From efd1a4f717afa83d3d3609f0d70e4da151a8dc9b Mon Sep 17 00:00:00 2001
From: tobi <31960611+tsmethurst@users.noreply.github.com>
Date: Mon, 16 Sep 2024 14:00:23 +0200
Subject: [bugfix] Use better plaintext representation of status for filtering
 (#3301)

* [bugfix] Use better plaintext representation of status for filtering

* add new deps to readme

* lint

* update tests

* update regexes

* address review comments

* remove now unused xxhash

* whoops, wrong logger

* Merge branch 'main' into status_filtering_bugfix

* put cache in caches struct

* pain
---
 internal/typeutils/util.go | 62 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)

(limited to 'internal/typeutils/util.go')

diff --git a/internal/typeutils/util.go b/internal/typeutils/util.go
index 3441e89a9..3a867ba35 100644
--- a/internal/typeutils/util.go
+++ b/internal/typeutils/util.go
@@ -27,6 +27,7 @@ import (
 	"strconv"
 	"strings"
 
+	"github.com/k3a/html2text"
 	apimodel "github.com/superseriousbusiness/gotosocial/internal/api/model"
 	"github.com/superseriousbusiness/gotosocial/internal/config"
 	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
@@ -284,3 +285,64 @@ func ContentToContentLanguage(
 
 	return contentStr, langTagStr
 }
+
+// filterableFields returns text fields from
+// a status that we might want to filter on:
+//
+//   - content warning
+//   - content (converted to plaintext from HTML)
+//   - media descriptions
+//   - poll options
+//
+// Each field should be filtered separately.
+// This avoids scenarios where false-positive
+// multiple-word matches can be made by matching
+// the last word of one field + the first word
+// of the next field together.
+func filterableFields(s *gtsmodel.Status) []string {
+	// Estimate length of fields.
+	fieldCount := 2 + len(s.Attachments)
+	if s.Poll != nil {
+		fieldCount += len(s.Poll.Options)
+	}
+	fields := make([]string, 0, fieldCount)
+
+	// Content warning / title.
+	if s.ContentWarning != "" {
+		fields = append(fields, s.ContentWarning)
+	}
+
+	// Status content. Though we have raw text
+	// available for statuses created on our
+	// instance, use the html2text version to
+	// remove markdown-formatting characters
+	// and ensure more consistent filtering.
+	if s.Content != "" {
+		text := html2text.HTML2TextWithOptions(
+			s.Content,
+			html2text.WithLinksInnerText(),
+			html2text.WithUnixLineBreaks(),
+		)
+		if text != "" {
+			fields = append(fields, text)
+		}
+	}
+
+	// Media descriptions.
+	for _, attachment := range s.Attachments {
+		if attachment.Description != "" {
+			fields = append(fields, attachment.Description)
+		}
+	}
+
+	// Poll options.
+	if s.Poll != nil {
+		for _, opt := range s.Poll.Options {
+			if opt != "" {
+				fields = append(fields, opt)
+			}
+		}
+	}
+
+	return fields
+}
-- 
cgit v1.2.3