aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--Basic/SourceManager.cpp147
-rw-r--r--clang.xcodeproj/project.pbxproj2
-rw-r--r--include/clang/Basic/SourceManager.h10
3 files changed, 106 insertions, 53 deletions
diff --git a/Basic/SourceManager.cpp b/Basic/SourceManager.cpp
index be8eeee408..05ead2d786 100644
--- a/Basic/SourceManager.cpp
+++ b/Basic/SourceManager.cpp
@@ -13,6 +13,7 @@
#include "clang/Basic/SourceManager.h"
#include "clang/Basic/FileManager.h"
+#include "llvm/Support/Compiler.h"
#include "llvm/Support/MemoryBuffer.h"
#include "llvm/System/Path.h"
#include <algorithm>
@@ -236,6 +237,50 @@ std::string SourceManager::getSourceName(SourceLocation Loc) {
return getFileInfo(FileID)->Buffer->getBufferIdentifier();
}
+static void ComputeLineNumbers(FileInfo *FI) DISABLE_INLINE;
+static void ComputeLineNumbers(FileInfo *FI) {
+ const MemoryBuffer *Buffer = FI->Buffer;
+
+ // Find the file offsets of all of the *physical* source lines. This does
+ // not look at trigraphs, escaped newlines, or anything else tricky.
+ std::vector<unsigned> LineOffsets;
+
+ // Line #1 starts at char 0.
+ LineOffsets.push_back(0);
+
+ const unsigned char *Buf = (const unsigned char *)Buffer->getBufferStart();
+ const unsigned char *End = (const unsigned char *)Buffer->getBufferEnd();
+ unsigned Offs = 0;
+ while (1) {
+ // Skip over the contents of the line.
+ // TODO: Vectorize this? This is very performance sensitive for programs
+ // with lots of diagnostics and in -E mode.
+ const unsigned char *NextBuf = (const unsigned char *)Buf;
+ while (*NextBuf != '\n' && *NextBuf != '\r' && *NextBuf != '\0')
+ ++NextBuf;
+ Offs += NextBuf-Buf;
+ Buf = NextBuf;
+
+ if (Buf[0] == '\n' || Buf[0] == '\r') {
+ // If this is \n\r or \r\n, skip both characters.
+ if ((Buf[1] == '\n' || Buf[1] == '\r') && Buf[0] != Buf[1])
+ ++Offs, ++Buf;
+ ++Offs, ++Buf;
+ LineOffsets.push_back(Offs);
+ } else {
+ // Otherwise, this is a null. If end of file, exit.
+ if (Buf == End) break;
+ // Otherwise, skip the null.
+ ++Offs, ++Buf;
+ }
+ }
+ LineOffsets.push_back(Offs);
+
+ // Copy the offsets into the FileInfo structure.
+ FI->NumLines = LineOffsets.size();
+ FI->SourceLineCache = new unsigned[LineOffsets.size()];
+ std::copy(LineOffsets.begin(), LineOffsets.end(), FI->SourceLineCache);
+}
/// getLineNumber - Given a SourceLocation, return the physical line number
/// for the position indicated. This requires building and caching a table of
@@ -244,66 +289,66 @@ std::string SourceManager::getSourceName(SourceLocation Loc) {
unsigned SourceManager::getLineNumber(SourceLocation Loc) {
unsigned FileID = Loc.getFileID();
if (FileID == 0) return 0;
- FileInfo *FileInfo = getFileInfo(FileID);
+ FileInfo *FileInfo;
+
+ if (LastLineNoFileIDQuery == FileID)
+ FileInfo = LastLineNoFileInfo;
+ else
+ FileInfo = getFileInfo(FileID);
// If this is the first use of line information for this buffer, compute the
- /// SourceLineCache for it on demand.
- if (FileInfo->SourceLineCache == 0) {
- const MemoryBuffer *Buffer = FileInfo->Buffer;
-
- // Find the file offsets of all of the *physical* source lines. This does
- // not look at trigraphs, escaped newlines, or anything else tricky.
