llvm-project/clang-tools-extra/clangd/SourceCode.cpp

234 lines
8.2 KiB
C++
Raw Normal View History

//===--- SourceCode.h - Manipulating source code as strings -----*- C++ -*-===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
#include "SourceCode.h"
#include "Logger.h"
#include "clang/AST/ASTContext.h"
#include "clang/Basic/SourceManager.h"
#include "clang/Lex/Lexer.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/Path.h"
namespace clang {
namespace clangd {
using namespace llvm;
[clangd] Fix unicode handling, using UTF-16 where LSP requires it. Summary: The Language Server Protocol unfortunately mandates that locations in files be represented by line/column pairs, where the "column" is actually an index into the UTF-16-encoded text of the line. (This is because VSCode is written in JavaScript, which is UTF-16-native). Internally clangd treats source files at UTF-8, the One True Encoding, and generally deals with byte offsets (though there are exceptions). Before this patch, conversions between offsets and LSP Position pretended that Position.character was UTF-8 bytes, which is only true for ASCII lines. Now we examine the text to convert correctly (but don't actually need to transcode it, due to some nice details of the encodings). The updated functions in SourceCode are the blessed way to interact with the Position.character field, and anything else is likely to be wrong. So I also updated the other accesses: - CodeComplete needs a "clang-style" line/column, with column in utf-8 bytes. This is now converted via Position -> offset -> clang line/column (a new function is added to SourceCode.h for the second conversion). - getBeginningOfIdentifier skipped backwards in UTF-16 space, which is will behave badly when it splits a surrogate pair. Skipping backwards in UTF-8 coordinates gives the lexer a fighting chance of getting this right. While here, I clarified(?) the logic comments, fixed a bug with identifiers containing digits, simplified the signature slightly and added a test. This seems likely to cause problems with editors that have the same bug, and treat the protocol as if columns are UTF-8 bytes. But we can find and fix those. Reviewers: hokein Subscribers: klimek, ilya-biryukov, ioeric, MaskRay, jkorous, cfe-commits Differential Revision: https://reviews.llvm.org/D46035 llvm-svn: 331029
2018-04-27 19:59:28 +08:00
// Here be dragons. LSP positions use columns measured in *UTF-16 code units*!
// Clangd uses UTF-8 and byte-offsets internally, so conversion is nontrivial.
// Iterates over unicode codepoints in the (UTF-8) string. For each,
// invokes CB(UTF-8 length, UTF-16 length), and breaks if it returns true.
// Returns true if CB returned true, false if we hit the end of string.
template <typename Callback>
static bool iterateCodepoints(StringRef U8, const Callback &CB) {
for (size_t I = 0; I < U8.size();) {
unsigned char C = static_cast<unsigned char>(U8[I]);
if (LLVM_LIKELY(!(C & 0x80))) { // ASCII character.
if (CB(1, 1))
return true;
++I;
continue;
}
// This convenient property of UTF-8 holds for all non-ASCII characters.
size_t UTF8Length = countLeadingOnes(C);
// 0xxx is ASCII, handled above. 10xxx is a trailing byte, invalid here.
// 11111xxx is not valid UTF-8 at all. Assert because it's probably our bug.
assert((UTF8Length >= 2 && UTF8Length <= 4) &&
"Invalid UTF-8, or transcoding bug?");
I += UTF8Length; // Skip over all trailing bytes.
// A codepoint takes two UTF-16 code unit if it's astral (outside BMP).
// Astral codepoints are encoded as 4 bytes in UTF-8 (11110xxx ...)
if (CB(UTF8Length, UTF8Length == 4 ? 2 : 1))
return true;
}
return false;
}
// Returns the offset into the string that matches \p Units UTF-16 code units.
// Conceptually, this converts to UTF-16, truncates to CodeUnits, converts back
// to UTF-8, and returns the length in bytes.
static size_t measureUTF16(StringRef U8, int U16Units, bool &Valid) {
size_t Result = 0;
Valid = U16Units == 0 || iterateCodepoints(U8, [&](int U8Len, int U16Len) {
Result += U8Len;
U16Units -= U16Len;
return U16Units <= 0;
});
if (U16Units < 0) // Offset was into the middle of a surrogate pair.
Valid = false;
// Don't return an out-of-range index if we overran.
return std::min(Result, U8.size());
}
// Counts the number of UTF-16 code units needed to represent a string.
