Revisions to Zim parsing, netflix parsing, and updates to html scraping to include synopsis

.gitignore

@@ -218,3 +218,5 @@ __marimo__/
 
 # Data Folder
 *.tsv
+
+data/

.idea/.gitignore

@@ -0,0 +1,8 @@
+# Default ignored files
+/shelf/
+/workspace.xml
+# Editor-based HTTP Client requests
+/httpRequests/
+# Datasource local storage ignored files
+/dataSources/
+/dataSources.local.xml

.idea/datamining_881_new.iml

@@ -0,0 +1,11 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<module type="PYTHON_MODULE" version="4">
+  <component name="NewModuleRootManager">
+    <content url="file://$MODULE_DIR$">
+      <excludeFolder url="file://$MODULE_DIR$/.venv" />
+      <excludeFolder url="file://$MODULE_DIR$/data/processed/wikipedia_html" />
+    </content>
+    <orderEntry type="jdk" jdkName="Python 3.13 (datamining_881_new)" jdkType="Python SDK" />
+    <orderEntry type="sourceFolder" forTests="false" />
+  </component>
+</module>

.idea/inspectionProfiles/Project_Default.xml

@@ -0,0 +1,13 @@
+<component name="InspectionProjectProfileManager">
+  <profile version="1.0">
+    <option name="myName" value="Project Default" />
+    <inspection_tool class="Eslint" enabled="true" level="WARNING" enabled_by_default="true" />
+    <inspection_tool class="PyUnresolvedReferencesInspection" enabled="true" level="WARNING" enabled_by_default="true">
+      <option name="ignoredIdentifiers">
+        <list>
+          <option value="dict.*" />
+        </list>
+      </option>
+    </inspection_tool>
+  </profile>
+</component>

.idea/inspectionProfiles/profiles_settings.xml

@@ -0,0 +1,6 @@
+<component name="InspectionProjectProfileManager">
+  <settings>
+    <option name="USE_PROJECT_PROFILE" value="false" />
+    <version value="1.0" />
+  </settings>
+</component>

.idea/misc.xml

@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="Black">
+    <option name="sdkName" value="Python 3.13 (datamining_881_new)" />
+  </component>
+  <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.13 (datamining_881_new)" project-jdk-type="Python SDK" />
+</project>

.idea/modules.xml

@@ -0,0 +1,8 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="ProjectModuleManager">
+    <modules>
+      <module fileurl="file://$PROJECT_DIR$/.idea/datamining_881_new.iml" filepath="$PROJECT_DIR$/.idea/datamining_881_new.iml" />
+    </modules>
+  </component>
+</project>

.idea/vcs.xml

@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="VcsDirectoryMappings">
+    <mapping directory="" vcs="Git" />
+    <mapping directory="$PROJECT_DIR$" vcs="Git" />
+  </component>
+</project>

scripts/dataset_create.py

@@ -2,45 +2,42 @@ import pandas as pd
 import os
 from scrape import extract_movie_info
 
-script_dir = os.path.dirname(os.path.abspath(__file__))
-# file_path = os.path.join(script_dir, "..", "sample_data.xlsx")
-# movie_data = pd.read_excel(file_path)
-# print(movie_data.columns)
-
 BASE_DIR = os.path.dirname(os.path.abspath(__file__))
-INPUT_DIR = r'C:\Users\Prabhaav\Projects\PyCharm\datamining_881\data\processed\wikipedia_html'
+INPUT_DIR = os.path.join(BASE_DIR, "../data/processed/wikipedia_html_test/")
 SPREADSHEET_DIR = os.path.join(BASE_DIR, "../data/processed/spreadsheets/")
 
-movie_data = pd.DataFrame(columns=['Title', 'Director', 'Cast', 'Genre', 'Plot', 'Release Date', 'Slug', 'Poster Filename'])
+rows = []
 
 for folder in os.listdir(INPUT_DIR):
     path = os.path.join(INPUT_DIR, folder)
-    script_dir = os.path.join(path, next((f for f in os.listdir(path) if f.endswith(".html")), None))
+    script_dir = next((f for f in os.listdir(path) if f.endswith(".html")), None)
     if not script_dir:
         continue
+    full_path = os.path.join(path, script_dir)
+    slug = os.path.splitext(script_dir)[0]
     try:
-        print(script_dir)
+        print(full_path)
-        title, directed_by, cast, genre, plot, year, poster_filename = extract_movie_info(script_dir)
+        title, directed_by, cast, genre, plot, year, poster_filename = extract_movie_info(full_path)
-        new_row = {
+        rows.append({
             "Title": title,
             "Director": directed_by,
             "Cast": ", ".join(cast),
             "Genre": genre,
             "Plot": plot,
             "Release Date": year,
-            "Slug": script_dir,
+            "Slug": slug,
             "Poster Filename": poster_filename
-        }
+        })
-        movie_data.loc[len(movie_data)] = new_row
 
