[clang-tools-extra] [llvm] [llvm] add support for mustache templating language (PR #105893)

[Paul Kirth via cfe-commits]

================
@@ -0,0 +1,489 @@
+//===-- Mustache.cpp ------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Support/Mustache.h"
+#include "llvm/Support/Error.h"
+#include <sstream>
+
+using namespace llvm;
+using namespace llvm::json;
+using namespace llvm::mustache;
+
+SmallString<128> escapeString(StringRef Input,
+                              DenseMap<char, StringRef> &Escape) {
+  SmallString<128> EscapedString("");
+  for (char C : Input) {
+    if (Escape.find(C) != Escape.end())
+      EscapedString += Escape[C];
+    else
+      EscapedString += C;
+  }
+  return EscapedString;
+}
+
+Accessor split(StringRef Str, char Delimiter) {
+  Accessor Tokens;
+  if (Str == ".") {
+    Tokens.emplace_back(Str);
+    return Tokens;
+  }
+  StringRef Ref(Str);
+  while (!Ref.empty()) {
+    StringRef Part;
+    std::tie(Part, Ref) = Ref.split(Delimiter);
+    Tokens.emplace_back(Part.trim());
+  }
+  return Tokens;
+}
+
+void addIndentation(llvm::SmallString<128> &PartialStr,
+                    size_t IndentationSize) {
+  std::string Indent(IndentationSize, ' ');
+  llvm::SmallString<128> Result;
+  for (size_t I = 0; I < PartialStr.size(); ++I) {
+    Result.push_back(PartialStr[I]);
+    if (PartialStr[I] == '\n' && I < PartialStr.size() - 1)
+      Result.append(Indent);
+  }
+  PartialStr = Result;
+}
+
+Token::Token(StringRef RawBody, StringRef InnerBody, char Identifier)
+    : RawBody(RawBody), TokenBody(InnerBody), Indentation(0) {
+  TokenType = getTokenType(Identifier);
+  if (TokenType == Type::Comment)
+    return;
+
+  StringRef AccessorStr =
+      TokenType == Type::Variable ? InnerBody : InnerBody.substr(1);
+
+  Accessor = split(AccessorStr.trim(), '.');
+}
+
+Token::Token(StringRef Str)
+    : TokenType(Type::Text), RawBody(Str), Accessor({}), TokenBody(Str),
+      Indentation(0) {}
+
+Token::Type Token::getTokenType(char Identifier) {
+  switch (Identifier) {
+  case '#':
+    return Type::SectionOpen;
+  case '/':
+    return Type::SectionClose;
+  case '^':
+    return Type::InvertSectionOpen;
+  case '!':
+    return Type::Comment;
+  case '>':
+    return Type::Partial;
+  case '&':
+    return Type::UnescapeVariable;
+  default:
+    return Type::Variable;
+  }
+}
+
+// Function to check if there's no meaningful text behind
+bool noTextBehind(size_t Idx, const SmallVector<Token, 0> &Tokens) {
+  if (Idx == 0 || Tokens[Idx - 1].getType() != Token::Type::Text)
+    return false;
+  const Token &PrevToken = Tokens[Idx - 1];
+  StringRef TokenBody = PrevToken.getRawBody().rtrim(" \t\v\t");
+  return TokenBody.ends_with("\n") || TokenBody.ends_with("\r\n") ||
+         (TokenBody.empty() && Idx == 1);
+}
+// Function to check if there's no meaningful text ahead
+bool noTextAhead(size_t Idx, const SmallVector<Token, 0> &Tokens) {
+  if (Idx >= Tokens.size() - 1 ||
+      Tokens[Idx + 1].getType() != Token::Type::Text)
+    return false;
+
+  const Token &NextToken = Tokens[Idx + 1];
+  StringRef TokenBody = NextToken.getRawBody().ltrim(" ");
+  return TokenBody.starts_with("\r\n") || TokenBody.starts_with("\n");
+}
+
+// Simple tokenizer that splits the template into tokens
+// the mustache spec allows {{{ }}} to unescape variables
+// but we don't support that here unescape variable
+// is represented only by {{& variable}}
+SmallVector<Token, 0> tokenize(StringRef Template) {
+  SmallVector<Token, 0> Tokens;
+  StringRef Open("{{");
+  StringRef Close("}}");
+  std::size_t Start = 0;
+  std::size_t DelimiterStart = Template.find(Open);
+  if (DelimiterStart == StringRef::npos) {
+    Tokens.emplace_back(Template);
+    return Tokens;
+  }
+  while (DelimiterStart != StringRef::npos) {
+    if (DelimiterStart != Start) {
+      Tokens.emplace_back(Template.substr(Start, DelimiterStart - Start));
+    }
+
+    size_t DelimiterEnd = Template.find(Close, DelimiterStart);
+    if (DelimiterEnd == StringRef::npos) {
+      break;
+    }
+
+    // Extract the Interpolated variable without {{ and }}
+    size_t InterpolatedStart = DelimiterStart + Open.size();
+    size_t InterpolatedEnd = DelimiterEnd - DelimiterStart - Close.size();
+    SmallString<128> Interpolated =
+        Template.substr(InterpolatedStart, InterpolatedEnd);
+    SmallString<128> RawBody;
+    RawBody += Open;
+    RawBody += Interpolated;
+    RawBody += Close;
----------------
ilovepi wrote:

I'm not sure this totally makes sense. You're going to use this in the `emplace_back()` call below, but have potentially many allocations here, when concatenating.

Perhaps it makes sense to change the constructor, so you can pass a `Twine` to initialize the SmallString? But even if you were going to leave this as is, this should probably use `append()` to concatenate all 3 strings at once. Or even better, just us an initializer list of string refs in the call to `emplace_back()`. https://llvm.org/doxygen/classllvm_1_1SmallString.html#af5506265736f3dd65b69c378a216aa6a

https://github.com/llvm/llvm-project/pull/105893