// Like most strings in clangd, the input is UTF-8 encoded.
static size_t utf16Len(StringRef U8) {
// A codepoint takes two UTF-16 code unit if it's astral (outside BMP).
// Astral codepoints are encoded as 4 bytes in UTF-8, starting with 11110xxx.
size_t Count = 0;
iterateCodepoints(U8, [&](int U8Len, int U16Len) {
Count += U16Len;
return false;
});
return Count;
}
llvm::Expected<size_t> positionToOffset(StringRef Code, Position P,
bool AllowColumnsBeyondLineLength) {
if (P.line < 0)
return llvm::make_error<llvm::StringError>(
llvm::formatv("Line value can't be negative ({0})", P.line),
llvm::errc::invalid_argument);
if (P.character < 0)
return llvm::make_error<llvm::StringError>(
llvm::formatv("Character value can't be negative ({0})", P.character),
llvm::errc::invalid_argument);
size_t StartOfLine = 0;
for (int I = 0; I != P.line; ++I) {
size_t NextNL = Code.find('\n', StartOfLine);
if (NextNL == StringRef::npos)
return llvm::make_error<llvm::StringError>(
llvm::formatv("Line value is out of range ({0})", P.line),
llvm::errc::invalid_argument);
StartOfLine = NextNL + 1;
}
size_t NextNL = Code.find('\n', StartOfLine);
if (NextNL == StringRef::npos)
NextNL = Code.size();
[clangd] Fix unicode handling, using UTF-16 where LSP requires it. Summary: The Language Server Protocol unfortunately mandates that locations in files be represented by line/column pairs, where the "column" is actually an index into the UTF-16-encoded text of the line. (This is because VSCode is written in JavaScript, which is UTF-16-native). Internally clangd treats source files at UTF-8, the One True Encoding, and generally deals with byte offsets (though there are exceptions). Before this patch, conversions between offsets and LSP Position pretended that Position.character was UTF-8 bytes, which is only true for ASCII lines. Now we examine the text to convert correctly (but don't actually need to transcode it, due to some nice details of the encodings). The updated functions in SourceCode are the blessed way to interact with the Position.character field, and anything else is likely to be wrong. So I also updated the other accesses: - CodeComplete needs a "clang-style" line/column, with column in utf-8 bytes. This is now converted via Position -> offset -> clang line/column (a new function is added to SourceCode.h for the second conversion). - getBeginningOfIdentifier skipped backwards in UTF-16 space, which is will behave badly when it splits a surrogate pair. Skipping backwards in UTF-8 coordinates gives the lexer a fighting chance of getting this right. While here, I clarified(?) the logic comments, fixed a bug with identifiers containing digits, simplified the signature slightly and added a test. This seems likely to cause problems with editors that have the same bug, and treat the protocol as if columns are UTF-8 bytes. But we can find and fix those. Reviewers: hokein Subscribers: klimek, ilya-biryukov, ioeric, MaskRay, jkorous, cfe-commits Differential Revision: https://reviews.llvm.org/D46035 llvm-svn: 331029
2018-04-27 19:59:28 +08:00
bool Valid;
size_t ByteOffsetInLine = measureUTF16(
Code.substr(StartOfLine, NextNL - StartOfLine), P.character, Valid);
if (!Valid && !AllowColumnsBeyondLineLength)
return llvm::make_error<llvm::StringError>(
[clangd] Fix unicode handling, using UTF-16 where LSP requires it. Summary: The Language Server Protocol unfortunately mandates that locations in files be represented by line/column pairs, where the "column" is actually an index into the UTF-16-encoded text of the line. (This is because VSCode is written in JavaScript, which is UTF-16-native). Internally clangd treats source files at UTF-8, the One True Encoding, and generally deals with byte offsets (though there are exceptions). Before this patch, conversions between offsets and LSP Position pretended that Position.character was UTF-8 bytes, which is only true for ASCII lines. Now we examine the text to convert correctly (but don't actually need to transcode it, due to some nice details of the encodings). The updated functions in SourceCode are the blessed way to interact with the Position.character field, and anything else is likely to be wrong. So I also updated the other accesses: - CodeComplete needs a "clang-style" line/column, with column in utf-8 bytes. This is now converted via Position -> offset -> clang line/column (a new function is added to SourceCode.h for the second conversion). - getBeginningOfIdentifier skipped backwards in UTF-16 space, which is will behave badly when it splits a surrogate pair. Skipping backwards in UTF-8 coordinates gives the lexer a fighting chance of getting this right. While here, I clarified(?) the logic comments, fixed a bug with identifiers containing digits, simplified the signature slightly and added a test. This seems likely to cause problems with editors that have the same bug, and treat the protocol as if columns are UTF-8 bytes. But we can find and fix those. Reviewers: hokein Subscribers: klimek, ilya-biryukov, ioeric, MaskRay, jkorous, cfe-commits Differential Revision: https://reviews.llvm.org/D46035 llvm-svn: 331029
