paperformatter/paperformatter.go

// Package paperformatter provides functionality to convert JSON files containing paper reviews
// into formatted Markdown documents. It organizes papers into accepted and rejected categories,
// with each entry containing a title, arXiv link, abstract, decision, and explanation.
//
// The package provides robust error handling with custom error types for validation,
// file operations, and JSON parsing issues. It supports case-insensitive decision values
// and properly formats multiline abstracts.
//
// Basic usage:
//
//	err := paperformatter.FormatPapers("input.json", "output.md")
//	if err != nil {
//	    log.Fatal(err)
//	}
//
// For detailed examples, see the examples in the documentation.
package paperformatter

import (
	"encoding/json"
	"fmt"
	"os"
	"strings"
)

// Paper represents a research paper with its basic metadata.
type Paper struct {
	Title    string `json:"title"`
	Abstract string `json:"abstract"`
	ArxivID  string `json:"arxiv_id"`
}

// Entry represents a paper review entry containing the paper details,
// the review decision, and an explanation for the decision.
type Entry struct {
	Paper       Paper  `json:"paper"`
	Decision    string `json:"decision"`
	Explanation string `json:"explanation"`
}

// FormatPapers reads paper reviews from a JSON file and generates a formatted Markdown document.
// The output document organizes papers into "Accepted Papers" and "Rejected Papers" sections.
//
// The input JSON file should contain an array of Entry objects. Each entry must have:
//   - A paper with a title, abstract, and arXiv ID
//   - A decision ("accept" or "reject", case-insensitive)
//   - An explanation for the decision
//
// The function returns an error if:
//   - The input file cannot be read (FileError)
//   - The JSON is invalid (JSONError)
//   - Any entry fails validation (ValidationError)
//   - The output file cannot be written (FileError)
func FormatPapers(inputFile, outputFile string) error {
	data, err := os.ReadFile(inputFile)
	if err != nil {
		return &FileError{
			Path:    inputFile,
			Op:      "read",
			Wrapped: err,
		}
	}

	var entries []Entry
	if err := json.Unmarshal(data, &entries); err != nil {
		return &JSONError{Wrapped: err}
	}

	// Validate all entries before processing
	for i, entry := range entries {
		if err := validateEntry(entry); err != nil {
			return fmt.Errorf("entry %d: %w", i+1, err)
		}
	}

	var accepted, rejected []Entry
	for _, entry := range entries {
		if strings.ToUpper(entry.Decision) == "ACCEPT" {
			accepted = append(accepted, entry)
		} else {
			rejected = append(rejected, entry)
		}
	}

	output := "# Accepted Papers\n\n"
	for _, entry := range accepted {
		output += formatEntry(entry)
	}

	output += "# Rejected Papers\n\n"
	for _, entry := range rejected {
		output += formatEntry(entry)
	}

	if err := os.WriteFile(outputFile, []byte(output), 0644); err != nil {
		return &FileError{
			Path:    outputFile,
			Op:      "write",
			Wrapped: err,
		}
	}

	return nil
}

func formatEntry(entry Entry) string {
	return fmt.Sprintf("## %s\n\n"+
		"[arXiv:%s](https://arxiv.org/abs/%s)\n\n"+
		"> %s\n\n"+
		"**Decision:** %s\n\n"+
		"**Explanation:** %s\n\n",
		entry.Paper.Title,
		entry.Paper.ArxivID,
		entry.Paper.ArxivID,
		strings.ReplaceAll(entry.Paper.Abstract, "\n", "\n> "),
		entry.Decision,
		entry.Explanation)
}