Merge pull request #2911 from fernewelten/NC_Unicode_Char

ivan-mogilko · web-flow · commit abd2efa75a8f · 2025-11-11T16:18:07.000+03:00
New Compiler: Accept UTF-8 within `'…'`
diff --git a/Compiler/script2/cs_parser.cpp b/Compiler/script2/cs_parser.cpp
@@ -7722,21 +7722,6 @@ void AGS::Parser::Parse()
     }
 }
 
-// Scan inpl into scan tokens, build a symbol table
-int cc_scan(std::string const &inpl, AGS::SrcList &src, AGS::ccCompiledScript &scrip, AGS::SymbolTable &symt, AGS::MessageHandler &mh)
-{
-    AGS::Scanner scanner = { inpl, src, scrip, symt, mh };
-    scanner.Scan();
-    return -static_cast<int>(mh.HasError());
-}
-
-int cc_parse(AGS::SrcList &src, AGS::FlagSet options, AGS::ccCompiledScript &scrip, AGS::SymbolTable &symt, AGS::MessageHandler &mh)
-{
-    AGS::Parser parser = { src, options, scrip, symt, mh };
-    parser.Parse();
-    return -static_cast<int>(mh.HasError());
-}
-
 int cc_compile(std::string const &inpl, AGS::FlagSet options, AGS::ccCompiledScript &scrip, AGS::MessageHandler &mh)
 {
     AGS::SymbolTable local_symt;
@@ -7754,9 +7739,16 @@ int cc_compile(std::string const &inpl, AGS::FlagSet options, AGS::ccCompiledScr
     src.NewSection("UnnamedSection");
     src.NewLine(1u);
 
-    int error_code = cc_scan(inpl, src, scrip, symt, mh);
-    if (error_code >= 0)
-        error_code = cc_parse(src, options, scrip, symt, mh);
+    AGS::Scanner scanner = { inpl, FlagIsSet(options, SCOPT_UTF8), src, scrip, symt, mh };
+    scanner.Scan();
+    if (mh.HasError())
+        return -1;
+    
+    AGS::Parser parser = { src, options, scrip, symt, mh };
+    parser.Parse();
+    if (mh.HasError())
+        return -2;
+
     sections = lh.CreateSectionList();
-    return error_code;
+    return 0;
 }
diff --git a/Compiler/script2/cs_scanner.cpp b/Compiler/script2/cs_scanner.cpp
@@ -26,8 +26,9 @@
 std::string const AGS::Scanner::kNewSectionLitPrefix = "__NEWSCRIPTSTART_";
 size_t const AGS::Scanner::kNewSectionLitPrefixSize = 17u;
 
-AGS::Scanner::Scanner(std::string const &input, SrcList &token_list, ccCompiledScript &string_collector, SymbolTable &symt, MessageHandler &messageHandler)
-    : _ocMatcher(*this)
+AGS::Scanner::Scanner(std::string const &input, bool const use_utf8, SrcList &token_list, ccCompiledScript &string_collector, SymbolTable &symt, MessageHandler &messageHandler)
+    : _useUtf8(use_utf8)
+    , _ocMatcher(*this)
     , _lineno(1u)
     , _tokenList(token_list)
     , _msgHandler(messageHandler)
@@ -345,7 +346,7 @@ void AGS::Scanner::ReadInNumberLit(std::string &symstring, ScanType &scan_type,
 
         errno = 0;
         char *endptr;
-        std::strtof(valstring.c_str(), &endptr);
+        (void) std::strtof(valstring.c_str(), &endptr);
         bool const can_be_a_floating_point =
             (errno == 0 || errno == ERANGE) &&                  // range errors will be treated below
             (valstring.length() == endptr - valstring.c_str()); // ensure that all chars are used up in the conversion
@@ -459,6 +460,11 @@ void AGS::Scanner::ReadInCharLit(std::string &symstring, CodeCell &value)
 