-    except Exception as e:
-        print("error:", e)
     except KeyboardInterrupt:
-        output_path = os.path.join(SPREADSHEET_DIR, "updated_data.xlsx")
+        movie_data = pd.DataFrame(rows)
-        print(output_path)
+        output_path = os.path.join(SPREADSHEET_DIR, "updated_datav_test.xlsx")
         movie_data.to_excel(output_path, index=False)
         quit()
+    except Exception as e:
+        print("error:", e)
 
-output_path = os.path.join(SPREADSHEET_DIR, "updated_data.xlsx")
+movie_data = pd.DataFrame(rows)
+output_path = os.path.join(SPREADSHEET_DIR, "updated_data_test.xlsx")
 print(output_path)
 movie_data.to_excel(output_path, index=False)

scripts/extract_wiki_zim.py

@@ -8,8 +8,8 @@ import csv
 from slugify import slugify
 
 BASE_DIR = os.path.dirname(os.path.abspath(__file__))
-INPUT_TSV = os.path.abspath(os.path.join(BASE_DIR, "../data/raw/imdb_datasets/title.basics.tsv"))
+INPUT_TSV = os.path.abspath(os.path.join(BASE_DIR, "../data/raw/imdb_datasets/title.basics.test.tsv"))
-OUTPUT_DIR = os.path.abspath(os.path.join(BASE_DIR, "../data/processed/wikipedia_html"))
+OUTPUT_DIR = os.path.abspath(os.path.join(BASE_DIR, "../data/processed/wikipedia_html_test"))
 ZIM_PATH = os.path.abspath(os.path.join(BASE_DIR, "../data/raw/wikipedia/wikipedia_en_all_maxi_2025-08.zim"))
 
 os.makedirs(OUTPUT_DIR, exist_ok=True)
@@ -21,38 +21,79 @@ print("The Zim file is now opened")
 def sanitize_slug(slug):
     return slugify(slug, separator="_", max_length=200) or "_unknown"
 
+
+def is_movie_page(html_content, primary_title, original_title, year):
+    soup = BeautifulSoup(html_content, "html.parser")
+    page_title = soup.find("h1", {"id": "firstHeading"})
+    if not page_title:
+        return False
+    page_title_text = page_title.get_text().lower()
+    if primary_title.lower() not in page_title_text and original_title.lower() not in page_title_text:
+        return False
+    infobox = soup.find("table", {"class": "infobox"})
+    if not infobox:
+        return False
+    infobox_text = infobox.get_text()
+    if "Directed by" not in infobox_text or ("Produced by" not in infobox_text and "Written by" not in infobox_text):
+        return False
+    # Also verify the year appears in the infobox
+    if year and year != "\\N" and year not in infobox_text:
+        return False
+    return True
+
+
 # Fetch the html AND the images and put them in a folder
-def fetch_wikipedia_html_with_images(query, save_dir):
+def fetch_wikipedia_html_with_images(query, save_dir, primary_title, original_title, year):
     q = Query().set_query(query)
     search = searcher.search(q)
     if search.getEstimatedMatches() == 0:
         return None
     results = list(search.getResults(0, 5))
-    best_path = results[0]
+
+    for best_path in results:
         try:
             entry = zim.get_entry_by_path(best_path)
             item = entry.get_item()
             html_content = bytes(item.content).decode("UTF-8")
         except Exception:
-        return None
-    soup = BeautifulSoup(html_content, "html.parser")
-    for img in soup.find_all("img"):
-        src = img.get("src")
-        if not src:
             continue
-        img_path = src.lstrip("/")
+
+        if not is_movie_page(html_content, primary_title, original_title, year):
+            continue
+
+        soup = BeautifulSoup(html_content, "html.parser")
+        poster_img = None
+        infobox = soup.find("table", class_="infobox")
+        if infobox:
+            poster_img = infobox.select_one("img")
+            if poster_img and poster_img.get("src"):
+                img_path = poster_img["src"].lstrip("/")
                 try:
                     img_entry = zim.get_entry_by_path(img_path)
                     img_bytes = bytes(img_entry.get_item().content)
-        except Exception:
-            continue
                     img_name = os.path.basename(img_path)
-        img_file_path = os.path.join(save_dir, img_name)
+                    with open(os.path.join(save_dir, img_name), "wb") as f:
-        with open(img_file_path, "wb") as f:
                         f.write(img_bytes)
-        img["src"] = img_name
+                    poster_img["src"] = img_name
+                except Exception:
+                    pass
+        for img in soup.find_all("img"):
+            if img is not poster_img:
+                img["src"] = ""
         return str(soup), best_path
 
