rootloader 1.2.7__tar.gz → 1.3.1__tar.gz

{rootloader-1.2.7 → rootloader-1.3.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rootloader
-Version: 1.2.7
+Version: 1.3.1
 Summary: Read simple root files into memory
 Project-URL: Homepage, https://github.com/ucn-triumf/rootloader
 Project-URL: Bug Tracker, https://github.com/ucn-triumf/rootloader/issues

{rootloader-1.2.7 → rootloader-1.3.1}/docs/rootloader/ttree.md

@@ -19,17 +19,19 @@
     - [ttree.min](#ttreemin)
     - [ttree.reset](#ttreereset)
     - [ttree.reset_columns](#ttreereset_columns)
-    - [ttree.rms](#ttreerms)
     - [ttree.set_filter](#ttreeset_filter)
     - [ttree.set_index](#ttreeset_index)
     - [ttree.size](#ttreesize)
     - [ttree.std](#ttreestd)
+    - [ttree.sum](#ttreesum)
+    - [ttree.to_array](#ttreeto_array)
     - [ttree.to_dataframe](#ttreeto_dataframe)
     - [ttree.to_dict](#ttreeto_dict)
+    - [ttree.values](#ttreevalues)
 
 ## ttree
 
-[Show source in ttree.py:14](../../rootloader/ttree.py#L14)
+[Show source in ttree.py:103](../../rootloader/ttree.py#L103)
 
 Extract ROOT.TTree with lazy operation. Looks like a dataframe in most ways
 
@@ -48,7 +50,7 @@ class ttree(object):
 
 ### ttree.__getitem__
 
-[Show source in ttree.py:79](../../rootloader/ttree.py#L79)
+[Show source in ttree.py:165](../../rootloader/ttree.py#L165)
 
 Fetch a new dataframe with fewer 'columns', as a memory view
 
@@ -60,7 +62,9 @@ def __getitem__(self, key): ...
 
 ### ttree.columns
 
-[Show source in ttree.py:318](../../rootloader/ttree.py#L318)
+[Show source in ttree.py:420](../../rootloader/ttree.py#L420)
+
+Return list of column (branch) names
 
 #### Signature
 
@@ -71,7 +75,9 @@ def columns(self): ...
 
 ### ttree.filters
 
-[Show source in ttree.py:321](../../rootloader/ttree.py#L321)
+[Show source in ttree.py:424](../../rootloader/ttree.py#L424)
+
+Return list of RDataFrame filters
 
 #### Signature
 
@@ -82,7 +88,7 @@ def filters(self): ...
 
 ### ttree.hist1d
 
-[Show source in ttree.py:129](../../rootloader/ttree.py#L129)
+[Show source in ttree.py:214](../../rootloader/ttree.py#L214)
 
 Return histogram of column
 
@@ -107,7 +113,7 @@ def hist1d(self, column=None, nbins=None, step=None, edges=None): ...
 
 ### ttree.hist2d
 
-[Show source in ttree.py:182](../../rootloader/ttree.py#L182)
+[Show source in ttree.py:267](../../rootloader/ttree.py#L267)
 
 Return histogram of two columns
 
@@ -131,7 +137,9 @@ def hist2d(
 
 ### ttree.index
 
-[Show source in ttree.py:324](../../rootloader/ttree.py#L324)
+[Show source in ttree.py:428](../../rootloader/ttree.py#L428)
+
+Return ttree of just the index data
 
 #### Signature
 
@@ -142,7 +150,9 @@ def index(self): ...
 
 ### ttree.index_name
 
-[Show source in ttree.py:327](../../rootloader/ttree.py#L327)
+[Show source in ttree.py:432](../../rootloader/ttree.py#L432)
+
+Return string of the name of the index branch
 
 #### Signature
 
@@ -153,7 +163,9 @@ def index_name(self): ...
 
 ### ttree.loc
 
-[Show source in ttree.py:330](../../rootloader/ttree.py#L330)
+[Show source in ttree.py:436](../../rootloader/ttree.py#L436)
+
+Return a ttree that can be indexed like a pandas dataframe
 
 #### Signature
 
@@ -164,7 +176,9 @@ def loc(self): ...
 
