llvm/lib/Support/FormattedStream.cpp

   1 //===-- llvm/Support/FormattedStream.cpp - Formatted streams ----*- C++ -*-===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8 //
   9 // This file contains the implementation of formatted_raw_ostream.
  10 //
  11 //===----------------------------------------------------------------------===//
  12
  13 #include "llvm/Support/FormattedStream.h"
  14 #include "llvm/Support/ConvertUTF.h"
  15 #include "llvm/Support/Debug.h"
  16 #include "llvm/Support/Unicode.h"
  17 #include "llvm/Support/raw_ostream.h"
  18 #include <algorithm>
  19
  20 using namespace llvm;
  21
  22 /// UpdatePosition - Examine the given char sequence and figure out which
  23 /// column we end up in after output, and how many line breaks are contained.
  24 /// This assumes that the input string is well-formed UTF-8, and takes into
  25 /// account Unicode characters which render as multiple columns wide.
  26 void formatted_raw_ostream::UpdatePosition(const char *Ptr, size_t Size) {
  27   unsigned &Column = Position.first;
  28   unsigned &Line = Position.second;
  29
  30   auto ProcessUTF8CodePoint = [&Line, &Column](StringRef CP) {
  31     int Width = sys::unicode::columnWidthUTF8(CP);
  32     if (Width != sys::unicode::ErrorNonPrintableCharacter)
  33       Column += Width;
  34
  35     // The only special whitespace characters we care about are single-byte.
  36     if (CP.size() > 1)
  37       return;
  38
  39     switch (CP[0]) {
  40     case '\n':
  41       Line += 1;
  42       [[fallthrough]];
  43     case '\r':
  44       Column = 0;
  45       break;
  46     case '\t':
  47       // Assumes tab stop = 8 characters.
  48       Column += (8 - (Column & 0x7)) & 0x7;
  49       break;
  50     }
  51   };
  52
  53   // If we have a partial UTF-8 sequence from the previous buffer, check that
  54   // first.
  55   if (PartialUTF8Char.size()) {
  56     size_t BytesFromBuffer =
  57         getNumBytesForUTF8(PartialUTF8Char[0]) - PartialUTF8Char.size();
  58     if (Size < BytesFromBuffer) {
  59       // If we still don't have enough bytes for a complete code point, just
  60       // append what we have.
  61       PartialUTF8Char.append(StringRef(Ptr, Size));
  62       return;
  63     } else {
  64       // The first few bytes from the buffer will complete the code point.
  65       // Concatenate them and process their effect on the line and column
  66       // numbers.
  67       PartialUTF8Char.append(StringRef(Ptr, BytesFromBuffer));
  68       ProcessUTF8CodePoint(PartialUTF8Char);
  69       PartialUTF8Char.clear();
  70       Ptr += BytesFromBuffer;
  71       Size -= BytesFromBuffer;
  72     }
  73   }
  74
  75   // Now scan the rest of the buffer.
  76   unsigned NumBytes;
  77   for (const char *End = Ptr + Size; Ptr < End; Ptr += NumBytes) {
  78     // Fast path for printable ASCII characters without special handling.
  79     if (*Ptr >= 0x20 && *Ptr <= 0x7e) {
  80       NumBytes = 1;
  81       ++Column;
  82       continue;
  83     }
  84
  85     NumBytes = getNumBytesForUTF8(*Ptr);
  86
  87     // The buffer might end part way through a UTF-8 code unit sequence for a
  88     // Unicode scalar value if it got flushed. If this happens, we can't know
  89     // the display width until we see the rest of the code point. Stash the
  90     // bytes we do have, so that we can reconstruct the whole code point later,
  91     // even if the buffer is being flushed.
  92     if ((unsigned)(End - Ptr) < NumBytes) {
  93       PartialUTF8Char = StringRef(Ptr, End - Ptr);
  94       return;
  95     }
  96
  97     ProcessUTF8CodePoint(StringRef(Ptr, NumBytes));
  98   }
  99 }
 100
 101 /// ComputePosition - Examine the current output and update line and column
 102 /// counts.
 103 void formatted_raw_ostream::ComputePosition(const char *Ptr, size_t Size) {
 104   if (DisableScan)
 105     return;
 106
 107   // If our previous scan pointer is inside the buffer, assume we already
 108   // scanned those bytes. This depends on raw_ostream to not change our buffer
 109   // in unexpected ways.
 110   if (Ptr <= Scanned && Scanned <= Ptr + Size)
 111     // Scan all characters added since our last scan to determine the new
 112     // column.
 113     UpdatePosition(Scanned, Size - (Scanned - Ptr));
 114   else
 115     UpdatePosition(Ptr, Size);
 116
 117   // Update the scanning pointer.
 118   Scanned = Ptr + Size;
 119 }
 120
 121 /// PadToColumn - Align the output to some column number.
 122 ///
 123 /// \param NewCol - The column to move to.
 124 ///
 125 formatted_raw_ostream &formatted_raw_ostream::PadToColumn(unsigned NewCol) {
 126   // Figure out what's in the buffer and add it to the column count.
 127   ComputePosition(getBufferStart(), GetNumBytesInBuffer());
 128
 129   // Output spaces until we reach the desired column.
 130   indent(std::max(int(NewCol - getColumn()), 1));
 131   return *this;
 132 }
 133
 134 void formatted_raw_ostream::write_impl(const char *Ptr, size_t Size) {
 135   // Figure out what's in the buffer and add it to the column count.
 136   ComputePosition(Ptr, Size);
 137
 138   // Write the data to the underlying stream (which is unbuffered, so
 139   // the data will be immediately written out).
 140   TheStream->write(Ptr, Size);
 141
 142   // Reset the scanning pointer.
 143   Scanned = nullptr;
 144 }
 145
 146 /// fouts() - This returns a reference to a formatted_raw_ostream for
 147 /// standard output.  Use it like: fouts() << "foo" << "bar";
 148 formatted_raw_ostream &llvm::fouts() {
 149   static formatted_raw_ostream S(outs());
 150   return S;
 151 }
 152
 153 /// ferrs() - This returns a reference to a formatted_raw_ostream for
 154 /// standard error.  Use it like: ferrs() << "foo" << "bar";
 155 formatted_raw_ostream &llvm::ferrs() {
 156   static formatted_raw_ostream S(errs());
 157   return S;
 158 }
 159
 160 /// fdbgs() - This returns a reference to a formatted_raw_ostream for
 161 /// the debug stream.  Use it like: fdbgs() << "foo" << "bar";
 162 formatted_raw_ostream &llvm::fdbgs() {
 163   static formatted_raw_ostream S(dbgs());
 164   return S;
 165 }