Experiments with Claude 3 and Student Newspapers

issue.py

import anthropic
import base64
import httpx
import json


class Issue:
    def __init__(self, iiif_manifest):
        self.manifest = iiif_manifest
        self.pages = self.__get_pages()

    def __get_pages(self):
        canvases = httpx.get(self.manifest).json()
        pages = []
        for item in canvases['items']:
            pid = item['items'][0]['id'].split('/')[-1].split(':')[1]
            pages.append(
                self.__switch_image_to_hash(
                    self.__convert(
                        f"https://digital.lib.utk.edu/collections/islandora/object/beacon%3A{pid}/datastream/JPG"
                    )
                )
            )
        return pages

    @staticmethod
    def __convert(image_url):
        return base64.b64encode(
            httpx.get(
                image_url
            ).content
        ).decode("utf-8")

    @staticmethod
    def __switch_image_to_hash(image):
        return {
            "type": "image",
            "source": {
                "type": "base64",
                "media_type": "image/jpeg",
                "data": image
            }
        }


class CostCalculator:
    def  __init__(self, input_tokens, output_tokens, model):
        self.input_tokens = input_tokens
        self.output_tokens = output_tokens
        self.model = model
        self.cost = self.__calculate_cost()

    def __calculate_cost(self):
        input_cost = self.input_tokens / 1000000  * .25
        output_cost = self.output_tokens / 1000000 * 1.25
        return f"Input cost: {input_cost}, Output cost: {output_cost}"


class ClaudeRequest:
    def __init__(self, model, key, prompt, pages):
        self.model = model
        self.prompt = prompt
        self.content = self.__add_prompt(pages, prompt)
        self.client = self.__create_client(key)
        self.output = self.__request()
        self.cost = self.__determine_cost()

    @staticmethod
    def __create_client(apikey):
        return anthropic.Anthropic(
            api_key=apikey,
        )

    @staticmethod
    def __add_prompt(all_pages, user_prompt):
        all_pages.append(
            {
                "type": "text",
                "text": user_prompt
            }
        )
        return all_pages

    def __request(self):
        response = self.client.messages.create(
            model=self.model,
            max_tokens=4096,
            messages=[
                {
                    "role": "user",
                    "content": self.content
                }
            ],
        )
        return json.loads(response.json())

    def __determine_cost(self):
        return CostCalculator(
            input_tokens=self.output['usage']['input_tokens'],
            output_tokens=self.output['usage']['output_tokens'],
            model=self.model
        ).cost

    def write_output(self, output_file, iiif_manifest):
        with open(output_file, 'w') as output_file:
            output_file.write(f"# Newspaper Issue [{iiif_manifest.split('/')[-2]}:{iiif_manifest.split('/')[-1]}]({iiif_manifest})\n\n")
            output_file.write(f"## Cost:\n\nCosts to run: \n\n")
            output_file.write(f"{self.cost}\n\n")
            text_response = json.loads(self.output['content'][0]['text'])
            output_file.write(f"## Articles and Ads Found:\n\n")
            for item in text_response['articles']:
                output_file.write(f"* {item}\n")
            output_file.write(f"\n\n## Keywords:\n\n")
            for item in text_response['keywords']:
                output_file.write(f"* {item}\n")


if __name__ == "__main__":
    manifest = "https://digital.lib.utk.edu/assemble/manifest/beacon/5521"
    key = "my-api-key"
    model = "claude-3-haiku-20240307"
    prompt = "The combined images above create an issue of a newspaper. Give me the title of the articles available in the issue and 10 keywords that describe the contents of the issue overall. Please respond with the message as JSON with the following format: {\"articles\": [\"article\"], \"keywords\": [\"keyword1\", \"keyword2\", \"keyword3\"]}."
    my_output_file = "harrison.md"
    x = Issue(manifest)
    content = x.pages
    y = ClaudeRequest(
        model=model,
        key=key,
        prompt=prompt,
        pages=content
    )
    y.write_output(my_output_file, manifest)