mattermost/server/platform/services/docextractor/documents.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
}