Refactor search and answer logic to improve modularity

Extracted article handling and text evaluation functions to a new `extractor` module to enhance separation of concerns. Updated the search logic to incorporate result picking using LLM and adjusted the `Answer` function to return structured responses, ensuring better maintainability and extensibility.

pkg/answer/answer.go

@@ -5,7 +5,6 @@ import (
 	"errors"
 	"fmt"
 	"log/slog"
-	"net/url"
 
 	"github.com/Edw590/go-wolfram"
 	"go.starlark.net/lib/math"
@@ -13,7 +12,6 @@ import (
 	"go.starlark.net/syntax"
 
 	"gitea.stevedudenhoeffer.com/steve/answer/pkg/cache"
-	"gitea.stevedudenhoeffer.com/steve/answer/pkg/extractor"
 	"gitea.stevedudenhoeffer.com/steve/answer/pkg/search"
 	gollm "gitea.stevedudenhoeffer.com/steve/go-llm"
 )
@@ -94,105 +92,13 @@ type article struct {
 	Body  string
 }
 
-func extractArticle(ctx context.Context, c cache.Cache, u *url.URL) (res article, err error) {
+type Response struct {
-	defer func() {
+	Text    string
-		e := recover()
+	Sources []string
-
-		if e != nil {
-			if e, ok := e.(error); ok {
-				err = fmt.Errorf("panic: %w", e)
-			} else {
-				err = fmt.Errorf("panic: %v", e)
-			}
-		}
-	}()
-
-	extractors := extractor.MultiExtractor(
-		extractor.CacheExtractor{
-			Cache:     c,
-			Tag:       "goose",
-			Extractor: extractor.GooseExtractor{},
-		},
-		extractor.CacheExtractor{
-			Cache:     c,
-			Tag:       "playwright",
-			Extractor: extractor.PlaywrightExtractor{},
-		},
-	)
-
-	a, err := extractors.Extract(ctx, u.String())
-
-	if err != nil {
-		return article{
-			URL:   "",
-			Title: "",
-			Body:  "",
-		}, err
-	}
-
-	return article{
-		URL:   a.URL,
-		Title: a.Title,
-		Body:  a.Body,
-	}, nil
 }
 
-func doesTextAnswerQuestion(ctx *gollm.Context, q Question, text string) (string, error) {
+func (o Options) Answer(ctx context.Context, q Question) (Response, error) {
-	fnAnswer := gollm.NewFunction(
+	var answer Response
-		"answer",
-		"The answer from the given text that answers the question.",
-		func(ctx *gollm.Context, args struct {
-			Answer string `description:"the answer to the question, the answer should come from the text"`
-		}) (string, error) {
-			return args.Answer, nil
-		})
-
-	fnNoAnswer := gollm.NewFunction(
-		"no_answer",
-		"Indicate that the text does not answer the question.",
-		func(ctx *gollm.Context, args struct {
-			Ignored string `description:"ignored, just here to make sure the function is called.  Fill with anything."`
-		}) (string, error) {
-			return "", nil
-		})
-
-	req := gollm.Request{
-		Messages: []gollm.Message{
-			{
-				Role: gollm.RoleSystem,
-				Text: "Evaluate the given text to see if it answers the question from the user.  The text is as follows:",
-			},
-			{
-				Role: gollm.RoleSystem,
-				Text: text,
-			},
-			{
-				Role: gollm.RoleUser,
-				Text: q.Question,
-			},
-		},
-		Toolbox: gollm.NewToolBox(fnAnswer, fnNoAnswer),
-	}
-
-	res, err := q.Model.ChatComplete(ctx, req)
-
-	if err != nil {
-		return "", err
-	}
-
-	if len(res.Choices) == 0 {
-		return "", nil
-	}
-
-	if len(res.Choices[0].Calls) == 0 {
-		return "", nil
-	}
-
-	return req.Toolbox.Execute(ctx, res.Choices[0].Calls[0])
-}
-
-func (o Options) Answer(ctx context.Context, q Question) (string, error) {
-	var answer string
 
 	fnSearch := gollm.NewFunction(
 		"search",
@@ -217,7 +123,8 @@ func (o Options) Answer(ctx context.Context, q Question) (string, error) {
 			Answer  string   `json:"answer" description:"the answer to the question"`
 			Sources []string `json:"sources" description:"the sources used to find the answer (e.g.: urls of sites from search)"`
 		}) (string, error) {
-			answer = args.Answer
+			answer.Text = args.Answer
+			answer.Sources = args.Sources
 			return args.Answer, nil
 		})
 
