//== HTMLRewrite.cpp - Translate source code into prettified HTML --*- C++ -*-//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file defines the HTMLRewriter clas, which is used to translate the
// text of a source file into prettified HTML.
//
//===----------------------------------------------------------------------===//
#include "clang/Lex/Preprocessor.h"
#include "clang/Rewrite/Rewriter.h"
#include "clang/Rewrite/HTMLRewrite.h"
#include "clang/Lex/Preprocessor.h"
#include "clang/Basic/SourceManager.h"
#include "llvm/ADT/SmallString.h"
#include "llvm/ADT/OwningPtr.h"
#include "llvm/Support/MemoryBuffer.h"
#include "llvm/Support/raw_ostream.h"
using namespace clang;
/// HighlightRange - Highlight a range in the source code with the specified
/// start/end tags. B/E must be in the same file. This ensures that
/// start/end tags are placed at the start/end of each line if the range is
/// multiline.
void html::HighlightRange(Rewriter &R, SourceLocation B, SourceLocation E,
const char *StartTag, const char *EndTag) {
SourceManager &SM = R.getSourceMgr();
B = SM.getInstantiationLoc(B);
E = SM.getInstantiationLoc(E);
FileID FID = SM.getFileID(B);
assert(SM.getFileID(E) == FID && "B/E not in the same file!");
unsigned BOffset = SM.getFullFilePos(B);
unsigned EOffset = SM.getFullFilePos(E);
// Include the whole end token in the range.
EOffset += Lexer::MeasureTokenLength(E, R.getSourceMgr());
HighlightRange(R.getEditBuffer(FID), BOffset, EOffset,
SM.getBufferData(FID).first, StartTag, EndTag);
}
/// HighlightRange - This is the same as the above method, but takes
/// decomposed file locations.
void html::HighlightRange(RewriteBuffer &RB, unsigned B, unsigned E,
const char *BufferStart,
const char *StartTag, const char *EndTag) {
// Insert the tag at the absolute start/end of the range.
RB.InsertTextAfter(B, StartTag, strlen(StartTag));
RB.InsertTextBefore(E, EndTag, strlen(EndTag));
// Scan the range to see if there is a \r or \n. If so, and if the line is
// not blank, insert tags on that line as well.
bool HadOpenTag = true;
unsigned LastNonWhiteSpace = B;
for (unsigned i = B; i != E; ++i) {
switch (BufferStart[i]) {
case '\r':
case '\n':
// Okay, we found a newline in the range. If we have an open tag, we need
// to insert a close tag at the first non-whitespace before the newline.
if (HadOpenTag)
RB.InsertTextBefore(LastNonWhiteSpace+1, EndTag, strlen(EndTag));
// Instead of inserting an open tag immediately after the newline, we
// wait until we see a non-whitespace character. This prevents us from
// inserting tags around blank lines, and also allows the open tag to
// be put *after* whitespace on a non-blank line.
HadOpenTag = false;
break;
case '\0':
case ' ':
case '\t':
case '\f':
case '\v':
// Ignore whitespace.
break;
default:
// If there is no tag open, do it now.
if (!HadOpenTag) {
RB.InsertTextAfter(i, StartTag, strlen(StartTag));
HadOpenTag = true;
}
// Remember this character.
LastNonWhiteSpace = i;
break;
}
}
}
void html::EscapeText(Rewriter &R, FileID FID,
bool EscapeSpaces, bool ReplaceTabs) {
const llvm::MemoryBuffer *Buf = R.getSourceMgr().getBuffer(FID);
const char* C = Buf->getBufferStart();
const char* FileEnd = Buf->getBufferEnd();
assert (C <= FileEnd);
RewriteBuffer &RB = R.getEditBuffer(FID);
unsigned ColNo = 0;
for (unsigned FilePos = 0; C != FileEnd ; ++C, ++FilePos) {
switch (*C) {
default: ++ColNo; break;
case '\n':
case '\r':
ColNo = 0;
break;
case ' ':
if (EscapeSpaces)
RB.ReplaceText(FilePos, 1, " ", 6);
++ColNo;
break;
case '\f':
RB.ReplaceText(FilePos, 1, "
", 4);
ColNo = 0;
break;
case '\t': {
if (!ReplaceTabs)
break;
unsigned NumSpaces = 8-(ColNo&7);
if (EscapeSpaces)
RB.ReplaceText(FilePos, 1, " "
" ", 6*NumSpaces);
else
RB.ReplaceText(FilePos, 1, " ", NumSpaces);
ColNo += NumSpaces;
break;
}
case '<':
RB.ReplaceText(FilePos, 1, "<", 4);
++ColNo;
break;
case '>':
RB.ReplaceText(FilePos, 1, ">", 4);
++ColNo;
break;
case '&':
RB.ReplaceText(FilePos, 1, "&", 5);
++ColNo;
break;
}
}
}
std::string html::EscapeText(const std::string& s, bool EscapeSpaces,
bool ReplaceTabs) {
unsigned len = s.size();
std::string Str;
llvm::raw_string_ostream os(Str);
for (unsigned i = 0 ; i < len; ++i) {
char c = s[i];
switch (c) {
default:
os << c; break;
case ' ':
if (EscapeSpaces) os << " ";
else os << ' ';
break;
case '\t':
if (ReplaceTabs) {
if (EscapeSpaces)
for (unsigned i = 0; i < 4; ++i)
os << " ";
else
for (unsigned i = 0; i < 4; ++i)
os << " ";
}
else
os << c;
break;
case '<': os << "<"; break;
case '>': os << ">"; break;
case '&': os << "&"; break;
}
}
return os.str();
}
static void AddLineNumber(RewriteBuffer &RB, unsigned LineNo,
unsigned B, unsigned E) {
llvm::SmallString<100> Str;
Str += "| ";
Str.append_uint(LineNo);
Str += " | ";
if (B == E) { // Handle empty lines.
