123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404 |
- #include "BXParser.h"
-
- #include <stdlib.h>
- #include <stdio.h>
- #include <string.h>
- #include <errno.h>
-
- int BXParser::get() {
- int ch = stream_.get();
- if (ch == '\n') {
- line_ += 1;
- ch_ = 1;
- } else {
- ch_ += 1;
- }
-
- return ch;
- }
-
- BXParser::Operator BXParser::readOperator() {
- int ch2 = peek2();
- if (peek() == ':' && ch2 == '=') {
- skip(); // ':'
- skip(); // '='
- return Operator::COLON_EQUALS;
- } else if (peek() == '+' && ch2 == '=') {
- skip(); // '+'
- skip(); // '='
- return Operator::PLUS_EQUALS;
- } else if (peek() == '=' && ch2 == '+') {
- skip(); // '='
- skip(); // '+'
- return Operator::EQUALS_PLUS;
- }
-
- return Operator::NONE;
- }
-
- void BXParser::skip(char expected) {
- int ch = get();
- if (ch == EOF) {
- error(std::string("Expected '") + expected + "', got EOF");
- } else if (ch != expected) {
- error(std::string("Expected '") + expected + "', got '" + (char)ch + "'");
- }
- }
-
- [[noreturn]] void BXParser::error(std::string msg) {
- throw BXParseError(std::to_string(line_) + ":" + std::to_string(ch_) + ": " + msg);
- }
-
- static bool isWhitespace(int ch) {
- if (ch == ' ' || ch == '\t' || ch == '\r' || ch == '\n')
- return true;
- return false;
- }
-
- void BXParser::skipWhitespace() {
- if (flags_ & FLAG_ONE_LINE) {
- int ch;
- while (isWhitespace(ch = peek()) && ch != '\r' && ch != '\n')
- get();
- } else {
- while (isWhitespace(peek()))
- get();
- }
- }
-
- char BXParser::parseEscape() {
- skip(); // '\'
- int ch;
- switch (ch = get()) {
- case EOF:
- error("Unexpected EOF");
-
- case 'n':
- return '\n';
-
- case 'r':
- return '\r';
-
- case 't':
- return '\t';
-
- default:
- return (char)ch;
- }
- }
-
- static void appendVariableToString(
- const BXVariables &vars, std::string &name,
- std::string &value) {
- if (name.size() == 0)
- return;
-
- auto it = vars.find(name);
- if (it == vars.end())
- return;
-
- auto &vec = it->second;
- bool first = true;
- for (auto &part: vec) {
- if (!first) {
- value += ' ';
- }
-
- first = false;
- value += part;
- }
- }
-
- static void appendVariableToArray(
- const BXVariables &vars, const std::string &name,
- std::vector<std::string> &values) {
- if (name.size() == 0)
- return;
-
- auto it = vars.find(name);
- if (it == vars.end())
- return;
-
- auto &vec = it->second;
- for (auto &part: vec) {
- values.push_back(part);
- }
- }
-
- void BXParser::parseExpansion(const BXVariables &vars, std::vector<std::string> &values) {
- skip(); // '$'
-
- std::string str;
- switch (peek()) {
- case '{':
- skip();
- parseString(vars, str, '}');
- skip('}');
- appendVariableToArray(vars, str, values);
- break;
-
- default:
- if (!parseIdentifier(str)) {
- error("No identifier after $.");
- }
-
- appendVariableToArray(vars, str, values);
- break;
- }
- }
-
- void BXParser::parseQuotedExpansion(const BXVariables &vars, std::string &content) {
- skip(); // '$'
-
- std::string str;
- switch (peek()) {
- case '{':
- skip();
- parseString(vars, str, '}');
- skip('}');
- appendVariableToString(vars, str, content);
- break;
-
- default:
- if (!parseIdentifier(str)) {
- error("No identifier after $.");
- }
-
- appendVariableToString(vars, str, content);
- break;
- }
- }
-
- void BXParser::parseQuotedString(const BXVariables &vars, std::string &content) {
- skip(); // '"'
-
- int ch;
- while ((ch = peek()) != EOF) {
- switch (ch) {
- case EOF:
- error("Unexpected EOF");
-
- case '\\':
- content.push_back(parseEscape());
- break;
-
- case '$':
- parseQuotedExpansion(vars, content);
- break;
-
- case '"':
- skip();
- return;
-
- default:
- content.push_back(get());
- break;
- }