             EscapedChar2Char(lit_char, symstring, lit_char);
         }
+        else if (UseUtf8() && (lit_char < 0 || lit_char > 0x7f))
+        {
+            // Accept an UTF-8 sequence, use the codepoint that corresponds to the sequence
+            ReadInCharLit_utf8(lit_char, symstring, lit_char);
+        }
 
         // Closing '\''
         int const closer = Get();
@@ -558,6 +564,91 @@ void AGS::Scanner::EscapedChar2Char(int first_char_after_backslash, std::string
     UserError("Unrecognized '\\%c' in character or string literal", first_char_after_backslash);
 }
 
+void AGS::Scanner::ReadInCharLit_utf8(int const first_char, std::string &symstring, int &converted)
+{
+    unsigned int const char0 = first_char & 0xff;
+    if (char0 <= 0x7f)
+    {
+        // No UTF-8 sequence
+        symstring.push_back(first_char);
+        converted = first_char;
+        return;
+    }
+
+    if (char0 < 0xc2 || char0 > 0xf4)
+        UserError("After the quote mark, found the byte '0x%02X'; "
+            "a legal UTF-8 sequence cannot start this way (file corrupt?)", char0);
+
+    do // exactly 1 time
+    {
+        int next_char = Get();
+        if (EOFReached() || Failed())
+            break; // to error msg
+        symstring.push_back(next_char);
+        unsigned int const char1 = next_char & 0xff;
+
+        // Strictly going by the algo, you can encode some characters
+        // in different ways, but the standard doesn't allow this:
+        // You must always use the shortest encoding possible.
+        // Thus, some combinations of first and second byte are forbidden.
+        if (char1 < 0x80 ||
+            char1 > 0xbf ||
+            char0 == 0xe0 && char1 <= 0x9f ||
+            char0 == 0xed && char1 >= 0xa0 ||
+            char0 == 0xf0 && char1 <= 0x8f ||
+            char0 == 0xf4 && char1 >= 0x90)
+            UserError("After the quote mark, found the bytes '0x%02X%02X'; "
+            "a legal UTF-8 sequence cannot start this way (file corrupt?)", char0, char1);
+
+        if (char0 <= 0xdf)
+        {
+            // 2-byte sequence
+            converted = ((char0 & 0x1f) << 6) + (char1 & 0x3f);
+            return;
+        }
+
+        next_char = Get();
+        if (EOFReached() || Failed())
+            break; // to error msg
+        unsigned int const char2 = next_char & 0xff;
+        if (char2 < 0x80 || char2 > 0xbf)
+            UserError("After the quote mark, found the bytes '0x%02X%02X%02X'; "
+                "a legal UTF-8 sequence cannot start this way (file corrupt?)", char0, char1, char2);
+        symstring.push_back(next_char);
+
+        if (char0 <= 0xef)
+        {
+            // 3-byte sequence
+            converted = ((char0 & 0x0F) << 12) |
+                        ((char1 & 0x3F) << 6) |
+                        (char2 & 0x3F);
+            return;
+        }
+
+        // 4-byte sequence
+        next_char = Get();
+        if (EOFReached() || Failed())
+            break; // to error msg
+        unsigned int const char3 = next_char & 0xff;
+        if (char3 < 0x80 || char3 > 0xbf)
+            UserError("After the quote mark, found the bytes '0x%02X%02X%02X%02X' "
+                "a legal UTF-8 sequence cannot start this way (file corrupt?)", char0, char1, char2, char3);
+        symstring.push_back(next_char);
+
+        converted = ((char0 & 0x07) << 18) |
+                    ((char1 & 0x3F) << 12) |
+                    ((char2 & 0x3F) << 6) |
+                    (char3 & 0x3F);
+        return;
+    } while (false);
+
+    // Here in case of reading errors
+    if (EOFReached())
+        UserError("Reached the end of the input inmidst of a UTF-8 sequence (file corrupt?)");
+    if (Failed())
+        UserError("Encountered a reading error inmidst of a UTF-8 sequence (file corrupt?)");
+}
+
 void AGS::Scanner::ReadInStringLit(std::string &symstring, std::string &valstring)
 {
     symstring = "\"";
diff --git a/Compiler/script2/cs_scanner.h b/Compiler/script2/cs_scanner.h
@@ -98,6 +98,7 @@ class Scanner
     // We need that, so we collect in _eofReached the fact that end-of-stream has ever been reached.
     bool _eofReached = false;
     bool _failed = false;
+    bool _useUtf8;
     std::size_t _lineno;
     std::string _section;
     SrcList &_tokenList;
@@ -131,6 +132,9 @@ class Scanner
     // Translate a '\\' combination into a character, backslash is already read in
     void EscapedChar2Char(int first_char_after_backslash, std::string &symstring, int &converted);
 
