| /* |
| * Copyright (C) 2017 The Android Open Source Project |
| * |
| * Licensed under the Apache License, Version 2.0 (the "License"); |
| * you may not use this file except in compliance with the License. |
| * You may obtain a copy of the License at |
| * |
| * http://www.apache.org/licenses/LICENSE-2.0 |
| * |
| * Unless required by applicable law or agreed to in writing, software |
| * distributed under the License is distributed on an "AS IS" BASIS, |
| * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| * See the License for the specific language governing permissions and |
| * limitations under the License. |
| */ |
| |
| #pragma once |
| |
| #include <stddef.h> |
| |
| #include <string> |
| #include <utility> |
| #include <vector> |
| |
| #include <android-base/logging.h> |
| #include <android-base/parseint.h> |
| #include <android-base/stringprintf.h> |
| #include <android-base/strings.h> |
| |
| using Range = std::pair<size_t, size_t>; |
| |
| class RangeSet { |
| public: |
| RangeSet() : blocks_(0) {} |
| |
| explicit RangeSet(std::vector<Range>&& pairs) { |
| CHECK_NE(pairs.size(), static_cast<size_t>(0)) << "Invalid number of tokens"; |
| |
| // Sanity check the input. |
| size_t result = 0; |
| for (const auto& range : pairs) { |
| CHECK_LT(range.first, range.second) |
| << "Empty or negative range: " << range.first << ", " << range.second; |
| size_t sz = range.second - range.first; |
| CHECK_LE(result, SIZE_MAX - sz) << "RangeSet size overflow"; |
| result += sz; |
| } |
| |
| ranges_ = pairs; |
| blocks_ = result; |
| } |
| |
| static RangeSet Parse(const std::string& range_text) { |
| std::vector<std::string> pieces = android::base::Split(range_text, ","); |
| CHECK_GE(pieces.size(), static_cast<size_t>(3)) << "Invalid range text: " << range_text; |
| |
| size_t num; |
| CHECK(android::base::ParseUint(pieces[0], &num, static_cast<size_t>(INT_MAX))) |
| << "Failed to parse the number of tokens: " << range_text; |
| |
| CHECK_NE(num, static_cast<size_t>(0)) << "Invalid number of tokens: " << range_text; |
| CHECK_EQ(num % 2, static_cast<size_t>(0)) << "Number of tokens must be even: " << range_text; |
| CHECK_EQ(num, pieces.size() - 1) << "Mismatching number of tokens: " << range_text; |
| |
| std::vector<Range> pairs; |
| for (size_t i = 0; i < num; i += 2) { |
| size_t first; |
| CHECK(android::base::ParseUint(pieces[i + 1], &first, static_cast<size_t>(INT_MAX))); |
| size_t second; |
| CHECK(android::base::ParseUint(pieces[i + 2], &second, static_cast<size_t>(INT_MAX))); |
| |
| pairs.emplace_back(first, second); |
| } |
| |
| return RangeSet(std::move(pairs)); |
| } |
| |
| std::string ToString() const { |
| if (ranges_.empty()) { |
| return ""; |
| } |
| std::string result = std::to_string(ranges_.size() * 2); |
| for (const auto& r : ranges_) { |
| result += android::base::StringPrintf(",%zu,%zu", r.first, r.second); |
| } |
| |
| return result; |
| } |
| |
| // Get the block number for the i-th (starting from 0) block in the RangeSet. |
| size_t GetBlockNumber(size_t idx) const { |
| CHECK_LT(idx, blocks_) << "Out of bound index " << idx << " (total blocks: " << blocks_ << ")"; |
| |
| for (const auto& range : ranges_) { |
| if (idx < range.second - range.first) { |
| return range.first + idx; |
| } |
| idx -= (range.second - range.first); |
| } |
| |
| CHECK(false) << "Failed to find block number for index " << idx; |
| return 0; // Unreachable, but to make compiler happy. |
| } |
| |
| // RangeSet has half-closed half-open bounds. For example, "3,5" contains blocks 3 and 4. So "3,5" |
| // and "5,7" are not overlapped. |
| bool Overlaps(const RangeSet& other) const { |
| for (const auto& range : ranges_) { |
| size_t start = range.first; |
| size_t end = range.second; |
| for (const auto& other_range : other.ranges_) { |
| size_t other_start = other_range.first; |
| size_t other_end = other_range.second; |
| // [start, end) vs [other_start, other_end) |
| if (!(other_start >= end || start >= other_end)) { |
| return true; |
| } |
| } |
| } |
| return false; |
| } |
| |
| // size() gives the number of Range's in this RangeSet. |
| size_t size() const { |
| return ranges_.size(); |
| } |
| |
| // blocks() gives the number of all blocks in this RangeSet. |
| size_t blocks() const { |
| return blocks_; |
| } |
| |
| // We provide const iterators only. |
| std::vector<Range>::const_iterator cbegin() const { |
| return ranges_.cbegin(); |
| } |
| |
| std::vector<Range>::const_iterator cend() const { |
| return ranges_.cend(); |
| } |
| |
| // Need to provide begin()/end() since range-based loop expects begin()/end(). |
| std::vector<Range>::const_iterator begin() const { |
| return ranges_.cbegin(); |
| } |
| |
| std::vector<Range>::const_iterator end() const { |
| return ranges_.cend(); |
| } |
| |
