X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Flexer%2Flexer.c;h=817a07baf555f9a3bf479c7640c969439b0394f2;hb=69bf3f901b0a949cfa957950f55df78d0c86a765;hp=1caadac1de9854433d499122686d1ffa052d9704;hpb=0838c7ce8528a241fd6bb422767e187af4b5d9a7;p=pspp diff --git a/src/language/lexer/lexer.c b/src/language/lexer/lexer.c index 1caadac1de..817a07baf5 100644 --- a/src/language/lexer/lexer.c +++ b/src/language/lexer/lexer.c @@ -1,5 +1,5 @@ /* PSPP - a program for statistical analysis. - Copyright (C) 1997-9, 2000, 2006, 2009, 2010, 2011 Free Software Foundation, Inc. + Copyright (C) 1997-9, 2000, 2006, 2009, 2010, 2011, 2013, 2016 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -30,8 +30,8 @@ #include #include -#include "data/file-name.h" #include "language/command.h" +#include "language/lexer/macro.h" #include "language/lexer/scan.h" #include "language/lexer/segment.h" #include "language/lexer/token.h" @@ -45,7 +45,7 @@ #include "libpspp/str.h" #include "libpspp/u8-istream.h" #include "output/journal.h" -#include "output/text-item.h" +#include "output/output-item.h" #include "gl/c-ctype.h" #include "gl/minmax.h" @@ -68,6 +68,7 @@ struct lex_token size_t token_len; /* Length of source for token in bytes. */ size_t line_pos; /* Start of line containing token_pos. */ int first_line; /* Line number at token_pos. */ + bool from_macro; }; /* A source of tokens, corresponding to a syntax file. @@ -78,6 +79,7 @@ struct lex_source { struct ll ll; /* In lexer's list of sources. */ struct lex_reader *reader; + struct lexer *lexer; struct segmenter segmenter; bool eof; /* True if T_STOP was read from 'reader'. */ @@ -100,21 +102,25 @@ struct lex_source struct lex_token *tokens; /* Lookahead tokens for parser. */ }; -static struct lex_source *lex_source_create (struct lex_reader *); +static struct lex_source *lex_source_create (struct lexer *, + struct lex_reader *); static void lex_source_destroy (struct lex_source *); /* Lexer. */ struct lexer { struct ll_list sources; /* Contains "struct lex_source"s. */ + struct macro_set *macros; }; static struct lex_source *lex_source__ (const struct lexer *); +static struct substring lex_source_get_syntax__ (const struct lex_source *, + int n0, int n1); static const struct lex_token *lex_next__ (const struct lexer *, int n); static void lex_source_push_endcmd__ (struct lex_source *); static void lex_source_pop__ (struct lex_source *); -static bool lex_source_get__ (const struct lex_source *); +static bool lex_source_get (const struct lex_source *); static void lex_source_error_valist (struct lex_source *, int n0, int n1, const char *format, va_list) PRINTF_FORMAT (4, 0); @@ -128,10 +134,12 @@ lex_reader_init (struct lex_reader *reader, const struct lex_reader_class *class) { reader->class = class; - reader->syntax = LEX_SYNTAX_AUTO; - reader->error = LEX_ERROR_INTERACTIVE; + reader->syntax = SEG_MODE_AUTO; + reader->error = LEX_ERROR_CONTINUE; reader->file_name = NULL; + reader->encoding = NULL; reader->line_number = 0; + reader->eof = false; } /* Frees any file name already in READER and replaces it by a copy of @@ -140,15 +148,18 @@ void lex_reader_set_file_name (struct lex_reader *reader, const char *file_name) { free (reader->file_name); - reader->file_name = file_name != NULL ? xstrdup (file_name) : NULL; + reader->file_name = xstrdup_if_nonnull (file_name); } /* Creates and returns a new lexer. */ struct lexer * lex_create (void) { - struct lexer *lexer = xzalloc (sizeof *lexer); - ll_init (&lexer->sources); + struct lexer *lexer = xmalloc (sizeof *lexer); + *lexer = (struct lexer) { + .sources = LL_INITIALIZER (lexer->sources), + .macros = macro_set_create (), + }; return lexer; } @@ -162,10 +173,19 @@ lex_destroy (struct lexer *lexer) ll_for_each_safe (source, next, struct lex_source, ll, &lexer->sources) lex_source_destroy (source); + macro_set_destroy (lexer->macros); free (lexer); } } +/* Adds M to LEXER's set of macros. M replaces any existing macro with the + same name. Takes ownership of M. */ +void +lex_define_macro (struct lexer *lexer, struct macro *m) +{ + macro_set_add (lexer->macros, m); +} + /* Inserts READER into LEXER so that the next token read by LEXER comes from READER. Before the caller, LEXER must either be empty or at a T_ENDCMD token. */ @@ -173,7 +193,7 @@ void lex_include (struct lexer *lexer, struct lex_reader *reader) { assert (ll_is_empty (&lexer->sources) || lex_token (lexer) == T_ENDCMD); - ll_push_head (&lexer->sources, &lex_source_create (reader)->ll); + ll_push_head (&lexer->sources, &lex_source_create (lexer, reader)->ll); } /* Appends READER to LEXER, so that it will be read after all other current @@ -181,10 +201,10 @@ lex_include (struct lexer *lexer, struct lex_reader *reader) void lex_append (struct lexer *lexer, struct lex_reader *reader) { - ll_push_tail (&lexer->sources, &lex_source_create (reader)->ll); + ll_push_tail (&lexer->sources, &lex_source_create (lexer, reader)->ll); } -/* Advacning. */ +/* Advancing. */ static struct lex_token * lex_push_token__ (struct lex_source *src) @@ -195,20 +215,21 @@ lex_push_token__ (struct lex_source *src) src->tokens = deque_expand (&src->deque, src->tokens, sizeof *src->tokens); token = &src->tokens[deque_push_front (&src->deque)]; - token_init (&token->token); + token->token = (struct token) { .type = T_STOP }; + token->from_macro = false; return token; } static void lex_source_pop__ (struct lex_source *src) { - token_destroy (&src->tokens[deque_pop_back (&src->deque)].token); + token_uninit (&src->tokens[deque_pop_back (&src->deque)].token); } static void lex_source_pop_front (struct lex_source *src) { - token_destroy (&src->tokens[deque_pop_front (&src->deque)].token); + token_uninit (&src->tokens[deque_pop_front (&src->deque)].token); } /* Advances LEXER to the next token, consuming the current token. */ @@ -225,7 +246,7 @@ lex_get (struct lexer *lexer) lex_source_pop__ (src); while (deque_is_empty (&src->deque)) - if (!lex_source_get__ (src)) + if (!lex_source_get (src)) { lex_source_destroy (src); src = lex_source__ (lexer); @@ -268,23 +289,40 @@ lex_next_error (struct lexer *lexer, int n0, int n1, const char *format, ...) va_end (args); } -/* Prints a syntax error message saying that OPTION0 or one of the other - strings following it, up to the first NULL, is expected. */ +/* Prints a syntax error message saying that one of the strings provided as + varargs, up to the first NULL, is expected. */ void -lex_error_expecting (struct lexer *lexer, const char *option0, ...) +(lex_error_expecting) (struct lexer *lexer, ...) { - enum { MAX_OPTIONS = 8 }; - const char *options[MAX_OPTIONS + 1]; va_list args; - int n; - va_start (args, option0); - options[0] = option0; - n = 0; - while (n + 1 < MAX_OPTIONS && options[n] != NULL) - options[++n] = va_arg (args, const char *); + va_start (args, lexer); + lex_error_expecting_valist (lexer, args); va_end (args); +} +/* Prints a syntax error message saying that one of the options provided in + ARGS, up to the first NULL, is expected. */ +void +lex_error_expecting_valist (struct lexer *lexer, va_list args) +{ + enum { MAX_OPTIONS = 9 }; + const char *options[MAX_OPTIONS]; + int n = 0; + while (n < MAX_OPTIONS) + { + const char *option = va_arg (args, const char *); + if (!option) + break; + + options[n++] = option; + } + lex_error_expecting_array (lexer, options, n); +} + +void +lex_error_expecting_array (struct lexer *lexer, const char **options, size_t n) +{ switch (n) { case 0: @@ -333,7 +371,7 @@ lex_error_expecting (struct lexer *lexer, const char *option0, ...) break; default: - NOT_REACHED (); + lex_error (lexer, NULL); } } @@ -427,14 +465,14 @@ lex_end_of_command (struct lexer *lexer) /* Returns true if the current token is a number. */ bool -lex_is_number (struct lexer *lexer) +lex_is_number (const struct lexer *lexer) { return lex_next_is_number (lexer, 0); } /* Returns true if the current token is a string. */ bool -lex_is_string (struct lexer *lexer) +lex_is_string (const struct lexer *lexer) { return lex_next_is_string (lexer, 0); } @@ -442,14 +480,14 @@ lex_is_string (struct lexer *lexer) /* Returns the value of the current token, which must be a floating point number. */ double -lex_number (struct lexer *lexer) +lex_number (const struct lexer *lexer) { return lex_next_number (lexer, 0); } /* Returns true iff the current token is an integer. */ bool -lex_is_integer (struct lexer *lexer) +lex_is_integer (const struct lexer *lexer) { return lex_next_is_integer (lexer, 0); } @@ -457,7 +495,7 @@ lex_is_integer (struct lexer *lexer) /* Returns the value of the current token, which must be an integer. */ long -lex_integer (struct lexer *lexer) +lex_integer (const struct lexer *lexer) { return lex_next_integer (lexer, 0); } @@ -471,7 +509,7 @@ lex_integer (struct lexer *lexer) /* Returns true if the token N ahead of the current token is a number. */ bool -lex_next_is_number (struct lexer *lexer, int n) +lex_next_is_number (const struct lexer *lexer, int n) { enum token_type next_token = lex_next_token (lexer, n); return next_token == T_POS_NUM || next_token == T_NEG_NUM; @@ -479,7 +517,7 @@ lex_next_is_number (struct lexer *lexer, int n) /* Returns true if the token N ahead of the current token is a string. */ bool -lex_next_is_string (struct lexer *lexer, int n) +lex_next_is_string (const struct lexer *lexer, int n) { return lex_next_token (lexer, n) == T_STRING; } @@ -487,7 +525,7 @@ lex_next_is_string (struct lexer *lexer, int n) /* Returns the value of the token N ahead of the current token, which must be a floating point number. */ double -lex_next_number (struct lexer *lexer, int n) +lex_next_number (const struct lexer *lexer, int n) { assert (lex_next_is_number (lexer, n)); return lex_next_tokval (lexer, n); @@ -495,7 +533,7 @@ lex_next_number (struct lexer *lexer, int n) /* Returns true if the token N ahead of the current token is an integer. */ bool -lex_next_is_integer (struct lexer *lexer, int n) +lex_next_is_integer (const struct lexer *lexer, int n) { double