From 4a5cd05bac1c715afe19de36a9963ff457f5aef0 Mon Sep 17 00:00:00 2001
From: dmt <>
Date: Thu, 3 Oct 2019 19:03:56 +0200
Subject: [PATCH] Define wrapper around pandas that represents a block of a
dataframe with addtional methods for finding duplicate value pairs getting
all these duplicates.
---
cml/ports/source_adapters.py | 46 ++++++++++++++++++++++++++++++++++++
1 file changed, 46 insertions(+)
diff --git a/cml/ports/source_adapters.py b/cml/ports/source_adapters.py
index 7b33f85..073601f 100644
--- a/cml/ports/source_adapters.py
+++ b/cml/ports/source_adapters.py
@@ -32,6 +32,52 @@ class Adapter(ABC):
@abstractmethod
def get_column_values_as_list(self, *args, **kwargs): pass
+ @abstractmethod
+ def get_block(self, *args, **kwargs): pass
+
+
+# TODO (dmt): Provide common base class or pandas operations.
+
+class PandasBlock:
+ def __init__(self, data_block):
+ self.__data_block = data_block
+
+ def __str__(self):
+ return str(self.__data_block)
+
+ def get_duplicated_pairs(self, *args):
+ bool_series = self.__data_block.duplicated(subset=[args[0], args[1]])
+
+ duplicates = self.__data_block[bool_series]
+ for i, j in zip(duplicates[args[0]], duplicates[args[1]]):
+ yield i, j
+
+ def get_values(self, **kwargs):
+ t, z, sigma = kwargs.get("T"), kwargs.get("Z"), kwargs.get("Sigma")
+ if t and z:
+ data_frame = self.__data_block.loc[
+ (self.__data_block["T"] == t) & (self.__data_block["Z"] == z)]
+
+ elif t and sigma:
+ data_frame = self.__data_block.loc[
+ (self.__data_block["T"] == t) & (
+ self.__data_block["Sigma"] == sigma)]
+
+ elif z and sigma:
+ data_frame = self.__data_block.loc[
+ (self.__data_block["Z"] == z) & (
+ self.__data_block["Sigma"] == sigma)]
+
+ else:
+ # TODO (dmt): Write proper error handling.
+ raise Exception()
+
+ return PandasBlock(data_frame)
+
+ @property
+ def length(self):
+ return len(self.__data_block)
+
class PandasAdapter:
def __init__(self, data_frame):
--
GitLab