scc/libs/pprocessor/src/parse.c

#include <ctype.h>
#include <lex_parser.h>
#include <pp_macro.h>
#include <pp_parse.h>
#include <pp_token.h>
#include <string.h>

static const struct {
    const char *name;
    scc_pp_token_t tok;
} keywords[] = {
#define X(name, type, tok) {#name, tok},
    SCC_PP_INST_TOKEN
#undef X
};

// 使用二分查找查找关键字
static inline int keyword_cmp(const char *name, int len) {
    int low = 0;
    int high = sizeof(keywords) / sizeof(keywords[0]) - 1;
    while (low <= high) {
        int mid = (low + high) / 2;
        const char *key = keywords[mid].name;
        int cmp = 0;

        // 自定义字符串比较逻辑
        for (int i = 0; i < len; i++) {
            if (name[i] != key[i]) {
                cmp = (unsigned char)name[i] - (unsigned char)key[i];
                break;
            }
            if (name[i] == '\0')
                break; // 遇到终止符提前结束
        }

        if (cmp == 0) {
            // 完全匹配检查（长度相同）
            if (key[len] == '\0')
                return mid;
            cmp = -1; // 当前关键词比输入长
        }

        if (cmp < 0) {
            high = mid - 1;
        } else {
            low = mid + 1;
        }
    }
    return -1; // Not a keyword.
}

static inline void try_to_cut_list(scc_pp_macro_list_t *list,
                                   scc_cstring_t *buff) {
    if (scc_cstring_len(buff) != 0) {
        scc_vec_push(*list, *buff);
        *buff = scc_cstring_new();
    }
}

static cbool parse_macro_replace_list(scc_probe_stream_t *stream,
                                      scc_pp_macro_list_t *list) {
    Assert(stream != null && list != null);
    scc_probe_stream_reset(stream);

    scc_vec_init(*list);
    scc_cstring_t replacement = scc_cstring_new();
    int ch;
    scc_pos_t pos = scc_pos_init();

    while ((ch = scc_probe_stream_peek(stream)) != scc_stream_eof) {
        if (scc_lex_parse_is_endline(ch)) {
            break;
        }
        if (scc_lex_parse_is_identifier_prefix(ch)) {
            try_to_cut_list(list, &replacement);
            cbool ret = scc_lex_parse_identifier(stream, &pos, &replacement);
            Assert(ret == true);
            try_to_cut_list(list, &replacement);
        } else if (ch == '#') {
            // 处理 # 和 ## 操作符
            scc_probe_stream_consume(stream);
            try_to_cut_list(list, &replacement);

            scc_cstring_append_ch(&replacement, '#');
            if (scc_probe_stream_peek(stream) == '#') {
                // ## 连接操作符
                scc_probe_stream_consume(stream);
                scc_cstring_append_ch(&replacement, '#');
            }
            // 我需要尽可能防止空白字符干扰解析
            scc_lex_parse_skip_whitespace(stream, &pos);
            try_to_cut_list(list, &replacement);
        } else if (scc_lex_parse_is_whitespace(ch)) {
            try_to_cut_list(list, &replacement);
            scc_lex_parse_skip_whitespace(stream, &pos);
            scc_cstring_append_ch(&replacement, ' ');
            try_to_cut_list(list, &replacement);
        } else {
            scc_probe_stream_consume(stream);
            scc_cstring_append_ch(&replacement, (char)ch);
        }
    }

    if (scc_cstring_len(&replacement) != 0) {
        scc_vec_push(*list, replacement);
        replacement = scc_cstring_new();
    }

    // for (usize i = 0; i < list->size; ++i) {
    //     LOG_DEBUG("list %d: %s", (int)i,
    //               scc_cstring_as_cstr(&scc_vec_at(*list, i)));
    // }
    return true;
}

// 解析宏参数列表
static cbool parse_macro_arguments(scc_probe_stream_t *stream,
                                   scc_pp_macro_list_t *args) {
    Assert(stream != null && args != null);

    scc_vec_init(*args);
    int ch;
    scc_probe_stream_reset(stream);