@@ -260,7 +167,7 @@ func (o Options) Answer(ctx context.Context, q Question) (string, error) {
 		func(ctx *gollm.Context, args struct {
 			Reason string `description:"the reason the system is giving up (e.g.: 'no results found')"`
 		}) (string, error) {
-			answer = "given up: " + args.Reason
+			answer.Text = "given up: " + args.Reason
 			return "given up", nil
 		})
 
@@ -422,14 +329,14 @@ func (o Options) Answer(ctx context.Context, q Question) (string, error) {
 		newReq, err := runAnswer(o, req)
 
 		if err != nil {
-			return "", err
+			return answer, err
 		}
 
 		if newReq == nil {
 			break
 		}
 
-		if answer != "" {
+		if answer.Text != "" {
 			break
 		}
 
@@ -439,6 +346,6 @@ func (o Options) Answer(ctx context.Context, q Question) (string, error) {
 	return answer, nil
 }
 
-func Answer(ctx context.Context, q Question) (string, error) {
+func Answer(ctx context.Context, q Question) (Response, error) {
 	return DefaultOptions.Answer(ctx, q)
 }

pkg/answer/extractor.go

@@ -0,0 +1,109 @@
+package answer
+
+import (
+	"context"
+	"fmt"
+	"net/url"
+
+	gollm "gitea.stevedudenhoeffer.com/steve/go-llm"
+
+	"gitea.stevedudenhoeffer.com/steve/answer/pkg/cache"
+	"gitea.stevedudenhoeffer.com/steve/answer/pkg/extractor"
+)
+
+func extractArticle(ctx context.Context, c cache.Cache, u *url.URL) (res article, err error) {
+	defer func() {
+		e := recover()
+
+		if e != nil {
+			if e, ok := e.(error); ok {
+				err = fmt.Errorf("panic: %w", e)
+			} else {
+				err = fmt.Errorf("panic: %v", e)
+			}
+		}
+	}()
+
+	extractors := extractor.MultiExtractor(
+		extractor.CacheExtractor{
+			Cache:     c,
+			Tag:       "goose",
+			Extractor: extractor.GooseExtractor{},
+		},
+		extractor.CacheExtractor{
+			Cache:     c,
+			Tag:       "playwright",
+			Extractor: extractor.PlaywrightExtractor{},
+		},
+	)
+
+	a, err := extractors.Extract(ctx, u.String())
+
+	if err != nil {
+		return article{
+			URL:   "",
+			Title: "",
+			Body:  "",
+		}, err
+	}
+
+	return article{
+		URL:   a.URL,
+		Title: a.Title,
+		Body:  a.Body,
+	}, nil
+}
+
+func doesTextAnswerQuestion(ctx *gollm.Context, q Question, text string) (string, error) {
+	fnAnswer := gollm.NewFunction(
+		"answer",
+		"The answer from the given text that answers the question.",
+		func(ctx *gollm.Context, args struct {
+			Answer string `description:"the answer to the question, the answer should come from the text"`
+		}) (string, error) {
+			return args.Answer, nil
+		})
+
+	fnNoAnswer := gollm.NewFunction(
+		"no_answer",
+		"Indicate that the text does not answer the question.",
+		func(ctx *gollm.Context, args struct {
+			Ignored string `description:"ignored, just here to make sure the function is called.  Fill with anything."`
+		}) (string, error) {
+			return "", nil
+		})
+
+	req := gollm.Request{
+		Messages: []gollm.Message{
+			{
+				Role: gollm.RoleSystem,
+				Text: "Evaluate the given text to see if it answers the question from the user.  The text is as follows:",
+			},
+			{
+				Role: gollm.RoleSystem,
+				Text: text,
+			},
+			{
+				Role: gollm.RoleUser,
+				Text: q.Question,
+			},
+		},
+		Toolbox: gollm.NewToolBox(fnAnswer, fnNoAnswer),
+	}
+
+	res, err := q.Model.ChatComplete(ctx, req)
+
+	if err != nil {
+		return "", err
+	}
+
+	if len(res.Choices) == 0 {
+		return "", nil
+	}
+
+	if len(res.Choices[0].Calls) == 0 {
+		return "", nil
+	}
+
+	return req.Toolbox.Execute(ctx, res.Choices[0].Calls[0])
+}

pkg/answer/search.go

@@ -6,6 +6,8 @@ import (
 	"net/url"
 	"strings"
 
