v0.2.21

mtbp3/__init__.py

@@ -9,3 +9,9 @@
 from importlib.metadata import version
 __version__ = version(__package__)
 
+
+import os
+
+_ROOT = os.path.abspath(os.path.dirname(__file__))
+def get_data(path):
+    return os.path.join(_ROOT, 'data', path)

mtbp3/health/ectd.py

@@ -15,16 +15,18 @@
 
 import pandas as pd
 import os
-from mtbp3.util import cdt, util
+import mtbp3
+from mtbp3 import util
 
 class ctoc_by_fda:
     def __init__(self, ectd_version="3.2.2", ctoc_version="2.3.3"):
         assert isinstance(ectd_version, str) and all(char.isdigit() or char == '.' for char in ectd_version), "Version must be a string with integers and dots"
         assert isinstance(ctoc_version, str) and all(char.isdigit() or char == '.' for char in ctoc_version), "Version must be a string with integers and dots"
         self.ectd_version = ectd_version
         self.ctoc_version = ctoc_version
-        self.folder_name = util.get_data(f'supp_ectd/fda_ectd{ectd_version}_ctocv{ctoc_version}.txt')
+        self.folder_name = mtbp3.get_data(f'supp_ectd/fda_ectd{ectd_version}_ctocv{ctoc_version}.txt')
         self.ctoc = self.__load_list()
+        print(self.folder_name)
 
     def __load_list(self):
         #file_path = f'./mtbp3/data/supp_ectd/fda_ectd{self.ectd_version}_ctocv{ctoc_version}.txt'
@@ -43,7 +45,7 @@ def show_ctoc_tree(self, module=None, to_right=False):
             module = 1
 
         filtered_ctoc = [item for item in self.ctoc if item.startswith(str(module))]
-        tree = cdt.ListTree(lst=filtered_ctoc, infmt='dotspace')
+        tree = mtbp3.util.cdt.ListTree(lst=filtered_ctoc, infmt='dotspace')
         return tree.list_tree(to_right=to_right)
 
     @staticmethod
@@ -109,9 +111,9 @@ def find_section_given_words(self, words, outfmt='simple', include='up', to_righ
                         out_colored.append(f"{first_part} {colored_second_part}")
                     else:
                         out_colored.append(row)
-                out_tree = cdt.ListTree(lst=out_colored, infmt='dotspace')
+                out_tree = mtbp3.util.cdt.ListTree(lst=out_colored, infmt='dotspace')
             else:
-                out_tree = cdt.ListTree(lst=out, infmt='dotspace')
+                out_tree = mtbp3.util.cdt.ListTree(lst=out, infmt='dotspace')
             return out_tree.list_tree(to_right=to_right)
         else:
             raise ValueError("Invalid value for outfmt. Supported values are 'simple' and 'tree'.")

mtbp3/health/emt.py

@@ -19,7 +19,8 @@
 import re
 import numpy as np
 import pandas as pd
-from mtbp3.util import util, lsr, cdt 
+import mtbp3
+from mtbp3 import util
 
 class Emt:
     """A class representing MedDRA terms.
@@ -46,7 +47,7 @@ def __init__(self, folder_name=''):
             self.folder_name = folder_name
             self.demo = False
         else:
-            self.folder_name = util.get_data('test_emt/MedDRA')
+            self.folder_name = mtbp3.get_data('test_emt/MedDRA')
             self.demo = True
 
         self.version_number = "00.0"
@@ -126,7 +127,7 @@ def find_files(self):
         Returns:
             list: A list of missing file names.
         """
-        lsrt = lsr.LsrTree(self.folder_name, outfmt="list")
+        lsrt = util.lsr.LsrTree(self.folder_name, outfmt="list")
         lsr_files = lsrt.list_files()
         support_doc_files, med_ascii_files, seq_ascii_files = self.expected_file_lists()
 
@@ -160,7 +161,7 @@ def list_files(self):
         Returns:
             list: A list of file names.
         """
-        lsrt = lsr.LsrTree(self.folder_name, outfmt="tree", with_counts=True)
+        lsrt = util.lsr.LsrTree(self.folder_name, outfmt="tree", with_counts=True)
         lsr_files = lsrt.list_files()
         return lsr_files
 
