mattermost-community-enterp.../platform/services/docextractor/documents.go
Claude ec1f89217a Merge: Complete Mattermost Server with Community Enterprise
Full Mattermost server source with integrated Community Enterprise features.
Includes vendor directory for offline/air-gapped builds.

Structure:
- enterprise-impl/: Enterprise feature implementations
- enterprise-community/: Init files that register implementations
- enterprise/: Bridge imports (community_imports.go)
- vendor/: All dependencies for offline builds

Build (online):
  go build ./cmd/mattermost

Build (offline/air-gapped):
  go build -mod=vendor ./cmd/mattermost

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2025-12-17 23:59:07 +09:00

60 lines
1.4 KiB
Go

// Copyright (c) 2015-present Mattermost, Inc. All Rights Reserved.
// See LICENSE.txt for license information.
package docextractor
import (
"errors"
"io"
"path"
"strings"
"code.sajari.com/docconv/v2"
)
type documentExtractor struct{}
var doconvConverterByExtensions = map[string]func(io.Reader) (string, map[string]string, error){
"doc": docconv.ConvertDoc,
"docx": docconv.ConvertDocx,
"pptx": docconv.ConvertPptx,
"odt": docconv.ConvertODT,
"html": func(r io.Reader) (string, map[string]string, error) { return docconv.ConvertHTML(r, true) },
// Temporarily disabled to avoid crashes on malicious .pages files
// "pages": docconv.ConvertPages,
"rtf": docconv.ConvertRTF,
"pdf": docconv.ConvertPDF,
}
func (de *documentExtractor) Name() string {
return "documentExtractor"
}
func (de *documentExtractor) Match(filename string) bool {
extension := strings.TrimPrefix(path.Ext(filename), ".")
_, ok := doconvConverterByExtensions[extension]
return ok
}
func (de *documentExtractor) Extract(filename string, r io.ReadSeeker) (out string, outErr error) {
defer func() {
if r := recover(); r != nil {
out = ""
outErr = errors.New("error extracting document text")
}
}()
extension := strings.TrimPrefix(path.Ext(filename), ".")
converter, ok := doconvConverterByExtensions[extension]
if !ok {
return "", errors.New("unknown converter")
}
text, _, err := converter(r)
if err != nil {
return "", err
}
return text, nil
}