edsl 0.1.38.dev2__py3-none-any.whl → 0.1.38.dev4__py3-none-any.whl

edsl/results/DatasetTree.py

@@ -1,10 +1,18 @@
-from typing import Dict, List, Any, Optional
+from typing import Dict, List, Any, Optional, List
 from docx import Document
 from docx.shared import Inches, Pt
 from docx.enum.text import WD_ALIGN_PARAGRAPH
 from docx.enum.style import WD_STYLE_TYPE
 
 
+def is_hashable(v):
+    try:
+        hash(v)
+        return True
+    except TypeError:
+        return False
+
+
 class TreeNode:
     def __init__(self, key=None, value=None):
         self.key = key
@@ -13,16 +21,21 @@ class TreeNode:
 
 
 class Tree:
-    def __init__(self, data: "Dataset"):
+    def __init__(self, data: "Dataset", node_order: Optional[List[str]] = None):
         d = {}
         for entry in data:
             d.update(entry)
         self.data = d
         self.root = None
 
+        self.node_order = node_order
+
+        self.construct_tree(node_order)
+
     def unique_values_by_keys(self) -> dict:
         unique_values = {}
-        for key, values in self.data.items():
+        for key, raw_values in self.data.items():
+            values = [v if is_hashable(v) else str(v) for v in raw_values]
             unique_values[key] = list(set(values))
         return unique_values
 
@@ -45,15 +58,25 @@ class Tree:
             current = self.root
             for level in node_order[:-1]:
                 value = self.data[level][i]
+                if not is_hashable(value):
+                    value = str(value)
                 if value not in current.children:
                     current.children[value] = TreeNode(key=level, value=value)
                 current = current.children[value]
 
             leaf_key = node_order[-1]
             leaf_value = self.data[leaf_key][i]
+            if not is_hashable(leaf_value):
+                leaf_value = str(leaf_value)
             if leaf_value not in current.children:
                 current.children[leaf_value] = TreeNode(key=leaf_key, value=leaf_value)
 
+    def __repr__(self):
+        if self.node_order is not None:
+            return f"Tree(Dataset({self.data}), node_order={self.node_order})"
+        else:
+            return f"Tree(Dataset({self.data}))"
+
     def print_tree(
         self, node: Optional[TreeNode] = None, level: int = 0, print_keys: bool = False
     ):
@@ -71,7 +94,10 @@ class Tree:
         for child in node.children.values():
             self.print_tree(child, level + 1, print_keys)
 
-    def to_docx(self, filename: str):
+    def to_docx(self, filename: Optional[str] = None):
+        if filename is None:
+            filename = "tree_structure.docx"
+
         doc = Document()
 
