From c95d6b05559457a4abb693f3fc2e3d5eb54ab05b Mon Sep 17 00:00:00 2001 From: Ákos Kőrösi Date: Mon, 18 May 2026 12:17:22 +0200 Subject: coalesce the preproc files --- preproc.c | 365 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 365 insertions(+) create mode 100644 preproc.c (limited to 'preproc.c') diff --git a/preproc.c b/preproc.c new file mode 100644 index 0000000..474a108 --- /dev/null +++ b/preproc.c @@ -0,0 +1,365 @@ +#include +#include +#include +#include +#include +#include + +#include "preproc.h" +#include "utils.h" +#include "defs.h" + + + +typedef struct { + char* str; + size_t pos; +} RawLine; + + +String collect_identifier(RawLine* line) { + String id_buf = make_String(256); + while (line->str[line->pos] == '_' || isalnum(line->str[line->pos])) { + push_to_String(&id_buf, line->str[line->pos++]); + } + return id_buf; +} + + + +TokenPtrVec lex_level(RawLine* line, char until, PreprocArena* arena); + + +void expect_char(RawLine* line, char expected) { + if (line->str[line->pos] != expected) { + exit(45); // TODO: nicer handling + } + line->pos++; +} + + +/* +MaybeChunk lex_expansion_chunk(RawLine* line, RegionVec* reg_arena, QuotationMode qmode) { + line->pos++; // Consume $ + if (line->str[line->pos] == '(') { + line->pos++; + if (line->str[line->pos] == '(') { + exit(1); + // TODO: this (and potentially the other parsers must be custom, we need to recognize arithm-internal parens from the terminating `))` ) + + + TokenVec inner_toks = lex_level(line, ')', reg_arena); + expect_char(line, ')'); + expect_char(line, ')'); + Region* reg_ptr = push_to_RegionVec( + reg_arena, + (Region){.mode = ARITHM, .tokens = inner_toks} + ); + return (MaybeChunk){ + .some = true, + .value = (Chunk){ + .qmode = qmode, + .kind = REGION_EXP, + .as.reg_ptr = reg_ptr + }, + }; + } else { + TokenVec inner_toks = lex_level(line, ')', reg_arena); + expect_char(line, ')'); + Region* reg_ptr = push_to_RegionVec( + reg_arena, + (Region){.mode = CMDS, .tokens = inner_toks} + ); + return (MaybeChunk) { + .some = true, + .value = (Chunk){ + .qmode = qmode, + .kind = REGION_EXP, + .as.reg_ptr = reg_ptr, + } + }; + } + } else { + String id_name = collect_identifier(line); + if (id_name.count > 0) { + return (MaybeChunk){ + .some = true, + .value = (Chunk){ + .qmode = qmode, + .kind = VAR_EXP, + .as.varname = inner(&id_name), + } + }; + } + } + return (MaybeChunk){.some = false}; +} +*/ + + +void lex_double_quoted_section(RawLine* line, Word* coll, PreprocArena* arena) { + line->pos++; + String accum = make_String(256); + while (line->str[line->pos] != '\"') { + if (line->str[line->pos] == '\0') exit(43); + + MaybeChunk exp_chunk = {.some = false}; + + if (line->str[line->pos] == '$') { + exit(1); + /* + exp_chunk = lex_expansion_chunk(line, reg_arena, DOUBLE_Q); + if (!exp_chunk.some) { + push_to_String(&accum, '$'); + } + */ + } else { + push_to_String(&accum, line->str[line->pos++]); + } + // TODO: other special cases beyond $ + + if (exp_chunk.some) { + if (accum.count > 0) { + Chunk accum_chunk = {.qmode = DOUBLE_Q, .kind = LITERAL, .as.literal_str = inner(&accum)}; + Chunk* accum_ptr = push_to_ChunkVec(&arena->chunks, accum_chunk); + push_to_Word(coll, accum_ptr); + accum = make_String(256); + } + Chunk* exp_ptr = push_to_ChunkVec(&arena->chunks, exp_chunk.value); + push_to_Word(coll, exp_ptr); + exp_chunk.some = false; + } + } + line->pos++; + if (accum.count > 