+    // Read a sequence of bytes in UTF-8 format, translate to the respective codepoint
+    void ReadInCharLit_utf8(int first_char, std::string &symstring, int &converted);
+
     // Read oct combination \777; backslash is already read in
     int OctDigits2Char(int first_digit_char, std::string &symstring);
 
@@ -192,11 +196,14 @@ class Scanner
     inline static void ReplaceToken(std::string &where, std::string const &token, std::string const &replacement) { where.replace(where.find(token), token.length(), replacement); }
 
 public:
-    Scanner(std::string const &input, SrcList &token_list, ccCompiledScript &string_collector, SymbolTable &symt, MessageHandler &messageHandler);
+    Scanner(std::string const &input, bool use_utf8, SrcList &token_list, ccCompiledScript &string_collector, SymbolTable &symt, MessageHandler &messageHandler);
 
     // Scan the input into token_list; symbols into symt; strings into _string_collector
     void Scan();
 
+    // Whether UTF-8 encoding is used
+    inline bool UseUtf8() const { return _useUtf8; }
+
     // Returns whether we've encountered EOF.
     inline bool EOFReached() const { return _eofReached; };
     // Returns whether we've encountered a read failure
diff --git a/Compiler/test2/cc_parser_test_0.cpp b/Compiler/test2/cc_parser_test_0.cpp
@@ -19,6 +19,7 @@
 
 #include "script2/cc_symboltable.h"
 #include "script2/cc_internallist.h"
+#include "script2/cs_scanner.h"
 #include "script2/cs_parser.h"
 
 #include "cc_parser_test_lib.h"
@@ -300,9 +301,12 @@ TEST_F(Compile0, EnumNegative) {
         };\
         ";
 
-    // Call cc_scan() and cc_parse() by hand so that we can see the symbol table
-    ASSERT_LE(0, cc_scan(inpl, targ, scrip, sym, mh));
-    int compile_result = cc_parse(targ, options, scrip, sym, mh);
+    // Call scanner and parser by hand so that we can see the symbol table
+    AGS::Scanner scanner = { inpl, false, targ, scrip, sym, mh };
+    scanner.Scan();
+    ASSERT_FALSE(mh.HasError());
+    AGS::Parser parser = { targ, options, scrip, sym, mh };
+    parser.Parse();
     std::string const &err_msg = mh.GetError().Message;
     size_t err_line = mh.GetError().Lineno;
     EXPECT_EQ(0u, mh.WarningsCount());
@@ -351,12 +355,14 @@ TEST_F(Compile0, DefaultParametersLargeInts) {
         ";
 
 
-    ASSERT_LE(0, cc_scan(inpl, targ, scrip, sym, mh));
-    int compile_result = cc_parse(targ, options, scrip, sym, mh);
+    AGS::Scanner scanner = { inpl, false, targ, scrip, sym, mh };
+    scanner.Scan();
+    ASSERT_FALSE(mh.HasError());
+    AGS::Parser parser = { targ, options, scrip, sym, mh };
+    parser.Parse();
     std::string const &err_msg = mh.GetError().Message;
     size_t err_line = mh.GetError().Lineno;
     EXPECT_EQ(0u, mh.WarningsCount());
-
     ASSERT_STREQ("Ok", mh.HasError() ? err_msg.c_str() : "Ok");
 
     AGS::Symbol const funcidx = sym.Find("importedfunc");
@@ -405,12 +411,14 @@ TEST_F(Compile0, ImportFunctionReturningDynamicArray) {
         };                                  \n\
         ";
 
