2009-07-09 02:44:05 +08:00
|
|
|
//===- FileCheck.cpp - Check that File's Contents match what is expected --===//
|
|
|
|
//
|
|
|
|
// The LLVM Compiler Infrastructure
|
|
|
|
//
|
|
|
|
// This file is distributed under the University of Illinois Open Source
|
|
|
|
// License. See LICENSE.TXT for details.
|
|
|
|
//
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
//
|
|
|
|
// FileCheck does a line-by line check of a file that validates whether it
|
|
|
|
// contains the expected content. This is useful for regression tests etc.
|
|
|
|
//
|
2017-03-14 18:51:14 +08:00
|
|
|
// This program exits with an exit status of 2 on error, exit status of 0 if
|
2009-07-09 02:44:05 +08:00
|
|
|
// the file matched the expected contents, and exit status of 1 if it did not
|
|
|
|
// contain the expected contents.
|
|
|
|
//
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
2012-12-04 18:37:14 +08:00
|
|
|
#include "llvm/ADT/SmallString.h"
|
|
|
|
#include "llvm/ADT/StringExtras.h"
|
|
|
|
#include "llvm/ADT/StringMap.h"
|
2013-11-10 10:04:09 +08:00
|
|
|
#include "llvm/ADT/StringSet.h"
|
2009-07-09 02:44:05 +08:00
|
|
|
#include "llvm/Support/CommandLine.h"
|
|
|
|
#include "llvm/Support/MemoryBuffer.h"
|
|
|
|
#include "llvm/Support/PrettyStackTrace.h"
|
2009-09-25 05:47:32 +08:00
|
|
|
#include "llvm/Support/Regex.h"
|
2012-12-04 18:37:14 +08:00
|
|
|
#include "llvm/Support/Signals.h"
|
2009-07-09 02:44:05 +08:00
|
|
|
#include "llvm/Support/SourceMgr.h"
|
|
|
|
#include "llvm/Support/raw_ostream.h"
|
2009-09-27 15:56:52 +08:00
|
|
|
#include <algorithm>
|
2013-10-12 08:55:57 +08:00
|
|
|
#include <cctype>
|
2012-12-02 05:54:48 +08:00
|
|
|
#include <map>
|
|
|
|
#include <string>
|
2014-06-13 01:38:55 +08:00
|
|
|
#include <system_error>
|
2012-12-02 05:54:48 +08:00
|
|
|
#include <vector>
|
2009-07-09 02:44:05 +08:00
|
|
|
using namespace llvm;
|
|
|
|
|
|
|
|
static cl::opt<std::string>
|
2016-12-11 17:54:36 +08:00
|
|
|
CheckFilename(cl::Positional, cl::desc("<check-file>"), cl::Required);
|
2009-07-09 02:44:05 +08:00
|
|
|
|
|
|
|
static cl::opt<std::string>
|
2016-12-11 17:54:36 +08:00
|
|
|
InputFilename("input-file", cl::desc("File to check (defaults to stdin)"),
|
|
|
|
cl::init("-"), cl::value_desc("filename"));
|
2009-07-09 02:44:05 +08:00
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
static cl::list<std::string> CheckPrefixes(
|
|
|
|
"check-prefix",
|
|
|
|
cl::desc("Prefix to use from check file (defaults to 'CHECK')"));
|
2016-06-14 22:28:04 +08:00
|
|
|
static cl::alias CheckPrefixesAlias(
|
|
|
|
"check-prefixes", cl::aliasopt(CheckPrefixes), cl::CommaSeparated,
|
|
|
|
cl::NotHidden,
|
|
|
|
cl::desc(
|
|
|
|
"Alias for -check-prefix permitting multiple comma separated values"));
|
2009-07-09 02:44:05 +08:00
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
static cl::opt<bool> NoCanonicalizeWhiteSpace(
|
|
|
|
"strict-whitespace",
|
|
|
|
cl::desc("Do not treat all horizontal whitespace as equivalent"));
|
2009-07-12 02:58:15 +08:00
|
|
|
|
2014-07-11 20:39:32 +08:00
|
|
|
static cl::list<std::string> ImplicitCheckNot(
|
|
|
|
"implicit-check-not",
|
|
|
|
cl::desc("Add an implicit negative check with this pattern to every\n"
|
|
|
|
"positive check. This can be used to ensure that no instances of\n"
|
|
|
|
"this pattern occur which are not matched by a positive pattern"),
|
|
|
|
cl::value_desc("pattern"));
|
|
|
|
|
2017-11-07 21:24:44 +08:00
|
|
|
static cl::list<std::string> GlobalDefines("D", cl::Prefix,
|
|
|
|
cl::desc("Define a variable to be used in capture patterns."),
|
|
|
|
cl::value_desc("VAR=VALUE"));
|
|
|
|
|
2014-08-08 02:40:37 +08:00
|
|
|
static cl::opt<bool> AllowEmptyInput(
|
|
|
|
"allow-empty", cl::init(false),
|
|
|
|
cl::desc("Allow the input file to be empty. This is useful when making\n"
|
|
|
|
"checks that some error message does not occur, for example."));
|
|
|
|
|
2016-02-12 00:46:09 +08:00
|
|
|
static cl::opt<bool> MatchFullLines(
|
|
|
|
"match-full-lines", cl::init(false),
|
|
|
|
cl::desc("Require all positive matches to cover an entire input line.\n"
|
|
|
|
"Allows leading and trailing whitespace if --strict-whitespace\n"
|
|
|
|
"is not also passed."));
|
|
|
|
|
2017-03-10 01:59:04 +08:00
|
|
|
static cl::opt<bool> EnableVarScope(
|
|
|
|
"enable-var-scope", cl::init(false),
|
|
|
|
cl::desc("Enables scope for regex variables. Variables with names that\n"
|
|
|
|
"do not start with '$' will be reset at the beginning of\n"
|
|
|
|
"each CHECK-LABEL block."));
|
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
typedef cl::list<std::string>::const_iterator prefix_iterator;
|
|
|
|
|
2009-09-25 04:39:13 +08:00
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Pattern Handling Code.
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
2013-09-18 06:30:02 +08:00
|
|
|
namespace Check {
|
2016-12-11 17:54:36 +08:00
|
|
|
enum CheckType {
|
|
|
|
CheckNone = 0,
|
|
|
|
CheckPlain,
|
|
|
|
CheckNext,
|
|
|
|
CheckSame,
|
|
|
|
CheckNot,
|
|
|
|
CheckDAG,
|
|
|
|
CheckLabel,
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Indicates the pattern only matches the end of file. This is used for
|
|
|
|
/// trailing CHECK-NOTs.
|
2016-12-11 17:54:36 +08:00
|
|
|
CheckEOF,
|
2016-12-11 18:16:21 +08:00
|
|
|
|
|
|
|
/// Marks when parsing found a -NOT check combined with another CHECK suffix.
|
2016-12-11 17:54:36 +08:00
|
|
|
CheckBadNot
|
|
|
|
};
|
2013-09-18 06:30:02 +08:00
|
|
|
}
|
|
|
|
|
2009-09-25 04:25:55 +08:00
|
|
|
class Pattern {
|
2009-09-26 01:29:36 +08:00
|
|
|
SMLoc PatternLoc;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// A fixed string to match as the pattern or empty if this pattern requires
|
|
|
|
/// a regex match.
|
2009-09-26 01:09:12 +08:00
|
|
|
StringRef FixedStr;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// A regex string to match as the pattern or empty if this pattern requires
|
|
|
|
/// a fixed string to match.
|
2009-09-26 01:23:43 +08:00
|
|
|
std::string RegExStr;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Entries in this vector map to uses of a variable in the pattern, e.g.
|
|
|
|
/// "foo[[bar]]baz". In this case, the RegExStr will contain "foobaz" and
|
|
|
|
/// we'll get an entry in this vector that tells us to insert the value of
|
|
|
|
/// bar at offset 3.
|
2016-12-11 17:54:36 +08:00
|
|
|
std::vector<std::pair<StringRef, unsigned>> VariableUses;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Maps definitions of variables to their parenthesized capture numbers.
|
|
|
|
///
|
|
|
|
/// E.g. for the pattern "foo[[bar:.*]]baz", VariableDefs will map "bar" to
|
|
|
|
/// 1.
|
2012-12-02 05:54:48 +08:00
|
|
|
std::map<StringRef, unsigned> VariableDefs;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-08-27 00:18:40 +08:00
|
|
|
Check::CheckType CheckTy;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Contains the number of line this pattern is in.
|
2016-08-27 00:18:40 +08:00
|
|
|
unsigned LineNumber;
|
|
|
|
|
|
|
|
public:
|
|
|
|
explicit Pattern(Check::CheckType Ty) : CheckTy(Ty) {}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Returns the location in source code.
|
2013-04-26 05:31:34 +08:00
|
|
|
SMLoc getLoc() const { return PatternLoc; }
|
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
bool ParsePattern(StringRef PatternStr, StringRef Prefix, SourceMgr &SM,
|
2013-11-10 10:04:09 +08:00
|
|
|
unsigned LineNumber);
|
2009-09-27 15:56:52 +08:00
|
|
|
size_t Match(StringRef Buffer, size_t &MatchLen,
|
|
|
|
StringMap<StringRef> &VariableTable) const;
|
2009-11-23 06:08:06 +08:00
|
|
|
void PrintFailureInfo(const SourceMgr &SM, StringRef Buffer,
|
|
|
|
const StringMap<StringRef> &VariableTable) const;
|
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
bool hasVariable() const {
|
|
|
|
return !(VariableUses.empty() && VariableDefs.empty());
|
|
|
|
}
|
2013-07-12 22:51:05 +08:00
|
|
|
|
2013-09-18 06:30:02 +08:00
|
|
|
Check::CheckType getCheckTy() const { return CheckTy; }
|
2013-05-15 04:34:12 +08:00
|
|
|
|
2009-09-26 01:23:43 +08:00
|
|
|
private:
|
2012-12-02 05:54:48 +08:00
|
|
|
bool AddRegExToRegEx(StringRef RS, unsigned &CurParen, SourceMgr &SM);
|
|
|
|
void AddBackrefToRegEx(unsigned BackrefNum);
|
2016-12-11 17:54:36 +08:00
|
|
|
unsigned
|
|
|
|
ComputeMatchDistance(StringRef Buffer,
|
|
|
|
const StringMap<StringRef> &VariableTable) const;
|
2012-11-15 05:07:37 +08:00
|
|
|
bool EvaluateExpression(StringRef Expr, std::string &Value) const;
|
2014-01-04 05:49:09 +08:00
|
|
|
size_t FindRegexVarEnd(StringRef Str, SourceMgr &SM);
|
2009-09-25 04:25:55 +08:00
|
|
|
};
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Parses the given string into the Pattern.
|
|
|
|
///
|
|
|
|
/// \p Prefix provides which prefix is being matched, \p SM provides the
|
|
|
|
/// SourceMgr used for error reports, and \p LineNumber is the line number in
|
|
|
|
/// the input file from which the pattern string was read. Returns true in
|
|
|
|
/// case of an error, false otherwise.