+    return None
+
+
+done_set = {
+    fname[:-5]
+    for d in os.listdir(OUTPUT_DIR)
+    if not d.startswith("_tmp_")
+    for fname in os.listdir(os.path.join(OUTPUT_DIR, d))
+    if fname.endswith(".html")
+}
+print(f"Found {len(done_set)} already processed")
+
 # Go through each row of the tsv file and try to get the movie on wiki
 with open(INPUT_TSV, encoding="utf-8") as f:
     reader = csv.DictReader(f, delimiter="\t")
@@ -64,12 +105,7 @@ with open(INPUT_TSV, encoding="utf-8") as f:
         if year is None or titleType != "movie":
             print("Skipping from TSV: ", title)
             continue
-        already_done = False
+        if tconst in done_set:
-        for d in os.listdir(OUTPUT_DIR):
-            if os.path.exists(os.path.join(OUTPUT_DIR, d, f"{tconst}.html")):
-                already_done = True
-                break
-        if already_done:
             print(f"Skipping already processed: {tconst}")
             continue
             # folder for each movie
@@ -77,7 +113,7 @@ with open(INPUT_TSV, encoding="utf-8") as f:
         os.makedirs(movie_dir, exist_ok=True)
         query = f"{title} ({year} film)" if year != "\\N" else title  # if year not empty
         print(f"fetching Wikipedia HTML + images for {tconst}: {query}")
-        result = fetch_wikipedia_html_with_images(query, movie_dir)
+        result = fetch_wikipedia_html_with_images(query, movie_dir, title, row["originalTitle"], row["startYear"])
         if result is None:
             print("Wikipedia fetch failed")
             shutil.rmtree(movie_dir, ignore_errors=True)
@@ -86,9 +122,6 @@ with open(INPUT_TSV, encoding="utf-8") as f:
             html_with_images, slug = result
         slug_dir = os.path.join(OUTPUT_DIR, sanitize_slug(slug))
         if html_with_images:
-            if "Directed by" not in html_with_images:
-                shutil.rmtree(movie_dir, ignore_errors=True)
-                continue
             if os.path.exists(slug_dir):
                 shutil.rmtree(movie_dir, ignore_errors=True)
             else:
@@ -98,6 +131,7 @@ with open(INPUT_TSV, encoding="utf-8") as f:
                 continue
             with open(outfile, "w", encoding="utf-8") as out:
                 out.write(html_with_images)
+                done_set.add(tconst)
         else:
             shutil.rmtree(movie_dir, ignore_errors=True)
             print(f"no Wikipedia page found for {query}")

