small fixes to group/ungropu buttons, node editor

stephenfreund · stephenfreund · commit 63a2e8a6fa14 · 2025-02-10T14:50:46.000-05:00
diff --git a/mxgraph_component/mxgraph_component/frontend/public/index.html b/mxgraph_component/mxgraph_component/frontend/public/index.html
@@ -260,10 +260,8 @@
 
 </head>
 <body>
-  <div id="toolbar" style="margin-top: 100px;">
-    <button id="groupBtn" style="display: none;">Group</button>
-    <button id="ungroupBtn" style="display: none;">Ungroup</button>
-  </div>
+    <button id="groupBtn" style="display: none; margin-top:80px;">Group</button>
+    <button id="ungroupBtn" style="display: none;  margin-top:80px;">Ungroup</button>
 
   <div id="graph-container" style="width: 2000px; height: 1600px; overflow: hidden;">
   </div>
diff --git a/src/flowco/builder/type_ops.py b/src/flowco/builder/type_ops.py
@@ -1,28 +1,21 @@
 from tempfile import TemporaryFile
-from typing import Any, Dict, List, Tuple
+from typing import Any
 
-import json
 import numpy as np
 import pandas as pd
-from typing import Any, Dict, List, Set, Tuple
+from typing import Any
 
 import pickle
 import base64
 import pandas as pd
 import numpy as np
 from typing import Any
 
-from regex import E
 
-from flowco.util.output import error, log, logger
 
 # These are for the typchecker in str_to_type
 import pandas as pd
 import numpy as np
-import matplotlib.pyplot as plt
-import seaborn as sns
-import sklearn
-import scipy
 from typing import get_origin, get_args
 
 
diff --git a/src/flowco/dataflow/extended_type.py b/src/flowco/dataflow/extended_type.py
@@ -1,16 +1,11 @@
 from __future__ import annotations
 import importlib
-import inspect
-import pprint
-import textwrap
-from typing import Any, Iterable, Set, Tuple, Union, Dict, List, Literal, Optional
+from typing import Any, Iterable, Union, Dict, List, Literal, Optional
 from pydantic import BaseModel, Field
 from abc import abstractmethod
 import numpy as np
 import pandas as pd
-from sklearn.linear_model import LinearRegression
 
-from flowco.builder import type_ops
 from flowco.util.output import error
 
 # Define TypeRepresentation before using it in classes
@@ -25,7 +20,7 @@
     "RecordType",
     "DictType",
     # "TupleType",
-    "SklearnClassType",
+    "LibraryClassType",
     "SetType",
     "PDDataFrameType",
     "PDSeriesType",
@@ -755,11 +750,11 @@ def type_schema(self) -> Dict[str, Any]:
         return schema
 
 
-class SklearnClassType(BaseType):
+class LibraryClassType(BaseType):
     type: Literal["class"]
     name: str = Field(
         ...,
-        description="The fully qualified name of any class in the sklearn library.",
+        description="The fully qualified name of any class in the sklearn or statsmodels library.",
     )  # Required field
     description: str = Field(
         ..., description="A human-readable description of the type."  # Required field
@@ -991,9 +986,9 @@ def infer_type(val) -> "TypeRepresentation":
                     description="Automatically inferred as NumpyNdarrayType.",
                 )
             elif not isinstance(val, type):
-                return SklearnClassType(
+                return LibraryClassType(
                     name=str(type(val)),
-                    description="Automatically inferred as SklearnClassType.",
+                    description="Automatically inferred as LibraryClassType.",
                 )
             return AnyType(description="Automatically inferred as AnyType.")
 