|
2016-12-11 17:54:36 +08:00
|
|
|
bool Pattern::ParsePattern(StringRef PatternStr, StringRef Prefix,
|
|
|
|
SourceMgr &SM, unsigned LineNumber) {
|
2016-02-12 00:46:09 +08:00
|
|
|
bool MatchFullLinesHere = MatchFullLines && CheckTy != Check::CheckNot;
|
|
|
|
|
2012-11-15 05:07:37 +08:00
|
|
|
this->LineNumber = LineNumber;
|
2009-09-26 01:29:36 +08:00
|
|
|
PatternLoc = SMLoc::getFromPointer(PatternStr.data());
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-19 04:45:59 +08:00
|
|
|
if (!(NoCanonicalizeWhiteSpace && MatchFullLines))
|
|
|
|
// Ignore trailing whitespace.
|
|
|
|
while (!PatternStr.empty() &&
|
|
|
|
(PatternStr.back() == ' ' || PatternStr.back() == '\t'))
|
|
|
|
PatternStr = PatternStr.substr(0, PatternStr.size() - 1);
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-25 04:39:13 +08:00
|
|
|
// Check that there is something on the line.
|
|
|
|
if (PatternStr.empty()) {
|
2011-10-16 13:43:57 +08:00
|
|
|
SM.PrintMessage(PatternLoc, SourceMgr::DK_Error,
|
2016-12-11 17:54:36 +08:00
|
|
|
"found empty check string with prefix '" + Prefix + ":'");
|
2009-09-25 04:39:13 +08:00
|
|
|
return true;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-26 01:09:12 +08:00
|
|
|
// Check to see if this is a fixed string, or if it has regex pieces.
|
2016-02-12 00:46:09 +08:00
|
|
|
if (!MatchFullLinesHere &&
|
|
|
|
(PatternStr.size() < 2 || (PatternStr.find("{{") == StringRef::npos &&
|
|
|
|
PatternStr.find("[[") == StringRef::npos))) {
|
2009-09-26 01:09:12 +08:00
|
|
|
FixedStr = PatternStr;
|
|
|
|
return false;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-02-12 00:46:09 +08:00
|
|
|
if (MatchFullLinesHere) {
|
|
|
|
RegExStr += '^';
|
|
|
|
if (!NoCanonicalizeWhiteSpace)
|
|
|
|
RegExStr += " *";
|
|
|
|
}
|
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// Paren value #0 is for the fully matched string. Any new parenthesized
|
2011-04-09 14:18:02 +08:00
|
|
|
// values add from there.
|
2009-09-27 15:56:52 +08:00
|
|
|
unsigned CurParen = 1;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-26 01:23:43 +08:00
|
|
|
// Otherwise, there is at least one regex piece. Build up the regex pattern
|
|
|
|
// by escaping scary characters in fixed strings, building up one big regex.
|
2009-09-25 05:47:32 +08:00
|
|
|
while (!PatternStr.empty()) {
|
2009-09-27 15:56:52 +08:00
|
|
|
// RegEx matches.
|
2011-04-09 14:18:02 +08:00
|
|
|
if (PatternStr.startswith("{{")) {
|
2012-11-30 22:22:14 +08:00
|
|
|
// This is the start of a regex match. Scan for the }}.
|
2009-09-27 15:56:52 +08:00
|
|
|
size_t End = PatternStr.find("}}");
|
|
|
|
if (End == StringRef::npos) {
|
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
|
2011-10-16 13:43:57 +08:00
|
|
|
SourceMgr::DK_Error,
|
|
|
|
"found start of regex string with no end '}}'");
|
2009-09-27 15:56:52 +08:00
|
|
|
return true;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2011-04-09 14:37:03 +08:00
|
|
|
// Enclose {{}} patterns in parens just like [[]] even though we're not
|
|
|
|
// capturing the result for any purpose. This is required in case the
|
|
|
|
// expression contains an alternation like: CHECK: abc{{x|z}}def. We
|
|
|
|
// want this to turn into: "abc(x|z)def" not "abcx|zdef".
|
|
|
|
RegExStr += '(';
|
|
|
|
++CurParen;
|
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
if (AddRegExToRegEx(PatternStr.substr(2, End - 2), CurParen, SM))
|
2009-09-27 15:56:52 +08:00
|
|
|
return true;
|
2011-04-09 14:37:03 +08:00
|
|
|
RegExStr += ')';
|
2011-04-09 14:18:02 +08:00
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
PatternStr = PatternStr.substr(End + 2);
|
2009-09-25 05:47:32 +08:00
|
|
|
continue;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// Named RegEx matches. These are of two forms: [[foo:.*]] which matches .*
|
|
|
|
// (or some other regex) and assigns it to the FileCheck variable 'foo'. The
|
|
|
|
// second form is [[foo]] which is a reference to foo. The variable name
|
2009-11-23 06:07:50 +08:00
|
|
|
// itself must be of the form "[a-zA-Z_][0-9a-zA-Z_]*", otherwise we reject
|
2009-09-27 15:56:52 +08:00
|
|
|
// it. This is to catch some common errors.
|
2011-04-09 14:18:02 +08:00
|
|
|
if (PatternStr.startswith("[[")) {
|
2012-12-03 00:02:41 +08:00
|
|
|
// Find the closing bracket pair ending the match. End is going to be an
|
|
|
|
// offset relative to the beginning of the match string.
|
2014-01-04 05:49:09 +08:00
|
|
|
size_t End = FindRegexVarEnd(PatternStr.substr(2), SM);
|
2012-12-03 00:02:41 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
if (End == StringRef::npos) {
|
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
|
2011-10-16 13:43:57 +08:00
|
|
|
SourceMgr::DK_Error,
|
|
|
|
"invalid named regex reference, no ]] found");
|
2009-09-27 15:56:52 +08:00
|
|
|
return true;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2012-12-03 00:02:41 +08:00
|
|
|
StringRef MatchStr = PatternStr.substr(2, End);
|
2016-12-11 17:54:36 +08:00
|
|
|
PatternStr = PatternStr.substr(End + 4);
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// Get the regex name (e.g. "foo").
|
|
|
|
size_t NameEnd = MatchStr.find(':');
|
|
|
|
StringRef Name = MatchStr.substr(0, NameEnd);
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
if (Name.empty()) {
|
2011-10-16 13:43:57 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Name.data()), SourceMgr::DK_Error,
|
|
|
|
"invalid name in named regex: empty name");
|
2009-09-27 15:56:52 +08:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2012-11-15 05:07:37 +08:00
|
|
|
// Verify that the name/expression is well formed. FileCheck currently
|
|
|
|
// supports @LINE, @LINE+number, @LINE-number expressions. The check here
|
|
|
|
// is relaxed, more strict check is performed in \c EvaluateExpression.
|
|
|
|
bool IsExpression = false;
|
|
|
|
for (unsigned i = 0, e = Name.size(); i != e; ++i) {
|
2017-03-10 01:59:04 +08:00
|
|
|
if (i == 0) {
|
|
|
|
if (Name[i] == '$') // Global vars start with '$'
|
|
|
|
continue;
|
|
|
|
if (Name[i] == '@') {
|
|
|
|
if (NameEnd != StringRef::npos) {
|
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
|
|
|
|
SourceMgr::DK_Error,
|
|
|
|
"invalid name in named regex definition");
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
IsExpression = true;
|
|
|
|
continue;
|
2012-11-15 05:07:37 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
if (Name[i] != '_' && !isalnum(Name[i]) &&
|
|
|
|
(!IsExpression || (Name[i] != '+' && Name[i] != '-'))) {
|
2016-12-11 17:54:36 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Name.data() + i),
|
2011-10-16 13:43:57 +08:00
|
|
|
SourceMgr::DK_Error, "invalid name in named regex");
|
2009-09-27 15:56:52 +08:00
|
|
|
return true;
|
|
|
|
}
|
2012-11-15 05:07:37 +08:00
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// Name can't start with a digit.
|
2013-02-13 05:21:59 +08:00
|
|
|
if (isdigit(static_cast<unsigned char>(Name[0]))) {
|
2011-10-16 13:43:57 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Name.data()), SourceMgr::DK_Error,
|
|
|
|
"invalid name in named regex");
|
2009-09-27 15:56:52 +08:00
|
|
|
return true;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// Handle [[foo]].
|
|
|
|
if (NameEnd == StringRef::npos) {
|
2012-12-02 05:54:48 +08:00
|
|
|
// Handle variables that were defined earlier on the same line by
|
|
|
|
// emitting a backreference.
|
|
|
|
if (VariableDefs.find(Name) != VariableDefs.end()) {
|
|
|
|
unsigned VarParenNum = VariableDefs[Name];
|
|
|
|
if (VarParenNum < 1 || VarParenNum > 9) {
|
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
|
|
|
|
SourceMgr::DK_Error,
|
|
|
|
"Can't back-reference more than 9 variables");
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
AddBackrefToRegEx(VarParenNum);
|
|
|
|
} else {
|
|
|
|
VariableUses.push_back(std::make_pair(Name, RegExStr.size()));
|
|
|
|
}
|
2009-09-27 15:56:52 +08:00
|
|
|
continue;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// Handle [[foo:.*]].
|
2012-12-02 05:54:48 +08:00
|
|
|
VariableDefs[Name] = CurParen;
|
2009-09-27 15:56:52 +08:00
|
|
|
RegExStr += '(';
|
|
|
|
++CurParen;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
if (AddRegExToRegEx(MatchStr.substr(NameEnd + 1), CurParen, SM))
|
2009-09-27 15:56:52 +08:00
|
|
|
return true;
|
|
|
|
|
|
|
|
RegExStr += ')';
|
2009-09-25 05:47:32 +08:00
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// Handle fixed string matches.
|
|
|
|
// Find the end, which is the start of the next regex.
|
|
|
|
size_t FixedMatchEnd = PatternStr.find("{{");
|
|
|
|
FixedMatchEnd = std::min(FixedMatchEnd, PatternStr.find("[["));
|
2013-12-12 08:06:41 +08:00
|
|
|
RegExStr += Regex::escape(PatternStr.substr(0, FixedMatchEnd));
|
2009-09-27 15:56:52 +08:00
|
|
|
PatternStr = PatternStr.substr(FixedMatchEnd);
|
2009-09-25 05:47:32 +08:00
|
|
|
}
|
|
|
|
|
2016-02-12 00:46:09 +08:00
|
|
|
if (MatchFullLinesHere) {
|
|
|
|
if (!NoCanonicalizeWhiteSpace)
|
|
|
|
RegExStr += " *";
|
|
|
|
RegExStr += '$';
|
|
|
|
}
|
|
|
|
|
2009-09-25 05:47:32 +08:00
|
|
|
return false;
|
|
|
|
}
|
2009-09-25 04:45:07 +08:00
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
bool Pattern::AddRegExToRegEx(StringRef RS, unsigned &CurParen, SourceMgr &SM) {
|
2012-12-02 05:54:48 +08:00
|
|
|
Regex R(RS);
|
2009-09-27 15:56:52 +08:00
|
|
|
std::string Error;
|
|
|
|
if (!R.isValid(Error)) {
|
2012-12-02 05:54:48 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(RS.data()), SourceMgr::DK_Error,
|
2011-10-16 13:43:57 +08:00
|
|
|
"invalid regex: " + Error);
|
2009-09-27 15:56:52 +08:00
|
|
|
return true;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2012-12-02 05:54:48 +08:00
|
|
|
RegExStr += RS.str();
|
2009-09-27 15:56:52 +08:00
|
|
|
CurParen += R.getNumMatches();
|
|
|
|
return false;
|
|
|
|
}
|
2009-09-26 01:23:43 +08:00
|
|
|
|
2012-12-02 05:54:48 +08:00
|
|
|
void Pattern::AddBackrefToRegEx(unsigned BackrefNum) {
|
|
|
|
assert(BackrefNum >= 1 && BackrefNum <= 9 && "Invalid backref number");
|
2016-12-11 17:54:36 +08:00
|
|
|
std::string Backref = std::string("\\") + std::string(1, '0' + BackrefNum);
|
2012-12-02 05:54:48 +08:00
|
|
|
RegExStr += Backref;
|
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Evaluates expression and stores the result to \p Value.