2018-04-27 19:59:28 +08:00
llvm::formatv("UTF-16 offset {0} is invalid for line {1}", P.character,
P.line),
llvm::errc::invalid_argument);
[clangd] Fix unicode handling, using UTF-16 where LSP requires it. Summary: The Language Server Protocol unfortunately mandates that locations in files be represented by line/column pairs, where the "column" is actually an index into the UTF-16-encoded text of the line. (This is because VSCode is written in JavaScript, which is UTF-16-native). Internally clangd treats source files at UTF-8, the One True Encoding, and generally deals with byte offsets (though there are exceptions). Before this patch, conversions between offsets and LSP Position pretended that Position.character was UTF-8 bytes, which is only true for ASCII lines. Now we examine the text to convert correctly (but don't actually need to transcode it, due to some nice details of the encodings). The updated functions in SourceCode are the blessed way to interact with the Position.character field, and anything else is likely to be wrong. So I also updated the other accesses: - CodeComplete needs a "clang-style" line/column, with column in utf-8 bytes. This is now converted via Position -> offset -> clang line/column (a new function is added to SourceCode.h for the second conversion). - getBeginningOfIdentifier skipped backwards in UTF-16 space, which is will behave badly when it splits a surrogate pair. Skipping backwards in UTF-8 coordinates gives the lexer a fighting chance of getting this right. While here, I clarified(?) the logic comments, fixed a bug with identifiers containing digits, simplified the signature slightly and added a test. This seems likely to cause problems with editors that have the same bug, and treat the protocol as if columns are UTF-8 bytes. But we can find and fix those. Reviewers: hokein Subscribers: klimek, ilya-biryukov, ioeric, MaskRay, jkorous, cfe-commits Differential Revision: https://reviews.llvm.org/D46035 llvm-svn: 331029
2018-04-27 19:59:28 +08:00
return StartOfLine + ByteOffsetInLine;
}
Position offsetToPosition(StringRef Code, size_t Offset) {
Offset = std::min(Code.size(), Offset);
StringRef Before = Code.substr(0, Offset);
int Lines = Before.count('\n');
size_t PrevNL = Before.rfind('\n');
size_t StartOfLine = (PrevNL == StringRef::npos) ? 0 : (PrevNL + 1);
Position Pos;
Pos.line = Lines;
[clangd] Fix unicode handling, using UTF-16 where LSP requires it. Summary: The Language Server Protocol unfortunately mandates that locations in files be represented by line/column pairs, where the "column" is actually an index into the UTF-16-encoded text of the line. (This is because VSCode is written in JavaScript, which is UTF-16-native). Internally clangd treats source files at UTF-8, the One True Encoding, and generally deals with byte offsets (though there are exceptions). Before this patch, conversions between offsets and LSP Position pretended that Position.character was UTF-8 bytes, which is only true for ASCII lines. Now we examine the text to convert correctly (but don't actually need to transcode it, due to some nice details of the encodings). The updated functions in SourceCode are the blessed way to interact with the Position.character field, and anything else is likely to be wrong. So I also updated the other accesses: - CodeComplete needs a "clang-style" line/column, with column in utf-8 bytes. This is now converted via Position -> offset -> clang line/column (a new function is added to SourceCode.h for the second conversion). - getBeginningOfIdentifier skipped backwards in UTF-16 space, which is will behave badly when it splits a surrogate pair. Skipping backwards in UTF-8 coordinates gives the lexer a fighting chance of getting this right. While here, I clarified(?) the logic comments, fixed a bug with identifiers containing digits, simplified the signature slightly and added a test. This seems likely to cause problems with editors that have the same bug, and treat the protocol as if columns are UTF-8 bytes. But we can find and fix those. Reviewers: hokein Subscribers: klimek, ilya-biryukov, ioeric, MaskRay, jkorous, cfe-commits Differential Revision: https://reviews.llvm.org/D46035 llvm-svn: 331029