value; @@ -509,7 +547,7 @@ lex_next_is_integer (struct lexer *lexer, int n) /* Returns the value of the token N ahead of the current token, which must be an integer. */ long -lex_next_integer (struct lexer *lexer, int n) +lex_next_integer (const struct lexer *lexer, int n) { assert (lex_next_is_integer (lexer, n)); return lex_next_tokval (lexer, n); @@ -587,7 +625,7 @@ lex_force_match_id (struct lexer *lexer, const char *identifier) return true; else { - lex_error_expecting (lexer, identifier, NULL_SENTINEL); + lex_error_expecting (lexer, identifier); return false; } } @@ -604,9 +642,16 @@ lex_force_match (struct lexer *lexer, enum token_type type) } else { - char *s = xasprintf ("`%s'", token_type_to_string (type)); - lex_error_expecting (lexer, s, NULL_SENTINEL); - free (s); + const char *type_string = token_type_to_string (type); + if (type_string) + { + char *s = xasprintf ("`%s'", type_string); + lex_error_expecting (lexer, s); + free (s); + } + else + lex_error_expecting (lexer, token_type_to_name (type)); + return false; } } @@ -637,7 +682,7 @@ lex_force_string (struct lexer *lexer) bool lex_force_string_or_id (struct lexer *lexer) { - return lex_is_integer (lexer) || lex_force_string (lexer); + return lex_token (lexer) == T_ID || lex_force_string (lexer); } /* If the current token is an integer, does nothing and returns true. @@ -654,6 +699,96 @@ lex_force_int (struct lexer *lexer) } } +/* If the current token is an integer in the range MIN...MAX (inclusive), does + nothing and returns true. Otherwise, reports an error and returns false. + If NAME is nonnull, then it is used in the error message. */ +bool +lex_force_int_range (struct lexer *lexer, const char *name, long min, long max) +{ + bool is_integer = lex_is_integer (lexer); + bool too_small = is_integer && lex_integer (lexer) < min; + bool too_big = is_integer && lex_integer (lexer) > max; + if (is_integer && !too_small && !too_big) + return true; + + if (min > max) + { + /* Weird, maybe a bug in the caller. Just report that we needed an + integer. */ + if (name) + lex_error (lexer, _("Integer expected for %s."), name); + else + lex_error (lexer, _("Integer expected.")); + } + else if (min == max) + { + if (name) + lex_error (lexer, _("Expected %ld for %s."), min, name); + else + lex_error (lexer, _("Expected %ld."), min); + } + else if (min + 1 == max) + { + if (name) + lex_error (lexer, _("Expected %ld or %ld for %s."), min, min + 1, name); + else + lex_error (lexer, _("Expected %ld or %ld."), min, min + 1); + } + else + { + bool report_lower_bound = (min > INT_MIN / 2) || too_small; + bool report_upper_bound = (max < INT_MAX / 2) || too_big; + + if (report_lower_bound && report_upper_bound) + { + if (name) + lex_error (lexer, + _("Expected integer between %ld and %ld for %s."), + min, max, name); + else + lex_error (lexer, _("Expected integer between %ld and %ld."), + min, max); + } + else if (report_lower_bound) + { + if (min == 0) + { + if (name) + lex_error (lexer, _("Expected non-negative integer for %s."), + name); + else + lex_error (lexer, _("Expected non-negative integer.")); + } + else if (min == 1) + { + if (name) + lex_error (lexer, _("Expected positive integer for %s."), + name); + else + lex_error (lexer, _("Expected positive integer.")); + } + } + else if (report_upper_bound) + { + if (name) + lex_error (lexer, + _("Expected integer less than or equal to %ld for %s."), + max, name); + else + lex_error (lexer, _("Expected integer less than or equal to %ld."), + max); + } + else + { + if (name) + lex_error (lexer, _("Integer expected for %s."), name); + else + lex_error (lexer, _("Integer expected.")); + } + } + return false; +} + /* If the current token is a number, does nothing and returns true. Otherwise, reports an error and returns false. */ bool @@ -744,13 +879,17 @@ lex_next__ (const struct lexer *lexer_, int n) return lex_source_next__ (src, n); else { - static const struct lex_token stop_token = - { TOKEN_INITIALIZER (T_STOP, 0.0, ""), 0, 0, 0, 0 }; - + static const struct lex_token stop_token = { .token = { .type = T_STOP } }; return &stop_token; } } +static const struct lex_token * +lex_source_front (const struct lex_source *src) +{ + return &src->tokens[deque_front (&src->deque, 0)]; +} + static const struct lex_token * lex_source_next__ (const struct lex_source *src, int n) { @@ -758,14 +897,12 @@ lex_source_next__ (const struct lex_source *src, int n) { if (!deque_is_empty (&src->deque)) { - struct lex_token *front; - - front = &src->tokens[deque_front (&src->deque, 0)]; + const struct lex_token *front = lex_source_front (src); if (front->token.type == T_STOP || front->token.type == T_ENDCMD) return front; } - lex_source_get__ (src); + lex_source_get (src); } return &src->tokens[deque_back (&src->deque, n)]; @@ -818,8 +955,8 @@ lex_next_tokcstr (const struct lexer *lexer, int n) The string is null-terminated (but the null terminator is not included in the returned substring's 'length'). - Only T_ID and T_STRING tokens have