From 2d636a05e41cef9131b1696624436358aba363b2 Mon Sep 17 00:00:00 2001
From: Tim Barron <tjbarron@google.com>
Date: Fri, 13 Aug 2021 11:28:00 -0700
Subject: Sync from upstream.

Descriptions:
==================
Add Initialization Count marker file to break out of crash loops.
==================
Delete the Simple Normalizer.
==================
Add submatch information to identify a submatch within a document.
==================

Bug: 195720764
Change-Id: I27bcc66c400da625d595609049d281b809dd6028
---
 proto/icing/proto/logging.proto |  6 +++++-
 proto/icing/proto/search.proto  | 32 +++++++++++++++++++++++++++++++-
 2 files changed, 36 insertions(+), 2 deletions(-)

(limited to 'proto')

diff --git a/proto/icing/proto/logging.proto b/proto/icing/proto/logging.proto
index 4dcfecf..7abbf4a 100644
--- a/proto/icing/proto/logging.proto
+++ b/proto/icing/proto/logging.proto
@@ -23,7 +23,7 @@ option java_multiple_files = true;
 option objc_class_prefix = "ICNG";
 
 // Stats of the top-level function IcingSearchEngine::Initialize().
-// Next tag: 11
+// Next tag: 12
 message InitializeStatsProto {
   // Overall time used for the function call.
   optional int32 latency_ms = 1;
@@ -92,6 +92,10 @@ message InitializeStatsProto {
 
   // Number of schema types currently in schema store.
   optional int32 num_schema_types = 10;
+
+  // Number of consecutive initialization failures that immediately preceded
+  // this initialization.
+  optional int32 num_previous_init_failures = 11;
 }
 
 // Stats of the top-level function IcingSearchEngine::Put().
diff --git a/proto/icing/proto/search.proto b/proto/icing/proto/search.proto
index 66fdbe6..544995e 100644
--- a/proto/icing/proto/search.proto
+++ b/proto/icing/proto/search.proto
@@ -136,27 +136,57 @@ message ResultSpecProto {
 }
 
 // The representation of a single match within a DocumentProto property.
-// Next tag: 10
+//
+// Example : A document whose content is "Necesito comprar comida mañana." and a
+// query for "mana" with window=15
+// Next tag: 12
 message SnippetMatchProto {
   // The index of the byte in the string at which the match begins and the
   // length in bytes of the match.
+  //
+  // For the example above, the values of these fields would be
+  // exact_match_byte_position=24, exact_match_byte_length=7 "mañana"
   optional int32 exact_match_byte_position = 2;
   optional int32 exact_match_byte_length = 3;
 
+  // The length in bytes of the subterm that matches the query. The beginning of
+  // the submatch is the same as exact_match_byte_position.
+  //
+  // For the example above, the value of this field would be 5. With
+  // exact_match_byte_position=24 above, it would produce the substring "maña"
+  optional int32 submatch_byte_length = 10;
+
   // The index of the UTF-16 code unit in the string at which the match begins
   // and the length in UTF-16 code units of the match. This is for use with
   // UTF-16 encoded strings like Java.lang.String.
+  //
+  // For the example above, the values of these fields would be
+  // exact_match_utf16_position=24, exact_match_utf16_length=6 "mañana"
   optional int32 exact_match_utf16_position = 6;
   optional int32 exact_match_utf16_length = 7;
 
+  // The length in UTF-16 code units of the subterm that matches the query. The
+  // beginning of the submatch is the same as exact_match_utf16_position. This
+  // is for use with UTF-16 encoded strings like Java.lang.String.
+  //
+  // For the example above, the value of this field would be 4. With
+  // exact_match_utf16_position=24 above, it would produce the substring "maña"
+  optional int32 submatch_utf16_length = 11;
+
   // The index of the byte in the string at which the suggested snippet window
   // begins and the length in bytes of the window.
+  //
+  // For the example above, the values of these fields would be
+  // window_byte_position=17, window_byte_length=15 "comida mañana."
   optional int32 window_byte_position = 4;
   optional int32 window_byte_length = 5;
 
   // The index of the UTF-16 code unit in the string at which the suggested
   // snippet window begins and the length in UTF-16 code units of the window.
   // This is for use with UTF-16 encoded strings like Java.lang.String.
+  //
+  // For the example above, the values of these fields would be
+  // window_utf16_position=17, window_utf16_length=14 "comida mañana."
   optional int32 window_utf16_position = 8;
   optional int32 window_utf16_length = 9;
 
-- 
cgit v1.2.3