    // 跳过 '('
    ch = scc_probe_stream_peek(stream);
    if (ch != '(') {
        return false;
    }
    scc_probe_stream_consume(stream); // 消费 '('

    int paren_depth = 1;
    scc_cstring_t current_arg = scc_cstring_new();
    scc_pos_t pos = scc_pos_init();

    while (paren_depth > 0) {
        ch = scc_probe_stream_peek(stream);
        if (ch == scc_stream_eof) {
            scc_cstring_free(&current_arg);
            scc_cstring_free(&pos.name);
            return false;
        }

        if (ch == '(') {
            paren_depth++;
            scc_cstring_append_ch(&current_arg, (char)ch);
            scc_probe_stream_consume(stream);
        } else if (ch == ')') {
            paren_depth--;
            if (paren_depth > 0) {
                scc_cstring_append_ch(&current_arg, (char)ch);
            }
            scc_probe_stream_consume(stream);
        } else if (ch == ',' && paren_depth == 1) {
            // 参数分隔符
            scc_vec_push(*args, current_arg);
            current_arg = scc_cstring_new();
            scc_probe_stream_consume(stream);
            // 跳过参数后的空白
            scc_lex_parse_skip_whitespace(stream, &pos);
        } else {
            scc_cstring_append_ch(&current_arg, (char)ch);
            scc_probe_stream_consume(stream);
        }
    }

    // 添加最后一个参数
    if (!scc_cstring_is_empty(&current_arg)) {
        scc_vec_push(*args, current_arg);
    } else {
        scc_cstring_free(&current_arg);
    }

    scc_cstring_free(&pos.name);
    return true;
}

static cbool safe_skip_backspace_if_endline(scc_probe_stream_t *stream,
                                            scc_pos_t *pos) {
    scc_probe_stream_reset(stream);
    int ch = scc_probe_stream_peek(stream);
    // FIXME maybe it not correct
    while (ch == '\r' || ch == '\n' || ch == ' ' || ch == '\t') {
        if (scc_lex_parse_is_endline(ch)) {
            scc_lex_parse_skip_endline(stream, pos);
            return true;
        }
        scc_probe_stream_consume(stream);
        ch = scc_probe_stream_peek(stream);
    }
    scc_probe_stream_reset(stream);
    return false;
}

void scc_pp_parse_directive(scc_probe_stream_t *stream, scc_pos_t *pos,
                            scc_pp_macro_table_t *macros) {
    Assert(stream != null);

    scc_probe_stream_reset(stream);
    // 跳过 '#' 和后续空白
    if (scc_probe_stream_peek(stream) != '#') {
        LOG_WARN("Invalid directive");
        return;
    }
    scc_pos_next(pos);
    scc_probe_stream_consume(stream);
    if (safe_skip_backspace_if_endline(stream, pos))
        return;

    // 解析指令名称
    scc_cstring_t directive = scc_cstring_new();
    if (!scc_lex_parse_identifier(stream, pos, &directive)) {
        goto ERR;
    }
    if (safe_skip_backspace_if_endline(stream, pos))
        goto FREE;

    scc_pp_token_t token = keyword_cmp(scc_cstring_as_cstr(&directive),
                                       scc_cstring_len(&directive));

    scc_cstring_t name = scc_cstring_new();
    switch (token) {
    case SCC_PP_TOK_DEFINE: {
        if (!scc_lex_parse_identifier(stream, pos, &name)) {
            scc_cstring_free(&name);
            goto ERR;
        }

        // 检查是否是函数宏：宏名后是否直接跟着 '('（没有空白字符）
        scc_probe_stream_reset(stream);
        int ch = scc_probe_stream_peek(stream);
        cbool has_whitespace = scc_lex_parse_is_whitespace(ch);
        if (has_whitespace && safe_skip_backspace_if_endline(stream, pos)) {
            goto FREE;
        }

        if (!has_whitespace && ch == '(') {
            // 函数宏
            scc_pp_macro_list_t params;
            if (!parse_macro_arguments(stream, &params)) {
                goto ERR;
            }

            ch = scc_probe_stream_peek(stream);
            if (ch == ')') {
                scc_probe_stream_consume(stream); // 消费 ')'
            }
            if (safe_skip_backspace_if_endline(stream, pos)) {
                goto FREE;
            }