Str += " |
";
RB.InsertTextBefore(B, &Str[0], Str.size());
} else {
RB.InsertTextBefore(B, &Str[0], Str.size());
RB.InsertTextBefore(E, "", strlen(""));
}
}
void html::AddLineNumbers(Rewriter& R, FileID FID) {
const llvm::MemoryBuffer *Buf = R.getSourceMgr().getBuffer(FID);
const char* FileBeg = Buf->getBufferStart();
const char* FileEnd = Buf->getBufferEnd();
const char* C = FileBeg;
RewriteBuffer &RB = R.getEditBuffer(FID);
assert (C <= FileEnd);
unsigned LineNo = 0;
unsigned FilePos = 0;
while (C != FileEnd) {
++LineNo;
unsigned LineStartPos = FilePos;
unsigned LineEndPos = FileEnd - FileBeg;
assert (FilePos <= LineEndPos);
assert (C < FileEnd);
// Scan until the newline (or end-of-file).
while (C != FileEnd) {
char c = *C;
++C;
if (c == '\n') {
LineEndPos = FilePos++;
break;
}
++FilePos;
}
AddLineNumber(RB, LineNo, LineStartPos, LineEndPos);
}
// Add one big table tag that surrounds all of the code.
RB.InsertTextBefore(0, "\n",
strlen("\n"));
RB.InsertTextAfter(FileEnd - FileBeg, "
", strlen("
"));
}
void html::AddHeaderFooterInternalBuiltinCSS(Rewriter& R, FileID FID,
const char *title) {
const llvm::MemoryBuffer *Buf = R.getSourceMgr().getBuffer(FID);
const char* FileStart = Buf->getBufferStart();
const char* FileEnd = Buf->getBufferEnd();
SourceLocation StartLoc = R.getSourceMgr().getLocForStartOfFile(FID);
SourceLocation EndLoc = StartLoc.getFileLocWithOffset(FileEnd-FileStart);
std::string s;
llvm::raw_string_ostream os(s);
os << "\n" // Use HTML 5 doctype
"\n\n";
if (title)
os << "" << html::EscapeText(title) << "\n";
os << "\n\n";
// Generate header
R.InsertStrBefore(StartLoc, os.str());
// Generate footer
R.InsertCStrAfter(EndLoc, "\n");
}
/// SyntaxHighlight - Relex the specified FileID and annotate the HTML with
/// information about keywords, macro expansions etc. This uses the macro
/// table state from the end of the file, so it won't be perfectly perfect,
/// but it will be reasonably close.
void html::SyntaxHighlight(Rewriter &R, FileID FID, Preprocessor &PP) {
RewriteBuffer &RB = R.getEditBuffer(FID);
const SourceManager &SourceMgr = PP.getSourceManager();
Lexer L(FID, SourceMgr, PP.getLangOptions());
const char *BufferStart = L.getBufferStart();
// Inform the preprocessor that we want to retain comments as tokens, so we
// can highlight them.
L.SetCommentRetentionState(true);
// Lex all the tokens in raw mode, to avoid entering #includes or expanding
// macros.
Token Tok;
L.LexFromRawLexer(Tok);
while (Tok.isNot(tok::eof)) {
// Since we are lexing unexpanded tokens, all tokens are from the main
// FileID.
unsigned TokOffs = SourceMgr.getFullFilePos(Tok.getLocation());
unsigned TokLen = Tok.getLength();
switch (Tok.getKind()) {
default: break;
case tok::identifier: {
// Fill in Result.IdentifierInfo, looking up the identifier in the
// identifier table.