|
|
|
|
///
|
|
|
|
/// Returns true on success and false when the expression has invalid syntax.
|
2012-11-15 05:07:37 +08:00
|
|
|
bool Pattern::EvaluateExpression(StringRef Expr, std::string &Value) const {
|
|
|
|
// The only supported expression is @LINE([\+-]\d+)?
|
|
|
|
if (!Expr.startswith("@LINE"))
|
|
|
|
return false;
|
|
|
|
Expr = Expr.substr(StringRef("@LINE").size());
|
|
|
|
int Offset = 0;
|
|
|
|
if (!Expr.empty()) {
|
|
|
|
if (Expr[0] == '+')
|
|
|
|
Expr = Expr.substr(1);
|
|
|
|
else if (Expr[0] != '-')
|
|
|
|
return false;
|
|
|
|
if (Expr.getAsInteger(10, Offset))
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
Value = llvm::itostr(LineNumber + Offset);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Matches the pattern string against the input buffer \p Buffer
|
|
|
|
///
|
|
|
|
/// This returns the position that is matched or npos if there is no match. If
|
|
|
|
/// there is a match, the size of the matched string is returned in \p
|
|
|
|
/// MatchLen.
|
|
|
|
///
|
|
|
|
/// The \p VariableTable StringMap provides the current values of filecheck
|
|
|
|
/// variables and is updated if this match defines new values.
|
2009-09-27 15:56:52 +08:00
|
|
|
size_t Pattern::Match(StringRef Buffer, size_t &MatchLen,
|
|
|
|
StringMap<StringRef> &VariableTable) const {
|
2010-10-16 01:47:12 +08:00
|
|
|
// If this is the EOF pattern, match it immediately.
|
2013-09-18 06:30:02 +08:00
|
|
|
if (CheckTy == Check::CheckEOF) {
|
2010-10-16 01:47:12 +08:00
|
|
|
MatchLen = 0;
|
|
|
|
return Buffer.size();
|
|
|
|
}
|
|
|
|
|
2009-09-26 01:09:12 +08:00
|
|
|
// If this is a fixed string pattern, just match it now.
|
|
|
|
if (!FixedStr.empty()) {
|
|
|
|
MatchLen = FixedStr.size();
|
|
|
|
return Buffer.find(FixedStr);
|
|
|
|
}
|
2009-09-27 15:56:52 +08:00
|
|
|
|
2009-09-26 01:23:43 +08:00
|
|
|
// Regex match.
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// If there are variable uses, we need to create a temporary string with the
|
|
|
|
// actual value.
|
|
|
|
StringRef RegExToMatch = RegExStr;
|
|
|
|
std::string TmpStr;
|
|
|
|
if (!VariableUses.empty()) {
|
|
|
|
TmpStr = RegExStr;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
unsigned InsertOffset = 0;
|
2015-09-10 19:59:55 +08:00
|
|
|
for (const auto &VariableUse : VariableUses) {
|
2009-09-27 15:56:52 +08:00
|
|
|
std::string Value;
|
2012-11-15 05:07:37 +08:00
|
|
|
|
2015-09-10 19:59:55 +08:00
|
|
|
if (VariableUse.first[0] == '@') {
|
|
|
|
if (!EvaluateExpression(VariableUse.first, Value))
|
2012-11-15 05:07:37 +08:00
|
|
|
return StringRef::npos;
|
|
|
|
} else {
|
|
|
|
StringMap<StringRef>::iterator it =
|
2015-09-10 19:59:55 +08:00
|
|
|
VariableTable.find(VariableUse.first);
|
2012-11-15 05:07:37 +08:00
|
|
|
// If the variable is undefined, return an error.
|
|
|
|
if (it == VariableTable.end())
|
|
|
|
return StringRef::npos;
|
|
|
|
|
2013-12-12 08:06:41 +08:00
|
|
|
// Look up the value and escape it so that we can put it into the regex.
|
|
|
|
Value += Regex::escape(it->second);
|
2012-11-15 05:07:37 +08:00
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// Plop it into the regex at the adjusted offset.
|
2015-09-10 19:59:55 +08:00
|
|
|
TmpStr.insert(TmpStr.begin() + VariableUse.second + InsertOffset,
|
2009-09-27 15:56:52 +08:00
|
|
|
Value.begin(), Value.end());
|
|
|
|
InsertOffset += Value.size();
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// Match the newly constructed regex.
|
|
|
|
RegExToMatch = TmpStr;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-26 01:23:43 +08:00
|
|
|
SmallVector<StringRef, 4> MatchInfo;
|
2009-09-27 15:56:52 +08:00
|
|
|
if (!Regex(RegExToMatch, Regex::Newline).match(Buffer, &MatchInfo))
|
2009-09-26 01:23:43 +08:00
|
|
|
return StringRef::npos;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-26 01:23:43 +08:00
|
|
|
// Successful regex match.
|
|
|
|
assert(!MatchInfo.empty() && "Didn't get any match");
|
|
|
|
StringRef FullMatch = MatchInfo[0];
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-27 15:56:52 +08:00
|
|
|
// If this defines any variables, remember their values.
|
2015-09-10 19:59:55 +08:00
|
|
|
for (const auto &VariableDef : VariableDefs) {
|
|
|
|
assert(VariableDef.second < MatchInfo.size() && "Internal paren error");
|
|
|
|
VariableTable[VariableDef.first] = MatchInfo[VariableDef.second];
|
2009-09-26 01:29:36 +08:00
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-26 01:23:43 +08:00
|
|
|
MatchLen = FullMatch.size();
|
2016-12-11 17:54:36 +08:00
|
|
|
return FullMatch.data() - Buffer.data();
|
2009-09-25 04:39:13 +08:00
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
|
|
|
|
/// Computes an arbitrary estimate for the quality of matching this pattern at
|
|
|
|
/// the start of \p Buffer; a distance of zero should correspond to a perfect
|
|
|
|
/// match.
|
2016-12-11 17:54:36 +08:00
|
|
|
unsigned
|
|
|
|
Pattern::ComputeMatchDistance(StringRef Buffer,
|
2009-11-23 06:59:26 +08:00
|
|
|
const StringMap<StringRef> &VariableTable) const {
|
|
|
|
// Just compute the number of matching characters. For regular expressions, we
|
|
|
|
// just compare against the regex itself and hope for the best.
|
|
|
|
//
|
|
|
|
// FIXME: One easy improvement here is have the regex lib generate a single
|
|
|
|
// example regular expression which matches, and use that as the example
|
|
|
|
// string.
|
|
|
|
StringRef ExampleString(FixedStr);
|
|
|
|
if (ExampleString.empty())
|
|
|
|
ExampleString = RegExStr;
|
|
|
|
|
2010-01-30 08:24:06 +08:00
|
|
|
// Only compare up to the first line in the buffer, or the string size.
|
|
|
|
StringRef BufferPrefix = Buffer.substr(0, ExampleString.size());
|
|
|
|
BufferPrefix = BufferPrefix.split('\n').first;
|
|
|
|
return BufferPrefix.edit_distance(ExampleString);
|
2009-11-23 06:59:26 +08:00
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Prints additional information about a failure to match involving this
|
|
|
|
/// pattern.
|
2016-12-11 17:54:36 +08:00
|
|
|
void Pattern::PrintFailureInfo(
|
|
|
|
const SourceMgr &SM, StringRef Buffer,
|
|
|
|
const StringMap<StringRef> &VariableTable) const {
|
2009-11-23 06:08:06 +08:00
|
|
|
// If this was a regular expression using variables, print the current
|
|
|
|
// variable values.
|
|
|
|
if (!VariableUses.empty()) {
|
2015-09-10 19:59:55 +08:00
|
|
|
for (const auto &VariableUse : VariableUses) {
|
2014-06-27 06:52:05 +08:00
|
|
|
SmallString<256> Msg;
|
|
|
|
raw_svector_ostream OS(Msg);
|
2015-09-10 19:59:55 +08:00
|
|
|
StringRef Var = VariableUse.first;
|
2012-11-15 05:07:37 +08:00
|
|
|
if (Var[0] == '@') {
|
|
|
|
std::string Value;
|
|
|
|
if (EvaluateExpression(Var, Value)) {
|
|
|
|
OS << "with expression \"";
|
|
|
|
OS.write_escaped(Var) << "\" equal to \"";
|
|
|
|
OS.write_escaped(Value) << "\"";
|
|
|
|
} else {
|
|
|
|
OS << "uses incorrect expression \"";
|
|
|
|
OS.write_escaped(Var) << "\"";
|
|
|
|
}
|
2009-11-23 06:08:06 +08:00
|
|
|
} else {
|
2012-11-15 05:07:37 +08:00
|
|
|
StringMap<StringRef>::const_iterator it = VariableTable.find(Var);
|
|
|
|
|
|
|
|
// Check for undefined variable references.
|
|
|
|
if (it == VariableTable.end()) {
|
|
|
|
OS << "uses undefined variable \"";
|
|
|
|
OS.write_escaped(Var) << "\"";
|
|
|
|
} else {
|
|
|
|
OS << "with variable \"";
|
|
|
|
OS.write_escaped(Var) << "\" equal to \"";
|
|
|
|
OS.write_escaped(it->second) << "\"";
|
|
|
|
}
|
2009-11-23 06:08:06 +08:00
|
|
|
}
|
|
|
|
|
2011-10-16 13:43:57 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), SourceMgr::DK_Note,
|
|
|
|
OS.str());
|
2009-11-23 06:08:06 +08:00
|
|
|
}
|
|
|
|
}
|
2009-11-23 06:59:26 +08:00
|
|
|
|
|
|
|
// Attempt to find the closest/best fuzzy match. Usually an error happens
|
|
|
|
// because some string in the output didn't exactly match. In these cases, we
|
|
|
|
// would like to show the user a best guess at what "should have" matched, to
|
|
|
|
// save them having to actually check the input manually.
|
|
|
|
size_t NumLinesForward = 0;
|
|
|
|
size_t Best = StringRef::npos;
|
|
|
|
double BestQuality = 0;
|
|
|
|
|
|
|
|
// Use an arbitrary 4k limit on how far we will search.
|
2010-01-30 05:57:46 +08:00
|
|
|
for (size_t i = 0, e = std::min(size_t(4096), Buffer.size()); i != e; ++i) {
|
2009-11-23 06:59:26 +08:00
|
|
|
if (Buffer[i] == '\n')
|
|
|
|
++NumLinesForward;
|
|
|
|
|
2010-01-30 05:55:16 +08:00
|
|
|
// Patterns have leading whitespace stripped, so skip whitespace when
|
|
|
|
// looking for something which looks like a pattern.
|
|
|
|
if (Buffer[i] == ' ' || Buffer[i] == '\t')
|
|
|
|
continue;
|
|
|
|
|
2009-11-23 06:59:26 +08:00
|
|
|
// Compute the "quality" of this match as an arbitrary combination of the
|
|
|
|
// match distance and the number of lines skipped to get to this match.
