diff --git a/src/__pycache__/search.cpython-310.pyc b/src/__pycache__/search.cpython-310.pyc new file mode 100644 index 0000000..c740282 Binary files /dev/null and b/src/__pycache__/search.cpython-310.pyc differ diff --git a/src/index.py b/src/index.py index 781b9b6..83d4342 100755 --- a/src/index.py +++ b/src/index.py @@ -23,9 +23,17 @@ def build_index(): if not word in ignored_words: if not word in dictionary: dictionary[word] = [] - if not url.strip() in dictionary[word]: - dictionary[word].append(url.strip()) - + matching_urls = list(filter(lambda entry: entry["url"] == url.strip(), dictionary[word])) + if len(matching_urls) == 0: +# if not url.strip() in dictionary[word]: + entries = dictionary[word] + entry = {"url": url.strip(), "count": 1} + dictionary[word].append(entry) + else: + entries = dictionary[word] + entry = matching_urls[0] + entry["count"] += 1 + entries.sort(reverse=True, key=lambda entry: entry["count"]) index.write(json.dumps(dictionary)) if __name__ == "__main__": diff --git a/src/search.py b/src/search.py index 1a71241..e5389f8 100755 --- a/src/search.py +++ b/src/search.py @@ -18,6 +18,7 @@ def search(query): q = q.lower() if q in index: result.append(index[q]) +# result.sort(reverse= True,key=lambda entry: int(entry.count)) print(result) return result