X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/0960a3d8eaea25ca71e3165556979b41ab10cb2b..39d3d9002fc5eb3a66e926af9e5350030762add9:/input.c

diff --git a/input.c b/input.c
index b31d583..337517d 100644
--- a/input.c
+++ b/input.c
@@ -103,6 +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;
@@ -219,6 +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 */
@@ -289,6 +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 */
@@ -422,6 +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 {
@@ -587,11 +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;
@@ -605,7 +610,7 @@ 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 */
@@ -627,7 +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();
@@ -804,6 +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.
@@ -894,7 +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
@@ -996,10 +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
@@ -1196,6 +1220,8 @@ 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);
@@ -1233,6 +1259,15 @@ 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;
@@ -1508,7 +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;
     }
@@ -1526,14 +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");
@@ -1543,10 +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;
 }