|
|
|
|
unsigned Distance = ComputeMatchDistance(Buffer.substr(i), VariableTable);
|
|
|
|
double Quality = Distance + (NumLinesForward / 100.);
|
|
|
|
|
|
|
|
if (Quality < BestQuality || Best == StringRef::npos) {
|
|
|
|
Best = i;
|
|
|
|
BestQuality = Quality;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-03-20 02:07:43 +08:00
|
|
|
// Print the "possible intended match here" line if we found something
|
|
|
|
// reasonable and not equal to what we showed in the "scanning from here"
|
|
|
|
// line.
|
|
|
|
if (Best && Best != StringRef::npos && BestQuality < 50) {
|
2016-12-11 17:54:36 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + Best),
|
|
|
|
SourceMgr::DK_Note, "possible intended match here");
|
2009-11-23 06:59:26 +08:00
|
|
|
|
|
|
|
// FIXME: If we wanted to be really friendly we would show why the match
|
|
|
|
// failed, as it can be hard to spot simple one character differences.
|
|
|
|
}
|
2009-11-23 06:08:06 +08:00
|
|
|
}
|
2009-09-25 04:39:13 +08:00
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Finds the closing sequence of a regex variable usage or definition.
|
|
|
|
///
|
|
|
|
/// \p Str has to point in the beginning of the definition (right after the
|
|
|
|
/// opening sequence). Returns the offset of the closing sequence within Str,
|
|
|
|
/// or npos if it was not found.
|
2014-01-04 05:49:09 +08:00
|
|
|
size_t Pattern::FindRegexVarEnd(StringRef Str, SourceMgr &SM) {
|
2012-12-03 00:02:41 +08:00
|
|
|
// Offset keeps track of the current offset within the input Str
|
|
|
|
size_t Offset = 0;
|
|
|
|
// [...] Nesting depth
|
|
|
|
size_t BracketDepth = 0;
|
|
|
|
|
|
|
|
while (!Str.empty()) {
|
|
|
|
if (Str.startswith("]]") && BracketDepth == 0)
|
|
|
|
return Offset;
|
|
|
|
if (Str[0] == '\\') {
|
|
|
|
// Backslash escapes the next char within regexes, so skip them both.
|
|
|
|
Str = Str.substr(2);
|
|
|
|
Offset += 2;
|
|
|
|
} else {
|
|
|
|
switch (Str[0]) {
|
2016-12-11 17:54:36 +08:00
|
|
|
default:
|
|
|
|
break;
|
|
|
|
case '[':
|
|
|
|
BracketDepth++;
|
|
|
|
break;
|
|
|
|
case ']':
|
|
|
|
if (BracketDepth == 0) {
|
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Str.data()),
|
|
|
|
SourceMgr::DK_Error,
|
|
|
|
"missing closing \"]\" for regex variable");
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
BracketDepth--;
|
|
|
|
break;
|
2012-12-03 00:02:41 +08:00
|
|
|
}
|
|
|
|
Str = Str.substr(1);
|
|
|
|
Offset++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return StringRef::npos;
|
|
|
|
}
|
|
|
|
|
2009-09-25 04:39:13 +08:00
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Check Strings.
|
|
|
|
//===----------------------------------------------------------------------===//
|
2009-09-25 04:25:55 +08:00
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// A check that we found in the input file.
|
2009-09-25 04:25:55 +08:00
|
|
|
struct CheckString {
|
2016-12-11 18:16:21 +08:00
|
|
|
/// The pattern to match.
|
2009-09-25 04:25:55 +08:00
|
|
|
Pattern Pat;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Which prefix name this check matched.
|
2013-11-10 10:04:09 +08:00
|
|
|
StringRef Prefix;
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// The location in the match file that the check string was specified.
|
2009-08-16 01:41:04 +08:00
|
|
|
SMLoc Loc;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// All of the strings that are disallowed from occurring between this match
|
|
|
|
/// string and the previous one (or start of file).
|
2013-05-15 04:34:12 +08:00
|
|
|
std::vector<Pattern> DagNotStrings;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-02-12 00:46:09 +08:00
|
|
|
CheckString(const Pattern &P, StringRef S, SMLoc L)
|
|
|
|
: Pat(P), Prefix(S), Loc(L) {}
|
2013-05-15 04:29:52 +08:00
|
|
|
|
2013-10-12 02:38:36 +08:00
|
|
|
size_t Check(const SourceMgr &SM, StringRef Buffer, bool IsLabelScanMode,
|
2013-07-12 22:51:05 +08:00
|
|
|
size_t &MatchLen, StringMap<StringRef> &VariableTable) const;
|
2013-05-15 04:29:52 +08:00
|
|
|
|
|
|
|
bool CheckNext(const SourceMgr &SM, StringRef Buffer) const;
|
2015-02-26 12:53:00 +08:00
|
|
|
bool CheckSame(const SourceMgr &SM, StringRef Buffer) const;
|
2013-05-15 04:29:52 +08:00
|
|
|
bool CheckNot(const SourceMgr &SM, StringRef Buffer,
|
2013-05-15 04:34:12 +08:00
|
|
|
const std::vector<const Pattern *> &NotStrings,
|
2013-05-15 04:29:52 +08:00
|
|
|
StringMap<StringRef> &VariableTable) const;
|
2013-05-15 04:34:12 +08:00
|
|
|
size_t CheckDag(const SourceMgr &SM, StringRef Buffer,
|
|
|
|
std::vector<const Pattern *> &NotStrings,
|
|
|
|
StringMap<StringRef> &VariableTable) const;
|
2009-08-16 01:41:04 +08:00
|
|
|
};
|
|
|
|
|
2016-12-11 17:50:05 +08:00
|
|
|
/// Canonicalize whitespaces in the file. Line endings are replaced with
|
|
|
|
/// UNIX-style '\n'.
|
2016-12-11 18:22:17 +08:00
|
|
|
static StringRef CanonicalizeFile(MemoryBuffer &MB,
|
2016-12-11 17:50:05 +08:00
|
|
|
SmallVectorImpl<char> &OutputBuffer) {
|
|
|
|
OutputBuffer.reserve(MB.getBufferSize());
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 17:50:05 +08:00
|
|
|
for (const char *Ptr = MB.getBufferStart(), *End = MB.getBufferEnd();
|
2009-09-25 04:45:07 +08:00
|
|
|
Ptr != End; ++Ptr) {
|
2010-11-14 11:28:22 +08:00
|
|
|
// Eliminate trailing dosish \r.
|
|
|
|
if (Ptr <= End - 2 && Ptr[0] == '\r' && Ptr[1] == '\n') {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2013-04-26 02:54:02 +08:00
|
|
|
// If current char is not a horizontal whitespace or if horizontal
|
2013-02-07 04:40:38 +08:00
|
|
|
// whitespace canonicalization is disabled, dump it to output as is.
|
2016-12-11 18:22:17 +08:00
|
|
|
if (NoCanonicalizeWhiteSpace || (*Ptr != ' ' && *Ptr != '\t')) {
|
2016-12-11 17:50:05 +08:00
|
|
|
OutputBuffer.push_back(*Ptr);
|
2009-09-25 04:45:07 +08:00
|
|
|
continue;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-09-25 04:45:07 +08:00
|
|
|
// Otherwise, add one space and advance over neighboring space.
|
2016-12-11 17:50:05 +08:00
|
|
|
OutputBuffer.push_back(' ');
|
2016-12-11 17:54:36 +08:00
|
|
|
while (Ptr + 1 != End && (Ptr[1] == ' ' || Ptr[1] == '\t'))
|
2009-09-25 04:45:07 +08:00
|
|
|
++Ptr;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 17:50:05 +08:00
|
|
|
// Add a null byte and then return all but that byte.
|
|
|
|
OutputBuffer.push_back('\0');
|
|
|
|
return StringRef(OutputBuffer.data(), OutputBuffer.size() - 1);
|
2009-09-25 04:45:07 +08:00
|
|
|
}
|
|
|
|
|
2013-09-18 06:30:02 +08:00
|
|
|
static bool IsPartOfWord(char c) {
|
|
|
|
return (isalnum(c) || c == '-' || c == '_');
|
|
|
|
}
|
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
// Get the size of the prefix extension.
|
|
|
|
static size_t CheckTypeSize(Check::CheckType Ty) {
|
|
|
|
switch (Ty) {
|
|
|
|
case Check::CheckNone:
|
2016-03-01 06:13:03 +08:00
|
|
|
case Check::CheckBadNot:
|
2013-11-10 10:04:09 +08:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
case Check::CheckPlain:
|
|
|
|
return sizeof(":") - 1;
|
|
|
|
|
|
|
|
case Check::CheckNext:
|
|
|
|
return sizeof("-NEXT:") - 1;
|
|
|
|
|
2015-02-26 12:53:00 +08:00
|
|
|
case Check::CheckSame:
|
|
|
|
return sizeof("-SAME:") - 1;
|
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
case Check::CheckNot:
|
|
|
|
return sizeof("-NOT:") - 1;
|
|
|
|
|
|
|
|
case Check::CheckDAG:
|
|
|
|
return sizeof("-DAG:") - 1;
|
|
|
|
|
|
|
|
case Check::CheckLabel:
|
|
|
|
return sizeof("-LABEL:") - 1;
|
|
|
|
|
|
|
|
case Check::CheckEOF:
|
|
|
|
llvm_unreachable("Should not be using EOF size");
|
|
|
|
}
|
|
|
|
|
|
|
|
llvm_unreachable("Bad check type");
|
|
|
|
}
|
|
|
|
|
|
|
|
static Check::CheckType FindCheckType(StringRef Buffer, StringRef Prefix) {
|
2013-09-18 06:45:57 +08:00
|
|
|
char NextChar = Buffer[Prefix.size()];
|
2013-09-18 06:30:02 +08:00
|
|
|
|
|
|
|
// Verify that the : is present after the prefix.
|
2013-11-10 10:04:09 +08:00
|
|
|
if (NextChar == ':')
|
2013-09-18 06:30:02 +08:00
|
|
|
return Check::CheckPlain;
|
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
if (NextChar != '-')
|
2013-09-18 06:30:02 +08:00
|
|
|
return Check::CheckNone;
|
|
|
|
|
2013-09-18 06:45:57 +08:00
|
|
|
StringRef Rest = Buffer.drop_front(Prefix.size() + 1);
|
2013-11-10 10:04:09 +08:00
|
|
|
if (Rest.startswith("NEXT:"))
|
2013-09-18 06:30:02 +08:00
|
|
|
return Check::CheckNext;
|
|
|
|
|
2015-02-26 12:53:00 +08:00
|
|
|
if (Rest.startswith("SAME:"))
|
|
|
|
return Check::CheckSame;
|
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
if (Rest.startswith("NOT:"))
|
2013-09-18 06:30:02 +08:00
|
|
|
return Check::CheckNot;
|
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
if (Rest.startswith("DAG:"))
|
2013-09-18 06:30:02 +08:00
|
|
|
return Check::CheckDAG;
|
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
if (Rest.startswith("LABEL:"))
|
2013-09-18 06:30:02 +08:00
|
|
|
return Check::CheckLabel;
|
|
|
|
|
2016-03-01 06:13:03 +08:00
|
|
|
// You can't combine -NOT with another suffix.