2018-04-27 19:59:28 +08:00
Pos.character = utf16Len(Before.substr(StartOfLine));
return Pos;
}
Position sourceLocToPosition(const SourceManager &SM, SourceLocation Loc) {
[clangd] Fix unicode handling, using UTF-16 where LSP requires it. Summary: The Language Server Protocol unfortunately mandates that locations in files be represented by line/column pairs, where the "column" is actually an index into the UTF-16-encoded text of the line. (This is because VSCode is written in JavaScript, which is UTF-16-native). Internally clangd treats source files at UTF-8, the One True Encoding, and generally deals with byte offsets (though there are exceptions). Before this patch, conversions between offsets and LSP Position pretended that Position.character was UTF-8 bytes, which is only true for ASCII lines. Now we examine the text to convert correctly (but don't actually need to transcode it, due to some nice details of the encodings). The updated functions in SourceCode are the blessed way to interact with the Position.character field, and anything else is likely to be wrong. So I also updated the other accesses: - CodeComplete needs a "clang-style" line/column, with column in utf-8 bytes. This is now converted via Position -> offset -> clang line/column (a new function is added to SourceCode.h for the second conversion). - getBeginningOfIdentifier skipped backwards in UTF-16 space, which is will behave badly when it splits a surrogate pair. Skipping backwards in UTF-8 coordinates gives the lexer a fighting chance of getting this right. While here, I clarified(?) the logic comments, fixed a bug with identifiers containing digits, simplified the signature slightly and added a test. This seems likely to cause problems with editors that have the same bug, and treat the protocol as if columns are UTF-8 bytes. But we can find and fix those. Reviewers: hokein Subscribers: klimek, ilya-biryukov, ioeric, MaskRay, jkorous, cfe-commits Differential Revision: https://reviews.llvm.org/D46035 llvm-svn: 331029
2018-04-27 19:59:28 +08:00
// We use the SourceManager's line tables, but its column number is in bytes.
FileID FID;
unsigned Offset;
std::tie(FID, Offset) = SM.getDecomposedSpellingLoc(Loc);
Position P;
[clangd] Fix unicode handling, using UTF-16 where LSP requires it. Summary: The Language Server Protocol unfortunately mandates that locations in files be represented by line/column pairs, where the "column" is actually an index into the UTF-16-encoded text of the line. (This is because VSCode is written in JavaScript, which is UTF-16-native). Internally clangd treats source files at UTF-8, the One True Encoding, and generally deals with byte offsets (though there are exceptions). Before this patch, conversions between offsets and LSP Position pretended that Position.character was UTF-8 bytes, which is only true for ASCII lines. Now we examine the text to convert correctly (but don't actually need to transcode it, due to some nice details of the encodings). The updated functions in SourceCode are the blessed way to interact with the Position.character field, and anything else is likely to be wrong. So I also updated the other accesses: - CodeComplete needs a "clang-style" line/column, with column in utf-8 bytes. This is now converted via Position -> offset -> clang line/column (a new function is added to SourceCode.h for the second conversion). - getBeginningOfIdentifier skipped backwards in UTF-16 space, which is will behave badly when it splits a surrogate pair. Skipping backwards in UTF-8 coordinates gives the lexer a fighting chance of getting this right. While here, I clarified(?) the logic comments, fixed a bug with identifiers containing digits, simplified the signature slightly and added a test. This seems likely to cause problems with editors that have the same bug, and treat the protocol as if columns are UTF-8 bytes. But we can find and fix those. Reviewers: hokein Subscribers: klimek, ilya-biryukov, ioeric, MaskRay, jkorous, cfe-commits Differential Revision: https://reviews.llvm.org/D46035 llvm-svn: 331029
2018-04-27 19:59:28 +08:00
P.line = static_cast<int>(SM.getLineNumber(FID, Offset)) - 1;
bool Invalid = false;
StringRef Code = SM.getBufferData(FID, &Invalid);
if (!Invalid) {
auto ColumnInBytes = SM.getColumnNumber(FID, Offset) - 1;
auto LineSoFar = Code.substr(Offset - ColumnInBytes, ColumnInBytes);
P.character = utf16Len(LineSoFar);
}
return P;
}
Range halfOpenToRange(const SourceManager &SM, CharSourceRange R) {
// Clang is 1-based, LSP uses 0-based indexes.
