pyreadstat 1.3.0__cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl → 1.3.1__cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl

pyreadstat/__init__.py

@@ -22,5 +22,5 @@ from .pyreadstat import read_file_in_chunks, read_file_multiprocessing
 from ._readstat_parser import ReadstatError, metadata_container
 from .pyfunctions import set_value_labels, set_catalog_to_sas
 
-__version__ = "1.3.0"
+__version__ = "1.3.1"
 

pyreadstat/pyfunctions.py

@@ -1,9 +1,10 @@
 """
 Functions written in pure python
 """
-from copy import deepcopy
+from copy import deepcopy, copy
+import warnings
 
-import pandas as pd
+import narwhals.stable.v2 as nw
 
 # Functions to deal with value labels
 
@@ -33,14 +34,42 @@ def set_value_labels(dataframe, metadata, formats_as_category=True, formats_as_o
             otherwise
     """
 
-    df_copy = dataframe.copy()
+    df_copy = nw.from_native(dataframe).clone()
 
     if metadata.value_labels and metadata.variable_to_label:
         for var_name, label_name in metadata.variable_to_label.items():
             labels = metadata.value_labels.get(label_name)
             if labels:
+                labels = deepcopy(labels)
                 if var_name in df_copy.columns:
-                    df_copy[var_name] = df_copy[var_name].apply(lambda x: labels.get(x, x))
+                    # unique does not work for polars Object
+                    if not df_copy.implementation.is_pandas() and df_copy[var_name].dtype == nw.Object:
+                        unvals = list(set(df_copy[var_name].to_list()))
+                    else:
+                        unvals = df_copy[var_name].unique()
+                    for uval in unvals:
+                        if uval not in labels:
+                            labels[uval] = uval
+                    # if all values are null, there will be nothing to replace. However we cannot do replace_strict on null dtype, it raises an error
+                    if not df_copy.implementation.is_pandas() and (len(df_copy[var_name])==df_copy[var_name].null_count()):
+                        continue
+                    # replace_strict requires that all the values are in the map. Could not get map_batches or when/then/otherwise to work
+                    elif not df_copy.implementation.is_pandas() and (df_copy[var_name].dtype==nw.Object or not all([type(v)==type(list(labels.values())[0]) for v in labels.values() if v is not None])):
+                        # polars is very difficult to convince to mix strings and numbers, so we have to do it this way
+                        temp = [labels[x] for x in df_copy[var_name]]
+                        newser = nw.new_series(name=var_name, values= temp, dtype=nw.Object, backend=df_copy.implementation) 
+                        df_copy = df_copy.with_columns(newser.alias(var_name))
+                        if formats_as_category or formats_as_ordered_category:
+                            msg = f"You requested formats_as_category=True or formats_as_ordered_category=True, but it was not possible to cast variable '{var_name}' to category"
+                            warnings.warn(msg, RuntimeWarning)
+                            continue
+                    # not sure if we get into this situation ever or what would exactly happen, maybe this is not needed?
+                    elif not df_copy.implementation.is_pandas() and df_copy[var_name].dtype==nw.Unknown:
+                        msg = f"It was not possible to apply value formats to variable '{var_name}' due to unknown/not supported data type"
+                        warnings.warn(msg, RuntimeWarning)
+                        continue
+                    else:
+                        df_copy = df_copy.with_columns(nw.col(var_name).replace_strict(labels))
                     if formats_as_ordered_category:
                         categories = list(set(labels.values()))
                         original_values = list(labels.keys())
@@ -51,16 +80,12 @@ def set_value_labels(dataframe, metadata, formats_as_category=True, formats_as_o
                             if not revdict.get(curcat):
                                 revdict[curcat] = orival
                         categories.sort(key=revdict.get)
-                        df_copy[var_name] = pd.Categorical(
-                            df_copy[var_name],
-                            ordered = True,
-                            categories = categories
-                        )
+                        df_copy = df_copy.with_columns(nw.col(var_name).cast(nw.Enum(categories)))
                     elif formats_as_category:
-                        df_copy[var_name] = df_copy[var_name].astype("category")
+                        df_copy = df_copy.with_columns(nw.col(var_name).cast(nw.Categorical))
 
 
-    return df_copy
+    return df_copy.to_native()
 
 def set_catalog_to_sas(sas_dataframe, sas_metadata, catalog_metadata, formats_as_category=True,  
                        formats_as_ordered_category=False):
@@ -108,7 +133,8 @@ def set_catalog_to_sas(sas_dataframe, sas_metadata, catalog_metadata, formats_as
         metadata.variable_value_labels = variable_value_labels
 
     else:
-        df_copy = sas_dataframe.copy()
+        #df_copy = sas_dataframe.copy()
+        df_copy = nw.from_native(sas_dataframe).clone().to_native()
         metadata = deepcopy(sas_metadata)
 