| // Reverse const iterators for MoveRange(). |
| std::vector<Range>::const_reverse_iterator crbegin() const { |
| return ranges_.crbegin(); |
| } |
| |
| std::vector<Range>::const_reverse_iterator crend() const { |
| return ranges_.crend(); |
| } |
| |
| const Range& operator[](size_t i) const { |
| return ranges_[i]; |
| } |
| |
| bool operator==(const RangeSet& other) const { |
| // The orders of Range's matter. "4,1,5,8,10" != "4,8,10,1,5". |
| return (ranges_ == other.ranges_); |
| } |
| |
| bool operator!=(const RangeSet& other) const { |
| return ranges_ != other.ranges_; |
| } |
| |
| protected: |
| // Actual limit for each value and the total number are both INT_MAX. |
| std::vector<Range> ranges_; |
| size_t blocks_; |
| }; |
| |
| static constexpr size_t kBlockSize = 4096; |
| |
| // The class is a sorted version of a RangeSet; and it's useful in imgdiff to split the input |
| // files when we're handling large zip files. Specifically, we can treat the input file as a |
| // continuous RangeSet (i.e. RangeSet("0-99") for a 100 blocks file); and break it down into |
| // several smaller chunks based on the zip entries. |
| |
| // For example, [source: 0-99] can be split into |
| // [split_src1: 10-29]; [split_src2: 40-49, 60-69]; [split_src3: 70-89] |
| // Here "10-29" simply means block 10th to block 29th with respect to the original input file. |
| // Also, note that the split sources should be mutual exclusive, but they don't need to cover |
| // every block in the original source. |
| class SortedRangeSet : public RangeSet { |
| public: |
| SortedRangeSet() {} |
| |
| // Ranges in the the set should be mutually exclusive; and they're sorted by the start block. |
| explicit SortedRangeSet(std::vector<Range>&& pairs) : RangeSet(std::move(pairs)) { |
| std::sort(ranges_.begin(), ranges_.end()); |
| } |
| |
| void Insert(const Range& to_insert) { |
| SortedRangeSet rs({ to_insert }); |
| Insert(rs); |
| } |
| |
| // Insert the input SortedRangeSet; keep the ranges sorted and merge the overlap ranges. |
| void Insert(const SortedRangeSet& rs) { |
| if (rs.size() == 0) { |
| return; |
| } |
| // Merge and sort the two RangeSets. |
| std::vector<Range> temp = std::move(ranges_); |
| std::copy(rs.begin(), rs.end(), std::back_inserter(temp)); |
| std::sort(temp.begin(), temp.end()); |
| |
| Clear(); |
| // Trim overlaps and insert the result back to ranges_. |
| Range to_insert = temp.front(); |
| for (auto it = temp.cbegin() + 1; it != temp.cend(); it++) { |
| if (it->first <= to_insert.second) { |
| to_insert.second = std::max(to_insert.second, it->second); |
| } else { |
| ranges_.push_back(to_insert); |
| blocks_ += (to_insert.second - to_insert.first); |
| to_insert = *it; |
| } |
| } |
| ranges_.push_back(to_insert); |
| blocks_ += (to_insert.second - to_insert.first); |
| } |
| |
| void Clear() { |
| blocks_ = 0; |
| ranges_.clear(); |
| } |
| |
| using RangeSet::Overlaps; |
| bool Overlaps(size_t start, size_t len) const { |
| RangeSet rs({ { start / kBlockSize, (start + len - 1) / kBlockSize + 1 } }); |
| return Overlaps(rs); |
| } |
| |
| // Compute the block range the file occupies, and insert that range. |
| void Insert(size_t start, size_t len) { |
| Range to_insert{ start / kBlockSize, (start + len - 1) / kBlockSize + 1 }; |
| Insert(to_insert); |
| } |
| |
| // Given an offset of the file, checks if the corresponding block (by considering the file as |
| // 0-based continuous block ranges) is covered by the SortedRangeSet. If so, returns the offset |
| // within this SortedRangeSet. |
| // |
| // For example, the 4106-th byte of a file is from block 1, assuming a block size of 4096-byte. |
| // The mapped offset within a SortedRangeSet("1-9 15-19") is 10. |
| // |
| // An offset of 65546 falls into the 16-th block in a file. Block 16 is contained as the 10-th |
| // item in SortedRangeSet("1-9 15-19"). So its data can be found at offset 40970 (i.e. 4096 * 10 |
| // + 10) in a range represented by this SortedRangeSet. |
| size_t GetOffsetInRangeSet(size_t old_offset) const { |
| size_t old_block_start = old_offset / kBlockSize; |
| size_t new_block_start = 0; |
| for (const auto& range : ranges_) { |
| // Find the index of old_block_start. |
| if (old_block_start >= range.second) { |
| new_block_start += (range.second - range.first); |
| } else if (old_block_start >= range.first) { |
| new_block_start += (old_block_start - range.first); |
| return (new_block_start * kBlockSize + old_offset % kBlockSize); |
| } else { |
| CHECK(false) << "block_start " << old_block_start |
| << " is missing between two ranges: " << this->ToString(); |
| return 0; |
| } |
| } |
| CHECK(false) << "block_start " << old_block_start |
| << " exceeds the limit of current RangeSet: " << this->ToString(); |
| return 0; |
| } |
| }; |