Merge pull request #17 from cs3216-a3-group-4/seeleng/article-preproc…

…essing

feat: add event preprocessing

backend/alembic/versions/423a1e6552f0_create_article_and_event_tables.py

@@ -0,0 +1,79 @@
+"""Create article and event tables
+
+Revision ID: 423a1e6552f0
+Revises: 681607ac9aeb
+Create Date: 2024-09-21 16:44:37.663284
+
+"""
+
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+from sqlalchemy.dialects import postgresql
+
+# revision identifiers, used by Alembic.
+revision: str = "423a1e6552f0"
+down_revision: Union[str, None] = "681607ac9aeb"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    sa.Enum("CNA", "GUARDIAN", name="articlesource").create(op.get_bind())
+    op.create_table(
+        "article",
+        sa.Column("id", sa.Integer(), nullable=False),
+        sa.Column("title", sa.String(), nullable=False),
+        sa.Column("summary", sa.String(), nullable=False),
+        sa.Column("body", sa.String(), nullable=False),
+        sa.Column("url", sa.String(), nullable=False),
+        sa.Column(
+            "source",
+            postgresql.ENUM("CNA", "GUARDIAN", name="articlesource", create_type=False),
+            nullable=False,
+        ),
+        sa.PrimaryKeyConstraint("id"),
+    )
+    op.create_table(
+        "category",
+        sa.Column("id", sa.Integer(), nullable=False),
+        sa.Column("name", sa.String(), nullable=False),
+        sa.PrimaryKeyConstraint("id"),
+    )
+    op.create_table(
+        "event",
+        sa.Column("id", sa.Integer(), nullable=False),
+        sa.Column("title", sa.String(), nullable=False),
+        sa.Column("description", sa.String(), nullable=False),
+        sa.Column("analysis", sa.String(), nullable=False),
+        sa.Column("duplicate", sa.Boolean(), nullable=False),
+        sa.Column("date", sa.DateTime(), nullable=False),
+        sa.PrimaryKeyConstraint("id"),
+    )
+    op.create_table(
+        "event_category",
+        sa.Column("event_id", sa.Integer(), nullable=False),
+        sa.Column("category_id", sa.Integer(), nullable=False),
+        sa.ForeignKeyConstraint(
+            ["category_id"],
+            ["category.id"],
+        ),
+        sa.ForeignKeyConstraint(
+            ["event_id"],
+            ["event.id"],
+        ),
+        sa.PrimaryKeyConstraint("event_id", "category_id"),
+    )
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_table("event_category")
+    op.drop_table("event")
+    op.drop_table("category")
+    op.drop_table("article")
+    sa.Enum("CNA", "GUARDIAN", name="articlesource").drop(op.get_bind())
+    # ### end Alembic commands ###

backend/alembic/versions/e84e13568072_add_columns_to_event_table.py

@@ -0,0 +1,37 @@
+"""Add columns to event table
+
+Revision ID: e84e13568072
+Revises: 423a1e6552f0
+Create Date: 2024-09-21 17:21:24.515814
+
+"""
+
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = "e84e13568072"
+down_revision: Union[str, None] = "423a1e6552f0"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.add_column("event", sa.Column("is_singapore", sa.Boolean(), nullable=False))
+    op.add_column(
+        "event", sa.Column("original_article_id", sa.Integer(), nullable=False)
+    )
+    op.create_foreign_key(None, "event", "article", ["original_article_id"], ["id"])
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_constraint(None, "event", type_="foreignkey")
+    op.drop_column("event", "original_article_id")
+    op.drop_column("event", "is_singapore")
+    # ### end Alembic commands ###

backend/alembic/versions/f3e847c3ee9d_add_article_date_column.py

@@ -0,0 +1,31 @@
+"""Add article date column
+
+Revision ID: f3e847c3ee9d
+Revises: e84e13568072
+Create Date: 2024-09-21 18:25:05.912853
+
+"""
+
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = "f3e847c3ee9d"
+down_revision: Union[str, None] = "e84e13568072"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.add_column("article", sa.Column("date", sa.DateTime(), nullable=False))
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_column("article", "date")
+    # ### end Alembic commands ###

