working on article example

examples/articles.py

@@ -0,0 +1,71 @@
+import datetime
+from pydantic import BaseModel
+import lxml
+from beakers import Recipe
+from beakers.filters import ConditionalFilter
+from beakers.http import HttpRequest
+
+
+class ArticleURL(BaseModel):
+    url: str
+    source: str
+
+
+class HttpResponse(BaseModel):
+    url: str
+    status: int
+    content: str
+    retrieved_at: datetime.datetime
+
+
+class Article(BaseModel):
+    title: str
+    text: str
+    image_urls: list[str]
+
+
+def is_npr(item) -> bool:
+    return item.url.source == "npr"
+
+
+def extract_npr_article(item) -> Article:
+    doc = lxml.html.fromstring(item.response.content)
+    title = doc.cssselect(".story-title")[0].text()
+    text = doc.cssselect(".paragraphs-container").text()
+    return Article(
+        title=title,
+        text=text,
+        image_urls=[],
+    )
+
+
+recipe = Recipe("newsface", "newsface.db")
+recipe.add_beaker("url", ArticleURL)
+recipe.add_beaker("response", HttpResponse)
+recipe.add_beaker("article", Article)
+recipe.add_transform("url", "response", HttpRequest)
+recipe.add_conditional(
+    "response",
+    is_npr,
+    "npr_article",
+)
+recipe.add_transform(
+    "npr_article",
+    "article",
+    extract_npr_article,
+)
+
+
+npr_examples = [
+    ArticleURL(url="https://text.npr.org/1186770075", source="npr"),
+    ArticleURL(url="https://text.npr.org/1186525577", source="npr"),
+    ArticleURL(url="https://text.npr.org/1185780577", source="npr"),
+]
+other = [
+    ArticleURL(url="https://nytimes.com", source="nytimes"),
+]
+
+# recipe.add_seed(
+#     "article_url",
+#     npr_examples + other,
+# )

src/beakers/__init__.py

@@ -0,0 +1 @@
+from .recipe import Recipe

src/beakers/filters.py

@@ -0,0 +1,9 @@
+class ConditionalFilter:
+    def __init__(self, condition):
+        self.condition = condition
+
+    def __call__(self, item):
+        if self.condition(item):
+            return item
+        else:
+            return None

src/beakers/http.py

@@ -0,0 +1,31 @@
+import httpx
+import pydantic
+import datetime
+
+
+class HttpResponse(pydantic.BaseModel):
+    url: str
+    status_code: int
+    response_body: str
+    retrieved_at: datetime.datetime = pydantic.Field(
+        default_factory=datetime.datetime.now
+    )
+
+
+class HttpRequest:
+    def __init__(self, beaker: str, field: str):
+        self.beaker = beaker
+        self.field = field
+
+    async def __call__(self, item) -> HttpResponse:
+        bkr = getattr(item, self.beaker)
+        url = getattr(bkr, self.field)
+
+        async with httpx.AsyncClient() as client:
+            response = await client.get(url)
+
+        return HttpResponse(
+            url=url,
+            status_code=response.status_code,
+            response_body=response.text,
+        )

src/beakers/recipe.py

@@ -48,9 +48,9 @@ class Recipe:
     def __repr__(self) -> str:
         return f"Recipe({self.name})"
 
-    def add_beaker(self, name: str, temp: bool = False) -> Beaker:
-        self.graph.add_node(name)
-        if temp:
+    def add_beaker(self, name: str, datatype: type | None) -> Beaker:
+        self.graph.add_node(name, datatype=datatype)
+        if datatype is None:
             self.beakers[name] = TempBeaker(name, self)
         else:
             self.beakers[name] = SqliteBeaker(name, self)
@@ -85,14 +85,14 @@ class Recipe:
         from_beaker: str,
         condition_func: callable,
         if_true: str,
-        if_false: str,
+        if_false: str = "",
     ) -> None:
         # first add a transform to evaluate the conditional
         if condition_func.__name__ == "<lambda>":
             cond_name = f"cond-{from_beaker}"
         else:
             cond_name = f"cond-{from_beaker}-{condition_func.__name__}"
-        self.add_beaker(cond_name, temp=True)
+        self.add_beaker(cond_name, None)
         self.add_transform(
             from_beaker,
             cond_name,
@@ -101,16 +101,18 @@ class Recipe:
         )
 
         # then add two filtered paths that remove the condition result
+        self.add_beaker(if_true, None)
         self.add_transform(
             cond_name,
             if_true,
             if_cond_true,
         )
-        self.add_transform(
-            cond_name,
-            if_false,
-            if_cond_false,
-        )
+        if if_false:
+            self.add_transform(
+                cond_name,
+                if_false,
+                if_cond_false,
+            )
 
     def get_metadata(self, table_name) -> dict:
         cursor = self.db.cursor()