@@ -1159,7 +1154,7 @@ def rep_to_summary(the_type: TypeRepresentation) -> str:
     RecordType,
     DictType,
     # TupleType,
-    SklearnClassType,
+    LibraryClassType,
     SetType,
     PDDataFrameType,
     PDSeriesType,
@@ -1179,7 +1174,7 @@ def rep_to_summary(the_type: TypeRepresentation) -> str:
 RecordType.model_rebuild()
 DictType.model_rebuild()  # Newly added
 # TupleType.model_rebuild()
-SklearnClassType.model_rebuild()
+LibraryClassType.model_rebuild()
 SetType.model_rebuild()
 PDDataFrameType.model_rebuild()
 PDSeriesType.model_rebuild()
@@ -1204,6 +1199,7 @@ class update_node(BaseModel):
 
 
 if __name__ == "__main__":
+    from sklearn.linear_model import LinearRegression
 
     from openai import OpenAI
     import openai
diff --git a/src/flowco/ui/dialogs/node_editor.py b/src/flowco/ui/dialogs/node_editor.py
@@ -262,12 +262,13 @@ def save(self):
             page: Page = ui_page.page()
             page.clean(node.id)  # !!! this can change the page.dfg
 
+            node = node.update(phase=Phase.requirements)
+
         dfg = ui_page.dfg()  # must reload
 
         for phase in visible_phases():
             node = node.update(cache=node.cache.update(phase=phase, node=node))
 
-        node = node.update(phase=min(node.phase, Phase.code))
 
         log(
             "Updating node",
diff --git a/src/flowco/util/prompts.yaml b/src/flowco/util/prompts.yaml
@@ -2,7 +2,7 @@ system-prompt: |
     You are an assistant to take a data science computation dataflow diagram 
     and generate code. 
 
-    You may only import matplotlib, seaborn, numpy, sklearn, and pandas.  
+    You may only import matplotlib, seaborn, numpy, sklearn, statsmodels, scipy, and pandas.  
 
 semantic-diff: |
     Here are the old and new values for parts of a computation step:
@@ -523,113 +523,6 @@ algorithm: |
 
 
 
-# assess: |
-#     Here are the preconditions for a set of variables, and a post-condition:
-#     ```
-#     {requirements}
-#     ```
-#     And an algorithm:
-#     ```
-#     {algorithm}
-#     ```
-#     Assume an expert data scientist will write code for this algorithm using Python and the standard
-#     numpy, pandas, matplotlib, seaborn, scipy, and sklearn libraries.  
-#     The expert has no other information about the function but can make reasonable choices about 
-#     how to implement parts of it.
-
-#     I want you to assess the algorithm.  Here are the criteria:
-#     * Are the postconditions unambiguous in their meaning?
-#     * Does the algorithm meet the post-conditions if the preconditions are all satisfied?
-#     * Is it a meaningful algorithm?
-#     * Are the results easy to interpret?
-
-#     Answer, `Good`, `Bad`, or `More Information Needed`. 
-    
-#     If your answer is `Bad`, provide a reason why the algorithm does not meet the post-condition.
-    
-#     If your answer is `More Information Needed`, provide the most important question that would help you make a decision:
-#     * If additional data is needed to perform the computation, ask for that data.
-#     * If more information about the algorithm is needed, ask the most important question related to that algorithm.
-
-
-# return-type: |
-#     Here are the pre and post conditions for a function
-#     ```
-#     {requirements}
-#     ```
-#     Create a return type for the function that meets the postconditions.
-    
-#     The return type's specification
-#     should be detailed enough that users of the function can understand what
-#     the function returns.  Do not include a specificaiton if the return
-#     type is None.
-    
-#     Assume the client has no other information about the function.
-    
-#     If this function returns data from a file, use the column labels 
-#     and types found in that file.
-
-#     The type should reflect what is described in the specification.
-#     The return type for functions generating plots should be None.
-
-# return-description: |
-#     Here are the pre and post conditions for a function
-#     ```
-#     {requirements}
-#     ```
-#     Give a one sentence description of the value returned by
-#     this function.
-
-# description: |
-#     Here are the preconditions for a set of variables used by 
-#     a computation:
-#     ```
-#     {preconditions}
-#     ```
-#     Here are the requirements for the return value `{function_return_var}`:
-#     ```
-#     {requirements}
-#     ```
-
-#     Create a prose description for this step of the computation.
-
-# parameters: |
-#     Identify the function_return_vars from all direct and indirect 
-#     predecessors that will be necessary to compute the results of this stage.
-
-#     Your answer should be a list of variables drawn from this list:
-#     ```
-#     {result_vars}
-#     ``` 
-
-
-# signatures: |
-#     Create a Python function signature for this step of the computation.
-
-#     The signature should include the function name, the parameters, and the return
-#     type for this step.  
-
-#     The function name should capture the intent of this step computation.
-    
-#     There is one parameter for each predecessor node.  For this node, 
-#     there are `{n_predecessors}` predecessors.  They are:
-#     ```
-#     {predecessors}
-#     ```
-
-#     The parameter types should match the return values for 
-#     the predecessors' functions.
-    
-#     The return type should be described with its type and any
-#     requirements for the value.  If a stage has no outgoing edges.
-
-#     Assume that the following modules have been imported:
-#     ```
-#     import pandas as pd
-#     import numpy as np
-#     import seaborn as sns
-#     import matplotlib.pyplot as plt
-#     ```
 