            scc_pp_macro_list_t replacement;
            parse_macro_replace_list(stream, &replacement);
            scc_pp_add_function_macro(macros, &name, &params, &replacement);
        } else {
            // 对象宏
            scc_pp_macro_list_t replacement;
            parse_macro_replace_list(stream, &replacement);
            scc_pp_add_object_macro(macros, &name, &replacement);
        }
        scc_cstring_free(&name);
        break;
    }
    case SCC_PP_TOK_UNDEF: {
        if (scc_lex_parse_identifier(stream, pos, &name)) {
            // TODO ret value
            scc_pp_macro_table_remove(macros, &name);
        }
        break;
    }
    case SCC_PP_TOK_INCLUDE:
    case SCC_PP_TOK_IF:
    case SCC_PP_TOK_IFDEF:
    case SCC_PP_TOK_IFNDEF:
    case SCC_PP_TOK_ELSE:
    case SCC_PP_TOK_ELIF:
    case SCC_PP_TOK_ELIFDEF:
    case SCC_PP_TOK_ELIFNDEF:
    case SCC_PP_TOK_ENDIF:
    case SCC_PP_TOK_LINE:
    case SCC_PP_TOK_EMBED:
    case SCC_PP_TOK_ERROR:
    case SCC_PP_TOK_WARNING:
    case SCC_PP_TOK_PRAMA:
        // 暂时跳过这一行
        scc_lex_parse_skip_line(stream, pos);
        break;
    default:
        LOG_WARN("Unknown preprocessor directive: %s",
                 scc_cstring_as_cstr(&directive));
        scc_lex_parse_skip_line(stream, pos);
    }
ERR:
    scc_lex_parse_skip_line(stream, pos);
FREE:
    scc_cstring_free(&directive);
    scc_cstring_free(&name);
}

static inline void scc_generate_cstr(scc_cstring_t *buff) {
    scc_cstring_t out_buff = scc_cstring_new();
    scc_cstring_append_ch(&out_buff, '\"');
    // TODO it is too simple
    scc_cstring_append(&out_buff, buff);
    scc_cstring_append_ch(&out_buff, '\"');

    // FIXME 可能有着更好的解决方案
    scc_cstring_clear(buff);
    scc_cstring_append(buff, &out_buff);
    scc_cstring_free(&out_buff);
}

#define SCC_PP_IS_LIST_BLANK(i)                                                \
    ((i) < list->size && scc_vec_at(*list, (i)).data[0] == ' ' &&              \
     scc_vec_at(*list, (i)).data[1] == '\0')
#define SCC_PP_IS_LIST_TO_STRING(i)                                            \
    ((i) < list->size && scc_vec_at(*list, (i)).data[0] == '#' &&              \
     scc_vec_at(*list, (i)).data[1] == '\0')
#define SCC_PP_IS_LIST_CONNECT(i)                                              \
    ((i) < list->size && scc_vec_at(*list, (i)).data[0] == '#' &&              \
     scc_vec_at(*list, (i)).data[1] == '#' &&                                  \
     scc_vec_at(*list, (i)).data[2] == '\0')
#define SCC_PP_USE_CONNECT(font, rear)                                         \
    if (rear < list->size) {                                                   \
        scc_cstring_append(out_buff, &scc_vec_at(*list, font));                \
        scc_cstring_append(out_buff, &scc_vec_at(*list, rear));                \
    } else {                                                                   \
        scc_cstring_append(out_buff, &scc_vec_at(*list, font));                \
    }
// for # ## to generator string
static inline cbool scc_pp_expand_string_unsafe(scc_pp_macro_list_t *list,
                                                scc_cstring_t *out_buff) {
    for (usize i = 0; i < list->size; ++i) {
        if (SCC_PP_IS_LIST_BLANK(i + 1)) {
            if (SCC_PP_IS_LIST_CONNECT(i + 2)) {
                SCC_PP_USE_CONNECT(i, i + 3);
                i += 3;
                continue;
            }
        } else if (SCC_PP_IS_LIST_CONNECT(i + 1)) {
            SCC_PP_USE_CONNECT(i, i + 2);
            i += 2;
            continue;
        } else if (SCC_PP_IS_LIST_TO_STRING(i)) {
            i += 1;
            if (i < list->size) {
                scc_generate_cstr(&scc_vec_at(*list, i));
            } else {
                LOG_WARN("# need a valid literator");
                break;
            }
        }
        scc_cstring_append(out_buff, &scc_vec_at(*list, i));
    }
    return true;
}

// 展开对象宏
cbool scc_pp_expand_object_macro(scc_pp_macro_t *macro,
                                 scc_cstring_t *out_buff) {
    Assert(macro->type == SCC_PP_MACRO_OBJECT && macro->params.size == 0);
    Assert(scc_cstring_is_empty(out_buff) == true);
    // 对象宏输出替换文本并进行递归展开
    scc_pp_expand_string_unsafe(&macro->replaces, out_buff);
    return true;
}