- std::vector<unsigned> LineOffsets;
-
- // Line #1 starts at char 0.
- LineOffsets.push_back(0);
-
- const unsigned char *Buf = (const unsigned char *)Buffer->getBufferStart();
- const unsigned char *End = (const unsigned char *)Buffer->getBufferEnd();
- unsigned Offs = 0;
- while (1) {
- // Skip over the contents of the line.
- // TODO: Vectorize this? This is very performance sensitive for programs
- // with lots of diagnostics and in -E mode.
- const unsigned char *NextBuf = (const unsigned char *)Buf;
- while (*NextBuf != '\n' && *NextBuf != '\r' && *NextBuf != '\0')
- ++NextBuf;
- Offs += NextBuf-Buf;
- Buf = NextBuf;
-
- if (Buf[0] == '\n' || Buf[0] == '\r') {
- // If this is \n\r or \r\n, skip both characters.
- if ((Buf[1] == '\n' || Buf[1] == '\r') && Buf[0] != Buf[1])
- ++Offs, ++Buf;
- ++Offs, ++Buf;
- LineOffsets.push_back(Offs);
- } else {
- // Otherwise, this is a null. If end of file, exit.
- if (Buf == End) break;
- // Otherwise, skip the null.
- ++Offs, ++Buf;
- }
- }
- LineOffsets.push_back(Offs);
-
- // Copy the offsets into the FileInfo structure.
- FileInfo->NumLines = LineOffsets.size();
- FileInfo->SourceLineCache = new unsigned[LineOffsets.size()];
- std::copy(LineOffsets.begin(), LineOffsets.end(),
- FileInfo->SourceLineCache);
- }
+ /// SourceLineCache for it on demand.
+ if (FileInfo->SourceLineCache == 0)
+ ComputeLineNumbers(FileInfo);
// Okay, we know we have a line number table. Do a binary search to find the
// line number that this character position lands on.
- unsigned NumLines = FileInfo->NumLines;
unsigned *SourceLineCache = FileInfo->SourceLineCache;
-
+ unsigned *SourceLineCacheStart = SourceLineCache;
+ unsigned *SourceLineCacheEnd = SourceLineCache + FileInfo->NumLines;
+
+ unsigned QueriedFilePos = getFullFilePos(Loc)+1;
+
+ // If the previous query was to the same file, we know both the file pos from
+ // that query and the line number returned. This allows us to narrow the
+ // search space from the entire file to something near the match.
+ if (LastLineNoFileIDQuery == FileID) {
+ if (QueriedFilePos >= LastLineNoFilePos) {
+ SourceLineCache = SourceLineCache+LastLineNoResult-1;
+
+ // The query is likely to be nearby the previous one. Here we check to
+ // see if it is within 5, 10 or 20 lines. It can be far away in cases
+ // where big comment blocks and vertical whitespace eat up lines but
+ // contribute no tokens.
+ if (SourceLineCache+5 < SourceLineCacheEnd) {
+ if (SourceLineCache[5] > QueriedFilePos)
+ SourceLineCacheEnd = SourceLineCache+5;
+ else if (SourceLineCache+10 < SourceLineCacheEnd) {
+ if (SourceLineCache[10] > QueriedFilePos)
+ SourceLineCacheEnd = SourceLineCache+10;
+ else if (SourceLineCache+20 < SourceLineCacheEnd) {
+ if (SourceLineCache[20] > QueriedFilePos)
+ SourceLineCacheEnd = SourceLineCache+20;
+ }
+ }
+ }
+ } else {
+ SourceLineCacheEnd = SourceLineCache+LastLineNoResult+1;
+ }
+ }
+
+ unsigned *Pos;
// TODO: If this is performance sensitive, we could try doing simple radix
// type approaches to make good (tight?) initial guesses based on the
// assumption that all lines are the same average size.