         # Create styles for headings
@@ -90,6 +116,110 @@ class Tree:
 
         self._add_to_docx(doc, self.root, 0)
         doc.save(filename)
+        from edsl.utilities.utilities import file_notice
+
+        file_notice(filename)
+
+    def _repr_html_(self):
+        """Returns an interactive HTML representation of the tree with collapsible sections."""
+
+        # Generate a unique ID for this tree instance
+        import uuid
+
+        tree_id = f"tree_{uuid.uuid4().hex[:8]}"
+
+        styles = f"""
+        <div class="{tree_id}">
+            <style>
+                .{tree_id} details {{
+                    margin-left: 20px;
+                }}
+                .{tree_id} summary {{
+                    cursor: pointer;
+                    margin: 2px 0;
+                }}
+                .{tree_id} .value {{
+                    font-family: monospace;
+                    background: #f5f5f5;
+                    padding: 2px 6px;
+                    border-radius: 3px;
+                    margin: 1px 0;
+                }}
+                .{tree_id} .key {{
+                    color: #666;
+                    font-style: italic;
+                }}
+            </style>
+        """
+
+        def node_to_html(node, level=0, print_keys=True):
+            if node is None:
+                return "Tree has not been constructed yet."
+
+            html = []
+
+            if node.value is not None:
+                # Create the node content
+                content = []
+                if print_keys and node.key is not None:
+                    content.append(f'<span class="key">{node.key}: </span>')
+                content.append(f'<span class="value">{node.value}</span>')
+                content_html = "".join(content)
+
+                if node.children:
+                    # Node with children
+                    html.append(f'<details {"open" if level < 1 else ""}>')
+                    html.append(f"<summary>{content_html}</summary>")
+                    for child in node.children.values():
+                        html.append(node_to_html(child, level + 1, print_keys))
+                    html.append("</details>")
+                else:
+                    # Leaf node
+                    html.append(f"<div>{content_html}</div>")
+            else:
+                # Root node with no value
+                if node.children:
+                    for child in node.children.values():
+                        html.append(node_to_html(child, level, print_keys))
+
+            return "\n".join(html)
+
+        tree_html = node_to_html(self.root)
+        return f"{styles}{tree_html}</div>"
+
+    # def _repr_html_(self):
+    #     """Returns an HTML representation of the tree, following the same logic as print_tree."""
+    #     styles = """
+    #     <style>
+    #         .tree-container {
+    #             font-family: monospace;
+    #             white-space: pre;
+    #             margin: 10px;
+    #         }
+    #     </style>
+    #     """
+
+    #     def node_to_html(node, level=0, print_keys=False):
+    #         if node is None:
+    #             node = self.root
+    #             if node is None:
+    #                 return "Tree has not been constructed yet."
+
+    #         html = []
+    #         if node.value is not None:
+    #             indent = "&nbsp;" * 2 * level  # Using &nbsp; for HTML spaces
+    #             if print_keys and node.key is not None:
+    #                 html.append(f"{indent}{node.key}: {node.value}<br>")
+    #             else:
+    #                 html.append(f"{indent}{node.value}<br>")
+
+    #         for child in node.children.values():
+    #             html.append(node_to_html(child, level + 1, print_keys))
+
+    #         return "".join(html)
+
+    #     tree_html = node_to_html(self.root)
+    #     return f'<div class="tree-container">{tree_html}</div>{styles}'
 
     def _add_to_docx(self, doc, node: TreeNode, level: int):
         if node.value is not None:

edsl/results/Result.py

@@ -137,6 +137,15 @@ class Result(Base, UserDict):
         self._combined_dict = None
         self._problem_keys = None
 
+    def _repr_html_(self):
+        # d = self.to_dict(add_edsl_version=False)
+        d = self.to_dict(add_edsl_version=False)
+        data = [[k, v] for k, v in d.items()]
+        from tabulate import tabulate
+
+        table = str(tabulate(data, headers=["keys", "values"], tablefmt="html"))
+        return f"<pre>{table}</pre>"
+
     ###############
     # Used in Results
     ###############
@@ -156,15 +165,15 @@ class Result(Base, UserDict):
             if key in self.question_to_attributes:
                 # You might be tempted to just use the naked key
                 # but this is a bad idea because it pollutes the namespace
-                question_text_dict[key + "_question_text"] = (
-                    self.question_to_attributes[key]["question_text"]
-                )
-                question_options_dict[key + "_question_options"] = (
-                    self.question_to_attributes[key]["question_options"]
-                )
-                question_type_dict[key + "_question_type"] = (
-                    self.question_to_attributes[key]["question_type"]
-                )
+                question_text_dict[
+                    key + "_question_text"
+                ] = self.question_to_attributes[key]["question_text"]
+                question_options_dict[
+                    key + "_question_options"
+                ] = self.question_to_attributes[key]["question_options"]
+                question_type_dict[
+                    key + "_question_type"
+                ] = self.question_to_attributes[key]["question_type"]
 
         return {
             "agent": self.agent.traits

edsl/results/Results.py

@@ -32,7 +32,7 @@ from edsl.results.ResultsDBMixin import ResultsDBMixin
 from edsl.results.ResultsGGMixin import ResultsGGMixin
 from edsl.results.ResultsFetchMixin import ResultsFetchMixin
 
-from edsl.utilities.decorators import add_edsl_version, remove_edsl_version
+from edsl.utilities.decorators import remove_edsl_version
 from edsl.utilities.utilities import dict_hash
 
 
@@ -46,6 +46,9 @@ class Mixins(
     ResultsGGMixin,
     ResultsToolsMixin,
 ):
+    def long(self):
+        return self.table().long()
+
     def print_long(self, max_rows: int = None) -> None:
         """Print the results in long format.
 
