breakpilot-compliance/ai-compliance-sdk/internal/api/handlers/iace_handler_documents.go

package handlers

import (
	"net/http"
	"sort"
	"strings"
	"sync"

	"github.com/breakpilot/ai-compliance-sdk/internal/ucca"
	"github.com/gin-gonic/gin"
)

// Cached CE corpus document index — built once on first request.
var (
	ceCorpusOnce  sync.Once
	ceCorpusDocs  []ucca.CEDocumentInfo
	ceCorpusErr   error
)

// ListCECorpusDocuments returns the deduplicated document index from bp_compliance_ce.
// GET /iace/ce-corpus-documents
func (h *IACEHandler) ListCECorpusDocuments(c *gin.Context) {
	ceCorpusOnce.Do(func() {
		ceCorpusDocs, ceCorpusErr = h.ragClient.ScrollDocumentIndex(
			c.Request.Context(), "bp_compliance_ce",
		)
		if ceCorpusErr == nil {
			sort.Slice(ceCorpusDocs, func(i, j int) bool {
				if ceCorpusDocs[i].Category != ceCorpusDocs[j].Category {
					return ceCorpusDocs[i].Category < ceCorpusDocs[j].Category
				}
				return ceCorpusDocs[i].RegulationID < ceCorpusDocs[j].RegulationID
			})
		}
	})

	if ceCorpusErr != nil {
		c.JSON(http.StatusInternalServerError, gin.H{
			"error": "failed to load CE corpus index: " + ceCorpusErr.Error(),
		})
		return
	}

	// Optional search filter
	query := strings.ToLower(c.Query("q"))
	category := c.Query("category")

	filtered := ceCorpusDocs
	if query != "" || category != "" {
		filtered = make([]ucca.CEDocumentInfo, 0, len(ceCorpusDocs))
		for _, d := range ceCorpusDocs {
			if category != "" && d.Category != category {
				continue
			}
			if query != "" {
				haystack := strings.ToLower(d.RegulationID + " " + d.NameDE + " " + d.NameEN + " " + d.SourceOrg)
				if !strings.Contains(haystack, query) {
					continue
				}
			}
			filtered = append(filtered, d)
		}
	}

	// Group by category for the response
	groups := make(map[string][]ucca.CEDocumentInfo)
	for _, d := range filtered {
		cat := d.Category
		if cat == "" {
			cat = "other"
		}
		groups[cat] = append(groups[cat], d)
	}

	totalChunks := 0
	for _, d := range filtered {
		totalChunks += d.ChunkCount
	}

	c.JSON(http.StatusOK, gin.H{
		"documents":    filtered,
		"groups":       groups,
		"total":        len(filtered),
		"total_chunks": totalChunks,
	})
}