-    ASSERT_LE(0, cc_scan(inpl, targ, scrip, sym, mh));
-    int compile_result = cc_parse(targ, options, scrip, sym, mh);
+    AGS::Scanner scanner = { inpl, false, targ, scrip, sym, mh };
+    scanner.Scan();
+    ASSERT_FALSE(mh.HasError());
+    AGS::Parser parser = { targ, options, scrip, sym, mh };
+    parser.Parse();
     std::string const &err_msg = mh.GetError().Message;
     size_t err_line = mh.GetError().Lineno;
     EXPECT_EQ(0u, mh.WarningsCount());
-
     ASSERT_STREQ("Ok", mh.HasError() ? err_msg.c_str() : "Ok");
 
     int funcidx;
@@ -847,8 +855,13 @@ TEST_F(Compile0, LocalGlobalSeq2) {
     AGS::SymbolTable sym;
     AGS::FlagSet const options = ~SCOPT_NOIMPORTOVERRIDE | SCOPT_LINENUMBERS;
 
-    ASSERT_LE(0, cc_scan(inpl, targ, scrip, sym, mh));
-    ASSERT_EQ(0, cc_parse(targ, options, scrip, sym, mh));
+    AGS::Scanner scanner = { inpl, false, targ, scrip, sym, mh };
+    scanner.Scan();
+    ASSERT_FALSE(mh.HasError());
+    AGS::Parser parser = { targ, options, scrip, sym, mh };
+    parser.Parse();
+    std::string const &err_msg = mh.GetError().Message;
+    ASSERT_STREQ("Ok", mh.HasError() ? err_msg.c_str() : "Ok");
 
     ASSERT_LE(1u, mh.GetMessages().size());
     EXPECT_EQ(7u, mh.GetMessages()[0].Lineno);
@@ -1850,8 +1863,13 @@ TEST_F(Compile0, Import2GlobalAllocation) {
     AGS::SymbolTable sym;
     AGS::FlagSet const options = ~SCOPT_NOIMPORTOVERRIDE | SCOPT_LINENUMBERS;
 
-    ASSERT_LE(0, cc_scan(inpl, targ, scrip, sym, mh));
-    ASSERT_EQ(0, cc_parse(targ, options, scrip, sym, mh));
+    AGS::Scanner scanner = { inpl, false, targ, scrip, sym, mh };
+    scanner.Scan();
+    ASSERT_FALSE(mh.HasError());
+    AGS::Parser parser = { targ, options, scrip, sym, mh };
+    parser.Parse();
+    std::string const &err_msg = mh.GetError().Message;
+    ASSERT_STREQ("Ok", mh.HasError() ? err_msg.c_str() : "Ok");
 
     AGS::Symbol const idx = sym.Find("J");
     ASSERT_LE(0, idx);
diff --git a/Compiler/test2/cc_parser_test_lib.h b/Compiler/test2/cc_parser_test_lib.h
@@ -22,21 +22,5 @@
 extern char kAgsHeaderString[];
 extern char kAgsHeaderBool[];
 
-// Only use this function for googletests. Scan and tokenize the input.
-extern int cc_scan(
-    std::string const &inpl,        // preprocessed text to be tokenized
-    AGS::SrcList &src,              // store for the tokenized text
-    AGS::ccCompiledScript &scrip,   // repository for the strings in the text
-    AGS::SymbolTable &symt,         // symbol table
-    AGS::MessageHandler &mh);       // warnings and the error
-
-// Only use this function for googletests. Parse the input
-extern int cc_parse(
-    AGS::SrcList &src,              // tokenized text
-    AGS::FlagSet options,           // as defined in cc_options 
-    AGS::ccCompiledScript &scrip,   // result of the compilation
-    AGS::SymbolTable &symt,         // symbol table
-    AGS::MessageHandler &mh);       // warnings and the error 
-
 
 #endif
diff --git a/Compiler/test2/cc_scanner_test.cpp b/Compiler/test2/cc_scanner_test.cpp