0) { + Chunk accum_chunk = {.qmode = DOUBLE_Q, .kind = LITERAL, .as.literal_str = inner(&accum)}; + Chunk* accum_ptr = push_to_ChunkVec(&arena->chunks, accum_chunk); + push_to_Word(coll, accum_ptr); + accum = make_String(256); + } +} + + +Chunk next_word_unquoted(RawLine* line, RegionVec* reg_arena) { + String buf = make_String(256); + for (char c; (c=line->str[line->pos], c != '$' && c != '\0' && c != ' '); line->pos++) { + push_to_String(&buf, c); + } + return (Chunk){.qmode = UN_Q, .kind = LITERAL, .as.literal_str = inner(&buf)}; +} + + +Chunk collect_single_quoted_section(RawLine* line) { + line->pos++; // Consume quote mark + String chunk_str = make_String(256); + while (line->str[line->pos] != '\'') { + if (line->str[line->pos] == '\0') exit(44); + push_to_String(&chunk_str, line->str[line->pos++]); + } + line->pos++; // Consume closing quote + return (Chunk){ + .qmode = SINGLE_Q, + .kind = LITERAL, + .as.literal_str = inner(&chunk_str), + }; +} + + +typedef struct { + Word word_germ; + String unq_liter_germ; +} LexState; + + +void close_word_and_lit_germs(LexState* state, TokenPtrVec* toks, PreprocArena* arena) { + if (state->unq_liter_germ.count > 0) { + Chunk chu = { + .qmode = UN_Q, + .kind = LITERAL, + .as.literal_str = inner(&state->unq_liter_germ) + }; + Chunk* chu_ptr = push_to_ChunkVec(&arena->chunks, chu); + push_to_Word(&state->word_germ, chu_ptr); + state->unq_liter_germ = make_String(256); + } + if (state->word_germ.count > 0) { + Token tok = {.kind = TOK_WORD, .as.word = state->word_germ}; + Token* tok_ptr = push_to_TokenVec(&arena->tokens, tok); + push_to_TokenPtrVec(toks, tok_ptr); + state->word_germ = make_Word(256); + } +} + +TokenPtrVec lex_level(RawLine* line, char until, PreprocArena* arena) { + + TokenPtrVec line_toks = make_TokenPtrVec(256); + LexState state = { + .word_germ = make_Word(256), + .unq_liter_germ = make_String(256) + }; + + while (line->str[line->pos] != until) { + + if (line->str[line->pos] == ' ') { + close_word_and_lit_germs(&state, &line_toks, arena); + while (line->str[line->pos] == ' ') line->pos++; + continue; + } + + switch (line->str[line->pos]) { + case '\'': + Chunk sqchunk = collect_single_quoted_section(line); + Chunk* sqptr = push_to_ChunkVec(&arena->chunks, sqchunk); + push_to_Word(&state.word_germ, sqptr); + break; + case '\"': + lex_double_quoted_section(line, &state.word_germ, arena); + break; + case '|': + line->pos++; + close_word_and_lit_germs(&state, &line_toks, arena); + Token* tok_ptr = push_to_TokenVec(&arena->tokens, (Token){.kind = TOK_SPECIAL, .as.spec = PIPE}); + push_to_TokenPtrVec(&line_toks, tok_ptr); + break; + case '$': + exit(1); + /* + line->pos++; + MaybeChunk mchunk = lex_expansion_chunk(line, reg_arena, UN_Q); + if (mchunk.some) { + push_to_Word(&state.word_germ, mchunk.value); + } else { + push_to_String(&state.unq_liter_germ, '$'); + } + */ + break; + case '>': + close_word_and_lit_germs(&state, &line_toks, arena); + line->pos++; + + switch (line->str[line->pos]) { + case '(': + line->pos++; + TokenPtrVec sub_regtoks = lex_level(line, ')', arena); + line->pos++; + Region subreg = {.mode = CMDS, .tokens = sub_regtoks}; + Region* subreg_ptr = push_to_RegionVec(&arena->regions, subreg); + Token tok = {.kind = TOK_PROCSUB_OUT, .as.procsub_region = subreg_ptr}; + Token* tok_ptr = push_to_TokenVec(&arena->tokens, tok); + push_to_TokenPtrVec(&line_toks, tok_ptr); + break; + case '>': + line->pos++; + Token* tok_ptr2 = push_to_TokenVec(&arena->tokens, (Token){.kind = TOK_SPECIAL, .as.spec = WRITE_REDIR_APPEND}); + push_to_TokenPtrVec(&line_toks, tok_ptr2); + break; + default: + Token* tok_ptr3 = push_to_TokenVec(&arena->tokens, (Token){.kind = TOK_SPECIAL, .as.spec = WRITE_REDIR_TRUNC}); + push_to_TokenPtrVec(&line_toks, tok_ptr3); + } + break; + case '<': + close_word_and_lit_germs(&state, &line_toks, arena); + line->pos++; + if (line->str[line->pos] == '(') { + line->pos++; + TokenPtrVec sub_regtoks = lex_level(line, ')', arena); + line->pos++; + Region subreg = {.mode = CMDS, .tokens = sub_regtoks}; + Region* reg_ptr = push_to_RegionVec(&arena->regions, subreg); + Token reg_tok = {.kind = TOK_PROCSUB_IN, .as.procsub_region = reg_ptr}; + Token* reg_tok_ptr = push_to_TokenVec(&arena->tokens, reg_tok); + push_to_TokenPtrVec(&line_toks, reg_tok_ptr); + } else { + Token* redir_ptr = push_to_TokenVec(&arena->tokens, (Token){.kind = TOK_SPECIAL, .as.spec = READ_REDIR}); + push_to_TokenPtrVec(&line_toks, redir_ptr); + } + break; + default: + push_to_String(&state.unq_liter_germ, line->str[line->pos++]); + } + } + // TODO: handle until char _and_ EOL + close_word_and_lit_germs(&state, &line_toks, arena); + return line_toks; +} + + +TokenPtrVec lex(char* input, PreprocArena* arena) { + RawLine line = {.str = input, .pos = 0}; + TokenPtrVec toks = lex_level(&line, '\0', arena); + Token* eol_ptr = push_to_TokenVec(&arena->tokens, (Token){.kind = TOK_SPECIAL, .as.spec = EOL}); + push_to_TokenPtrVec(&toks, eol_ptr); + return toks; +} + + + + +// Parsing + +Connector parse_connector(Token* toks, size_t* tok_pos) { + Token tok = toks[*tok_pos]; + if (tok.kind != TOK_SPECIAL) exit(67); + switch (tok.as.spec) { + case PIPE: + (*tok_pos)++; + return CONN_PIPE; + case EOL: + (*tok_pos)++; + return CONN_EOL; + default: exit(76); + } +} + + +Command parse_command(Token* toks, size_t* tok_pos) { + TokenVec args = make_TokenVec(256); + RedirectVec redirects = make_RedirectVec(256); + while (true) { + Token curr_tok = toks[*tok_pos]; + if (curr_tok.kind == TOK_SPECIAL &&(curr_tok.as.spec == PIPE || curr_tok.as.spec == EOL)) { + break; + } + + switch (curr_tok.kind) { + case TOK_SPECIAL: + (*tok_pos)++; + Token name_tok = toks[*tok_pos]; + switch (curr_tok.as.spec) { + case WRITE_REDIR_TRUNC: + if(name_tok.kind != TOK_WORD) exit(1); + push_to_RedirectVec(&redirects, (Redirect){.kind = STDOUT_REDIR_TRUNC, .as.std_redir_filename = name_tok.as.word}); + break; + case WRITE_REDIR_APPEND: + if(name_tok.kind != TOK_WORD) exit(1); + push_to_RedirectVec(&redirects, (Redirect){.kind = STDOUT_REDIR_APPEND, .as.std_redir_filename = name_tok.as.word}); + break; + case READ_REDIR: + if(name_tok.kind != TOK_WORD) exit(1); + push_to_RedirectVec(&redirects, (Redirect){.kind = STDOUT_REDIR_APPEND, .as.std_redir_filename = name_tok.as.word}); + break; + default: + exit(69); + } + break; + default: + push_to_TokenVec(&args, curr_tok); + } + (*tok_pos)++; + } + return (Command){.args = args, .redirects = redirects}; +} + + +Pipeline parse_tokstream(Token* toks) { + size_t tok_pos = 0; + Pipeline result = make_Pipeline(256); + while (true) { + Command cmd = parse_command(toks, &tok_pos); + Connector conn = parse_connector(toks, &tok_pos); + push_to_Pipeline(&result, (PipelineElement){.cmd = cmd, .conn = conn}); + if (conn == CONN_EOL) break; + } + return result; +} -- cgit v1.2.3