// 展开函数宏
cbool scc_pp_expand_function_macro(scc_pp_macro_t *macro,
                                   scc_pp_macro_list_t *params,
                                   scc_cstring_t *out_buff) {
    Assert(macro->type == SCC_PP_MACRO_FUNCTION);
    Assert(out_buff != null);
    Assert(scc_cstring_is_empty(out_buff) == true);
    for (usize i = 0; i < macro->replaces.size; ++i) {
        // TODO ... __VA_ARGS__
        for (usize j = 0; j < macro->params.size; ++j) {
            if (scc_strcmp(
                    scc_cstring_as_cstr(&scc_vec_at(macro->replaces, i)),
                    scc_cstring_as_cstr(&scc_vec_at(macro->params, j))) == 0) {
                scc_cstring_free(&scc_vec_at(macro->replaces, i));
                scc_cstring_append(&scc_vec_at(macro->replaces, i),
                                   &scc_vec_at(*params, j));
                continue;
            }
        }
    }
    scc_pp_expand_string_unsafe(&macro->replaces, out_buff);
    return true;
}

cbool scc_pp_expand_macro(scc_probe_stream_t *stream,
                          scc_pp_macro_table_t *macros,
                          scc_pp_macro_table_t *expand_stack,
                          scc_probe_stream_t **out_stream, int depth) {
    // TODO self position and it maybe is a stack on #include ?
    // 递归扫描
    if (depth <= 0) {
        *out_stream = null;
        return false;
    }
    Assert(stream != null && macros != null && out_stream != null);

    scc_cstring_t identifier = scc_cstring_new();
    scc_pos_t pos = scc_pos_init();

    cbool ret;
    ret = scc_lex_parse_identifier(stream, &pos, &identifier);
    Assert(ret == true);

    scc_pp_macro_t *macro = scc_pp_macro_table_get(macros, &identifier);
    // 1. 不是宏，直接输出标识符
    // 2. 检查到重复展开跳过
    if (macro == null ||
        scc_pp_macro_table_get(expand_stack, &macro->name) != null) {
        *out_stream =
            scc_mem_probe_stream_new(scc_cstring_as_cstr(&identifier),
                                     scc_cstring_len(&identifier), false);
        return true;
    } else {
        scc_cstring_free(&identifier);
    }

    // 根据宏类型展开
    scc_cstring_t tmp_buff = scc_cstring_new();
    if (macro->type == SCC_PP_MACRO_OBJECT) {
        cbool ret = scc_pp_expand_object_macro(macro, &tmp_buff);
        Assert(ret == true);
    } else if (macro->type == SCC_PP_MACRO_FUNCTION) {
        // FIXME 是否需要忽略空白字符？
        scc_lex_parse_skip_whitespace(stream, &pos);
        if (scc_probe_stream_peek(stream) != '(') {
            LOG_ERROR("Not a function and skip it");
            goto ERR;
        }
        scc_pp_macro_list_t params;
        ret = parse_macro_arguments(stream, &params);
        Assert(ret == true);
        scc_pp_expand_function_macro(macro, &params, &tmp_buff);
        Assert(ret == true);
    }

    // 已经展开的将被标记并入栈
    scc_pp_macro_table_set(expand_stack,
                           scc_pp_macro_new(&macro->name, macro->type));

    // 将展开内容变换成stream
    scc_probe_stream_t *tmp_stream = scc_mem_probe_stream_new(
        scc_cstring_as_cstr(&tmp_buff), scc_cstring_len(&tmp_buff), false);
    int ch;
    scc_cstring_t real_buff = scc_cstring_new();

    while ((ch = scc_probe_stream_peek(tmp_stream)) != scc_stream_eof) {
        if (scc_lex_parse_is_identifier_prefix(ch)) {
            // 递归检查
            scc_probe_stream_t *tmp_out_stream;
            if (scc_pp_expand_macro(tmp_stream, macros, expand_stack,
                                    &tmp_out_stream, depth - 1) == false) {
                return false;
            }
            // scc_cstring_append_cstr();
            Assert(tmp_out_stream != null);
            while (scc_probe_stream_peek(tmp_out_stream) != scc_stream_eof) {
                scc_cstring_append_ch(&real_buff,
                                      scc_probe_stream_consume(tmp_out_stream));
            }
            Assert(tmp_out_stream != null && tmp_out_stream->drop != null);
            scc_probe_stream_drop(tmp_out_stream);
        } else {
            scc_cstring_append_ch(&real_buff,
                                  scc_probe_stream_consume(tmp_stream));
        }
    }
    scc_cstring_free(&tmp_buff);
    scc_probe_stream_drop(tmp_stream);
    *out_stream = scc_mem_probe_stream_new(scc_cstring_as_cstr(&real_buff),
                                           scc_cstring_len(&real_buff), false);

    // 已经展开的将被标记并出栈
    scc_pp_macro_table_remove(expand_stack, &macro->name);
    return true;
ERR:
    *out_stream = null;
    return false;
}