backend/src/events/models.py

@@ -0,0 +1,63 @@
+from enum import Enum
+from sqlalchemy.orm import Mapped, mapped_column, relationship
+from sqlalchemy import ForeignKey
+from datetime import datetime
+from src.common.base import Base
+
+
+class ArticleSource(str, Enum):
+    CNA = "CNA"
+    GUARDIAN = "GUARDIAN"
+
+
+class Article(Base):
+    __tablename__ = "article"
+
+    id: Mapped[int] = mapped_column(primary_key=True)
+    title: Mapped[str]
+    summary: Mapped[str]
+    body: Mapped[str]
+    url: Mapped[str]
+    source: Mapped[ArticleSource]
+    date: Mapped[datetime]
+
+    event: Mapped["Event"] = relationship(back_populates="original_article")
+
+
+class Event(Base):
+    __tablename__ = "event"
+
+    id: Mapped[int] = mapped_column(primary_key=True)
+    title: Mapped[str]
+    description: Mapped[str]
+    analysis: Mapped[str]
+    duplicate: Mapped[bool]
+    date: Mapped[datetime]
+    is_singapore: Mapped[bool]
+    original_article_id: Mapped[int] = mapped_column(ForeignKey("article.id"))
+
+    categories: Mapped[list["Category"]] = relationship(
+        back_populates="events", secondary="event_category"
+    )
+
+    original_article: Mapped[Article] = relationship(back_populates="event")
+
+
+class Category(Base):
+    __tablename__ = "category"
+
+    id: Mapped[int] = mapped_column(primary_key=True)
+    name: Mapped[str]
+
+    events: Mapped[list[Event]] = relationship(
+        secondary="event_category", back_populates="categories"
+    )
+
+
+class EventCategory(Base):
+    __tablename__ = "event_category"
+
+    event_id: Mapped[int] = mapped_column(ForeignKey("event.id"), primary_key=True)
+    category_id: Mapped[int] = mapped_column(
+        ForeignKey("category.id"), primary_key=True
+    )

backend/src/scrapers/cna/process.py

@@ -0,0 +1,96 @@
+import argparse
+import asyncio
+import json
+from sqlalchemy.orm import Session
+from src.events.models import Article, ArticleSource
+from src.common.database import engine
+from pydantic import BaseModel, ConfigDict
+from bs4 import BeautifulSoup
+import os
+
+
+parser = argparse.ArgumentParser()
+parser.add_argument("-i", "--input", help="input folder path")
+
+args = parser.parse_args()
+folder_path = args.input
+
+
+processed_ids = set()
+
+CATEGORIES = [
+    "Asia",
+    "East Asia",
+    "Singapore",
+    "World",
+    "Commentary",
+    "CNA Explains",
+    "Business",
+    "Sport",
+    "CNA Insider",
+]
+
+
+class CNAArticle(BaseModel):
+    model_config = ConfigDict(extra="allow")
+
+    type: str
+    uuid: str
+    nid: str
+    title: str
+    title_url: str
+    absolute_url: str
+    field_summary: str | None = None
+    description: str | None = None
+    release_date: str
+
+
+async def process(category: str):
+    with open(f"./src/scrapers/cna/data/{category}.json") as f:
+        data = json.load(f)
+    for page_index, page in enumerate(data):
+        for index, item in enumerate(page):
+            try:
+                article = CNAArticle.model_validate(item)
+
+                if article.type != "article":
+                    continue
+                if article.uuid in processed_ids:
+                    continue
+                processed_ids.add(article.uuid)
+
+                # Read body text from scrape.py
+                with open(
+                    os.path.join(folder_path, f"{article.uuid}_{category}.txt")
+                ) as f:
+                    body = f.read()
+
+                if not body.strip():
+                    continue
+
+                # Add to database
+                article_orm = Article(
+                    title=article.title,
+                    summary=BeautifulSoup(article.description).getText()
+                    if article.description
+                    else "",
+                    url=article.absolute_url,
+                    source=ArticleSource.CNA,
+                    body=body.strip(),
+                    date=article.release_date,
+                )
+                with Session(engine) as session:
+                    session.add(article_orm)
+                    session.commit()
+
+            except Exception as e:
+                print(f"{category}: something went wrong with {page_index}, {index}")
+                print(e)
+
+
+async def process_all_categories():
+    asyncio.gather(*[process(category) for category in CATEGORIES])
+
+
+if __name__ == "__main__":
+    asyncio.run(process_all_categories())

backend/src/scrapers/guardian/process.py

@@ -0,0 +1,43 @@
+import argparse
+
+import json
+from sqlalchemy.orm import Session
+from src.events.models import Article, ArticleSource
+from src.common.database import engine
+from pydantic import BaseModel, ConfigDict
+
+
+parser = argparse.ArgumentParser()
+parser.add_argument("-i", "--input", help="input folder path")
+args = parser.parse_args()
+
+
+class GuardianArticleFields(BaseModel):
+    model_config = ConfigDict(extra="allow")
+    bodyText: str
+    trailText: str | None = None
+
+
+class GuardianArticle(BaseModel):
+    model_config = ConfigDict(extra="allow")
+    fields: GuardianArticleFields
+    webUrl: str
+    webTitle: str
+    webPublicationDate: str
+
+
+with open(args.input) as f:
+    data = json.load(f)
+    for row in data:
+        article = GuardianArticle.model_validate(row)
+        article_orm = Article(
+            title=article.webTitle,
+            summary=article.fields.trailText if article.fields.trailText else "",
+            url=article.webUrl,
+            source=ArticleSource.GUARDIAN,
+            body=article.fields.bodyText,
+            date=article.webPublicationDate,
+        )
+        with Session(engine) as session:
+            session.add(article_orm)
+            session.commit()

backend/src/utils/models.py

@@ -1,3 +1,4 @@
 # this is just to load models in alembic
 
-from src.auth import models  # noqa: F401
+from src.auth import models as auth_models  # noqa: F401
+from src.events import models as event_models  # noqa: F401