     return df_copy, metadata

{pyreadstat-1.3.0.dist-info → pyreadstat-1.3.1.dist-info}/METADATA

@@ -1,22 +1,22 @@
 Metadata-Version: 2.4
 Name: pyreadstat
-Version: 1.3.0
-Summary: Reads and Writes SAS, SPSS and Stata files into/from pandas data frames.
+Version: 1.3.1
+Summary: Reads and Writes SAS, SPSS and Stata files into/from pandas and polars data frames.
 Home-page: https://github.com/Roche/pyreadstat
 Download-URL: https://github.com/Roche/pyreadstat/dist
 Author: Otto Fajardo
 Author-email: pleasecontactviagithub@notvalid.com
-License: Apache License Version 2.0
+License: Apache-2.0
 Classifier: Programming Language :: Python
 Classifier: Programming Language :: Cython
 Classifier: Programming Language :: C
-Classifier: License :: OSI Approved :: Apache Software License
 Classifier: Intended Audience :: Science/Research
 Classifier: Topic :: Scientific/Engineering
 Classifier: Environment :: Console
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: pandas>=1.2.0
+Requires-Dist: narwhals>=2.0
+Requires-Dist: numpy
 Dynamic: author
 Dynamic: author-email
 Dynamic: classifier
@@ -30,7 +30,7 @@ Dynamic: requires-dist
 Dynamic: summary
 
  A Python package to read and write SAS
-(sas7bdat, sas7bcat, xport/xpt), SPSS (sav, zsav, por) and Stata (dta) files into/from pandas data frames. It is a wrapper
+(sas7bdat, sas7bcat, xport/xpt), SPSS (sav, zsav, por) and Stata (dta) files into/from pandas and polars data frames. It is a wrapper
 around the C library readstat.<br>
 Please visit out project home page for more information:<br>
 https://github.com/Roche/pyreadstat

pyreadstat-1.3.1.dist-info/RECORD

@@ -0,0 +1,11 @@
+pyreadstat/__init__.py,sha256=v8Bii3pv9gaU6zX6BDQgQHtPLhPxu0RGTJDUtInmPKg,1232
+pyreadstat/_readstat_parser.cpython-311-aarch64-linux-gnu.so,sha256=4cb8f1xPOJqeRJ1NB-kF1sruJYtEzKXXwObX0TDw-ZQ,632152
+pyreadstat/_readstat_writer.cpython-311-aarch64-linux-gnu.so,sha256=KpevgVSi8UiyEsVnGh1pvXNsKWA0iGS3KKeMN9qDpVY,625920
+pyreadstat/pyfunctions.py,sha256=JkCsGUT71iLU5onnIhTXzMbzjCjGcAz2HMDB0b3ptIU,7577
+pyreadstat/pyreadstat.cpython-311-aarch64-linux-gnu.so,sha256=Kj_Me2tOmzRPjFvzbeLgR1GJCvD61zO4wuPIaTj37cc,627392
+pyreadstat/worker.py,sha256=DHA7KXzZ3KSUSiYrepD7RpABPSLCYDq3-hOcoaFNBaI,972
+pyreadstat-1.3.1.dist-info/METADATA,sha256=SnLBSBGd78hVp0WsWeG787hT5FReEHJmJlqPQjL7ZpQ,1245
+pyreadstat-1.3.1.dist-info/WHEEL,sha256=QJg38rE8f0PT7_ZWlFpvwOoUFGenUbSJhXM-6SbDiao,153
+pyreadstat-1.3.1.dist-info/top_level.txt,sha256=7LlluhR4SADp00dJTEVpKMet_Jki7JHA6abJ-wu831E,11
+pyreadstat-1.3.1.dist-info/RECORD,,
+pyreadstat-1.3.1.dist-info/licenses/LICENSE,sha256=Tjohfl1RlkuDoTF5ctnLvkGnr8TU27PEy7PhOHjRz5c,12903

pyreadstat-1.3.0.dist-info/RECORD

@@ -1,11 +0,0 @@
-pyreadstat/__init__.py,sha256=wIENGF3x1VBer_Tp0r5e8PxGlcz5jDU_Wu2T0qHKya4,1232
-pyreadstat/_readstat_parser.cpython-311-aarch64-linux-gnu.so,sha256=s9Yv7IalNPuMtvG7d-Eu8lqFewTS23JDY79xxO-sg5s,630336
-pyreadstat/_readstat_writer.cpython-311-aarch64-linux-gnu.so,sha256=3ZkVQWom4D2E0d5EJFnkzSW92RsjYuBX3MEiVpyoJts,692800
-pyreadstat/pyfunctions.py,sha256=wnlWbD5o1knLWX28s9ve8jWWv_MFDUTQ7vQUiyNFYmk,5172
-pyreadstat/pyreadstat.cpython-311-aarch64-linux-gnu.so,sha256=qHeQzIKzdLfP4MdLRZ4gVLLDP9YU3z_oxmPjrMcNoo8,626928
-pyreadstat/worker.py,sha256=DHA7KXzZ3KSUSiYrepD7RpABPSLCYDq3-hOcoaFNBaI,972
-pyreadstat-1.3.0.dist-info/METADATA,sha256=kqbzYyVqeHb4XpJzpmbwa0YIyvftxZtdcYt0uUEVMCo,1281
-pyreadstat-1.3.0.dist-info/WHEEL,sha256=QJg38rE8f0PT7_ZWlFpvwOoUFGenUbSJhXM-6SbDiao,153
-pyreadstat-1.3.0.dist-info/top_level.txt,sha256=7LlluhR4SADp00dJTEVpKMet_Jki7JHA6abJ-wu831E,11
-pyreadstat-1.3.0.dist-info/RECORD,,
-pyreadstat-1.3.0.dist-info/licenses/LICENSE,sha256=Tjohfl1RlkuDoTF5ctnLvkGnr8TU27PEy7PhOHjRz5c,12903