|
|
|
|
if (Rest.startswith("DAG-NOT:") || Rest.startswith("NOT-DAG:") ||
|
|
|
|
Rest.startswith("NEXT-NOT:") || Rest.startswith("NOT-NEXT:") ||
|
|
|
|
Rest.startswith("SAME-NOT:") || Rest.startswith("NOT-SAME:"))
|
|
|
|
return Check::CheckBadNot;
|
|
|
|
|
2013-09-18 06:30:02 +08:00
|
|
|
return Check::CheckNone;
|
|
|
|
}
|
2009-07-09 02:44:05 +08:00
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
// From the given position, find the next character after the word.
|
|
|
|
static size_t SkipWord(StringRef Str, size_t Loc) {
|
|
|
|
while (Loc < Str.size() && IsPartOfWord(Str[Loc]))
|
|
|
|
++Loc;
|
|
|
|
return Loc;
|
|
|
|
}
|
|
|
|
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
/// Search the buffer for the first prefix in the prefix regular expression.
|
|
|
|
///
|
|
|
|
/// This searches the buffer using the provided regular expression, however it
|
|
|
|
/// enforces constraints beyond that:
|
|
|
|
/// 1) The found prefix must not be a suffix of something that looks like
|
|
|
|
/// a valid prefix.
|
|
|
|
/// 2) The found prefix must be followed by a valid check type suffix using \c
|
|
|
|
/// FindCheckType above.
|
|
|
|
///
|
|
|
|
/// The first match of the regular expression to satisfy these two is returned,
|
|
|
|
/// otherwise an empty StringRef is returned to indicate failure.
|
|
|
|
///
|
|
|
|
/// If this routine returns a valid prefix, it will also shrink \p Buffer to
|
|
|
|
/// start at the beginning of the returned prefix, increment \p LineNumber for
|
|
|
|
/// each new line consumed from \p Buffer, and set \p CheckTy to the type of
|
|
|
|
/// check found by examining the suffix.
|
|
|
|
///
|
|
|
|
/// If no valid prefix is found, the state of Buffer, LineNumber, and CheckTy
|
|
|
|
/// is unspecified.
|
|
|
|
static StringRef FindFirstMatchingPrefix(Regex &PrefixRE, StringRef &Buffer,
|
2013-11-10 10:04:09 +08:00
|
|
|
unsigned &LineNumber,
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
Check::CheckType &CheckTy) {
|
|
|
|
SmallVector<StringRef, 2> Matches;
|
2013-11-10 10:04:09 +08:00
|
|
|
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
while (!Buffer.empty()) {
|
|
|
|
// Find the first (longest) match using the RE.
|
|
|
|
if (!PrefixRE.match(Buffer, &Matches))
|
|
|
|
// No match at all, bail.
|
2013-11-10 10:04:09 +08:00
|
|
|
return StringRef();
|
|
|
|
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
StringRef Prefix = Matches[0];
|
|
|
|
Matches.clear();
|
|
|
|
|
|
|
|
assert(Prefix.data() >= Buffer.data() &&
|
|
|
|
Prefix.data() < Buffer.data() + Buffer.size() &&
|
|
|
|
"Prefix doesn't start inside of buffer!");
|
|
|
|
size_t Loc = Prefix.data() - Buffer.data();
|
|
|
|
StringRef Skipped = Buffer.substr(0, Loc);
|
|
|
|
Buffer = Buffer.drop_front(Loc);
|
|
|
|
LineNumber += Skipped.count('\n');
|
|
|
|
|
|
|
|
// Check that the matched prefix isn't a suffix of some other check-like
|
|
|
|
// word.
|
|
|
|
// FIXME: This is a very ad-hoc check. it would be better handled in some
|
|
|
|
// other way. Among other things it seems hard to distinguish between
|
|
|
|
// intentional and unintentional uses of this feature.
|
|
|
|
if (Skipped.empty() || !IsPartOfWord(Skipped.back())) {
|
|
|
|
// Now extract the type.
|
|
|
|
CheckTy = FindCheckType(Buffer, Prefix);
|
|
|
|
|
|
|
|
// If we've found a valid check type for this prefix, we're done.
|
|
|
|
if (CheckTy != Check::CheckNone)
|
|
|
|
return Prefix;
|
|
|
|
}
|
2013-11-10 10:04:09 +08:00
|
|
|
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
// If we didn't successfully find a prefix, we need to skip this invalid
|
|
|
|
// prefix and continue scanning. We directly skip the prefix that was
|
|
|
|
// matched and any additional parts of that check-like word.
|
|
|
|
Buffer = Buffer.drop_front(SkipWord(Buffer, Prefix.size()));
|
2013-11-10 10:04:09 +08:00
|
|
|
}
|
|
|
|
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
// We ran out of buffer while skipping partial matches so give up.
|
2013-11-10 10:04:09 +08:00
|
|
|
return StringRef();
|
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Read the check file, which specifies the sequence of expected strings.
|
|
|
|
///
|
|
|
|
/// The strings are added to the CheckStrings vector. Returns true in case of
|
|
|
|
/// an error, false otherwise.
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
static bool ReadCheckFile(SourceMgr &SM, StringRef Buffer, Regex &PrefixRE,
|
2009-08-16 01:41:04 +08:00
|
|
|
std::vector<CheckString> &CheckStrings) {
|
2014-07-11 20:39:32 +08:00
|
|
|
std::vector<Pattern> ImplicitNegativeChecks;
|
|
|
|
for (const auto &PatternString : ImplicitCheckNot) {
|
|
|
|
// Create a buffer with fake command line content in order to display the
|
|
|
|
// command line option responsible for the specific implicit CHECK-NOT.
|
2015-11-18 03:00:52 +08:00
|
|
|
std::string Prefix = (Twine("-") + ImplicitCheckNot.ArgStr + "='").str();
|
2014-07-11 20:39:32 +08:00
|
|
|
std::string Suffix = "'";
|
2014-08-28 04:03:13 +08:00
|
|
|
std::unique_ptr<MemoryBuffer> CmdLine = MemoryBuffer::getMemBufferCopy(
|
|
|
|
Prefix + PatternString + Suffix, "command line");
|
|
|
|
|
2014-07-11 20:39:32 +08:00
|
|
|
StringRef PatternInBuffer =
|
|
|
|
CmdLine->getBuffer().substr(Prefix.size(), PatternString.size());
|
2014-08-22 04:44:56 +08:00
|
|
|
SM.AddNewSourceBuffer(std::move(CmdLine), SMLoc());
|
2014-07-11 20:39:32 +08:00
|
|
|
|
|
|
|
ImplicitNegativeChecks.push_back(Pattern(Check::CheckNot));
|
|
|
|
ImplicitNegativeChecks.back().ParsePattern(PatternInBuffer,
|
|
|
|
"IMPLICIT-CHECK", SM, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
std::vector<Pattern> DagNotMatches = ImplicitNegativeChecks;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2012-11-30 22:22:14 +08:00
|
|
|
// LineNumber keeps track of the line on which CheckPrefix instances are
|
|
|
|
// found.
|
2012-11-15 05:07:37 +08:00
|
|
|
unsigned LineNumber = 1;
|
|
|
|
|
2009-07-09 02:44:05 +08:00
|
|
|
while (1) {
|
2013-11-10 10:04:09 +08:00
|
|
|
Check::CheckType CheckTy;
|
|
|
|
|
|
|
|
// See if a prefix occurs in the memory buffer.
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
StringRef UsedPrefix = FindFirstMatchingPrefix(PrefixRE, Buffer, LineNumber,
|
|
|
|
CheckTy);
|
2013-11-10 10:04:09 +08:00
|
|
|
if (UsedPrefix.empty())
|
2009-07-09 02:44:05 +08:00
|
|
|
break;
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
assert(UsedPrefix.data() == Buffer.data() &&
|
|
|
|
"Failed to move Buffer's start forward, or pointed prefix outside "
|
|
|
|
"of the buffer!");
|
2012-11-15 05:07:37 +08:00
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
// Location to use for error messages.
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
const char *UsedPrefixStart = UsedPrefix.data();
|
2012-11-15 05:07:37 +08:00
|
|
|
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
// Skip the buffer to the end.
|
2013-11-10 10:04:09 +08:00
|
|
|
Buffer = Buffer.drop_front(UsedPrefix.size() + CheckTypeSize(CheckTy));
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-03-01 06:13:03 +08:00
|
|
|
// Complain about useful-looking but unsupported suffixes.
|
|
|
|
if (CheckTy == Check::CheckBadNot) {
|
2016-12-11 17:54:36 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), SourceMgr::DK_Error,
|
2016-03-01 06:13:03 +08:00
|
|
|
"unsupported -NOT combo on prefix '" + UsedPrefix + "'");
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2013-09-18 06:30:02 +08:00
|
|
|
// Okay, we found the prefix, yay. Remember the rest of the line, but ignore
|
2016-12-18 17:41:20 +08:00
|
|
|
// leading whitespace.
|
2016-12-19 04:45:59 +08:00
|
|
|
if (!(NoCanonicalizeWhiteSpace && MatchFullLines))
|
|
|
|
Buffer = Buffer.substr(Buffer.find_first_not_of(" \t"));
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-07-09 02:44:05 +08:00
|
|
|
// Scan ahead to the end of line.
|
2009-09-21 06:11:44 +08:00
|
|
|
size_t EOL = Buffer.find_first_of("\n\r");
|
2009-09-25 04:39:13 +08:00
|
|
|
|
2010-01-30 05:53:18 +08:00
|
|
|
// Remember the location of the start of the pattern, for diagnostics.
|
|
|
|
SMLoc PatternLoc = SMLoc::getFromPointer(Buffer.data());
|
|
|
|
|
2009-09-25 04:39:13 +08:00
|
|
|
// Parse the pattern.
|
2013-09-18 06:30:02 +08:00
|
|
|
Pattern P(CheckTy);
|
2013-11-10 10:04:09 +08:00
|
|
|
if (P.ParsePattern(Buffer.substr(0, EOL), UsedPrefix, SM, LineNumber))
|
2009-07-09 02:44:05 +08:00
|
|
|
return true;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2013-07-12 22:51:05 +08:00
|
|
|
// Verify that CHECK-LABEL lines do not define or use variables
|
2013-09-18 06:30:02 +08:00
|
|
|
if ((CheckTy == Check::CheckLabel) && P.hasVariable()) {
|
2016-12-11 17:54:36 +08:00
|
|
|
SM.PrintMessage(
|
|
|
|
SMLoc::getFromPointer(UsedPrefixStart), SourceMgr::DK_Error,
|
|
|
|
"found '" + UsedPrefix + "-LABEL:'"
|
|
|
|
" with variable definition or use");
|
2013-07-12 22:51:05 +08:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2009-09-25 04:39:13 +08:00
|
|
|
Buffer = Buffer.substr(EOL);
|
|
|
|
|
2009-08-16 02:32:21 +08:00
|
|
|
// Verify that CHECK-NEXT lines have at least one CHECK line before them.
|
2015-02-26 12:53:00 +08:00
|
|
|
if ((CheckTy == Check::CheckNext || CheckTy == Check::CheckSame) &&
|
|
|
|
CheckStrings.empty()) {
|
|
|
|
StringRef Type = CheckTy == Check::CheckNext ? "NEXT" : "SAME";
|
2013-11-10 10:04:09 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(UsedPrefixStart),
|
2011-10-16 13:43:57 +08:00
|
|
|
SourceMgr::DK_Error,
|
2016-12-11 17:54:36 +08:00
|
|
|
"found '" + UsedPrefix + "-" + Type +
|
|
|
|
"' without previous '" + UsedPrefix + ": line");
|
2009-08-16 02:32:21 +08:00
|
|
|
return true;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2013-05-15 04:34:12 +08:00
|
|
|
// Handle CHECK-DAG/-NOT.
