From fa54eafa4d37f5af0c7e57e6fec2015585c6a238 Mon Sep 17 00:00:00 2001
From: am9zZWY <46693545+am9zZWY@users.noreply.github.com>
Date: Thu, 11 Jul 2024 17:14:33 +0200
Subject: [PATCH] Add position of word in inverse index

---
 engine/custom_db.py | 30 ++++++++++++++++++++++++++----
 1 file changed, 26 insertions(+), 4 deletions(-)

diff --git a/engine/custom_db.py b/engine/custom_db.py
index 93cfc70..ec22b03 100644
--- a/engine/custom_db.py
+++ b/engine/custom_db.py
@@ -26,8 +26,32 @@ def save_pages():
 def add_document_to_index(doc_id, words: list[str]):
     print(f"Adding stuff")
     global inverted_index
-    for word in set(words):
-        inverted_index[word].append(doc_id)
+    for word_index, word in enumerate(words):
+        position = word_index + 1
+        # If the word is not in the invertex index, simply add it
+        if word not in inverted_index:
+            print(f"Adding new word {word}")
+            inverted_index[word] = [[doc_id, [position]]]
+        else:
+            # There are two cases.
+            # The word is in the inverted index:
+            # - but the document is not in the list
+            # - and the document is in the list but the position is not
+
+            # Check if the document is in the list
+            doc_ids = [doc[0] for doc in inverted_index[word]]
+
+            if doc_id not in doc_ids:
+                print(f"Adding new document {doc_id} to word {word}")
+                inverted_index[word].append([doc_id, [position]])
+            else:
+                # The document is in the list, but the position is not
+                # Get the index of the document
+                doc_index = doc_ids.index(doc_id)
+                # Check if the position is in the list
+                if position not in inverted_index[word][doc_index][1]:
+                    print(f"Adding new position {position} to word {word} in document {doc_id}")
+                    inverted_index[word][doc_index][1].append(position)
 
 
 def index_pages():
@@ -42,5 +66,3 @@ def access_index():
 
 
 # Convert the inverted index to a DataFrame
-
-