pending commit oll changes

Author: Jesús Cid Sueiro

.DS_Store

@@ -348,7 +348,6 @@ def run_labeler(project_path, url, transfer_mode, user, export_labels,
 
     # Load data from the standard datasets.
     df_labels, df_preds, labelhistory = data_mgr.loadData()
-    breakpoint()
 
     # Load new labels and predictions from the input folder
     log.info("-- Loading new data from the input folder")

labelfactory/.DS_Store

@@ -14,8 +14,6 @@
 else:
     import Tkinter as tk
 
-import ipdb
-
 
 class LabelGUIController(object):
 
@@ -30,9 +28,9 @@ class LabelGUIController(object):
     def __init__(self, newurls, newwids, newqueries, preds, labels, urls,
                  categories, alphabet, datatype='url', cat_model='single',
                  parent_cat={}, text2label=None):
-
-        """ This method initialize the sampler object. As part of this process
-        it creates the AL objects required for the sample generation.
+        """
+        This method initialize the sampler object. As part of this process it
+        creates the AL objects required for the sample generation.
 
         :Attributes:
             :newurls:    A list of urls to label
@@ -71,15 +69,14 @@ def __init__(self, newurls, newwids, newqueries, preds, labels, urls,
         self.text2label = text2label
 
     def takeandshow_sample(self):
+        """
+        Gets next sample id from the list and visualize the sample.
+        The type of visualization depends of the type of sample:
+            if sample id is a url, a browser is opened.
+            if sample id is not a url, data is printed.
 
-        """ Gets next sample id from the list and visualize the sample.
-            The type of visualization depends of the type of sample:
-                if sample id is a url, a browser is opened.
-                if sample id is not a url, data is printed.
-
-            Note that the sample identifiers are stored in variable self.url
-            for historical reasons. This variable does not necessarily stores
-            urls.
+        Note that the sample identifiers are stored in variable self.url for
+        historical reasons. This variable does not necessarily stores urls.
         """
 
         self.url = None

labelfactory/labelfactory.py

@@ -65,7 +65,6 @@ class BaseDM(object):
     def __init__(self, source_type, dest_type, file_info, db_info,
                  categories, parentcat, ref_class, alphabet,
                  compute_wid='yes', unknown_pred=0):
-
         """
         Stores files, folder and path names into the data struture of the
         DataManager object.
@@ -290,17 +289,17 @@ def df2data(self, df_labels, df_preds):
         return data
 
     def importData(self):
+        """
+        Read data from the input folder.
+        Only labels with positive or negative label are loaded
 
-        """ Read data from the input folder.
-            Only labels with positive or negative label are loaded
-
-            :Args:
-                :-: None. File locations and the set of categories are taken
-                          from the class attributes
+        :Args:
+            :-: None. File locations and the set of categories are taken
+                      from the class attributes
 
-            :Returns:
-                :df_labels: Pandas dataframe of labels
-                :df_preds: Pandas dataframe of predictions
+        :Returns:
+            :df_labels: Pandas dataframe of labels
+            :df_preds: Pandas dataframe of predictions
         """
 
         # Warning.
@@ -373,8 +372,11 @@ def importData(self):
                            'category {2}       \r').format(nk, ntot, cat),
                           end="")
 
-                urls_dict[wid] = preds[cat][wid]['url']
-                pred_dict[wid] = preds[cat][wid]['pred']
+                if preds[cat][wid] is not None:
+                    urls_dict[wid] = preds[cat][wid]['url']
+                    pred_dict[wid] = preds[cat][wid]['pred']
+                else:
+                    print(f"{wid}")
 
             df2_preds['url'].update(pd.Series(urls_dict))
             df2_preds[cat].update(pd.Series(pred_dict))
@@ -434,7 +436,7 @@ def importData(self):
         # Make sure taht any changes here are also done there
         # (I know, this is not a good programming style..)
         info = ['marker', 'relabel', 'weight', 'userId', 'date']
-        arrays = [len(info)*['info'] + len(self.categories)*['label'],
+        arrays = [len(info) * ['info'] + len(self.categories) * ['label'],
                   info + self.categories]
         tuples = list(zip(*arrays))
         mindex = pd.MultiIndex.from_tuples(tuples)
@@ -508,21 +510,20 @@ def importLabels(self, category):
         return labels
 
     def importPredicts(self, category=None):
+        """
+        Get dictionary of predictions relative to a given category
 
-        """ Get dictionary of predictions relative to a given category
-
-            :Args:
-                :category: The category to load (from a pkl file)
-                           If None, all categories are read from a unique
-                           csv file
+        :Args:
+            :category: The category to load (from a pkl file)
+                       If None, all categories are read from a unique csv file
 
-            :Returns:
-                :preds: Dictionary of predictions
-                        - If category is not None, preds[wid] has the
-                          prediction for url wid about the given category.
-                        - If category is not None, preds is a dataframe
-                          with the wid as uid column and one column with
-                          predictions for each category.
+        :Returns:
+            :preds: Dictionary of predictions
+                    - If category is not None, preds[wid] has the prediction
+                      for url wid about the given category.
+                    - If category is not None, preds is a dataframe with the
+                      wid as uid column and one column with predictions for
+                      each category.
         """
 
         # The default category is the reference class used by the

labelfactory/labeling/LabelGUIController.py

@@ -192,6 +192,9 @@ def get_urls_batch(self, max_urls=10,):
         print([newwids[k] for k in range(n) if relabels[k] == 1 and
                markers[k] == 1])
 
+        print(f"Predictions for the reference class, {self.ref_class}:")
+        print({w: self.preds[self.ref_class][w] for w in newwids})
+
         return newurls, newwids, newqueries
 
     def get_single_url(self, target_wid):
@@ -212,6 +215,7 @@ def get_single_url(self, target_wid):
         # Make sure that the target url has been previously labeled.
         # The code is not ready to label urls without a previous label,
         # because that would requiere to change the active learning weights
+
         if target_wid not in self.markers:
             sys.exit("The target url has no label. Labeling of a specific " +
                      "url should be done to revise existing labels only")
@@ -225,6 +229,8 @@ def get_single_url(self, target_wid):
 
         # Print the four types of webs to label
         print("Relabeling {0} ".format(newurls[0]))
+        print(f"Predictions for the reference class, {self.ref_class}:")
+        print({w: self.preds[self.ref_class][w] for w in newqueries})
 
         return newurls, newwids, newqueries