tr>
             <tr>
                 <td>Interview ID (index in results)</td>
                 <td>{{ index }}</td>
             </tr>
             <tr>
-                <td>Question name (question_name)</td>
+                <td>Question name</td>
                 <td>{{ question }}</td>
             </tr>
- 
             <tr>
-                <td>Question type (question_type)</td>
+                <td>Question type</td>
                 <td>{{ exception_message.question_type }}</td>
             </tr>
- 
             <tr>
                 <td>Human-readable question</td>
                 <td>{{ interview.survey._get_question_by_name(question).html(
                     scenario = interview.scenario, 
                     agent = interview.agent,
-                    answers = exception_message.answers) 
-                
+                    answers = exception_message.answers
+                    ) 
                 }}</td>
             </tr>
+            <tr>
+                <td>User Prompt</td>
+                <td><pre>{{ exception_message.rendered_prompts['user_prompt'] }}</pre></td>
+            </tr>
             <tr>
                 <td>Scenario</td>
                 <td>{{ interview.scenario.__repr__() }}</td>
@@ -47,24 +71,20 @@
                 <td>{{ interview.agent.__repr__() }}</td>
             </tr>
             <tr>
-                <td>Model name</td>
-                <td>{{ interview.model.model }}</td>
+                <td>System Prompt</td>
+                <td><pre>{{ exception_message.rendered_prompts['system_prompt'] }}</pre></td>
             </tr>
             <tr>
                 <td>Inference service</td>
                 <td>{{ interview.model._inference_service_ }}</td>
             </tr>
             <tr>
-                <td>Model parameters</td>
-                <td>{{ interview.model.__repr__() }}</td>
-            </tr>
-            <tr>
-                <td>User Prompt</td>
-                <td><pre>{{ exception_message.rendered_prompts['user_prompt'] }}</pre></td>
+                <td>Model name</td>
+                <td>{{ interview.model.model }}</td>
             </tr>
             <tr>
-                <td>System Prompt</td>
-                <td><pre>{{ exception_message.rendered_prompts['system_prompt'] }}</pre></td>
+                <td>Model parameters</td>
+                <td>{{ interview.model.__repr__() }}</td>
             </tr>
             <tr>
                 <td>Raw model response</td>
@@ -77,7 +97,7 @@
                 </td>
             </tr>
             <tr>
-            <td>Code to (likely) reproduce the error</td>
+            <td>Code likely to reproduce the error</td>
             <td>
                 <textarea id="codeToCopy" rows="10" cols="90">{{ exception_message.code_to_reproduce }}</textarea>
                 <button onclick="copyCode()">Copy</button>
@@ -85,32 +105,26 @@
             </tr>
 
         </table>
-
-        
-            {% if exception_message.exception.__class__.__name__ == 'QuestionAnswerValidationError' %}        
-            <h3>Answer validation details</h3>
-            <table border="1">
-                <tr>
-                    <th>Field</th>
-                    <th>Value</th>
-                </tr>
-                {% for field, (explanation, open_tag, close_tag, value) in exception_message.exception.to_html_dict().items() %}
-                
-                <tr>
-                    <td>{{ field }}: ({{ explanation }})</td>
-                    <td><{{open_tag}}> {{ value | escape }} <{{close_tag}}></td>
-                </tr>
-                {% endfor %}
-            </table>
-        {% endif %}
-
-            <div class="exception-time">Time: {{ exception_message.time }}</div>          
-            <div class="exception-traceback">Traceback: 
-                <text>
-                <pre>{{ exception_message.traceback }}</pre>
-                </text>
-            </div>
+    
+        {% if exception_message.exception.__class__.__name__ == 'QuestionAnswerValidationError' %}        
+        <h3>Answer validation details</h3>
+        <table border="1">
+            {% for field, (open_tag, close_tag, value) in exception_message.exception.to_html_dict().items() %}
+            <tr>
+                <td>{{ field }}</td>
+                <td><{{ open_tag }}> {{ value | escape }} <{{ close_tag }}></td>
+            </tr>
+            {% endfor %}
+        </table>
+    {% endif %}
+        <br><br>
+        <div class="exception-time">Time: {{ exception_message.time }}</div>          
+        <div class="exception-traceback">Traceback: 
+            <text>
+            <pre>{{ exception_message.traceback }}</pre>
+            </text>
         </div>
     </div>
+</div>
 
 {% endfor %}

edsl/templates/error_reporting/interviews.html

@@ -1,19 +1,6 @@
-
-{% if interviews|length > max_interviews %}
-    <h1>Only showing the first {{ max_interviews }} interviews with errors</h1>
-{% else %}
-    <h1>Showing all interviews</h1>    
-{% endif %}
-
+<h2>Exceptions Details</h2>
 {% for index, interview in interviews.items() %}
-    {% if index < max_interviews %}
-        {% if interview.exceptions != {} %}
-        <div class="interview">Interview: {{ index }} </div>
-        Model: {{ interview.model.model }}
-        <h1>Failing questions</h1>
-        {% endif %}
-        {% for question, exceptions in interview.exceptions.items() %}
-            {% include 'interview_details.html' %}
-        {% endfor %}
-    {% endif %}            
+    {% for question, exceptions in interview.exceptions.items() %}
+        {% include 'interview_details.html' %}
+    {% endfor %}
 {% endfor %}

edsl/templates/error_reporting/overview.html

@@ -1,5 +1,31 @@
-<h1>Overview</h1>
-<p>There were {{ interviews|length }} total interview(s). An 'interview' is the result of one survey, taken by one agent, with one model, with one scenario.</p>    
-The number of interviews with any exceptions was {{ num_exceptions }}.</p>
-<p>For advice on dealing with exceptions on Expected Parrot, 
-see <a href="https://docs.expectedparrot.com/en/latest/exceptions.html">here</a>.</p>
+<style>
+    td {
+        padding: 0 10px; /* This applies the padding uniformly to all td elements */
+    }
+</style>
+
+<h1>Exceptions Report</h1>
+<p>
+    This report summarizes exceptions encountered in the job that was run.
+</p>
+<p>
+    For advice on dealing with exceptions, please see the EDSL <a href="https://docs.expectedparrot.com/en/latest/exceptions.html">documentation</a> page. <br>
+    You can also post a question at the Expected Parrot <a href="https://discord.com/invite/mxAYkjfy9m">Discord channel</a>, open an issue on <a href="https://github.com/expectedparrot/edsl">GitHub</a>, or send an email to <a href="mailto:info@expectedparrot.com">info@expectedparrot.com</a>.
+</p>
+
+<h2>Overview</h2>
+<table border="1">
+    <tbody>
+        <tr>
+            <td>Total interviews</td>
+            <td>{{ interviews|length }}</td>
+        </tr>
+        <tr>
+            <td>Interviews with exceptions</td>
+            <td>{{ num_exceptions }}</td>
+        </tr>
+    </tbody>
+</table>
+<p>
+    An "interview" is the result of one survey, taken by one agent, with one model and one scenario (if any).
+</p>

edsl/templates/error_reporting/performance_plot.html

@@ -1,2 +1,2 @@
-<h1>Performance Plot</h1>
+<h2>Performance Plot</h2>
 {{ performance_plot_html }}

{edsl-0.1.40.dev2.dist-info → edsl-0.1.42.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: edsl
-Version: 0.1.40.dev2
+Version: 0.1.42
 Summary: Create and analyze LLM-based surveys
 Home-page: https://www.expectedparrot.com/
 License: MIT