Position Begin = sourceLocToPosition(SM, R.getBegin());
Position End = sourceLocToPosition(SM, R.getEnd());
return {Begin, End};
}
[clangd] Fix unicode handling, using UTF-16 where LSP requires it. Summary: The Language Server Protocol unfortunately mandates that locations in files be represented by line/column pairs, where the "column" is actually an index into the UTF-16-encoded text of the line. (This is because VSCode is written in JavaScript, which is UTF-16-native). Internally clangd treats source files at UTF-8, the One True Encoding, and generally deals with byte offsets (though there are exceptions). Before this patch, conversions between offsets and LSP Position pretended that Position.character was UTF-8 bytes, which is only true for ASCII lines. Now we examine the text to convert correctly (but don't actually need to transcode it, due to some nice details of the encodings). The updated functions in SourceCode are the blessed way to interact with the Position.character field, and anything else is likely to be wrong. So I also updated the other accesses: - CodeComplete needs a "clang-style" line/column, with column in utf-8 bytes. This is now converted via Position -> offset -> clang line/column (a new function is added to SourceCode.h for the second conversion). - getBeginningOfIdentifier skipped backwards in UTF-16 space, which is will behave badly when it splits a surrogate pair. Skipping backwards in UTF-8 coordinates gives the lexer a fighting chance of getting this right. While here, I clarified(?) the logic comments, fixed a bug with identifiers containing digits, simplified the signature slightly and added a test. This seems likely to cause problems with editors that have the same bug, and treat the protocol as if columns are UTF-8 bytes. But we can find and fix those. Reviewers: hokein Subscribers: klimek, ilya-biryukov, ioeric, MaskRay, jkorous, cfe-commits Differential Revision: https://reviews.llvm.org/D46035 llvm-svn: 331029
2018-04-27 19:59:28 +08:00
std::pair<size_t, size_t> offsetToClangLineColumn(StringRef Code,
size_t Offset) {
Offset = std::min(Code.size(), Offset);
StringRef Before = Code.substr(0, Offset);
int Lines = Before.count('\n');
size_t PrevNL = Before.rfind('\n');
size_t StartOfLine = (PrevNL == StringRef::npos) ? 0 : (PrevNL + 1);
return {Lines + 1, Offset - StartOfLine + 1};
}
std::pair<llvm::StringRef, llvm::StringRef>
splitQualifiedName(llvm::StringRef QName) {
size_t Pos = QName.rfind("::");
if (Pos == llvm::StringRef::npos)
return {StringRef(), QName};
return {QName.substr(0, Pos + 2), QName.substr(Pos + 2)};
}
TextEdit replacementToEdit(StringRef Code, const tooling::Replacement &R) {
Range ReplacementRange = {
offsetToPosition(Code, R.getOffset()),
offsetToPosition(Code, R.getOffset() + R.getLength())};
return {ReplacementRange, R.getReplacementText()};
}
std::vector<TextEdit> replacementsToEdits(StringRef Code,
const tooling::Replacements &Repls) {
std::vector<TextEdit> Edits;
for (const auto &R : Repls)
Edits.push_back(replacementToEdit(Code, R));
return Edits;
}
llvm::Optional<std::string> getRealPath(const FileEntry *F,
const SourceManager &SourceMgr) {
// Ideally, we get the real path from the FileEntry object.
SmallString<128> FilePath = F->tryGetRealPathName();
if (!FilePath.empty()) {
return FilePath.str().str();
}
// Otherwise, we try to compute ourselves.
vlog("FileEntry for {0} did not contain the real path.", F->getName());
llvm::SmallString<128> Path = F->getName();
if (!llvm::sys::path::is_absolute(Path)) {
if (!SourceMgr.getFileManager().makeAbsolutePath(Path)) {
log("Could not turn relative path to absolute: {0}", Path);
return llvm::None;
}
}
llvm::SmallString<128> RealPath;
if (SourceMgr.getFileManager().getVirtualFileSystem()->getRealPath(
Path, RealPath)) {
log("Could not compute real path: {0}", Path);
return Path.str().str();
}
return RealPath.str().str();
}
TextEdit toTextEdit(const FixItHint &FixIt, const SourceManager &M,
const LangOptions &L) {
TextEdit Result;
Result.range =
halfOpenToRange(M, Lexer::makeFileCharRange(FixIt.RemoveRange, M, L));
Result.newText = FixIt.CodeToInsert;
return Result;
}
bool IsRangeConsecutive(const Range &Left, const Range &Right) {
return Left.end.line == Right.start.line &&
Left.end.character == Right.start.character;
}
} // namespace clangd
} // namespace clang