@@ -73,6 +76,8 @@ class Results(UserList, Mixins, Base):
     It also has a list of created_columns, which are columns that have been created with `mutate` and are not part of the original data.
     """
 
+    __documentation__ = "https://docs.expectedparrot.com/en/latest/results.html"
+
     known_data_types = [
         "answer",
         "scenario",
@@ -121,13 +126,55 @@ class Results(UserList, Mixins, Base):
         if hasattr(self, "_add_output_functions"):
             self._add_output_functions()
 
+    def _summary(self) -> dict:
+        import reprlib
+
+        # import yaml
+
+        d = {
+            "EDSL Class": "Results",
+            # "docs_url": self.__documentation__,
+            "# of agents": len(set(self.agents)),
+            "# of distinct models": len(set(self.models)),
+            "# of observations": len(self),
+            "# Scenarios": len(set(self.scenarios)),
+            "Survey Length (# questions)": len(self.survey),
+            "Survey question names": reprlib.repr(self.survey.question_names),
+            "Object hash": hash(self),
+        }
+        return d
+
+    def compute_job_cost(self, include_cached_responses_in_cost=False) -> float:
+        """
+        Computes the cost of a completed job in USD.
+        """
+        total_cost = 0
+        for result in self:
+            for key in result.raw_model_response:
+                if key.endswith("_cost"):
+                    result_cost = result.raw_model_response[key]
+
+                    question_name = key.removesuffix("_cost")
+                    cache_used = result.cache_used_dict[question_name]
+
+                    if isinstance(result_cost, (int, float)):
+                        if include_cached_responses_in_cost:
+                            total_cost += result_cost
+                        elif not include_cached_responses_in_cost and not cache_used:
+                            total_cost += result_cost
+
+        return total_cost
+
     def leaves(self):
         leaves = []
         for result in self:
             leaves.extend(result.leaves())
         return leaves
 
-    def tree(
+    def tree(self, node_list: Optional[List[str]] = None):
+        return self.to_scenario_list().tree(node_list)
+
+    def interactive_tree(
         self,
         fold_attributes: Optional[List[str]] = None,
         drop: Optional[List[str]] = None,
@@ -260,13 +307,67 @@ class Results(UserList, Mixins, Base):
 
         return f"Results(data = {reprlib.repr(self.data)}, survey = {repr(self.survey)}, created_columns = {self.created_columns})"
 
+    def table(
+        self,
+        # selector_string: Optional[str] = "*.*",
+        *fields,
+        tablefmt: Optional[str] = None,
+        pretty_labels: Optional[dict] = None,
+        print_parameters: Optional[dict] = None,
+    ):
+        new_fields = []
+        for field in fields:
+            if "." in field:
+                data_type, key = field.split(".")
+                if data_type not in self.known_data_types:
+                    raise ResultsInvalidNameError(
+                        f"{data_type} is not a valid data type. Must be in {self.known_data_types}"
+                    )
+                if key == "*":
+                    for k in self._data_type_to_keys[data_type]:
+                        new_fields.append(k)
+                else:
+                    if key not in self._key_to_data_type:
+                        raise ResultsColumnNotFoundError(
+                            f"{key} is not a valid key. Must be in {self._key_to_data_type}"
+                        )
+                    new_fields.append(key)
+            else:
+                new_fields.append(field)
+
+        return (
+            self.to_scenario_list()
+            .to_dataset()
+            .table(
+                *new_fields,
+                tablefmt=tablefmt,
+                pretty_labels=pretty_labels,
+                print_parameters=print_parameters,
+            )
+        )
+        # return (
+        #     self.select(f"{selector_string}")
+        #     .to_scenario_list()
+        #     .table(*fields, tablefmt=tablefmt)
+        # )
+
     def _repr_html_(self) -> str:
-        # from IPython.display import HTML
+        d = self._summary()
+        from edsl import Scenario
 
-        json_str = json.dumps(self.to_dict(add_edsl_version=False)["data"], indent=4)
-        return f"<pre>{json_str}</pre>"
+        footer = f"<a href={self.__documentation__}>(docs)</a>"
 