scripts/fuse_with_netflix.py

@@ -0,0 +1,91 @@
+import pandas as pd
+import os
+from rapidfuzz import fuzz
+
+BASE_DIR = os.path.dirname(os.path.abspath(__file__))
+NETFLIX_DIR = os.path.join(BASE_DIR, "../data/raw/netflix/")
+MOVIE_EXCEL = os.path.join(BASE_DIR, "../data/processed/spreadsheets/updated_data_test.xlsx")
+MOVIE_TITLES = os.path.join(NETFLIX_DIR, "movie_titles.csv")
+COMBINED_FILES = [os.path.join(NETFLIX_DIR, f"combined_data_{i}.txt") for i in range(1, 5)]
+OUTPUT = os.path.join(BASE_DIR, "../data/processed/spreadsheets/fused_gtruth_test.csv")
+
+TITLE_THRESHOLD = 85   # fuzzy search
+
+main_data = pd.read_excel(MOVIE_EXCEL)
+main_data["title_lower"] = main_data["Title"].str.lower().str.strip()
+main_data["director_lower"] = main_data["Director"].fillna("").str.lower().str.strip()
+
+
+records = []
+with open(MOVIE_TITLES, encoding="latin-1") as f:
+    for line in f:
+        line = line.strip()
+        parts = line.split(",", 2)
+        if len(parts) == 3:
+            records.append({"netflix_id": int(parts[0]), "year": parts[1], "title": parts[2].strip()})
+
+titles_df = pd.DataFrame(records)
+titles_df["title_lower"] = titles_df["title"].str.lower().str.strip()
+netflix_id_to_tt = {}   # netflix_id -> tt_id
+
+for _, nrow in titles_df.iterrows():
+    best_score = 0
+    best_meta = None
+
+    #https://github.com/rapidfuzz/RapidFuzz docs
+    for _, mrow in main_data.iterrows():
+        score = fuzz.ratio(nrow["title_lower"], mrow["title_lower"])
+        if score > best_score:
+            best_score = score
+            best_meta = mrow
+
+    if best_score < TITLE_THRESHOLD or best_meta is None:
+        continue
+
+    # Director match
+    confirmed = best_score >= TITLE_THRESHOLD
+    print(best_score)
+    if best_meta["director_lower"] and best_score >= 70:
+        # year relese year match
+        try:
+            meta_year = str(best_meta["Release Date"])
+            nf_year = str(int(nrow["year"])) if pd.notna(nrow["year"]) else ""
+            if nf_year and nf_year in meta_year:
+                confirmed = True
+        except Exception:
+            pass
+
+    if confirmed:
+        netflix_id_to_tt[int(nrow["netflix_id"])] = best_meta["Slug"]
+
+print(f"Matched {len(netflix_id_to_tt)} Netflix movies to tt Ids")
+
+valid_netflix_ids = set(netflix_id_to_tt.keys())
+rows = []
+current_movie_id = None
+
+for filepath in COMBINED_FILES:
+    print(f"Reading {os.path.basename(filepath)}...")
+    with open(filepath, encoding="latin-1") as f:
+        for line in f:
+            line = line.strip()
+            if line.endswith(":"):
+                current_movie_id = int(line[:-1])
+            elif current_movie_id in valid_netflix_ids:
+                parts = line.split(",")
+                if len(parts) == 3:
+                    customer_id, rating, date = parts
+                    rows.append({
+                        "customer_id": int(customer_id),
+                        "tt_id": netflix_id_to_tt[current_movie_id],
+                        "rating": int(rating),
+                        "date": date,
+                    })
+
+print(f"Found {len(rows):,} rating")
+print(f"Found {len(valid_netflix_ids):,} movies ground truth")
+
+
+df = pd.DataFrame(rows)
+df.to_csv(OUTPUT, index=False)
+print(f"Written to {OUTPUT}")

scripts/scrape.py

@@ -1,11 +1,8 @@
 from bs4 import BeautifulSoup
 import os
 
-# script_dir = os.path.dirname(os.path.abspath(__file__))
-# file_path = os.path.join(script_dir, "..", "data", "tt0074888.html")
 
 def extract_movie_info(file_path):
-
     with open(file_path, "r", encoding="utf-8") as f:
         html = f.read()
 
@@ -66,17 +63,25 @@ def extract_movie_info(file_path):
     # -----------------------------
     plot = ""
 
-    plot_header = soup.find(id="Plot")
+    plot_header = soup.find(id="Plot") or soup.find(id="Synopsis")
 
     if plot_header:
         current = plot_header.parent
-
         for sibling in current.find_next_siblings():
             if sibling.name == "div" and "mw-heading2" in sibling.get("class", []):
                 break
             if sibling.name == "p":
                 plot += sibling.get_text(" ", strip=True) + " "
 
+    if not plot and content:
+        for el in content.find_all(["p", "div"], recursive=False):
+            if el.name == "div" and el.find(["h2"]):
+                break
+            if el.name == "p":
+                text = el.get_text(" ", strip=True)
+                if text:
+                    plot += text + " "
+
     plot = plot.strip()
 
     return title, directed_by, cast, genre, plot, year, poster_filename