|
2013-09-18 06:30:02 +08:00
|
|
|
if (CheckTy == Check::CheckDAG || CheckTy == Check::CheckNot) {
|
2013-05-15 04:34:12 +08:00
|
|
|
DagNotMatches.push_back(P);
|
2009-09-25 04:39:13 +08:00
|
|
|
continue;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-07-09 02:44:05 +08:00
|
|
|
// Okay, add the string we captured to the output vector and move on.
|
2016-02-12 00:46:09 +08:00
|
|
|
CheckStrings.emplace_back(P, UsedPrefix, PatternLoc);
|
2013-05-15 04:34:12 +08:00
|
|
|
std::swap(DagNotMatches, CheckStrings.back().DagNotStrings);
|
2014-07-11 20:39:32 +08:00
|
|
|
DagNotMatches = ImplicitNegativeChecks;
|
2009-07-09 02:44:05 +08:00
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
// Add an EOF pattern for any trailing CHECK-DAG/-NOTs, and use the first
|
|
|
|
// prefix as a filler for the error message.
|
2013-05-15 04:34:12 +08:00
|
|
|
if (!DagNotMatches.empty()) {
|
2015-05-30 03:43:39 +08:00
|
|
|
CheckStrings.emplace_back(Pattern(Check::CheckEOF), *CheckPrefixes.begin(),
|
2016-02-12 00:46:09 +08:00
|
|
|
SMLoc::getFromPointer(Buffer.data()));
|
2013-05-15 04:34:12 +08:00
|
|
|
std::swap(DagNotMatches, CheckStrings.back().DagNotStrings);
|
2010-10-16 01:47:12 +08:00
|
|
|
}
|
|
|
|
|
2009-07-09 02:44:05 +08:00
|
|
|
if (CheckStrings.empty()) {
|
2013-11-10 10:04:09 +08:00
|
|
|
errs() << "error: no check strings found with prefix"
|
|
|
|
<< (CheckPrefixes.size() > 1 ? "es " : " ");
|
2015-04-30 05:45:24 +08:00
|
|
|
prefix_iterator I = CheckPrefixes.begin();
|
|
|
|
prefix_iterator E = CheckPrefixes.end();
|
|
|
|
if (I != E) {
|
|
|
|
errs() << "\'" << *I << ":'";
|
|
|
|
++I;
|
2013-11-10 10:04:09 +08:00
|
|
|
}
|
2016-02-12 00:46:09 +08:00
|
|
|
for (; I != E; ++I)
|
2015-04-30 05:45:24 +08:00
|
|
|
errs() << ", \'" << *I << ":'";
|
2013-11-10 10:04:09 +08:00
|
|
|
|
|
|
|
errs() << '\n';
|
2009-07-09 02:44:05 +08:00
|
|
|
return true;
|
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-07-09 02:44:05 +08:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
static void PrintCheckFailed(const SourceMgr &SM, SMLoc Loc, const Pattern &Pat,
|
|
|
|
StringRef Buffer,
|
2009-11-23 06:08:06 +08:00
|
|
|
StringMap<StringRef> &VariableTable) {
|
2009-08-16 02:32:21 +08:00
|
|
|
// Otherwise, we have an error, emit an error message.
|
2013-05-15 04:34:12 +08:00
|
|
|
SM.PrintMessage(Loc, SourceMgr::DK_Error,
|
2011-10-16 13:43:57 +08:00
|
|
|
"expected string not found in input");
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-08-16 02:32:21 +08:00
|
|
|
// Print the "scanning from here" line. If the current position is at the
|
|
|
|
// end of a line, advance to the start of the next line.
|
2009-09-21 06:11:44 +08:00
|
|
|
Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r"));
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2011-10-16 13:43:57 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), SourceMgr::DK_Note,
|
|
|
|
"scanning from here");
|
2009-11-23 06:08:06 +08:00
|
|
|
|
|
|
|
// Allow the pattern to print additional information if desired.
|
2013-05-15 04:34:12 +08:00
|
|
|
Pat.PrintFailureInfo(SM, Buffer, VariableTable);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void PrintCheckFailed(const SourceMgr &SM, const CheckString &CheckStr,
|
|
|
|
StringRef Buffer,
|
|
|
|
StringMap<StringRef> &VariableTable) {
|
|
|
|
PrintCheckFailed(SM, CheckStr.Loc, CheckStr.Pat, Buffer, VariableTable);
|
2009-08-16 02:32:21 +08:00
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Count the number of newlines in the specified range.
|
2014-04-08 01:09:53 +08:00
|
|
|
static unsigned CountNumNewlinesBetween(StringRef Range,
|
|
|
|
const char *&FirstNewLine) {
|
2009-08-16 02:32:21 +08:00
|
|
|
unsigned NumNewLines = 0;
|
2009-09-21 06:42:44 +08:00
|
|
|
while (1) {
|
2009-08-16 02:32:21 +08:00
|
|
|
// Scan for newline.
|
2009-09-21 06:42:44 +08:00
|
|
|
Range = Range.substr(Range.find_first_of("\n\r"));
|
2016-12-11 17:54:36 +08:00
|
|
|
if (Range.empty())
|
|
|
|
return NumNewLines;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-08-16 02:32:21 +08:00
|
|
|
++NumNewLines;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2009-08-16 02:32:21 +08:00
|
|
|
// Handle \n\r and \r\n as a single newline.
|
2016-12-11 17:54:36 +08:00
|
|
|
if (Range.size() > 1 && (Range[1] == '\n' || Range[1] == '\r') &&
|
2009-09-21 06:42:44 +08:00
|
|
|
(Range[0] != Range[1]))
|
|
|
|
Range = Range.substr(1);
|
|
|
|
Range = Range.substr(1);
|
2014-04-08 01:09:53 +08:00
|
|
|
|
|
|
|
if (NumNewLines == 1)
|
|
|
|
FirstNewLine = Range.begin();
|
2009-08-16 02:32:21 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Match check string and its "not strings" and/or "dag strings".
|
2013-05-15 04:29:52 +08:00
|
|
|
size_t CheckString::Check(const SourceMgr &SM, StringRef Buffer,
|
2013-10-12 02:38:36 +08:00
|
|
|
bool IsLabelScanMode, size_t &MatchLen,
|
2013-05-15 04:29:52 +08:00
|
|
|
StringMap<StringRef> &VariableTable) const {
|
2013-05-15 04:34:12 +08:00
|
|
|
size_t LastPos = 0;
|
|
|
|
std::vector<const Pattern *> NotStrings;
|
|
|
|
|
2013-10-12 02:38:36 +08:00
|
|
|
// IsLabelScanMode is true when we are scanning forward to find CHECK-LABEL
|
|
|
|
// bounds; we have not processed variable definitions within the bounded block
|
|
|
|
// yet so cannot handle any final CHECK-DAG yet; this is handled when going
|
|
|
|
// over the block again (including the last CHECK-LABEL) in normal mode.
|
|
|
|
if (!IsLabelScanMode) {
|
|
|
|
// Match "dag strings" (with mixed "not strings" if any).
|
|
|
|
LastPos = CheckDag(SM, Buffer, NotStrings, VariableTable);
|
|
|
|
if (LastPos == StringRef::npos)
|
|
|
|
return StringRef::npos;
|
|
|
|
}
|
2013-05-15 04:34:12 +08:00
|
|
|
|
|
|
|
// Match itself from the last position after matching CHECK-DAG.
|
|
|
|
StringRef MatchBuffer = Buffer.substr(LastPos);
|
|
|
|
size_t MatchPos = Pat.Match(MatchBuffer, MatchLen, VariableTable);
|
2013-05-15 04:29:52 +08:00
|
|
|
if (MatchPos == StringRef::npos) {
|
2013-05-15 04:34:12 +08:00
|
|
|
PrintCheckFailed(SM, *this, MatchBuffer, VariableTable);
|
2013-05-15 04:29:52 +08:00
|
|
|
return StringRef::npos;
|
|
|
|
}
|
|
|
|
|
2013-10-12 02:38:36 +08:00
|
|
|
// Similar to the above, in "label-scan mode" we can't yet handle CHECK-NEXT
|
|
|
|
// or CHECK-NOT
|
|
|
|
if (!IsLabelScanMode) {
|
2013-07-12 22:51:05 +08:00
|
|
|
StringRef SkippedRegion = Buffer.substr(LastPos, MatchPos);
|
2013-05-15 04:29:52 +08:00
|
|
|
|
2013-07-12 22:51:05 +08:00
|
|
|
// If this check is a "CHECK-NEXT", verify that the previous match was on
|
|
|
|
// the previous line (i.e. that there is one newline between them).
|
|
|
|
if (CheckNext(SM, SkippedRegion))
|
|
|
|
return StringRef::npos;
|
2013-05-15 04:29:52 +08:00
|
|
|
|
2015-02-26 12:53:00 +08:00
|
|
|
// If this check is a "CHECK-SAME", verify that the previous match was on
|
|
|
|
// the same line (i.e. that there is no newline between them).
|
|
|
|
if (CheckSame(SM, SkippedRegion))
|
|
|
|
return StringRef::npos;
|
|
|
|
|
2013-07-12 22:51:05 +08:00
|
|
|
// If this match had "not strings", verify that they don't exist in the
|
|
|
|
// skipped region.
|
|
|
|
if (CheckNot(SM, SkippedRegion, NotStrings, VariableTable))
|
|
|
|
return StringRef::npos;
|
|
|
|
}
|
2013-05-15 04:29:52 +08:00
|
|
|
|
2015-03-12 08:07:29 +08:00
|
|
|
return LastPos + MatchPos;
|
2013-05-15 04:29:52 +08:00
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Verify there is a single line in the given buffer.
|
2013-05-15 04:29:52 +08:00
|
|
|
bool CheckString::CheckNext(const SourceMgr &SM, StringRef Buffer) const {
|
2016-02-12 00:46:09 +08:00
|
|
|
if (Pat.getCheckTy() != Check::CheckNext)
|
2013-05-15 04:29:52 +08:00
|
|
|
return false;
|
|
|
|
|
|
|
|
// Count the number of newlines between the previous match and this one.