-    def to_dict(self, sort=False, add_edsl_version=False) -> dict[str, Any]:
+        s = Scenario(d)
+        td = s.to_dataset().table(tablefmt="html")
+        return td._repr_html_() + footer
+
+    def to_dict(
+        self,
+        sort=False,
+        add_edsl_version=False,
+        include_cache=False,
+        include_task_history=False,
+    ) -> dict[str, Any]:
         from edsl.data.Cache import Cache
 
         if sort:
@@ -280,13 +381,21 @@ class Results(UserList, Mixins, Base):
             ],
             "survey": self.survey.to_dict(add_edsl_version=add_edsl_version),
             "created_columns": self.created_columns,
-            "cache": (
-                Cache()
-                if not hasattr(self, "cache")
-                else self.cache.to_dict(add_edsl_version=add_edsl_version)
-            ),
-            "task_history": self.task_history.to_dict(),
         }
+        if include_cache:
+            d.update(
+                {
+                    "cache": (
+                        Cache()
+                        if not hasattr(self, "cache")
+                        else self.cache.to_dict(add_edsl_version=add_edsl_version)
+                    )
+                }
+            )
+
+        if self.task_history.has_unfixed_exceptions or include_task_history:
+            d.update({"task_history": self.task_history.to_dict()})
+
         if add_edsl_version:
             from edsl import __version__
 
@@ -375,7 +484,11 @@ class Results(UserList, Mixins, Base):
                 cache=(
                     Cache.from_dict(data.get("cache")) if "cache" in data else Cache()
                 ),
-                task_history=TaskHistory.from_dict(data.get("task_history")),
+                task_history=(
+                    TaskHistory.from_dict(data.get("task_history"))
+                    if "task_history" in data
+                    else TaskHistory(interviews=[])
+                ),
             )
         except Exception as e:
             raise ResultsDeserializationError(f"Error in Results.from_dict: {e}")
@@ -875,31 +988,19 @@ class Results(UserList, Mixins, Base):
 
         >>> r = Results.example()
         >>> r.sort_by('how_feeling', reverse=False).select('how_feeling').print()
-        ┏━━━━━━━━━━━━━━┓
-        ┃ answer       ┃
-        ┃ .how_feeling ┃
-        ┡━━━━━━━━━━━━━━┩
-        │ Great        │
-        ├──────────────┤
-        │ OK           │
-        ├──────────────┤
-        │ OK           │
-        ├──────────────┤
-        │ Terrible     │
-        └──────────────┘
+        answer.how_feeling
+        --------------------
+        Great
+        OK
+        OK
+        Terrible
         >>> r.sort_by('how_feeling', reverse=True).select('how_feeling').print()
-        ┏━━━━━━━━━━━━━━┓
-        ┃ answer       ┃
-        ┃ .how_feeling ┃
-        ┡━━━━━━━━━━━━━━┩
-        │ Terrible     │
-        ├──────────────┤
-        │ OK           │
-        ├──────────────┤
-        │ OK           │
-        ├──────────────┤
-        │ Great        │
-        └──────────────┘
+        answer.how_feeling
+        --------------------
+        Terrible
+        OK
+        OK
+        Great
         """
 
         def to_numeric_if_possible(v):
@@ -932,12 +1033,9 @@ class Results(UserList, Mixins, Base):
 
         >>> r = Results.example()
         >>> r.filter("how_feeling == 'Great'").select('how_feeling').print()
-        ┏━━━━━━━━━━━━━━┓
-        ┃ answer       ┃
-        ┃ .how_feeling ┃
-        ┡━━━━━━━━━━━━━━┩
-        │ Great        │
-        └──────────────┘
+        answer.how_feeling
+        --------------------
+        Great
 
         Example usage: Using an OR operator in the filter expression.
 
@@ -948,14 +1046,10 @@ class Results(UserList, Mixins, Base):
         ...
 
         >>> r.filter("how_feeling == 'Great' or how_feeling == 'Terrible'").select('how_feeling').print()
-        ┏━━━━━━━━━━━━━━┓
-        ┃ answer       ┃
-        ┃ .how_feeling ┃
-        ┡━━━━━━━━━━━━━━┩
-        │ Great        │
-        ├──────────────┤
-        │ Terrible     │
-        └──────────────┘
+        answer.how_feeling
+        --------------------
+        Great
+        Terrible
         """
 
         def has_single_equals(string):