- }
- }
-
- bool BXParser::parseString(const BXVariables &vars, std::string &content, int sep) {
- bool success = false;
- int ch;
- while (1) {
- ch = peek();
- if ((sep > 0 && ch == sep) || isWhitespace(ch)) {
- return success;
- }
-
- switch (ch) {
- case EOF:
- return success;
-
- case '\\':
- content.push_back(parseEscape());
- success = true;
- break;
-
- case '$':
- parseQuotedExpansion(vars, content);
- success = true;
- break;
-
- case '"':
- parseQuotedString(vars, content);
- success = true;
- break;
-
- default:
- if (ch == ':' && peek2() == '=')
- return success;
-
- content.push_back(get());
- success = true;
- break;
- }
- }
- }
-
- bool BXParser::parseIdentifier(std::string &content) {
- int ch = peek();
- if (!(
- (ch >= 'a' && ch <= 'z') ||
- (ch >= 'A' && ch <= 'Z') ||
- (ch == '_'))) {
- return false;
- }
-
- content += get();
-
- while (1) {
- ch = peek();
- if (!(
- (ch >= '0' && ch <= '9') ||
- (ch >= 'a' && ch <= 'z') ||
- (ch >= 'A' && ch <= 'Z') ||
- (ch == '_'))) {
- return true;
- }
-
- content += get();
- }
- }
-
- void BXParser::parse(BXVariables &vars) {
- std::string key, value;
- std::vector<std::string> values;
-
- skipWhitespace();
- if (!parseString(vars, key)) {
- return;
- }
-
- skipWhitespace();
- Operator prevOper = readOperator();
- if (prevOper == Operator::NONE) {
- error("Expected operator.");
- }
-
- auto doAssignment = [&] {
- switch (prevOper) {
- case Operator::COLON_EQUALS:
- vars[key] = std::move(values);
- values.clear();
- break;
-
- case Operator::PLUS_EQUALS:
- {
- auto &vec = vars[key];
- vec.reserve(vec.size() + values.size());
- for (size_t i = 0; i < values.size(); ++i) {
- vec.push_back(std::move(values[i]));
- }
- }
- values.clear();
- break;
-
- case Operator::EQUALS_PLUS:
- {
- auto &vec = vars[key];
- vec.reserve(vec.size() + values.size());
- for (size_t i = 0; i < vec.size(); ++i) {
- values.push_back(std::move(vec[i]));
- }
- vec = std::move(values);
- }
- values.clear();
- break;
-
- case Operator::NONE:
- break;
- }
- };
-
- while (true) {
- skipWhitespace();
-
- // Parse next value
- if (peek() == '$') {
- parseExpansion(vars, values);
- value.clear();
- continue; // We can't have an assignment after an expansion
- } else if (!parseString(vars, value)) {
- break;
- }
-
- skipWhitespace();
-
- // If there's an operator next, the value we just read was a actually a key.
- // Otherwise, it was just another value.
- Operator op = readOperator();
- if (op == Operator::NONE) {
- values.push_back(std::move(value));
- value.clear();
- } else {
- if (value.size() == 0) {
- error("Expected string before assignment operator");
- }
-
- doAssignment();
- prevOper = op;
- key = std::move(value);
- value.clear();
- }
- }
-
- doAssignment();
- }
-
- void BXParser::parseList(const BXVariables &vars, std::vector<std::string> &values) {
- while (true) {
- skipWhitespace();
- std::string value;
- if (!parseString(vars, value)) {
- break;
- }
-
- values.push_back(std::move(value));
- }
- }
-
- void BXWriter::put(char ch) {
- ch_ += 1;
- stream_ << ch;
- }
-
- void BXWriter::put(const std::string &str) {
- ch_ += str.size();
- stream_ << str;
- }
-
- void BXWriter::newline() {
- ch_ = 1;
- line_ += 1;
- stream_ << '\n';
- }
-
- void BXWriter::escape(const std::string &str) {
- put('"');
- for (char ch: str) {
- if (ch == '$' || ch == '"' || ch == '\\') {
- put('\\');
- }
- put(ch);
- }
- put('"');
- }
-
- void BXWriter::write(const BXVariables &vars) {
- for (const auto &pair: vars) {
- put(pair.first);
- put(" :=");
- for (auto &val: pair.second) {
- if (ch_ >= 80) {
- newline();
- put('\t');
- } else {
- put(' ');
- }
-
- escape(val);
- }
- newline();
- }
- }
|