modified

Author: ayushianan

Storage.py

@@ -1,6 +1,8 @@
 import sys, os
 projectpath = os.path.dirname(os.path.realpath('Storage.py'))
+#directory path
 libpath = projectpath + '/lib'
+#lib path
 sys.path.append(libpath)
 os.chdir(projectpath)
 import parsing
@@ -12,8 +14,10 @@
 index = {}
 # What collection to index?
 collection = 'New Testament'
+#mongo folder
 # Indicate the path where relative to the collection
 os.chdir(projectpath + '/data/' + collection)
+#added to project path
 # List all files in the collection
 files = [file for file in os.listdir('.') if os.path.isfile(file)]
 # Iterate through every file
@@ -23,12 +27,13 @@
     # Normalize the content
     words = parsing.clean(data)
     # Remove the extension from the file for storage
+    #start of folder
     name = re.match('(^[^.]*)', file).group(0)
     # Add the words to the index
     parsing.index(name, words, index)
 print("Indexation took " + str(time.time() - startTime) + " seconds.")
 
 # Storage
 startTime = time.time()       
-parsing.store(index, collection)
+parsing.store(index, collection)#in mongo
 print("Storage took " + str(time.time() - startTime) + " seconds.")

cd

@@ -0,0 +1,2 @@
+python3 Storage.py
+sudo python3 main.py

lib/__pycache__/browser.cpython-36.pyc

@@ -1,11 +1,4 @@
-# -*- coding: utf-8 -*-
-
-# Form implementation generated from reading ui file 'browser.ui'
-#
-# Created: Wed Feb  4 12:43:32 2015
-#      by: PyQt4 UI code generator 4.10.4
-#
-# WARNING! All changes made in this file will be lost!
+
 
 from PyQt4 import QtCore, QtGui
 

lib/__pycache__/parsing.cpython-36.pyc

@@ -26,6 +26,7 @@ def clean(data):
 def index(file, words, index):
     for position in range(len(words)):
         word = words[position]     
+        #each word has frequency and has document and in tht document frequency and position
         # If the word is not in the index
         if words[position] not in index:
             index[word] = {'term frequency' : 1,
@@ -52,6 +53,6 @@ def index(file, words, index):
 def store(index, folder):
     collection = db[folder]
     for word in index:
-        collection.save({'_id' : word, 'info' : index[word]})
+        collection.save({'_id' : word, 'info' : index[word]})#storing in mongo db
 
 

lib/__pycache__/querying.cpython-36.pyc

@@ -15,6 +15,7 @@
 folder = 'New Testament'
 collection = db[folder]
 
+
 class browser(QtGui.QMainWindow):
     def __init__(self, parent = None):
         QtGui.QWidget.__init__(self, parent)
@@ -33,8 +34,13 @@ def query(self):
             index[word] = collection.find({'_id' : word})[0]['info']
         # Rank the documents according to the query
         results = rankDocuments(index, words)
+        i=0
         for result in results:
-            self.ui.listWidget.addItem(result[0]+' : '+str(round(result[1], 2)))
+            if(i<10):
+                self.ui.listWidget.addItem(result[0]+' : '+str(round(result[1], 2)))
+            i=i+1
+	    
+	    
 
 if __name__ == "__main__":
     app = QtGui.QApplication(sys.argv)