 compile: |
     Write the code for this step as a single function.
@@ -649,7 +542,7 @@ compile: |
     * Be optimized for readability, clarity, and simplicity.
     * Do not include pydoc strings.
     
-    You may only import matplotlib, seaborn, numpy, sklearn, scipy, and pandas.  
+    You may only import matplotlib, seaborn, numpy, sklearn, statsmodels, scipy, and pandas.  
 
     Import any classes referenced in the return type.
     
@@ -1046,7 +939,7 @@ ama_node_editor: |
     * Be optimized for readability, clarity, and simplicity.
     * Do not include pydoc strings.
     
-    You may only import matplotlib, seaborn, numpy, sklearn, scipy, and pandas.  
+    You may only import matplotlib, seaborn, numpy, sklearn, statsmodels, scipy, and pandas.  
 
     Import any classes referenced in the return type.
     
@@ -1207,8 +1100,8 @@ repair-syntax: |
     * Have no side effects.
     * Be optimized for readability, clarity, and simplicity.
     * Do not include pydoc strings.
-    
-    You may only import matplotlib, seaborn, numpy, sklearn, scipy, and pandas.  
+
+    You may only import matplotlib, seaborn, numpy, sklearn, statsmodels, scipy, and pandas.  
 
     Import any classes referenced in the return type.
     
@@ -1280,7 +1173,7 @@ repair-node-run: |
     * Be optimized for readability, clarity, and simplicity.
     * Do not include pydoc strings.
     
-    You may only import matplotlib, seaborn, numpy, sklearn, scipy, and pandas.  
+    You may only import matplotlib, seaborn, numpy, sklearn, statsmodels, scipy, and pandas.  
 
     Import any classes referenced in the return type.
     
@@ -1681,7 +1574,8 @@ full-compile-code: |
     * Have no side effects.
     * Be optimized for readability, clarity, and simplicity.
     
-    You may only import matplotlib, seaborn, numpy, sklearn, and pandas.  Uses seaborn for all plots, with
+    You may only import matplotlib, seaborn, numpy, sklearn, statsmodels, scipy, and pandas.  
+    Use seaborn for all plots, with
     no changes to the base style.  Do not call `plt.close()` and *do not* save them to disk.
 
     Do not include other functions.
diff --git a/src/flowthon/__main__.py b/src/flowthon/__main__.py
@@ -12,7 +12,6 @@
 import webbrowser
 
 import nbformat
-from sklearn import base
 
 from flowthon.nbflowthon import convert_notebook_to_flowthon
 from flowthon.nbsplit import split_notebook_by_heading_level