meaningful strings. For other tokens - this functions this function will always return NULL. + Only T_ID, T_MACRO_ID, T_STRING tokens have meaningful strings. For other + tokens this functions this function will always return NULL. The UTF-8 encoding of the returned string is correct for variable names and other identifiers. Use filename_to_utf8() to use it as a filename. Use @@ -830,99 +967,70 @@ lex_next_tokss (const struct lexer *lexer, int n) return lex_next (lexer, n)->string; } -/* If LEXER is positioned at the (pseudo)identifier S, skips it and returns - true. Otherwise, returns false. - - S may consist of an arbitrary number of identifiers, integers, and - punctuation e.g. "KRUSKAL-WALLIS", "2SLS", or "END INPUT PROGRAM". - Identifiers may be abbreviated to their first three letters. Currently only - hyphens, slashes, and equals signs are supported as punctuation (but it - would be easy to add more). +struct substring +lex_next_representation (const struct lexer *lexer, int n0, int n1) +{ + return lex_source_get_syntax__ (lex_source__ (lexer), n0, n1); +} - S must be an ASCII string. */ bool -lex_match_phrase (struct lexer *lexer, const char *s) +lex_next_is_from_macro (const struct lexer *lexer, int n) { - int tok_idx; - - for (tok_idx = 0; ; tok_idx++) - { - enum token_type token; - unsigned char c; - - while (c_isspace (*s)) - s++; - - c = *s; - if (c == '\0') - { - int i; - - for (i = 0; i < tok_idx; i++) - lex_get (lexer); - return true; - } - - token = lex_next_token (lexer, tok_idx); - switch (c) - { - case '-': - if (token != T_DASH) - return false; - s++; - break; - - case '/': - if (token != T_SLASH) - return false; - s++; - break; + return lex_next__ (lexer, n)->from_macro; +} - case '=': - if (token != T_EQUALS) - return false; - s++; - break; +static bool +lex_tokens_match (const struct token *actual, const struct token *expected) +{ + if (actual->type != expected->type) + return false; - case '0': case '1': case '2': case '3': case '4': - case '5': case '6': case '7': case '8': case '9': - { - unsigned int value; + switch (actual->type) + { + case T_POS_NUM: + case T_NEG_NUM: + return actual->number == expected->number; - if (token != T_POS_NUM) - return false; + case T_ID: + return lex_id_match (expected->string, actual->string); - value = 0; - do - { - value = value * 10 + (*s++ - '0'); - } - while (c_isdigit (*s)); + case T_STRING: + return (actual->string.length == expected->string.length + && !memcmp (actual->string.string, expected->string.string, + actual->string.length)); - if (lex_next_tokval (lexer, tok_idx) != value) - return false; - } - break; + default: + return true; + } +} - default: - if (lex_is_id1 (c)) - { - int len; +/* If LEXER is positioned at the sequence of tokens that may be parsed from S, + skips it and returns true. Otherwise, returns false. - if (token != T_ID) - return false; + S may consist of an arbitrary sequence of tokens, e.g. "KRUSKAL-WALLIS", + "2SLS", or "END INPUT PROGRAM". Identifiers may be abbreviated to their + first three letters. */ +bool +lex_match_phrase (struct lexer *lexer, const char *s) +{ + struct string_lexer slex; + struct token token; + int i; - len = lex_id_get_length (ss_cstr (s)); - if (!lex_id_match (ss_buffer (s, len), - lex_next_tokss (lexer, tok_idx))) - return false; + i = 0; + string_lexer_init (&slex, s, strlen (s), SEG_MODE_INTERACTIVE); + while (string_lexer_next (&slex, &token)) + if (token.type != SCAN_SKIP) + { + bool match = lex_tokens_match (lex_next (lexer, i++), &token); + token_uninit (&token); + if (!match) + return false; + } - s += len; - } - else - NOT_REACHED (); - } - } + while (i-- > 0) + lex_get (lexer); + return true; } static int @@ -1079,22 +1187,29 @@ lex_get_file_name (const struct lexer *lexer) return src == NULL ? NULL : src->reader->file_name; } +const char * +lex_get_encoding (const struct lexer *lexer) +{ + struct lex_source *src = lex_source__ (lexer); + return src == NULL ? NULL : src->reader->encoding; +} + /* Returns the syntax mode for the syntax file from which the current drawn is - drawn. Returns LEX_SYNTAX_AUTO for a T_STOP token or if the command's - source does not have line numbers. + drawn. Returns SEG_MODE_AUTO for a T_STOP token or if the command's source + does not have line numbers. There is no version of this function that takes an N argument because lookahead only works to the end of a command and any given command is always within a single syntax file. */ -enum lex_syntax_mode +enum segmenter_mode lex_get_syntax_mode (const struct lexer *lexer) { struct lex_source *src = lex_source__ (lexer); - return src == NULL ? LEX_SYNTAX_AUTO : src->reader->syntax; + return src == NULL ? SEG_MODE_AUTO : src->reader->syntax; } /* Returns the error mode for the syntax file from which the current drawn is - drawn. Returns LEX_ERROR_INTERACTIVE for a T_STOP token or if the command's + drawn. Returns LEX_ERROR_TERMINAL for a T_STOP token or if the command's source does not have line numbers. There is no