1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
|
/* Distributed under the OSI-approved BSD 3-Clause License. See accompanying
file LICENSE.rst or https://cmake.org/licensing for details. */
#include "cmDocumentationFormatter.h"
#include <algorithm> // IWYU pragma: keep
#include <cassert>
#include <iomanip>
#include <iterator>
#include <ostream>
#include <string>
#include <vector>
#include <cm/string_view>
#include <cmext/string_view>
#include "cmDocumentationEntry.h"
#include "cmDocumentationSection.h"
#include "cmStringAlgorithms.h"
namespace {
auto const EOL = "\n"_s;
auto const SPACE = " "_s;
auto const TWO_SPACES = " "_s;
auto const MAX_WIDTH_PADDING =
std::string(cmDocumentationFormatter::TEXT_WIDTH, ' ');
void FormatLine(std::back_insert_iterator<std::vector<cm::string_view>> outIt,
cm::string_view const text, cm::string_view const padding)
{
auto tokens = cmTokenizedView(text, ' ', cmTokenizerMode::New);
if (tokens.empty()) {
return;
}
// Push padding in front of a first line
if (!padding.empty()) {
outIt = padding;
}
auto currentWidth = padding.size();
auto newSentence = false;
for (auto token : tokens) {
// It's no need to add a space if this is a very first
// word on a line.
auto const needSpace = currentWidth > padding.size();
// Evaluate the size of a current token + possibly spaces before it.
auto const tokenWithSpaceSize = token.size() + std::size_t(needSpace) +
std::size_t(needSpace && newSentence);
// Check if a current word fits on a line.
// Also, take in account:
// - extra space if not a first word on a line
// - extra space if last token ends w/ a period
if (currentWidth + tokenWithSpaceSize <=
cmDocumentationFormatter::TEXT_WIDTH) {
// If not a first word on a line...
if (needSpace) {
// ... add a space after the last token +
// possibly one more space if the last token
// ends with a period (means, end of a sentence).
outIt = newSentence ? TWO_SPACES : SPACE;
}
outIt = token;
currentWidth += tokenWithSpaceSize;
} else {
// Start a new line!
outIt = EOL;
if (!padding.empty()) {
outIt = padding;
}
outIt = token;
currentWidth = padding.size() + token.size();
}
// Start a new sentence if the current word ends with period
newSentence = token.back() == '.';
}
// Always add EOL at the end of formatted text
outIt = EOL;
}
} // anonymous namespace
std::string cmDocumentationFormatter::Format(std::string text) const
{
// Exit early on empty text
if (text.empty()) {
return {};
}
assert(this->TextIndent < this->TEXT_WIDTH);
auto const padding =
cm::string_view(MAX_WIDTH_PADDING.c_str(), this->TextIndent);
std::vector<cm::string_view> tokens;
auto outIt = std::back_inserter(tokens);
auto prevWasPreFormatted = false;
// NOTE Can't use `cmTokenizedView()` cuz every sequential EOL does matter
// (and `cmTokenizedView()` will squeeze 'em)
for ( // clang-format off
std::string::size_type start = 0
, end = text.find('\n')
; start < text.size()
; start = end + ((end != std::string::npos) ? 1 : 0)
, end = text.find('\n', start)
) // clang-format on
{
auto const isLastLine = end == std::string::npos;
auto const line = isLastLine
? cm::string_view{ text.c_str() + start }
: cm::string_view{ text.c_str() + start, end - start };
if (!line.empty() && line.front() == ' ') {
// Preformatted lines go as is w/ a leading padding
if (!padding.empty()) {
outIt = padding;
}
outIt = line;
prevWasPreFormatted = true;
} else {
// Separate a normal paragraph from a pre-formatted
// w/ an extra EOL
if (prevWasPreFormatted) {
outIt = EOL;
}
if (line.empty()) {
if (!isLastLine) {
outIt = EOL;
}
} else {
FormatLine(outIt, line, padding);
}
prevWasPreFormatted = false;
}
if (!isLastLine) {
outIt = EOL;
}
}
if (prevWasPreFormatted) {
outIt = EOL;
}
return cmJoinStrings(tokens, {}, {});
}
void cmDocumentationFormatter::PrintSection(
std::ostream& os, cmDocumentationSection const& section)
{
std::size_t const PREFIX_SIZE =
sizeof(cmDocumentationEntry::CustomNamePrefix) + 1u;
// length of the "= " literal (see below)
std::size_t const SUFFIX_SIZE = 2u;
// legacy magic number ;-)
std::size_t const NAME_SIZE = 29u;
std::size_t const PADDING_SIZE = PREFIX_SIZE + SUFFIX_SIZE;
std::size_t const TITLE_SIZE = NAME_SIZE + PADDING_SIZE;
auto const savedIndent = this->TextIndent;
os << section.GetName() << '\n';
for (cmDocumentationEntry const& entry : section.GetEntries()) {
if (!entry.Name.empty()) {
this->TextIndent = TITLE_SIZE;
os << std::setw(PREFIX_SIZE) << std::left << entry.CustomNamePrefix
<< std::setw(int(std::max(NAME_SIZE, entry.Name.size())))
<< entry.Name;
if (entry.Name.size() > NAME_SIZE) {
os << '\n' << std::setw(int(this->TextIndent - PREFIX_SIZE)) << ' ';
}
os << "= " << this->Format(entry.Brief).substr(this->TextIndent);
} else {
this->TextIndent = 0u;
os << '\n' << this->Format(entry.Brief);
}
}
os << '\n';
this->TextIndent = savedIndent;
}
|