IdentifierInfo *II = PP.LookUpIdentifierInfo(Tok, BufferStart+TokOffs);
// If this is a pp-identifier, for a keyword, highlight it as such.
if (II->getTokenID() != tok::identifier)
HighlightRange(RB, TokOffs, TokOffs+TokLen, BufferStart,
"", "");
break;
}
case tok::comment:
HighlightRange(RB, TokOffs, TokOffs+TokLen, BufferStart,
"");
break;
case tok::wide_string_literal:
// Chop off the L prefix
++TokOffs;
--TokLen;
// FALL THROUGH.
case tok::string_literal:
HighlightRange(RB, TokOffs, TokOffs+TokLen, BufferStart,
"", "");
break;
case tok::hash: {
// If this is a preprocessor directive, all tokens to end of line are too.
if (!Tok.isAtStartOfLine())
break;
// Eat all of the tokens until we get to the next one at the start of
// line.
unsigned TokEnd = TokOffs+TokLen;
L.LexFromRawLexer(Tok);
while (!Tok.isAtStartOfLine() && Tok.isNot(tok::eof)) {
TokEnd = SourceMgr.getFullFilePos(Tok.getLocation())+Tok.getLength();
L.LexFromRawLexer(Tok);
}
// Find end of line. This is a hack.
HighlightRange(RB, TokOffs, TokEnd, BufferStart,
"", "");
// Don't skip the next token.
continue;
}
}
L.LexFromRawLexer(Tok);
}
}
/// HighlightMacros - This uses the macro table state from the end of the
/// file, to reexpand macros and insert (into the HTML) information about the
/// macro expansions. This won't be perfectly perfect, but it will be
/// reasonably close.
void html::HighlightMacros(Rewriter &R, FileID FID, Preprocessor& PP) {
RewriteBuffer &RB = R.getEditBuffer(FID);
// Inform the preprocessor that we don't want comments.
PP.SetCommentRetentionState(false, false);
// Start parsing the specified input file.
PP.EnterMainSourceFile();
// Lex all the tokens.
const SourceManager &SourceMgr = PP.getSourceManager();
Token Tok;
PP.Lex(Tok);
while (Tok.isNot(tok::eof)) {
// Ignore non-macro tokens.
if (!Tok.getLocation().isMacroID()) {
PP.Lex(Tok);
continue;
}
// Ignore tokens whose instantiation location was not the main file.
SourceLocation LLoc = SourceMgr.getInstantiationLoc(Tok.getLocation());
std::pair LLocInfo = SourceMgr.getDecomposedLoc(LLoc);
if (LLocInfo.first != FID) {
PP.Lex(Tok);
continue;
}
// Okay, we have the first token of a macro expansion: highlight the
// instantiation.
// Get the size of current macro call itself.
// FIXME: This should highlight the args of a function-like
// macro, using a heuristic.
unsigned TokLen = Lexer::MeasureTokenLength(LLoc, SourceMgr);
unsigned TokOffs = LLocInfo.second;
// Highlight the macro invocation itself.
RB.InsertTextAfter(TokOffs, "",
strlen(""));
RB.InsertTextBefore(TokOffs+TokLen, "", strlen(""));
std::string Expansion = PP.getSpelling(Tok);
unsigned LineLen = Expansion.size();
// Okay, eat this token, getting the next one.
PP.Lex(Tok);
// Skip all the rest of the tokens that are part of this macro
// instantiation. It would be really nice to pop up a window with all the
// spelling of the tokens or something.
while (!Tok.is(tok::eof) &&
SourceMgr.getInstantiationLoc(Tok.getLocation()) == LLoc) {
// Insert a newline if the macro expansion is getting large.
if (LineLen > 60) {
Expansion += "
";
LineLen = 0;
}
LineLen -= Expansion.size();
// Escape any special characters in the token text.
Expansion += ' ' + EscapeText(PP.getSpelling(Tok));
LineLen += Expansion.size();
PP.Lex(Tok);
}
// Insert the information about the expansion inside the macro span.
Expansion = "" + Expansion + "";
RB.InsertTextBefore(TokOffs+TokLen, Expansion.c_str(), Expansion.size());
}
}
void html::HighlightMacros(Rewriter &R, FileID FID,
PreprocessorFactory &PPF) {
llvm::OwningPtr PP(PPF.CreatePreprocessor());
HighlightMacros(R, FID, *PP);
}