version of this function that takes an N argument because @@ -1104,13 +1219,12 @@ enum lex_error_mode lex_get_error_mode (const struct lexer *lexer) { struct lex_source *src = lex_source__ (lexer); - return src == NULL ? LEX_ERROR_INTERACTIVE : src->reader->error; + return src == NULL ? LEX_ERROR_TERMINAL : src->reader->error; } /* If the source that LEXER is currently reading has error mode - LEX_ERROR_INTERACTIVE, discards all buffered input and tokens, so that the - next token to be read comes directly from whatever is next read from the - stream. + LEX_ERROR_TERMINAL, discards all buffered input and tokens, so that the next + token to be read comes directly from whatever is next read from the stream. It makes sense to call this function after encountering an error in a command entered on the console, because usually the user would prefer not to @@ -1119,7 +1233,7 @@ void lex_interactive_reset (struct lexer *lexer) { struct lex_source *src = lex_source__ (lexer); - if (src != NULL && src->reader->error == LEX_ERROR_INTERACTIVE) + if (src != NULL && src->reader->error == LEX_ERROR_TERMINAL) { src->head = src->tail = 0; src->journal_pos = src->seg_pos = src->line_pos = 0; @@ -1141,7 +1255,7 @@ lex_discard_rest_of_command (struct lexer *lexer) } /* Discards all lookahead tokens in LEXER, then discards all input sources - until it encounters one with error mode LEX_ERROR_INTERACTIVE or until it + until it encounters one with error mode LEX_ERROR_TERMINAL or until it runs out of input sources. */ void lex_discard_noninteractive (struct lexer *lexer) @@ -1153,7 +1267,7 @@ lex_discard_noninteractive (struct lexer *lexer) while (!deque_is_empty (&src->deque)) lex_source_pop__ (src); - for (; src != NULL && src->reader->error != LEX_ERROR_INTERACTIVE; + for (; src != NULL && src->reader->error != LEX_ERROR_TERMINAL; src = lex_source__ (lexer)) lex_source_destroy (src); } @@ -1208,33 +1322,20 @@ lex_source_read__ (struct lex_source *src) { do { - size_t head_ofs; - size_t space; - size_t n; - lex_source_expand__ (src); - head_ofs = src->head - src->tail; - space = src->allocated - head_ofs; - n = src->reader->class->read (src->reader, &src->buffer[head_ofs], - space, - segmenter_get_prompt (&src->segmenter)); + size_t head_ofs = src->head - src->tail; + size_t space = src->allocated - head_ofs; + enum prompt_style prompt = segmenter_get_prompt (&src->segmenter); + size_t n = src->reader->class->read (src->reader, &src->buffer[head_ofs], + space, prompt); assert (n <= space); if (n == 0) { - /* End of input. - - Ensure that the input always ends in a new-line followed by a null - byte, as required by the segmenter library. */ - - if (src->head == src->tail - || src->buffer[src->head - src->tail - 1] != '\n') - src->buffer[src->head++ - src->tail] = '\n'; - + /* End of input. */ + src->reader->eof = true; lex_source_expand__ (src); - src->buffer[src->head++ - src->tail] = '\0'; - return; } @@ -1252,16 +1353,24 @@ lex_source__ (const struct lexer *lexer) } static struct substring -lex_source_get_syntax__ (const struct lex_source *src, int n0, int n1) +lex_tokens_get_syntax__ (const struct lex_source *src, + const struct lex_token *token0, + const struct lex_token *token1) { - const struct lex_token *token0 = lex_source_next__ (src, n0); - const struct lex_token *token1 = lex_source_next__ (src, MAX (n0, n1)); size_t start = token0->token_pos; size_t end = token1->token_pos + token1->token_len; return ss_buffer (&src->buffer[start - src->tail], end - start); } +static struct substring +lex_source_get_syntax__ (const struct lex_source *src, int n0, int n1) +{ + return lex_tokens_get_syntax__ (src, + lex_source_next__ (src, n0), + lex_source_next__ (src, MAX (n0, n1))); +} + static void lex_ellipsize__ (struct substring in, char *out, size_t out_size) { @@ -1270,10 +1379,14 @@ lex_ellipsize__ (struct substring in, char *out, size_t out_size) int mblen; assert (out_size >= 16); - out_maxlen = out_size - (in.length >= out_size ? 3 : 0) - 1; + out_maxlen = out_size - 1; + if (in.length > out_maxlen - 3) + out_maxlen -= 3; + for (out_len = 0; out_len < in.length; out_len += mblen) { if (in.string[out_len] == '\n' + || in.string[out_len] == '\0' || (in.string[out_len] == '\r' && out_len + 1 < in.length && in.string[out_len + 1] == '\n')) @@ -1281,6 +1394,10 @@ lex_ellipsize__ (struct substring in, char *out, size_t out_size) mblen = u8_mblen (CHAR_CAST (const uint8_t *, in.string + out_len), in.length - out_len); + + if (mblen < 0) + break; + if (out_len + mblen > out_maxlen) break; } @@ -1295,13 +1412,22 @@ lex_source_error_valist (struct lex_source *src, int n0, int n1, { const struct lex_token *token; struct string s; - struct msg m; ds_init_empty (&s); token = lex_source_next__ (src, n0); if (token->token.type == T_ENDCMD) ds_put_cstr (&s, _("Syntax error at end of command")); + else if (token->from_macro) + { + /* XXX this isn't ideal, we should get the actual syntax */ + char *syntax = token_to_string (&token->token); + if (syntax) + ds_put_format (&s, _("Syntax error at `%s'"), syntax); + else + ds_put_cstr (&s, _("Syntax error")); + free (syntax); + } else { struct substring syntax = lex_source_get_syntax__ (src, n0, n1); @@ -1321,16 +1447,19 @@ lex_source_error_valist (struct lex_source *src, int n0, int n1, ds_put_cstr (&s, ": "); ds_put_vformat (&s, format, args); } - ds_put_byte (&s, '.'); - - m.category = MSG_C_SYNTAX; - m.severity = MSG_S_ERROR; - m.file_name = src->reader->file_name; - m.first_line = lex_source_get_first_line_number (src, n0); - m.last_line = lex_source_get_last_line_number (src, n1); - m.first_column = lex_source_get_first_column (src, n0); - m.last_column = lex_source_get_last_column (src, n1); - m.text = ds_steal_cstr (&s); + if (ds_last (&s) != '.') + ds_put_byte (&s, '.'); + + struct msg m = { + .category = MSG_C_SYNTAX, + .severity = MSG_S_ERROR, + .file_name = src->reader->file_name, + .first_line = lex_source_get_first_line_number (src, n0), + .last_line = lex_source_get_last_line_number (src, n1), + .first_column = lex_source_get_first_column (src, n0), + .last_column = lex_source_get_last_column (src, n1), + .text = ds_steal_cstr (&s), + }; msg_emit (&m); } @@ -1349,37 +1478,39 @@ lex_get_error (struct lex_source *src, const char *format, ...) va_end (args); } +/* Attempts to append an additional token into SRC's deque, reading more from + the underlying lex_reader if necessary. Returns true if a new token was + added to SRC's deque, false otherwise. */ static bool -lex_source_get__ (const struct lex_source *src_) +lex_source_try_get (struct lex_source *src) { - struct lex_source *src = CONST_CAST (struct lex_source *, src_); - + /* State maintained while scanning tokens. Usually we only need a single + state, but scanner_push() can return SCAN_SAVE to indicate that the state + needs to be saved and possibly restored later with SCAN_BACK. */ struct state { struct segmenter segmenter; enum segment_type last_segment; - int newlines; + int newlines; /* Number of newlines encountered so far. */ + /* Maintained here so we can update lex_source's similar members when we + finish. */ size_t line_pos; size_t seg_pos; }; - struct state state, saved; - enum scan_result result; - struct scanner scanner; - struct lex_token *token; - int n_lines; - int i; - - if (src->eof) - return false; - - state.segmenter = src->segmenter; - state.newlines = 0; - state.seg_pos = src->seg_pos; - state.line_pos = src->line_pos; - saved = state; + /* Initialize state. */ + struct state state = + { + .segmenter = src->segmenter, + .newlines = 0, + .seg_pos = src->seg_pos, + .line_pos = src->line_pos, + }; + struct state saved = state; - token = lex_push_token__ (src); + /* Append a new token to SRC and initialize it. */ + struct lex_token *token = lex_push_token__ (src); + struct scanner scanner; scanner_init (&scanner, &token->token); token->line_pos = src->line_pos; token->token_pos = src->seg_pos; @@ -1388,22 +1519,25 @@ lex_source_get__ (const struct lex_source *src_) else token->first_line = 0; + /* Extract segments and pass them through the scanner until we obtain a + token. */ for (;;) { + /* Extract a segment. */ + const char *segment = &src->buffer[state.seg_pos - src->tail]; + size_t seg_maxlen = src->head - state.seg_pos; enum segment_type type; - const char *segment; - size_t seg_maxlen; - int seg_len; - - segment = &src->buffer[state.seg_pos - src->tail]; - seg_maxlen = src->head - state.seg_pos; - seg_len = segmenter_push (&state.segmenter, segment, seg_maxlen, &type); + int seg_len = segmenter_push (&state.segmenter, segment, seg_maxlen, + src->reader->eof, &type); if (seg_len < 0) { + /* The segmenter needs more input to produce a segment. */ + assert (!src->reader->eof); lex_source_read__ (src); continue; } + /* Update state based on the segment. */ state.last_segment = type; state.seg_pos += seg_len; if (type == SEG_NEWLINE) @@ -1412,8 +1546,10 @@ lex_source_get__ (const struct lex_source *src_) state.line_pos = state.seg_pos; } - result = scanner_push (&scanner, type, ss_buffer (segment, seg_len), - &token->token); + /* Pass the segment into the scanner and try to get a token out. */ + enum scan_result result = scanner_push (&scanner, type, + ss_buffer (segment, seg_len), + &token->token); if (result == SCAN_SAVE) saved = state; else if (result == SCAN_BACK) @@ -1425,7 +1561,9 @@ lex_source_get__ (const struct lex_source *src_) break; } - n_lines = state.newlines; + /* If we've reached the end of a line, or the end of a command, then pass + the line to the output engine as a syntax text item. */ + int n_lines = state.newlines; if (state.last_segment == SEG_END_COMMAND && !src->suppress_next_newline) { n_lines++; @@ -1436,27 +1574,34 @@ lex_source_get__ (const struct lex_source *src_) n_lines--; src->suppress_next_newline = false; } - for (i = 0; i < n_lines; i++) + for (int i = 0; i < n_lines; i++) { - const char *newline; - const char *line; - size_t line_len; - char *syntax; - - line = &src->buffer[src->journal_pos - src->tail]; - newline = rawmemchr (line, '\n'); - line_len = newline - line; - if (line_len > 0 && line[line_len - 1] == '\r') - line_len--; - - syntax = malloc (line_len + 2); - memcpy (syntax, line, line_len); - syntax[line_len] = '\n'; - syntax[line_len + 1] = '\0'; - - text_item_submit (text_item_create_nocopy (TEXT_ITEM_SYNTAX, syntax)); - - src->journal_pos += newline - line + 1; + /* Beginning of line. */ + const char *line = &src->buffer[src->journal_pos - src->tail]; + + /* Calculate line length, including \n or \r\n end-of-line if present. + + We use src->head even though that may be beyond what we've actually + converted to tokens (which is only through state.line_pos). That's + because, if we're emitting the line due to SEG_END_COMMAND, we want to + take the whole line through the newline, not just through the '.'. */ + size_t max_len = src->head - src->journal_pos; + const char *newline = memchr (line, '\n', max_len); + size_t line_len = newline ? newline - line + 1 : max_len; + + /* Calculate line length excluding end-of-line. */ + size_t copy_len = line_len; + if (copy_len > 0 && line[copy_len - 1] == '\n') + copy_len--; + if (copy_len > 0 && line[copy_len - 1] == '\r') + copy_len--; + + /* Submit the line as syntax. */ + output_item_submit (text_item_create_nocopy (TEXT_ITEM_SYNTAX, + xmemdup0 (line, copy_len), + NULL)); + + src->journal_pos += line_len; } token->token_len = state.seg_pos - src->seg_pos; @@ -1469,95 +1614,162 @@ lex_source_get__ (const struct lex_source *src_) switch (token->token.type) { default: - break; + return true; case T_STOP: token->token.type = T_ENDCMD; src->eof = true; - break; + return true; case SCAN_BAD_HEX_LENGTH: lex_get_error (src, _("String of hex digits has %d characters, which " "is not a multiple of 2"), (int) token->token.number); - break; + return false; case SCAN_BAD_HEX_DIGIT: case SCAN_BAD_UNICODE_DIGIT: lex_get_error (src, _("`%c' is not a valid hex digit"), (int) token->token.number); - break; + return false; case SCAN_BAD_UNICODE_LENGTH: lex_get_error (src, _("Unicode string contains %d bytes, which is " "not in the valid range of 1 to 8 bytes"), (int) token->token.number); - break; + return false; case SCAN_BAD_UNICODE_CODE_POINT: lex_get_error (src, _("U+%04X is not a valid Unicode code point"), (int) token->token.number); - break; + return false; case SCAN_EXPECTED_QUOTE: lex_get_error (src, _("Unterminated string constant")); - break; + return false; case SCAN_EXPECTED_EXPONENT: lex_get_error (src, _("Missing exponent following `%s'"), token->token.string.string); - break; - - case SCAN_UNEXPECTED_DOT: - lex_get_error (src, _("Unexpected `.' in middle of command")); - break; + return false; case SCAN_UNEXPECTED_CHAR: { char c_name[16]; lex_get_error (src, _("Bad character %s in input"), uc_name (token->token.number, c_name)); + return false; } - break; case SCAN_SKIP: lex_source_pop_front (src); - break; + return false; + } + + NOT_REACHED (); +} + +static bool +lex_source_get__ (struct lex_source *src) +{ + for (;;) + { + if (src->eof) + return false; + else if (lex_source_try_get (src)) + return true; + } +} + +static bool +lex_source_get (const struct lex_source *src_) +{ + struct lex_source *src = CONST_CAST (struct lex_source *, src_); + + size_t old_count = deque_count (&src->deque); + if (!lex_source_get__ (src)) + return false; + + if (!settings_get_mexpand ()) + return true; + + struct macro_expander *me; + int retval = macro_expander_create (src->lexer->macros, + &lex_source_front (src)->token, + &me); + while (!retval) + { + if (!lex_source_get__ (src)) + { + /* This should not be reachable because we always get a T_ENDCMD at + the end of an input file (transformed from T_STOP by + lex_source_try_get()) and the macro_expander should always + terminate expansion on T_ENDCMD. */ + NOT_REACHED (); + } + + const struct lex_token *front = lex_source_front (src); + const struct macro_token mt = { + .token = front->token, + .representation = lex_tokens_get_syntax__ (src, front, front) + }; + retval = macro_expander_add (me, &mt); + } + if (retval < 0) + { + /* XXX handle case where there's a macro invocation starting from some + later token we've already obtained */ + macro_expander_destroy (me); + return true; + } + + /* XXX handle case where the macro invocation doesn't use all the tokens */ + while (deque_count (&src->deque) > old_count) + lex_source_pop_front (src); + + struct macro_tokens expansion = { .n = 0 }; + macro_expander_get_expansion (me, &expansion); + macro_expander_destroy (me); + + if (settings_get_mprint ()) + { + struct string mprint = DS_EMPTY_INITIALIZER; + macro_tokens_to_representation (&expansion, &mprint); + output_item_submit (text_item_create (TEXT_ITEM_LOG, ds_cstr (&mprint), + _("Macro Expansion"))); + ds_destroy (&mprint); } + for (size_t i = 0; i < expansion.n; i++) + { + *lex_push_token__ (src) = (struct lex_token) { + .token = expansion.mts[i].token, + .from_macro = true, + /* XXX the rest */ + }; + + ss_dealloc (&expansion.mts[i].representation); /* XXX should feed into lexer */ + } + free (expansion.mts); + return true; } static void lex_source_push_endcmd__ (struct lex_source *src) { - struct lex_token *token = lex_push_token__ (src); - token->token.type = T_ENDCMD; - token->token_pos = 0; - token->token_len = 0; - token->line_pos = 0; - token->first_line = 0; + *lex_push_token__ (src) = (struct lex_token) { .token = { .type = T_ENDCMD } }; } static struct lex_source * -lex_source_create (struct lex_reader *reader) +lex_source_create (struct lexer *lexer, struct lex_reader *reader) { struct lex_source *src; - enum segmenter_mode mode; src = xzalloc (sizeof *src); src->reader = reader; - - if (reader->syntax == LEX_SYNTAX_AUTO) - mode = SEG_MODE_AUTO; - else if (reader->syntax == LEX_SYNTAX_INTERACTIVE) - mode = SEG_MODE_INTERACTIVE; - else if (reader->syntax == LEX_SYNTAX_BATCH) - mode = SEG_MODE_BATCH; - else - NOT_REACHED (); - segmenter_init (&src->segmenter, mode); - + segmenter_init (&src->segmenter, reader->syntax); + src->lexer = lexer; src->tokens = deque_init (&src->deque, 4, sizeof *src->tokens); lex_source_push_endcmd__ (src); @@ -1569,9 +1781,11 @@ static void lex_source_destroy (struct lex_source *src) { char *file_name = src->reader->file_name; + char *encoding = src->reader->encoding; if (src->reader->class->destroy != NULL) src->reader->class->destroy (src->reader); free (file_name); + free (encoding); free (src->buffer); while (!deque_is_empty (&src->deque)) lex_source_pop__ (src); @@ -1584,7 +1798,6 @@ struct lex_file_reader { struct lex_reader reader; struct u8_istream *istream; - char *file_name; }; static struct lex_reader_class lex_file_reader_class; @@ -1598,7 +1811,7 @@ static struct lex_reader_class lex_file_reader_class; Returns a null pointer if FILE_NAME cannot be opened. */ struct lex_reader * lex_reader_for_file (const char *file_name, const char *encoding, - enum lex_syntax_mode syntax, + enum segmenter_mode syntax, enum lex_error_mode error) { struct lex_file_reader *r; @@ -1618,9 +1831,9 @@ lex_reader_for_file (const char *file_name, const char *encoding, r->reader.syntax = syntax; r->reader.error = error; r->reader.file_name = xstrdup (file_name); + r->reader.encoding = xstrdup_if_nonnull (encoding); r->reader.line_number = 1; r->istream = istream; - r->file_name = xstrdup (file_name); return &r->reader; } @@ -1639,7 +1852,7 @@ lex_file_read (struct lex_reader *r_, char *buf, size_t n, ssize_t n_read = u8_istream_read (r->istream, buf, n); if (n_read < 0) { - msg (ME, _("Error reading `%s': %s."), r->file_name, strerror (errno)); + msg (ME, _("Error reading `%s': %s."), r_->file_name, strerror (errno)); return 0; } return n_read; @@ -1653,12 +1866,11 @@ lex_file_close (struct lex_reader *r_) if (u8_istream_fileno (r->istream) != STDIN_FILENO) { if (u8_istream_close (r->istream) != 0) - msg (ME, _("Error closing `%s': %s."), r->file_name, strerror (errno)); + msg (ME, _("Error closing `%s': %s."), r_->file_name, strerror (errno)); } else u8_istream_free (r->istream); - free (r->file_name); free (r); } @@ -1678,16 +1890,17 @@ struct lex_string_reader static struct lex_reader_class lex_string_reader_class; /* Creates and returns a new lex_reader for the contents of S, which must be - encoded in UTF-8. The new reader takes ownership of S and will free it + encoded in the given ENCODING. The new reader takes ownership of S and will free it with ss_dealloc() when it is closed. */ struct lex_reader * -lex_reader_for_substring_nocopy (struct substring s) +lex_reader_for_substring_nocopy (struct substring s, const char *encoding) { struct lex_string_reader *r; r = xmalloc (sizeof *r); lex_reader_init (&r->reader, &lex_string_reader_class); - r->reader.syntax = LEX_SYNTAX_INTERACTIVE; + r->reader.syntax = SEG_MODE_AUTO; + r->reader.encoding = xstrdup_if_nonnull (encoding); r->s = s; r->offset = 0; @@ -1695,25 +1908,25 @@ lex_reader_for_substring_nocopy (struct substring s) } /* Creates and returns a new lex_reader for a copy of null-terminated string S, - which must be encoded in UTF-8. The caller retains ownership of S. */ + which must be encoded in ENCODING. The caller retains ownership of S. */ struct lex_reader * -lex_reader_for_string (const char *s) +lex_reader_for_string (const char *s, const char *encoding) { struct substring ss; ss_alloc_substring (&ss, ss_cstr (s)); - return lex_reader_for_substring_nocopy (ss); + return lex_reader_for_substring_nocopy (ss, encoding); } /* Formats FORMAT as a printf()-like format string and creates and returns a new lex_reader for the formatted result. */ struct lex_reader * -lex_reader_for_format (const char *format, ...) +lex_reader_for_format (const char *format, const char *encoding, ...) { struct lex_reader *r; va_list args; - va_start (args, format); - r = lex_reader_for_substring_nocopy (ss_cstr (xvasprintf (format, args))); + va_start (args, encoding); + r = lex_reader_for_substring_nocopy (ss_cstr (xvasprintf (format, args)), encoding); va_end (args); return r;