+	"gitea.stevedudenhoeffer.com/steve/answer/pkg/search"
+
 	gollm "gitea.stevedudenhoeffer.com/steve/go-llm"
 )
 
@@ -23,6 +25,92 @@ func (s searchResults) String() (string, error) {
 	return string(b), nil
 }
 
+func pickResult(ctx *gollm.Context, results []search.Result, q Question) (*search.Result, error) {
+	// if there's only one result, return it
+	if len(results) == 1 {
+		return &results[0], nil
+	}
+
+	// if there are no results, return nil
+	if len(results) == 0 {
+		return nil, nil
+	}
+
+	var pick *search.Result
+	var refused bool
+	// finally, if there are multiple results then ask the LLM to pick one to read next
+	fnPick := gollm.NewFunction(
+		"pick",
+		"The search result to read next.",
+		func(ctx *gollm.Context, args struct {
+			URL string `description:"the url to read next"`
+		}) (string, error) {
+			for _, r := range results {
+				if r.URL == args.URL {
+					pick = &r
+					break
+				}
+			}
+
+			return "", nil
+		})
+
+	fnNoPick := gollm.NewFunction(
+		"no_pick",
+		"Indicate that there are no results worth reading.",
+		func(ctx *gollm.Context, args struct {
+			Ignored string `description:"ignored, just here to make sure the function is called.  Fill with anything."`
+		}) (string, error) {
+			refused = true
+			return "", nil
+		})
+
+	req := gollm.Request{
+		Messages: []gollm.Message{
+			{
+				Role: gollm.RoleSystem,
+				Text: `You are being given results from a web search.  Please select the result you would like to read next to answer the question.  Try to pick the most reputable and relevant result.
+The results will be in the JSON format of: {"Url": "https://url.here", "Title": "Title Of Search", "Description": "description here"}`,
+			},
+			{
+				Role: gollm.RoleSystem,
+				Text: "The question you are trying to answer is: " + q.Question,
+			},
+		},
+		Toolbox: gollm.NewToolBox(fnPick, fnNoPick),
+	}
+
+	for _, r := range results {
+		b, _ := json.Marshal(r)
+		req.Messages = append(req.Messages, gollm.Message{
+			Role: gollm.RoleUser,
+			Text: string(b),
+		})
+	}
+
+	res, err := q.Model.ChatComplete(ctx, req)
+
+	if err != nil {
+		return nil, err
+	}
+
+	if len(res.Choices) == 0 {
+		return nil, nil
+	}
+
+	if len(res.Choices[0].Calls) == 0 {
+		return nil, nil
+	}
+
+	_, _ = req.Toolbox.Execute(ctx, res.Choices[0].Calls[0])
+
+	if refused {
+		return nil, nil
+	}
+
+	return pick, nil
+}
+
 func internalSearch(ctx *gollm.Context, q Question, searchTerm string) (searchResults, error) {
 	slog.Info("searching", "search", searchTerm, "question", q)
 	results, err := q.Search.Search(ctx, searchTerm)
@@ -34,9 +122,26 @@ func internalSearch(ctx *gollm.Context, q Question, searchTerm string) (searchRe
 		return searchResults{Url: "not-found", Answer: "no search results found"}, nil
 	}
 
-	// first pass try to see if any provide the result without needing archive
+	for len(results) > 0 {
-	for _, r := range results {
+		var pick *search.Result
-		trimmed := strings.TrimSpace(r.URL)
+		if len(results) == 1 {
+			pick = &results[0]
+			results = results[1:]
+		} else {
+			var err error
+			pick, err = pickResult(ctx, results, q)
+
+			slog.Info("picked result", "result", pick, "error", err)
+			if err != nil {
+				return searchResults{}, err
+			}
+
+			if pick == nil {
+				break
+			}
+		}
+
+		trimmed := strings.TrimSpace(pick.URL)
 		if trimmed == "" {
 
 		}