X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/6ff15f2bcd24c07d07e00c2e4edce9df11af3b40..8f664e7e91c918cd13248f6b684580c4dd2cdb31:/input.c

diff --git a/input.c b/input.c
index 4eb1d8c..1410022 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;
@@ -119,7 +119,8 @@ static int get(input *in, filepos *pos, rdstringc *rsc) {
 	    int c = getc(in->currfp);
 
 	    if (c == EOF) {
-		fclose(in->currfp);
+		if (in->wantclose)
+		    fclose(in->currfp);
 		in->currfp = NULL;
 		return EOF;
 	    }
@@ -221,6 +222,7 @@ 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 +293,7 @@ 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 +427,7 @@ 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 +593,11 @@ 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;
@@ -607,7 +611,7 @@ 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 */
@@ -629,7 +633,6 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 
     t.text = NULL;
     t.origtext = NULL;
-    macros = newtree234(macrocmp);
     already = FALSE;
 
     crossparastk = stk_new();
@@ -806,6 +809,11 @@ 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.
@@ -828,8 +836,10 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		needkw = 4;
 		break;
 	      case c__comment:
-		if (isbrace(in))
+		if (isbrace(in)) {
+		    needkw = -1;
 		    break;	       /* `\#{': isn't a comment para */
+		}
 		do {
 		    dtor(t), t = get_token(in);
 		} while (t.type != tok_eop && t.type != tok_eof);
@@ -896,7 +906,21 @@ 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
@@ -954,7 +978,8 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			    rdadd(&macrotext, L'\n');
 			rdadds(&macrotext, t.text);
 			dtor(t), t = get_token(in);
-			if (t.type == tok_eop) break;
+			if (t.type == tok_eop || t.type == tok_eof)
+                            break;
 		    }
 		    macrodef(macros, rs.text, macrotext.text, fp);
 		    continue;	       /* next paragraph */
@@ -998,10 +1023,11 @@ 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
@@ -1198,6 +1224,8 @@ 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);
@@ -1235,6 +1263,15 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			sitem = snew(struct stack_item);
 			sitem->fpos = t.pos;
 			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;
@@ -1510,7 +1547,17 @@ 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;
     }
@@ -1528,27 +1575,89 @@ 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;
+    int binary;
+    void (*reader)(input *);
+} magics[] = {
+    { "%!FontType1-",     12, FALSE, &read_pfa_file },
+    { "%!PS-AdobeFont-",  15, FALSE, &read_pfa_file },
+    { "\x80\x01",          2, TRUE,  &read_pfb_file },
+    { "StartFontMetrics", 16, FALSE, &read_afm_file },
+    { "\x00\x01\x00\x00",  4, TRUE,  &read_sfnt_file },
+    { "true",		   4, TRUE,  &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;
+    int binary;
+    void (*reader)(input *);
+
+    macros = newtree234(macrocmp);
 
     while (in->currindex < in->nfiles) {
-	in->currfp = fopen(in->filenames[in->currindex], "r");
+	setpos(in, in->filenames[in->currindex]);
+	in->charset = in->defcharset;
+	in->csstate = charset_init_state;
+	in->wcpos = in->nwc = 0;
+	in->pushback_chars = NULL;
+
+	if (!in->filenames[in->currindex]) {
+	    in->currfp = stdin;
+	    in->wantclose = FALSE;     /* don't fclose stdin */
+	    /*
+	     * When reading standard input, we always expect to see
+	     * an actual Halibut file and not any of the unusual
+	     * input types like fonts.
+	     */
+	    reader = NULL;
+	} else {
+	    /*
+	     * Open the file in binary mode to look for magic
+	     * numbers. We'll switch to text mode if we find we're
+	     * looking at a text file type.
+	     */
+	    in->currfp = fopen(in->filenames[in->currindex], "rb");
+	    binary = FALSE; /* default to Halibut source, which is text */
+	    if (in->currfp) {
+		in->wantclose = TRUE;
+		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;
+			binary = magics[i].binary;
+			break;
+		    }
+		}
+		rewind(in->currfp);
+	    }
+	    if (!binary) {
+		fclose(in->currfp);
+		in->currfp = fopen(in->filenames[in->currindex], "r");
+	    }
+	}
 	if (in->currfp) {
-	    setpos(in, in->filenames[in->currindex]);
-	    in->charset = in->defcharset;
-	    in->csstate = charset_init_state;
-	    in->wcpos = in->nwc = 0;
-	    in->pushback_chars = NULL;
-	    read_file(&hptr, in, idx);
+	    if (reader == NULL) {
+		read_file(&hptr, in, idx, macros);
+	    } else {
+		(*reader)(in);
+	    }
 	}
 	in->currindex++;
     }
 
+    macrocleanup(macros);
+
     return head;
 }