X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/f15300499bce37cd28ea2ace0f2bd1c364fc835e..d274495f584e38e4210ea8616120b2c3e3e3e8b8:/input.c diff --git a/input.c b/input.c index 59ea326..337517d 100644 --- a/input.c +++ b/input.c @@ -86,9 +86,7 @@ static void input_configure(input *in, paragraph *cfg) { assert(cfg->type == para_Config); if (!ustricmp(cfg->keyword, L"input-charset")) { - char *csname = utoa_dup(uadv(cfg->keyword), CS_ASCII); - in->charset = charset_from_localenc(csname); - sfree(csname); + in->charset = charset_from_ustr(&cfg->fpos, uadv(cfg->keyword)); } } @@ -105,6 +103,8 @@ static int get(input *in, filepos *pos, rdstringc *rsc) { } else if (in->stack) { wchar_t c = in->stack->text[in->stack->ptr]; + if (pos) + *pos = in->stack->pos; if (in->stack->text[++in->stack->ptr] == L'\0') { macrostack *tmp = in->stack; in->stack = tmp->next; @@ -221,6 +221,7 @@ enum { c_c, /* code */ c_cfg, /* configuration directive */ c_copyright, /* copyright statement */ + c_cq, /* quoted code (sugar for \q{\cw{x}}) */ c_cw, /* weak code */ c_date, /* document processing date */ c_dd, /* description list: description */ @@ -291,6 +292,7 @@ static void match_kw(token *tok) { {"c", c_c}, /* code */ {"cfg", c_cfg}, /* configuration directive */ {"copyright", c_copyright}, /* copyright statement */ + {"cq", c_cq}, /* quoted code (sugar for \q{\cw{x}}) */ {"cw", c_cw}, /* weak code */ {"date", c_date}, /* document processing date */ {"dd", c_dd}, /* description list: description */ @@ -424,6 +426,7 @@ token get_token(input *in) { c == '#' || c == '{' || c == '}' || c == '.') { /* single-char command */ rdadd(&rs, c); + prevpos = rsc.pos; } else if (c == 'u') { int len = 0; do { @@ -589,11 +592,11 @@ static paragraph *addpara(paragraph newpara, paragraph ***hptrptr) { /* * Reads a single file (ie until get() returns EOF) */ -static void read_file(paragraph ***ret, input *in, indexdata *idx) { +static void read_file(paragraph ***ret, input *in, indexdata *idx, + tree234 *macros) { token t; paragraph par; word wd, **whptr, **idximplicit; - tree234 *macros; wchar_t utext[2], *wdtext; int style, spcstyle; int already; @@ -607,11 +610,12 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) { stack_style = 2, /* \e, \c, \cw */ stack_idx = 4, /* \I, \i, \ii */ stack_hyper = 8, /* \W */ - stack_quote = 16, /* \q */ + stack_quote = 16 /* \q */ } type; word **whptr; /* to restore from \u alternatives */ word **idximplicit; /* to restore from \u alternatives */ filepos fpos; + int in_code; } *sitem; stack parsestk; struct crossparaitem { @@ -628,7 +632,6 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) { t.text = NULL; t.origtext = NULL; - macros = newtree234(macrocmp); already = FALSE; crossparastk = stk_new(); @@ -805,6 +808,11 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) { continue; } + while (t.type == tok_cmd && + macrolookup(macros, in, t.text, &t.pos)) { + dtor(t), t = get_token(in); + } + /* * This token begins a paragraph. See if it's one of the * special commands that define a paragraph type. @@ -895,7 +903,21 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) { /* Get keywords. */ dtor(t), t = get_token(in); fp = t.pos; - while (t.type == tok_lbrace) { + while (t.type == tok_lbrace || + (t.type == tok_white && (needkw & 24))) { + /* + * In paragraph types which can't accept any + * body text (such as \cfg), we are lenient + * about whitespace between keywords. This is + * important for \cfg in particular since it + * can often have many keywords which are long + * pieces of text, so it's useful to permit the + * user to wrap the line between them. + */ + if (t.type == tok_white) { + dtor(t), t = get_token(in); /* eat the space */ + continue; + } /* This is a keyword. */ nkeys++; /* FIXME: there will be bugs if anyone specifies an @@ -997,10 +1019,11 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) { * Mid-paragraph commands: * * \K \k - * \c \cw + * \c \cw \cq * \e * \i \ii * \I + * \q * \u * \W * \date @@ -1197,25 +1220,54 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) { } break; case c_q: + case c_cq: + type = t.cmd; dtor(t), t = get_token(in); if (t.type != tok_lbrace) { error(err_explbr, &t.pos); } else { - wd.text = NULL; - wd.type = toquotestyle(style); - wd.alt = NULL; - wd.aux = quote_Open; - wd.fpos = t.pos; - wd.breaks = FALSE; - if (!indexing || index_visible) - addword(wd, &whptr); - if (indexing) { - rdadd(&indexstr, L'"'); - addword(wd, &idximplicit); + /* + * Enforce that \q may not be used anywhere + * within \c. (It shouldn't be necessary + * since the whole point of \c should be + * that the user wants to exercise exact + * control over the glyphs used, and + * forbidding it has the useful effect of + * relieving some backends of having to + * make difficult decisions.) + */ + int stype; + + if (style != word_Code && style != word_WeakCode) { + wd.text = NULL; + wd.type = toquotestyle(style); + wd.alt = NULL; + wd.aux = quote_Open; + wd.fpos = t.pos; + wd.breaks = FALSE; + if (!indexing || index_visible) + addword(wd, &whptr); + if (indexing) { + rdadd(&indexstr, L'"'); + addword(wd, &idximplicit); + } + stype = stack_quote; + } else { + error(err_codequote, &t.pos); + stype = stack_nop; } sitem = snew(struct stack_item); sitem->fpos = t.pos; - sitem->type = stack_quote; + sitem->type = stype; + if (type == c_cq) { + if (style != word_Normal) { + error(err_nestedstyles, &t.pos); + } else { + style = word_WeakCode; + spcstyle = tospacestyle(style); + sitem->type |= stack_style; + } + } stk_push(parsestk, sitem); } break; @@ -1491,7 +1543,17 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) { } stk_free(parsestk); prev_para_type = par.type; - addpara(par, ret); + /* + * Before we add the paragraph to the output list, we + * should check that there was any text in it at all; there + * might not be if (for example) the paragraph contained + * nothing but an unrecognised command sequence, and if we + * put an empty paragraph on the list it may confuse the + * back ends later on. + */ + if (par.words) { + addpara(par, ret); + } if (t.type == tok_eof) already = TRUE; } @@ -1509,14 +1571,32 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) { * this cleanup doesn't happen. */ dtor(t); - macrocleanup(macros); stk_free(crossparastk); } +struct { + char const *magic; + size_t nmagic; + void (*reader)(input *); +} magics[] = { + { "%!FontType1-", 12, &read_pfa_file }, + { "%!PS-AdobeFont-", 15, &read_pfa_file }, + { "\x80\x01", 2, &read_pfb_file }, + { "StartFontMetrics", 16, &read_afm_file }, + { "\x00\x01\x00\x00", 4, &read_sfnt_file }, + { "true", 4, &read_sfnt_file }, +}; + paragraph *read_input(input *in, indexdata *idx) { paragraph *head = NULL; paragraph **hptr = &head; + tree234 *macros; + char mag[16]; + size_t len, i; + void (*reader)(input *); + + macros = newtree234(macrocmp); while (in->currindex < in->nfiles) { in->currfp = fopen(in->filenames[in->currindex], "r"); @@ -1526,10 +1606,25 @@ paragraph *read_input(input *in, indexdata *idx) { in->csstate = charset_init_state; in->wcpos = in->nwc = 0; in->pushback_chars = NULL; - read_file(&hptr, in, idx); + reader = NULL; + len = fread(mag, 1, sizeof(mag), in->currfp); + for (i = 0; i < lenof(magics); i++) { + if (len >= magics[i].nmagic && + memcmp(mag, magics[i].magic, magics[i].nmagic) == 0) { + reader = magics[i].reader; + break; + } + } + rewind(in->currfp); + if (reader == NULL) + read_file(&hptr, in, idx, macros); + else + (*reader)(in); } in->currindex++; } + macrocleanup(macros); + return head; }