Implement Image and Infobox counting (#171)

This feature counts the Image and Infobox from the submitted article and
saves it into the database.

Author: Aditya0545

backend/alembic/versions/f1a2b3c4d5e6_add_image_and_infobox_counts_to_submissions.py

@@ -0,0 +1,45 @@
+"""Add image_count and infobox_count columns to submissions table
+
+Revision ID: f1a2b3c4d5e6
+Revises: e4e56960f418
+
+"""
+from alembic import op
+import sqlalchemy as sa
+from sqlalchemy import inspect
+
+
+revision = "f1a2b3c4d5e6"
+down_revision = "e4e56960f418"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    conn = op.get_bind()
+    inspector = inspect(conn)
+    columns = [col["name"] for col in inspector.get_columns("submissions")]
+
+    if "image_count" not in columns:
+        op.add_column(
+            "submissions",
+            sa.Column("image_count", sa.Integer(), nullable=True),
+        )
+
+    if "infobox_count" not in columns:
+        op.add_column(
+            "submissions",
+            sa.Column("infobox_count", sa.Integer(), nullable=True),
+        )
+
+
+def downgrade() -> None:
+    conn = op.get_bind()
+    inspector = inspect(conn)
+    columns = [col["name"] for col in inspector.get_columns("submissions")]
+
+    if "infobox_count" in columns:
+        op.drop_column("submissions", "infobox_count")
+
+    if "image_count" in columns:
+        op.drop_column("submissions", "image_count")

backend/app/models/submission.py

@@ -70,6 +70,12 @@ class Submission(BaseModel):
     # Can be negative if article was reduced in size
     article_expansion_bytes = db.Column(db.Integer, nullable=True)
 
+    # Image count
+    image_count = db.Column(db.Integer, nullable=True)
+
+    # Infobox count
+    infobox_count = db.Column(db.Integer, nullable=True)
+
     # Template enforcement tracking
     # True if template was automatically added to the article during submission
     template_added = db.Column(db.Boolean, nullable=True, default=False)
@@ -173,6 +179,8 @@ def __init__(
         template_added=False,
         categories_added=None,
         category_error=None,
+        image_count=None,
+        infobox_count=None,
     ):
         """
         Initialize a new Submission instance
@@ -192,6 +200,8 @@ def __init__(
             template_added: Whether template was automatically added to article (optional)
             categories_added: List of category names that were automatically added (optional, stored as JSON)
             category_error: Error message if category attachment failed (optional)
+            image_count: Number of images in the article (optional)
+            infobox_count: Number of infoboxes in the article (optional)
         """
         # Set required fields
         self.user_id = user_id
@@ -219,6 +229,8 @@ def __init__(
         else:
             self.categories_added = None
         self.category_error = category_error
+        self.image_count = image_count
+        self.infobox_count = infobox_count
         self.reviewed_by = None
         self.reviewed_at = None
         self.review_comment = None
@@ -530,6 +542,8 @@ def to_dict(self, include_user_info=False):
             "article_page_id": self.article_page_id,
             "article_size_at_start": self.article_size_at_start,
             "article_expansion_bytes": self.article_expansion_bytes,
+            "image_count": self.image_count,
+            "infobox_count": self.infobox_count,
             "template_added": self.template_added,
             "categories_added": self.get_categories_added(),
             "category_error": self.category_error,

backend/app/routes/contest_routes.py

@@ -31,6 +31,8 @@
     check_article_has_category,
     append_categories_to_article,
     get_article_reference_count,
+    get_article_image_count,
+    get_article_infobox_count,
     MEDIAWIKI_API_TIMEOUT,
 )
 from app.services.outreach_dashboard import (
@@ -1314,6 +1316,8 @@ def submit_to_contest(contest_id):  # pylint: disable=too-many-return-statements
     article_size_at_start = None
     article_expansion_bytes = None
     article_reference_count = None
+    image_count = None
+    infobox_count = None
 