|
|
|
|
assert(Buffer.data() !=
|
2016-12-11 17:54:36 +08:00
|
|
|
SM.getMemoryBuffer(SM.FindBufferContainingLoc(
|
|
|
|
SMLoc::getFromPointer(Buffer.data())))
|
|
|
|
->getBufferStart() &&
|
2013-05-15 04:29:52 +08:00
|
|
|
"CHECK-NEXT can't be the first check in a file");
|
|
|
|
|
2014-06-09 06:29:17 +08:00
|
|
|
const char *FirstNewLine = nullptr;
|
2014-04-08 01:09:53 +08:00
|
|
|
unsigned NumNewLines = CountNumNewlinesBetween(Buffer, FirstNewLine);
|
2013-05-15 04:29:52 +08:00
|
|
|
|
|
|
|
if (NumNewLines == 0) {
|
2016-12-11 17:54:36 +08:00
|
|
|
SM.PrintMessage(Loc, SourceMgr::DK_Error,
|
|
|
|
Prefix + "-NEXT: is on the same line as previous match");
|
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.end()), SourceMgr::DK_Note,
|
|
|
|
"'next' match was here");
|
2013-05-15 04:29:52 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), SourceMgr::DK_Note,
|
|
|
|
"previous match ended here");
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (NumNewLines != 1) {
|
2016-12-11 17:54:36 +08:00
|
|
|
SM.PrintMessage(Loc, SourceMgr::DK_Error,
|
|
|
|
Prefix +
|
|
|
|
"-NEXT: is not on the line after the previous match");
|
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.end()), SourceMgr::DK_Note,
|
|
|
|
"'next' match was here");
|
2013-05-15 04:29:52 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), SourceMgr::DK_Note,
|
|
|
|
"previous match ended here");
|
2014-04-08 01:09:53 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(FirstNewLine), SourceMgr::DK_Note,
|
|
|
|
"non-matching line after previous match is here");
|
2013-05-15 04:29:52 +08:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Verify there is no newline in the given buffer.
|
2015-02-26 12:53:00 +08:00
|
|
|
bool CheckString::CheckSame(const SourceMgr &SM, StringRef Buffer) const {
|
2016-02-12 00:46:09 +08:00
|
|
|
if (Pat.getCheckTy() != Check::CheckSame)
|
2015-02-26 12:53:00 +08:00
|
|
|
return false;
|
|
|
|
|
|
|
|
// Count the number of newlines between the previous match and this one.
|
|
|
|
assert(Buffer.data() !=
|
|
|
|
SM.getMemoryBuffer(SM.FindBufferContainingLoc(
|
|
|
|
SMLoc::getFromPointer(Buffer.data())))
|
|
|
|
->getBufferStart() &&
|
|
|
|
"CHECK-SAME can't be the first check in a file");
|
|
|
|
|
|
|
|
const char *FirstNewLine = nullptr;
|
|
|
|
unsigned NumNewLines = CountNumNewlinesBetween(Buffer, FirstNewLine);
|
|
|
|
|
|
|
|
if (NumNewLines != 0) {
|
|
|
|
SM.PrintMessage(Loc, SourceMgr::DK_Error,
|
|
|
|
Prefix +
|
|
|
|
"-SAME: is not on the same line as the previous match");
|
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.end()), SourceMgr::DK_Note,
|
|
|
|
"'next' match was here");
|
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), SourceMgr::DK_Note,
|
|
|
|
"previous match ended here");
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Verify there's no "not strings" in the given buffer.
|
2013-05-15 04:29:52 +08:00
|
|
|
bool CheckString::CheckNot(const SourceMgr &SM, StringRef Buffer,
|
2013-05-15 04:34:12 +08:00
|
|
|
const std::vector<const Pattern *> &NotStrings,
|
2013-05-15 04:29:52 +08:00
|
|
|
StringMap<StringRef> &VariableTable) const {
|
2015-09-10 19:59:55 +08:00
|
|
|
for (const Pattern *Pat : NotStrings) {
|
2013-09-18 06:30:02 +08:00
|
|
|
assert((Pat->getCheckTy() == Check::CheckNot) && "Expect CHECK-NOT!");
|
2013-05-15 04:34:12 +08:00
|
|
|
|
2013-05-15 04:29:52 +08:00
|
|
|
size_t MatchLen = 0;
|
2013-05-15 04:34:12 +08:00
|
|
|
size_t Pos = Pat->Match(Buffer, MatchLen, VariableTable);
|
2013-05-15 04:29:52 +08:00
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
if (Pos == StringRef::npos)
|
|
|
|
continue;
|
2013-05-15 04:29:52 +08:00
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + Pos),
|
|
|
|
SourceMgr::DK_Error, Prefix + "-NOT: string occurred!");
|
2013-05-15 04:34:12 +08:00
|
|
|
SM.PrintMessage(Pat->getLoc(), SourceMgr::DK_Note,
|
2013-11-10 10:04:09 +08:00
|
|
|
Prefix + "-NOT: pattern specified here");
|
2013-05-15 04:29:52 +08:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2016-12-11 18:16:21 +08:00
|
|
|
/// Match "dag strings" and their mixed "not strings".
|
2013-05-15 04:34:12 +08:00
|
|
|
size_t CheckString::CheckDag(const SourceMgr &SM, StringRef Buffer,
|
|
|
|
std::vector<const Pattern *> &NotStrings,
|
|
|
|
StringMap<StringRef> &VariableTable) const {
|
|
|
|
if (DagNotStrings.empty())
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
size_t LastPos = 0;
|
|
|
|
size_t StartPos = LastPos;
|
|
|
|
|
2015-09-10 19:59:55 +08:00
|
|
|
for (const Pattern &Pat : DagNotStrings) {
|
2013-09-18 06:30:02 +08:00
|
|
|
assert((Pat.getCheckTy() == Check::CheckDAG ||
|
|
|
|
Pat.getCheckTy() == Check::CheckNot) &&
|
2013-05-15 04:34:12 +08:00
|
|
|
"Invalid CHECK-DAG or CHECK-NOT!");
|
|
|
|
|
2013-09-18 06:30:02 +08:00
|
|
|
if (Pat.getCheckTy() == Check::CheckNot) {
|
2013-05-15 04:34:12 +08:00
|
|
|
NotStrings.push_back(&Pat);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2013-09-18 06:30:02 +08:00
|
|
|
assert((Pat.getCheckTy() == Check::CheckDAG) && "Expect CHECK-DAG!");
|
2013-05-15 04:34:12 +08:00
|
|
|
|
|
|
|
size_t MatchLen = 0, MatchPos;
|
|
|
|
|
|
|
|
// CHECK-DAG always matches from the start.
|
|
|
|
StringRef MatchBuffer = Buffer.substr(StartPos);
|
|
|
|
MatchPos = Pat.Match(MatchBuffer, MatchLen, VariableTable);
|
|
|
|
// With a group of CHECK-DAGs, a single mismatching means the match on
|
|
|
|
// that group of CHECK-DAGs fails immediately.
|
|
|
|
if (MatchPos == StringRef::npos) {
|
|
|
|
PrintCheckFailed(SM, Pat.getLoc(), Pat, MatchBuffer, VariableTable);
|
|
|
|
return StringRef::npos;
|
|
|
|
}
|
|
|
|
// Re-calc it as the offset relative to the start of the original string.
|
|
|
|
MatchPos += StartPos;
|
|
|
|
|
|
|
|
if (!NotStrings.empty()) {
|
|
|
|
if (MatchPos < LastPos) {
|
|
|
|
// Reordered?
|
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + MatchPos),
|
|
|
|
SourceMgr::DK_Error,
|
2013-11-10 10:04:09 +08:00
|
|
|
Prefix + "-DAG: found a match of CHECK-DAG"
|
2016-12-11 17:54:36 +08:00
|
|
|
" reordering across a CHECK-NOT");
|
2013-05-15 04:34:12 +08:00
|
|
|
SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + LastPos),
|
|
|
|
SourceMgr::DK_Note,
|
2013-11-10 10:04:09 +08:00
|
|
|
Prefix + "-DAG: the farthest match of CHECK-DAG"
|
2016-12-11 17:54:36 +08:00
|
|
|
" is found here");
|
2013-05-15 04:34:12 +08:00
|
|
|
SM.PrintMessage(NotStrings[0]->getLoc(), SourceMgr::DK_Note,
|
2013-11-10 10:04:09 +08:00
|
|
|
Prefix + "-NOT: the crossed pattern specified"
|
2016-12-11 17:54:36 +08:00
|
|
|
" here");
|
2013-05-15 04:34:12 +08:00
|
|
|
SM.PrintMessage(Pat.getLoc(), SourceMgr::DK_Note,
|
2013-11-10 10:04:09 +08:00
|
|
|
Prefix + "-DAG: the reordered pattern specified"
|
2016-12-11 17:54:36 +08:00
|
|
|
" here");
|
2013-05-15 04:34:12 +08:00
|
|
|
return StringRef::npos;
|
|
|
|
}
|
|
|
|
// All subsequent CHECK-DAGs should be matched from the farthest
|
|
|
|
// position of all precedent CHECK-DAGs (including this one.)
|
|
|
|
StartPos = LastPos;
|
|
|
|
// If there's CHECK-NOTs between two CHECK-DAGs or from CHECK to
|
|
|
|
// CHECK-DAG, verify that there's no 'not' strings occurred in that
|
|
|
|
// region.
|
2017-06-07 20:06:45 +08:00
|
|
|
StringRef SkippedRegion = Buffer.slice(LastPos, MatchPos);
|
2013-08-02 19:32:50 +08:00
|
|
|
if (CheckNot(SM, SkippedRegion, NotStrings, VariableTable))
|
2013-05-15 04:34:12 +08:00
|
|
|
return StringRef::npos;
|
|
|
|
// Clear "not strings".
|
|
|
|
NotStrings.clear();
|
|
|
|
}
|
|
|
|
|
|
|
|
// Update the last position with CHECK-DAG matches.
|
|
|
|
LastPos = std::max(MatchPos + MatchLen, LastPos);
|
|
|
|
}
|
|
|
|
|
|
|
|
return LastPos;
|
|
|
|
}
|
|
|
|
|
2013-11-10 10:04:09 +08:00
|
|
|
// A check prefix must contain only alphanumeric, hyphens and underscores.
|
|
|
|
static bool ValidateCheckPrefix(StringRef CheckPrefix) {
|
|
|
|
Regex Validator("^[a-zA-Z0-9_-]*$");
|
|
|
|
return Validator.match(CheckPrefix);
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool ValidateCheckPrefixes() {
|
|
|
|
StringSet<> PrefixSet;
|
|
|
|
|
2015-09-10 19:59:55 +08:00
|
|
|
for (StringRef Prefix : CheckPrefixes) {
|
2014-07-30 04:30:53 +08:00
|
|
|
// Reject empty prefixes.
|
|
|
|
if (Prefix == "")
|
|
|
|
return false;
|
|
|
|
|
2014-11-19 10:56:00 +08:00
|
|
|
if (!PrefixSet.insert(Prefix).second)
|
2013-11-10 10:04:09 +08:00
|
|
|
return false;
|
|
|
|
|
|
|
|
if (!ValidateCheckPrefix(Prefix))
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
// Combines the check prefixes into a single regex so that we can efficiently
|
|
|
|
// scan for any of the set.
|
|
|
|
//
|
|
|
|
// The semantics are that the longest-match wins which matches our regex
|
|
|
|
// library.
|
|
|
|
static Regex buildCheckPrefixRegex() {
|
|
|
|
// I don't think there's a way to specify an initial value for cl::list,
|
|
|
|
// so if nothing was specified, add the default
|
2013-11-10 10:04:09 +08:00
|
|
|
if (CheckPrefixes.empty())
|
|
|
|
CheckPrefixes.push_back("CHECK");
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
|
|
|
|
// We already validated the contents of CheckPrefixes so just concatenate
|
|
|
|
// them as alternatives.