- unsigned *Pos = std::lower_bound(SourceLineCache, SourceLineCache+NumLines,
- getFullFilePos(Loc)+1);
- return Pos-SourceLineCache;
+ Pos = std::lower_bound(SourceLineCache, SourceLineCacheEnd, QueriedFilePos);
+ unsigned LineNo = Pos-SourceLineCacheStart;
+
+ LastLineNoFileIDQuery = FileID;
+ LastLineNoFileInfo = FileInfo;
+ LastLineNoFilePos = QueriedFilePos;
+ LastLineNoResult = LineNo;
+ return LineNo;
}
/// PrintStats - Print statistics to stderr.
diff --git a/clang.xcodeproj/project.pbxproj b/clang.xcodeproj/project.pbxproj
index bb772fa182..ff84ac7c23 100644
--- a/clang.xcodeproj/project.pbxproj
+++ b/clang.xcodeproj/project.pbxproj
@@ -191,7 +191,7 @@
1A869AA70BA21ABA008DA07A /* LiteralSupport.cpp */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.cpp.cpp; path = LiteralSupport.cpp; sourceTree = "<group>"; };
84D9A8870C1A57E100AC7ABC /* AttributeList.cpp */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.cpp.cpp; name = AttributeList.cpp; path = Parse/AttributeList.cpp; sourceTree = "<group>"; };
84D9A88B0C1A581300AC7ABC /* AttributeList.h */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.c.h; name = AttributeList.h; path = clang/Parse/AttributeList.h; sourceTree = "<group>"; };
- 8DD76F6C0486A84900D96B5E /* clang */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = clang; sourceTree = BUILT_PRODUCTS_DIR; };
+ 8DD76F6C0486A84900D96B5E /* clang */ = {isa = PBXFileReference; includeInIndex = 0; lastKnownFileType = "compiled.mach-o.executable"; path = clang; sourceTree = BUILT_PRODUCTS_DIR; };
DE01DA480B12ADA300AC22CE /* PPCallbacks.h */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.c.h; path = PPCallbacks.h; sourceTree = "<group>"; };
DE06756B0C051CFE00EBBFD8 /* ParseExprCXX.cpp */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.cpp.cpp; name = ParseExprCXX.cpp; path = Parse/ParseExprCXX.cpp; sourceTree = "<group>"; };
DE06B73D0A8307640050E87E /* LangOptions.h */ = {isa = PBXFileReference; fileEncoding = 30; lastKnownFileType = sourcecode.c.h; path = LangOptions.h; sourceTree = "<group>"; };
diff --git a/include/clang/Basic/SourceManager.h b/include/clang/Basic/SourceManager.h
index f06a6fbf54..dd8a86cc7e 100644
--- a/include/clang/Basic/SourceManager.h
+++ b/include/clang/Basic/SourceManager.h
@@ -154,13 +154,21 @@ class SourceManager {
/// MacroIDs - Information about each MacroID.
std::vector<SrcMgr::MacroIDInfo> MacroIDs;
+ /// LastLineNo - These ivars serve as a cache used in the getLineNumber
+ /// method which is used to speedup getLineNumber calls to nearby locations.
+ unsigned LastLineNoFileIDQuery;
+ SrcMgr::FileInfo *LastLineNoFileInfo;
+ unsigned LastLineNoFilePos;
+ unsigned LastLineNoResult;
public:
- SourceManager() {}
+ SourceManager() : LastLineNoFileIDQuery(~0U) {}
~SourceManager();
void clearIDTables() {
FileIDs.clear();
MacroIDs.clear();
+ LastLineNoFileIDQuery = ~0U;
+ LastLineNoFileInfo = 0;
}
/// createFileID - Create a new FileID that represents the specified file
generated by cgit v1.2.3-18-g5258 (git 2.32.0) at 2025-07-01 16:01:10 +0000