@@ -624,7 +625,7 @@ def find_terms_given_smq_sub(self, subset_df, keep_columns, llt_only, llt_curren
     def load_fmq_default(self):
         if self.fmq_list_default is None:
             try:
-                tmp = pd.read_csv(os.path.join(util.get_data('test_emt/FMQ'), "FMQ_Consolidated_List.csv"), delimiter=',', header=0)
+                tmp = pd.read_csv(os.path.join(mtbp3.get_data('test_emt/FMQ'), "FMQ_Consolidated_List.csv"), delimiter=',', header=0)
                 tmp = tmp.iloc[:, :-1]
                 tmp.columns = ['fmq', 'pt', 'fmq_pt', 'classification']
                 self.fmq_list_default = tmp
@@ -758,7 +759,7 @@ def show_fmq_tree(self, fmq=[], with_soc=False, ignore_case=False, to_right=Fals
             list1 = pt_df['fmq_class'].unique().tolist()
             lists = pt_df['fmq_class_soc'].unique().tolist()
             list2 = pt_df['fmq_class_soc_pt'].unique().tolist()
-            tree = lsr.ListTree(lst = ['FMQ/']+list0+list1+lists+list2)
+            tree = util.lsr.ListTree(lst = ['FMQ/']+list0+list1+lists+list2)
             return tree.list_tree(to_right=to_right)
         else:
             pt_df = pt_df.sort_values(by=['fmq', 'classification'])
@@ -775,7 +776,7 @@ def show_fmq_tree(self, fmq=[], with_soc=False, ignore_case=False, to_right=Fals
 
             list1 = pt_df['fmq_class'].unique().tolist()
             list2 = pt_df['fmq_class_pt'].unique().tolist()
-            tree = lsr.ListTree(lst = ['FMQ/']+list0+list1+list2)
+            tree = util.lsr.ListTree(lst = ['FMQ/']+list0+list1+list2)
             return tree.list_tree(to_right=to_right)
 
 if __name__ == "__main__":

mtbp3/util/__init__.py

@@ -1,4 +1,9 @@
 from .lsr import *
 from .cdt import *
 from .cdtg import *
-from .util import *
+
+import os
+
+_ROOT = os.path.abspath(os.path.dirname(__file__))
+def get_data(path):
+    return os.path.join(_ROOT, '../data', path)

mtbp3/util/cdt.py

@@ -172,7 +172,7 @@ def __init__(self, lst=[], label=[], infmt='path'):
         self.infmt = infmt
         self.df = pd.DataFrame()
         self.prelst = pd.DataFrame()
-        self.out = pd.DataFrame()
+        self.tree = pd.DataFrame()
 
     def __list_tree_df(self):
         if not isinstance(self.lst, list):
@@ -290,20 +290,55 @@ def __list_tree_pre(self, to_right=False):
         self.prelst = prelst
 
     def list_tree(self, to_right=False):
+        """
+        Returns a DataFrame representing the tree structure of the object.
+
+        Parameters:
+        - to_right (bool): If True, aligns the tree structure to the right by padding with spaces.
+
+        Returns:
+        - tree (DataFrame): DataFrame representing the tree structure.
+        """
         self.__list_tree_pre(to_right=to_right)
 
         if self.prelst.empty:
-            self.out = pd.DataFrame()
-            return self.out
+            self.tree = pd.DataFrame()
+            return self.tree
 
         out_joined = self.prelst.apply(lambda row: ''.join(row), axis=1)
 
         if to_right:
             max_length = out_joined.str.len().max()
             out_joined = out_joined.apply(lambda x: x.rjust(max_length))
 
-        self.out = out_joined
-        return self.out
+        self.tree = out_joined
+        return self.tree
+
+    def list_tree_with_keyword(self, keywords, neighbor = 0, outfmt='simple'):
+        """
+        Search for keywords in the list of items.
+
+        Args:
+            keywords (list): List of keywords to search for.
+
+        Returns:
+            list: List of items that contain the keywords.
+
+        """
+        if not isinstance(keywords, list):
+            raise ValueError("keywords should be a list.")
+
+        if not self.lst:
+            return []
+
+        if outfmt == 'simple':
+            result = [item for item in self.lst if any(keyword in item for keyword in keywords)]
+        elif outfmt == 'subtree':
+            for item in result:
+                if any(keyword in item for keyword in keywords):
+                    result.append(item)
+        else: 
+            return []
 
 if __name__ == "__main__":
     pass

mtbp3/util/cdtg.py

@@ -77,6 +77,7 @@ def __init__(self, df, y_col, group_col=None, grid_col=None, grid_wrap=None, x_c
             self.x_order = []
         if grid_kws is None:
             grid_kws = {}
+
         self.grid_kws = grid_kws
         sns.set_style("ticks", {'axes.grid': True})
 

mtbp3/util/util.py

@@ -17,7 +17,7 @@
 
 _ROOT = os.path.abspath(os.path.dirname(__file__))
 def get_data(path):
-    return os.path.join(_ROOT, 'data', path)
+    return os.path.join(_ROOT, '../data', path)
 
 if __name__ == "__main__":
     pass

tests/test_health_emt.py

@@ -3,12 +3,12 @@
 
 import unittest
 import pandas as pd
-from mtbp3.health.emt import Emt
+from mtbp3.health import emt
 
 class TestEmt(unittest.TestCase):
 
     def setUp(self):
-        self.emt = Emt()
+        self.emt = emt.Emt()
         self.emt.find_files()
 
     def test_expected_file_lists(self):