From 42f88e3b7250b978f544854c825dbf7f71ed44c7 Mon Sep 17 00:00:00 2001 From: levlam Date: Mon, 11 May 2020 01:31:29 +0300 Subject: [PATCH] Remove whitespaces at the beginning of some entities. GitOrigin-RevId: b0597524571c15b3ffbfec613cb856bfa6e7a224 --- td/telegram/MessageEntity.cpp | 19 ++++++++++++++ test/message_entities.cpp | 48 +++++++++++++++++++++++++---------- 2 files changed, 54 insertions(+), 13 deletions(-) diff --git a/td/telegram/MessageEntity.cpp b/td/telegram/MessageEntity.cpp index 87289999..007af6cd 100644 --- a/td/telegram/MessageEntity.cpp +++ b/td/telegram/MessageEntity.cpp @@ -3543,6 +3543,25 @@ static std::pair remove_invalid_entities(const string &text, vect break; } + if (!nested_entities_stack.empty() && nested_entities_stack.back()->offset == utf16_offset && + (text[pos] == '\n' || text[pos] == ' ')) { + // entities was fixed, so there can't be more than one splittable entity of each type, one blockquote and + // one continuous entity for the given offset + for (size_t i = nested_entities_stack.size(); i > 0; i--) { + auto *entity = nested_entities_stack[i - 1]; + if (entity->offset != utf16_offset || entity->type == MessageEntity::Type::TextUrl || + entity->type == MessageEntity::Type::MentionName || is_pre_entity(entity->type)) { + break; + } + entity->offset++; + entity->length--; + if (entity->length == 0) { + CHECK(i == nested_entities_stack.size()); + nested_entities_stack.pop_back(); + } + } + } + auto c = static_cast(text[pos]); switch (c) { case '\n': diff --git a/test/message_entities.cpp b/test/message_entities.cpp index e29e3059..e32ee0f8 100644 --- a/test/message_entities.cpp +++ b/test/message_entities.cpp @@ -625,26 +625,35 @@ TEST(MessageEntities, fix_formatted_text) { str += "a \r\n "; fixed_str += "a \n "; + for (td::int32 i = 33; i <= 35; i++) { + td::vector entities; + entities.emplace_back(td::MessageEntity::Type::Pre, 0, i); + + td::vector fixed_entities = entities; + fixed_entities.back().length = i - 1; + check_fix_formatted_text(str, entities, fixed_str, fixed_entities, true, false, false, true); + + td::string expected_str = fixed_str.substr(0, 33); + fixed_entities.back().length = i == 33 ? 32 : 33; + check_fix_formatted_text(str, entities, expected_str, fixed_entities, false, false, false, false); + } + for (td::int32 i = 33; i <= 35; i++) { td::vector entities; entities.emplace_back(td::MessageEntity::Type::Bold, 0, i); td::vector fixed_entities; if (i != 33) { - fixed_entities = entities; - fixed_entities.back().length = i - 1; + fixed_entities.emplace_back(td::MessageEntity::Type::Bold, 32, i - 33); } check_fix_formatted_text(str, entities, fixed_str, fixed_entities, true, false, false, true); td::string expected_str; if (i != 33) { - fixed_entities = entities; - fixed_entities.back().length = 33; - expected_str = fixed_str.substr(0, 33); - } else { - fixed_entities.clear(); - expected_str = "a"; + fixed_entities.back().offset = 0; + fixed_entities.back().length = 1; } + expected_str = "a"; check_fix_formatted_text(str, entities, expected_str, fixed_entities, false, false, false, false); } @@ -702,6 +711,12 @@ TEST(MessageEntities, fix_formatted_text) { while (static_cast(fixed_offset + fixed_length) > fixed_str.size()) { fixed_length--; } + if (type == td::MessageEntity::Type::Bold || type == td::MessageEntity::Type::Url) { + while (fixed_length > 0 && (fixed_str[fixed_offset] == ' ' || fixed_str[fixed_offset] == '\n')) { + fixed_offset++; + fixed_length--; + } + } td::vector entities; entities.emplace_back(type, offset, length); @@ -727,22 +742,29 @@ TEST(MessageEntities, fix_formatted_text) { for (td::int32 offset = -10; offset <= 10; offset++) { td::vector entities; entities.emplace_back(td::MessageEntity::Type::Bold, offset, length); - td::vector fixed_entities; if (length < 0 || offset < 0 || (length > 0 && static_cast(length + offset) > str.size())) { check_fix_formatted_text(str, entities, true, false, false, false); check_fix_formatted_text(str, entities, false, false, false, true); continue; } - if (length > 0 && (length >= 2 || offset != 3)) { - fixed_entities.emplace_back(td::MessageEntity::Type::Bold, offset, length); + td::vector fixed_entities; + if (length > 0) { + if (offset == 3) { + if (length >= 2) { + fixed_entities.emplace_back(td::MessageEntity::Type::Bold, offset + 1, length - 1); + } + } else { + fixed_entities.emplace_back(td::MessageEntity::Type::Bold, offset, length); + } } + check_fix_formatted_text(str, entities, str, fixed_entities, true, false, false, false); check_fix_formatted_text(str, entities, str, fixed_entities, false, false, false, true); } } - str = "aba caba"; + str = "abadcaba"; for (td::int32 length = 1; length <= 7; length++) { for (td::int32 offset = 0; offset <= 8 - length; offset++) { for (td::int32 length2 = 1; length2 <= 7; length2++) { @@ -926,7 +948,7 @@ TEST(MessageEntities, fix_formatted_text) { check_fix_formatted_text("example.com a", {{td::MessageEntity::Type::Italic, 0, 13}}, "example.com a", {{td::MessageEntity::Type::Url, 0, 11}, {td::MessageEntity::Type::Italic, 0, 11}, - {td::MessageEntity::Type::Italic, 11, 2}}); + {td::MessageEntity::Type::Italic, 12, 1}}); check_fix_formatted_text("a example.com", {{td::MessageEntity::Type::Italic, 0, 13}}, "a example.com", {{td::MessageEntity::Type::Italic, 0, 2}, {td::MessageEntity::Type::Url, 2, 11},