     # --- Fetch Article Information from MediaWiki API ---
     # MediaWiki API fetching has deep nesting due to complex error handling
@@ -1699,6 +1703,31 @@ def submit_to_contest(contest_id):  # pylint: disable=too-many-return-statements
             pass
         article_reference_count = None
 
+    # --- Fetch Image and Infobox Counts ---
+    # These richness metrics are stored for future use but are not used in
+    # validation or scoring yet.
+    try:
+        image_count = get_article_image_count(article_link)
+    except Exception as img_error:  # pylint: disable=broad-exception-caught
+        try:
+            current_app.logger.warning(
+                "Failed to fetch image count: %s", str(img_error)
+            )
+        except Exception:  # pylint: disable=broad-exception-caught
+            pass
+        image_count = None
+
+    try:
+        infobox_count = get_article_infobox_count(article_link)
+    except Exception as ibx_error:  # pylint: disable=broad-exception-caught
+        try:
+            current_app.logger.warning(
+                "Failed to fetch infobox count: %s", str(ibx_error)
+            )
+        except Exception:  # pylint: disable=broad-exception-caught
+            pass
+        infobox_count = None
+
     # --- Validate Article Requirements ---
     # Validate article byte count against contest requirements
     # This check happens after fetching article information from MediaWiki API
@@ -2060,6 +2089,8 @@ def submit_to_contest(contest_id):  # pylint: disable=too-many-return-statements
             template_added=template_added,
             categories_added=categories_added,
             category_error=category_error,
+            image_count=image_count,
+            infobox_count=infobox_count,
         )
 
         submission.save()

backend/app/utils/__init__.py

@@ -44,6 +44,8 @@
     "append_categories_to_article",
     "get_article_reference_count",
     "get_mediawiki_user_edit_count",
+    "get_article_image_count",
+    "get_article_infobox_count",
 ]
 
 
@@ -948,6 +950,63 @@ def _fetch_footnotes_count(api_url: str, page_title: str, headers: dict) -> int:
         return 0
 
 
+def _log_warning(message: str, error: Exception) -> None:
+    """Best-effort logging helper that uses Flask current_app when available.
+
+    This keeps network helpers free from hard Flask dependencies while still
+    providing useful diagnostics in a running application.
+    """
+    try:
+        from flask import current_app
+
+        current_app.logger.warning("%s: %s", message, str(error))
+    except Exception:  # pylint: disable=broad-exception-caught
+        # Logging must never break core logic, so ignore any logging failures
+        pass
+
+
+def get_article_image_count(article_url: str) -> Optional[int]:
+    """
+    The count is approximate and based purely on wikitext patterns; it does
+    not guarantee that every match results in a rendered image, but it
+    generally tracks user-added content images.
+    """
+    try:
+        wikitext = get_article_wikitext(article_url)
+        if wikitext is None:
+            return None
+
+        # Match explicit file/image links like [[File:Example.jpg|...]] or
+        # [[Image:Example.png|...]] in a case-insensitive way.
+        matches = re.findall(r'\[\[(?:File|Image):', wikitext, flags=re.IGNORECASE)
+        return len(matches)
+
+    except Exception as error:  # pylint: disable=broad-exception-caught
+        _log_warning("Failed to fetch image count", error)
+        return None
+
+
+def get_article_infobox_count(article_url: str) -> Optional[int]:
+    """Count approximate number of infobox templates in article wikitext.
+
+    Detection is done via a simple regex scan for ``{{infobox ...}}`` in the
+    raw wikitext. This is an approximation and may over-count or under-count
+    in edge cases (e.g. nested templates, unusual formatting), but is
+    sufficient for high-level richness metrics.
+    """
+    try:
+        wikitext = get_article_wikitext(article_url)
+        if wikitext is None:
+            return None
+
+        matches = re.findall(r"\{\{\s*infobox\b", wikitext, flags=re.IGNORECASE)
+        return len(matches)
+
+    except Exception as error:  # pylint: disable=broad-exception-caught
+        _log_warning("Failed to fetch infobox count", error)
+        return None
+
+
 def get_article_reference_count(article_url: str) -> Optional[int]:
     """
     Get the total number of references in a MediaWiki article.