 ### ttree.max
 
-[Show source in ttree.py:354](../../rootloader/ttree.py#L354)
+[Show source in ttree.py:473](../../rootloader/ttree.py#L473)
+
+Return the max value of the tree, for each branch
 
 #### Signature
 
@@ -174,7 +188,9 @@ def max(self): ...
 
 ### ttree.mean
 
-[Show source in ttree.py:359](../../rootloader/ttree.py#L359)
+[Show source in ttree.py:476](../../rootloader/ttree.py#L476)
+
+Return the mean value of the tree, for each branch
 
 #### Signature
 
@@ -184,7 +200,9 @@ def mean(self): ...
 
 ### ttree.min
 
-[Show source in ttree.py:349](../../rootloader/ttree.py#L349)
+[Show source in ttree.py:470](../../rootloader/ttree.py#L470)
+
+Return the min value of the tree, for each branch
 
 #### Signature
 
@@ -194,7 +212,7 @@ def min(self): ...
 
 ### ttree.reset
 
-[Show source in ttree.py:254](../../rootloader/ttree.py#L254)
+[Show source in ttree.py:339](../../rootloader/ttree.py#L339)
 
 Make a new tree
 
@@ -206,7 +224,7 @@ def reset(self): ...
 
 ### ttree.reset_columns
 
-[Show source in ttree.py:258](../../rootloader/ttree.py#L258)
+[Show source in ttree.py:343](../../rootloader/ttree.py#L343)
 
 Include all columns again
 
@@ -216,19 +234,9 @@ Include all columns again
 def reset_columns(self): ...
 ```
 
-### ttree.rms
-
-[Show source in ttree.py:364](../../rootloader/ttree.py#L364)
-
-#### Signature
-
-```python
-def rms(self): ...
-```
-
 ### ttree.set_filter
 
-[Show source in ttree.py:268](../../rootloader/ttree.py#L268)
+[Show source in ttree.py:353](../../rootloader/ttree.py#L353)
 
 Set a filter on the dataframe to select a subset of the data
 
@@ -240,7 +248,7 @@ def set_filter(self, expression, inplace=False): ...
 
 ### ttree.set_index
 
-[Show source in ttree.py:262](../../rootloader/ttree.py#L262)
+[Show source in ttree.py:347](../../rootloader/ttree.py#L347)
 
 Set the index column name
 
@@ -252,7 +260,9 @@ def set_index(self, column): ...
 
 ### ttree.size
 
-[Show source in ttree.py:333](../../rootloader/ttree.py#L333)
+[Show source in ttree.py:440](../../rootloader/ttree.py#L440)
+
+Return the number of rows in the ttree
 
 #### Signature
 
@@ -263,7 +273,9 @@ def size(self): ...
 
 ### ttree.std
 
-[Show source in ttree.py:369](../../rootloader/ttree.py#L369)
+[Show source in ttree.py:482](../../rootloader/ttree.py#L482)
+
+Return the standard deviationif the of values the tree, for each branch
 
 #### Signature
 
@@ -271,11 +283,35 @@ def size(self): ...
 def std(self): ...
 ```
 
+### ttree.sum
+
+[Show source in ttree.py:479](../../rootloader/ttree.py#L479)
+
+Return the sum of the values of the tree, for each branch
+
+#### Signature
+
+```python
+def sum(self): ...
+```
+
+### ttree.to_array
+
+[Show source in ttree.py:364](../../rootloader/ttree.py#L364)
+
+Return ttree data as 1D or 2D numpy array (depending on number of columns)
+
+#### Signature
+
+```python
+def to_array(self): ...
+```
+
 ### ttree.to_dataframe
 
-[Show source in ttree.py:280](../../rootloader/ttree.py#L280)
+[Show source in ttree.py:375](../../rootloader/ttree.py#L375)
 
-Return pandas dataframe of the data
+Return ttree data as pandas dataframe
 
 #### Signature
 
@@ -285,10 +321,25 @@ def to_dataframe(self): ...
 