|
|
|
|
SmallString<32> PrefixRegexStr;
|
|
|
|
for (StringRef Prefix : CheckPrefixes) {
|
|
|
|
if (Prefix != CheckPrefixes.front())
|
|
|
|
PrefixRegexStr.push_back('|');
|
|
|
|
|
|
|
|
PrefixRegexStr.append(Prefix);
|
|
|
|
}
|
|
|
|
|
|
|
|
return Regex(PrefixRegexStr);
|
2013-08-13 07:05:59 +08:00
|
|
|
}
|
|
|
|
|
2016-05-28 05:23:25 +08:00
|
|
|
static void DumpCommandLine(int argc, char **argv) {
|
|
|
|
errs() << "FileCheck command line: ";
|
|
|
|
for (int I = 0; I < argc; I++)
|
|
|
|
errs() << " " << argv[I];
|
|
|
|
errs() << "\n";
|
|
|
|
}
|
|
|
|
|
2017-03-10 01:59:04 +08:00
|
|
|
// Remove local variables from \p VariableTable. Global variables
|
|
|
|
// (start with '$') are preserved.
|
|
|
|
static void ClearLocalVars(StringMap<StringRef> &VariableTable) {
|
|
|
|
SmallVector<StringRef, 16> LocalVars;
|
|
|
|
for (const auto &Var : VariableTable)
|
|
|
|
if (Var.first()[0] != '$')
|
|
|
|
LocalVars.push_back(Var.first());
|
|
|
|
|
|
|
|
for (const auto &Var : LocalVars)
|
|
|
|
VariableTable.erase(Var);
|
|
|
|
}
|
|
|
|
|
2016-12-11 17:50:05 +08:00
|
|
|
/// Check the input to FileCheck provided in the \p Buffer against the \p
|
|
|
|
/// CheckStrings read from the check file.
|
|
|
|
///
|
|
|
|
/// Returns false if the input fails to satisfy the checks.
|
|
|
|
bool CheckInput(SourceMgr &SM, StringRef Buffer,
|
|
|
|
ArrayRef<CheckString> CheckStrings) {
|
|
|
|
bool ChecksFailed = false;
|
2014-08-22 04:44:56 +08:00
|
|
|
|
|
|
|
/// VariableTable - This holds all the current filecheck variables.
|
|
|
|
StringMap<StringRef> VariableTable;
|
|
|
|
|
2017-11-07 21:24:44 +08:00
|
|
|
for (const auto& Def : GlobalDefines)
|
|
|
|
VariableTable.insert(StringRef(Def).split('='));
|
|
|
|
|
2013-07-12 22:51:05 +08:00
|
|
|
unsigned i = 0, j = 0, e = CheckStrings.size();
|
|
|
|
while (true) {
|
|
|
|
StringRef CheckRegion;
|
|
|
|
if (j == e) {
|
|
|
|
CheckRegion = Buffer;
|
|
|
|
} else {
|
|
|
|
const CheckString &CheckLabelStr = CheckStrings[j];
|
2016-02-12 00:46:09 +08:00
|
|
|
if (CheckLabelStr.Pat.getCheckTy() != Check::CheckLabel) {
|
2013-07-12 22:51:05 +08:00
|
|
|
++j;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Scan to next CHECK-LABEL match, ignoring CHECK-NOT and CHECK-DAG
|
|
|
|
size_t MatchLabelLen = 0;
|
2016-12-11 17:54:36 +08:00
|
|
|
size_t MatchLabelPos =
|
|
|
|
CheckLabelStr.Check(SM, Buffer, true, MatchLabelLen, VariableTable);
|
2016-12-11 17:50:05 +08:00
|
|
|
if (MatchLabelPos == StringRef::npos)
|
|
|
|
// Immediately bail of CHECK-LABEL fails, nothing else we can do.
|
|
|
|
return false;
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2013-07-12 22:51:05 +08:00
|
|
|
CheckRegion = Buffer.substr(0, MatchLabelPos + MatchLabelLen);
|
|
|
|
Buffer = Buffer.substr(MatchLabelPos + MatchLabelLen);
|
|
|
|
++j;
|
|
|
|
}
|
2009-08-16 02:32:21 +08:00
|
|
|
|
2017-03-10 01:59:04 +08:00
|
|
|
if (EnableVarScope)
|
|
|
|
ClearLocalVars(VariableTable);
|
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
for (; i != j; ++i) {
|
2013-07-12 22:51:05 +08:00
|
|
|
const CheckString &CheckStr = CheckStrings[i];
|
|
|
|
|
|
|
|
// Check each string within the scanned region, including a second check
|
|
|
|
// of any final CHECK-LABEL (to verify CHECK-NOT and CHECK-DAG)
|
|
|
|
size_t MatchLen = 0;
|
2016-12-11 17:54:36 +08:00
|
|
|
size_t MatchPos =
|
|
|
|
CheckStr.Check(SM, CheckRegion, false, MatchLen, VariableTable);
|
2013-07-12 22:51:05 +08:00
|
|
|
|
|
|
|
if (MatchPos == StringRef::npos) {
|
2016-12-11 17:50:05 +08:00
|
|
|
ChecksFailed = true;
|
2013-07-12 22:51:05 +08:00
|
|
|
i = j;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
CheckRegion = CheckRegion.substr(MatchPos + MatchLen);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (j == e)
|
|
|
|
break;
|
2009-07-09 02:44:05 +08:00
|
|
|
}
|
2010-08-21 01:38:38 +08:00
|
|
|
|
2016-12-11 17:50:05 +08:00
|
|
|
// Success if no checks failed.
|
|
|
|
return !ChecksFailed;
|
|
|
|
}
|
|
|
|
|
|
|
|
int main(int argc, char **argv) {
|
|
|
|
sys::PrintStackTraceOnErrorSignal(argv[0]);
|
|
|
|
PrettyStackTraceProgram X(argc, argv);
|
|
|
|
cl::ParseCommandLineOptions(argc, argv);
|
|
|
|
|
|
|
|
if (!ValidateCheckPrefixes()) {
|
|
|
|
errs() << "Supplied check-prefix is invalid! Prefixes must be unique and "
|
|
|
|
"start with a letter and contain only alphanumeric characters, "
|
|
|
|
"hyphens and underscores\n";
|
|
|
|
return 2;
|
|
|
|
}
|
|
|
|
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
Regex PrefixRE = buildCheckPrefixRegex();
|
|
|
|
std::string REError;
|
|
|
|
if (!PrefixRE.isValid(REError)) {
|
|
|
|
errs() << "Unable to combine check-prefix strings into a prefix regular "
|
|
|
|
"expression! This is likely a bug in FileCheck's verification of "
|
|
|
|
"the check-prefix strings. Regular expression parsing failed "
|
|
|
|
"with the following error: "
|
|
|
|
<< REError << "\n";
|
|
|
|
return 2;
|
|
|
|
}
|
2016-12-11 17:50:05 +08:00
|
|
|
|
|
|
|
SourceMgr SM;
|
|
|
|
|
|
|
|
// Read the expected strings from the check file.
|
|
|
|
ErrorOr<std::unique_ptr<MemoryBuffer>> CheckFileOrErr =
|
|
|
|
MemoryBuffer::getFileOrSTDIN(CheckFilename);
|
|
|
|
if (std::error_code EC = CheckFileOrErr.getError()) {
|
|
|
|
errs() << "Could not open check file '" << CheckFilename
|
|
|
|
<< "': " << EC.message() << '\n';
|
|
|
|
return 2;
|
|
|
|
}
|
|
|
|
MemoryBuffer &CheckFile = *CheckFileOrErr.get();
|
|
|
|
|
|
|
|
SmallString<4096> CheckFileBuffer;
|
2016-12-11 18:22:17 +08:00
|
|
|
StringRef CheckFileText = CanonicalizeFile(CheckFile, CheckFileBuffer);
|
2016-12-11 17:50:05 +08:00
|
|
|
|
|
|
|
SM.AddNewSourceBuffer(MemoryBuffer::getMemBuffer(
|
|
|
|
CheckFileText, CheckFile.getBufferIdentifier()),
|
|
|
|
SMLoc());
|
|
|
|
|
|
|
|
std::vector<CheckString> CheckStrings;
|
[FileCheck] Re-implement the logic to find each check prefix in the
check file to not be unreasonably slow in the face of multiple check
prefixes.
The previous logic would repeatedly scan potentially large portions of
the check file looking for alternative prefixes. In the worst case this
would scan most of the file looking for a rare prefix between every
single occurance of a common prefix. Even if we bounded the scan, this
would do bad things if the order of the prefixes was "unlucky" and the
distant prefix was scanned for first.
None of this is necessary. It is straightforward to build a state
machine that recognizes the first, longest of the set of alternative
prefixes. That is in fact exactly whan a regular expression does.
This patch builds a regular expression once for the set of prefixes and
then uses it to search incrementally for the next prefix. This requires
some threading of state but actually makes the code dramatically
simpler. I've also added a big comment describing the algorithm as it
was not at all obvious to me when I started.
With this patch, several previously pathological test cases in
test/CodeGen/X86 are 5x and more faster. Overall, running all tests
under test/CodeGen/X86 uses 10% less CPU after this, and because all the
slowest tests were hitting this, finishes in 40% less wall time on my
system (going from just over 5.38s to just over 3.23s) on a release
build! This patch substantially improves the time of all 7 X86 tests
that were in the top 20 reported by --time-tests, 5 of them are
completely off the list and the remaining 2 are much lower. (Sadly, the
new tests on the list include 2 new X86 ones that are slow for unrelated
reasons, so the count stays at 4 of the top 20.)
It isn't clear how much this helps debug builds in aggregate in part
because of the noise, but it again makes mane of the slowest x86 tests
significantly faster (10% or more improvement).
llvm-svn: 289382
2016-12-11 20:49:05 +08:00
|
|
|
if (ReadCheckFile(SM, CheckFileText, PrefixRE, CheckStrings))
|
2016-12-11 17:50:05 +08:00
|
|
|
return 2;
|
|
|
|
|
|
|
|
// Open the file to check and add it to SourceMgr.
|
|
|
|
ErrorOr<std::unique_ptr<MemoryBuffer>> InputFileOrErr =
|
|
|
|
MemoryBuffer::getFileOrSTDIN(InputFilename);
|
|
|
|
if (std::error_code EC = InputFileOrErr.getError()) {
|
|
|
|
errs() << "Could not open input file '" << InputFilename
|
|
|
|
<< "': " << EC.message() << '\n';
|
|
|
|
return 2;
|
|
|
|
}
|
|
|
|
MemoryBuffer &InputFile = *InputFileOrErr.get();
|
|
|
|
|
|
|
|
if (InputFile.getBufferSize() == 0 && !AllowEmptyInput) {
|
|
|
|
errs() << "FileCheck error: '" << InputFilename << "' is empty.\n";
|
|
|
|
DumpCommandLine(argc, argv);
|
|
|
|
return 2;
|
|
|
|
}
|
|
|
|
|
|
|
|
SmallString<4096> InputFileBuffer;
|
2016-12-11 18:22:17 +08:00
|
|
|
StringRef InputFileText = CanonicalizeFile(InputFile, InputFileBuffer);
|
2016-12-11 17:50:05 +08:00
|
|
|
|
2016-12-11 17:54:36 +08:00
|
|
|
SM.AddNewSourceBuffer(MemoryBuffer::getMemBuffer(
|
|
|
|
InputFileText, InputFile.getBufferIdentifier()),
|
|
|
|
SMLoc());
|
2016-12-11 17:50:05 +08:00
|
|
|
|
|
|
|
return CheckInput(SM, InputFileText, CheckStrings) ? EXIT_SUCCESS : 1;
|
2009-07-09 02:44:05 +08:00
|
|
|
}
|