 ### ttree.to_dict
 
-[Show source in ttree.py:307](../../rootloader/ttree.py#L307)
+[Show source in ttree.py:409](../../rootloader/ttree.py#L409)
+
+Return ttree data as dict of numpy arrays
 
 #### Signature
 
 ```python
 def to_dict(self): ...
+```
+
+### ttree.values
+
+[Show source in ttree.py:444](../../rootloader/ttree.py#L444)
+
+Convert ttree 1D or 2D numpy array (depending on number of columns)
+
+#### Signature
+
+```python
+@property
+def values(self): ...
 ```

{rootloader-1.2.7 → rootloader-1.3.1}/rootloader/ttree.py

@@ -11,94 +11,13 @@ import os
 import ROOT
 ROOT.EnableImplicitMT()
 
-# pre-compile stats functions to avoid memory creep during JIT compilations
-
-cpp_code = """
-Double_t RDF_min_d(ROOT::RDataFrame df, string col){
-    return df.Min<Double_t>(col).GetValue();
-}
-
-Double_t RDF_max_d(ROOT::RDataFrame df, string col){
-    return df.Max<Double_t>(col).GetValue();
-}
-
-Double_t RDF_mean_d(ROOT::RDataFrame df, string col){
-    return df.Mean<Double_t>(col).GetValue();
-}
-
-Double_t RDF_std_d(ROOT::RDataFrame df, string col){
-    return df.StdDev<Double_t>(col).GetValue();
-}
-
-Double_t RDF_sum_d(ROOT::RDataFrame df, string col){
-    return df.Sum<Double_t>(col).GetValue();
-}
-
-Int_t RDF_min_i(ROOT::RDataFrame df, string col){
-    return df.Min<Int_t>(col).GetValue();
-}
-
-Int_t RDF_max_i(ROOT::RDataFrame df, string col){
-    return df.Max<Int_t>(col).GetValue();
-}
-
-Int_t RDF_mean_i(ROOT::RDataFrame df, string col){
-    return df.Mean<Int_t>(col).GetValue();
-}
-
-Int_t RDF_std_i(ROOT::RDataFrame df, string col){
-    return df.StdDev<Int_t>(col).GetValue();
-}
-
-Int_t RDF_sum_i(ROOT::RDataFrame df, string col){
-    return df.Sum<Int_t>(col).GetValue();
-}
-
-/////////////////////////////////////////////////////////////////
-
-Double_t RDF_min_d(ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void> df, string col){
-    return df.Min<Double_t>(col).GetValue();
-}
-
-Double_t RDF_max_d(ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void> df, string col){
-    return df.Max<Double_t>(col).GetValue();
-}
-
-Double_t RDF_mean_d(ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void> df, string col){
-    return df.Mean<Double_t>(col).GetValue();
+# template to pre-compile stats functions to avoid memory creep during JIT compilations
+# used in ttree._get_stat
+cpp_template = """
+{TYPE2} RDF_{FNNAME}_{TYPE2}({DTYPE} df, string col){
+    return df.{FNNAME}<{TYPE2}>(col).GetValue();
 }
-
-Double_t RDF_std_d(ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void> df, string col){
-    return df.StdDev<Double_t>(col).GetValue();
-}
-
-Double_t RDF_sum_d(ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void> df, string col){
-    return df.Sum<Double_t>(col).GetValue();
-}
-
-Int_t RDF_min_i(ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void> df, string col){
-    return df.Min<Int_t>(col).GetValue();
-}
-
-Int_t RDF_max_i(ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void> df, string col){
-    return df.Max<Int_t>(col).GetValue();
-}
-
-Int_t RDF_mean_i(ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void> df, string col){
-    return df.Mean<Int_t>(col).GetValue();
-}
-
-Int_t RDF_std_i(ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void> df, string col){
-    return df.StdDev<Int_t>(col).GetValue();
-}
-
-Int_t RDF_sum_i(ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void> df, string col){
-    return df.Sum<Int_t>(col).GetValue();
-}
-
-
 """
-ROOT.gInterpreter.Declare(cpp_code)
 
 class ttree(object):
     """Extract ROOT.TTree with lazy operation. Looks like a dataframe in most ways
@@ -361,8 +280,19 @@ class ttree(object):
             new.set_filter(expression, inplace=True)
             return new
 
+    def to_array(self):
+        """Return ttree data as 1D or 2D numpy array (depending on number of columns)"""
+        df = self.to_dict()
+        output = np.stack([df[k] for k in df.keys()])
+
+        # single array output for one column
+        if output.shape[0] == 1:
+            return output[0]
+        else:
+            return output
+
     def to_dataframe(self):
-        """Return pandas dataframe of the data"""
+        """Return ttree data as pandas dataframe"""
 
         df = self.to_dict()
 
@@ -396,7 +326,7 @@ class ttree(object):
         return df
 
     def to_dict(self):
-
+        """Return ttree data as dict of numpy arrays"""
         # ensure index is loaded
         if self._index not in self._columns and self._index is not None:
             columns = [*self._columns, self._index]
@@ -408,37 +338,51 @@ class ttree(object):
     # PROPERTIES ===========================
     @property
     def columns(self):
+        """Return list of column (branch) names"""
         return self._columns
     @property
     def filters(self):
+        """Return list of RDataFrame filters"""
         return self._filters
     @property
     def index(self):
+        """Return ttree of just the index data"""
         return self[self._index]
     @property
     def index_name(self):
+        """Return string of the name of the index branch"""
         return self._index
     @property
     def loc(self):
+        """Return a ttree that can be indexed like a pandas dataframe"""
         return _ttree_indexed(self)
     @property
     def size(self):
+        """Return the number of rows in the ttree"""
         return self._rdf.Count().GetValue()
-
+    @property
+    def values(self):
+        """Convert ttree 1D or 2D numpy array (depending on number of columns)"""
+        return self.to_array()
+    
     # STATS ================================
     def _getstat(self, fnname):
-
-        double_fn = getattr(ROOT, f'RDF_{fnname}_d')
-        int_fn = getattr(ROOT, f'RDF_{fnname}_i')
-
         vals = []
         for col in self._columns:
             dtype = self._rdf.GetColumnType(col)
 
-            if dtype == 'Double_t':
-                vals.append(double_fn(self._rdf, col))
-            elif dtype == 'Int_t':
-                vals.append(int_fn(self._rdf, col))
+            # pre-compile function 
+            fn_name = f'RDF_{fnname}_{dtype}'
+            if not hasattr(ROOT, fn_name):
+                for type1 in [r'ROOT::RDataFrame', 
+                              r'ROOT::RDF::RInterface<ROOT::Detail::RDF::RJittedFilter,void>']:
+                    cpp_code = cpp_template.replace('{DTYPE}', type1)
+                    cpp_code = cpp_code.replace('{TYPE2}', dtype)       
+                    cpp_code = cpp_code.replace('{FNNAME}', fnname)       
+                    ROOT.gInterpreter.Declare(cpp_code)
+                    
+            # do the computation
+            vals.append(getattr(ROOT, fn_name)(self._rdf, col))
 
         # for len = 1 outputs return only the value
         if len(vals) == 1:
@@ -446,11 +390,21 @@ class ttree(object):
         else:
             return pd.Series(vals, index=self._columns)
 
-    def min(self):  return self._getstat('min')
-    def max(self):  return self._getstat('max')
-    def mean(self): return self._getstat('mean')
-    def sum(self):  return self._getstat('sum')
-    def std(self):  return self._getstat('std')
+    def min(self):  
+        """Return the min value of the tree, for each branch"""
+        return self._getstat('Min')
+    def max(self):  
+        """Return the max value of the tree, for each branch"""
+        return self._getstat('Max')
+    def mean(self): 
+        """Return the mean value of the tree, for each branch"""
+        return self._getstat('Mean')
+    def sum(self):  
+        """Return the sum of the values of the tree, for each branch"""
+        return self._getstat('Sum')
+    def std(self):  
+        """Return the standard deviationif the of values the tree, for each branch"""
+        return self._getstat('StdDev')
 
 # ttree but slice on time
 class _ttree_indexed(object):

rootloader-1.3.1/rootloader/version.py

@@ -0,0 +1 @@
+__version__ = '1.3.1'

rootloader-1.2.7/rootloader/version.py

@@ -1 +0,0 @@
-__version__ = '1.2.7'