X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/b774705014b73abf1052a937deef534d34ca259d..refs/heads/master:/input.c

diff --git a/input.c b/input.c
index 365fec0..1ec733c 100644
--- a/input.c
+++ b/input.c
@@ -49,7 +49,7 @@ static void macrodef(tree234 *macros, wchar_t *name, wchar_t *text,
     m->name = name;
     m->text = text;
     if (add234(macros, m) != m) {
-	error(err_macroexists, &fpos, name);
+	err_macroexists(&fpos, name);
 	sfree(name);
 	sfree(text);
     }
@@ -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;
@@ -117,7 +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;
 	    }
@@ -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();
@@ -690,7 +693,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			   wtype == word_WeakCode) {
 		    wtype = word_Emph;
 		} else {
-		    error(err_brokencodepara, &t.pos);
+		    err_brokencodepara(&t.pos);
 		    prev_para_type = par.type;
 		    addpara(par, ret);
 		    while (t.type != tok_eop)   /* error recovery: */
@@ -719,7 +722,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 	     */
 	    dtor(t), t = get_token(in);
 	    if (t.type != tok_lbrace) {
-		error(err_explbr, &t.pos);
+		err_explbr(&t.pos);
 		continue;
 	    }
 
@@ -762,7 +765,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		     * don't give a cascade error.
 		     */
 		    sitem->type = -1;
-		    error(err_misplacedlcont, &t.pos);
+		    err_misplacedlcont(&t.pos);
 		}
 	    } else {
 		/*
@@ -787,7 +790,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 	} else if (t.type == tok_rbrace) {
 	    struct crossparaitem *sitem = stk_pop(crossparastk);
 	    if (!sitem)
-		error(err_unexbrace, &t.pos);
+		err_unexbrace(&t.pos);
 	    else {
 		switch (sitem->type) {
 		  case c_lcont:
@@ -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.
@@ -824,12 +832,14 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		needkw = -1;
 		break;
 	      case c__invalid:
-		error(err_badparatype, t.text, &t.pos);
+		err_badparatype(t.text, &t.pos);
 		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);
@@ -881,8 +891,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		par.type == para_UnnumberedChapter) {
 		struct crossparaitem *sitem = stk_top(crossparastk);
 		if (sitem && (sitem->seen_lcont || sitem->seen_quote)) {
-		    error(err_sectmarkerinblock,
-			  &t.pos,
+		    err_sectmarkerinblock(			  &t.pos,
 			  (sitem->seen_lcont ? "lcont" : "quote"));
 		}
 	    }
@@ -896,7 +905,21 @@ 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
@@ -921,7 +944,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			}
 		    }
 		    if (t.type != tok_rbrace) {
-			error(err_kwunclosed, &t.pos);
+			err_kwunclosed(&t.pos);
 			continue;
 		    }
 		    rdadd(&rs, 0);     /* add string terminator */
@@ -934,11 +957,11 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 
 		/* See whether we have the right number of keywords. */
 		if ((needkw & 48) && nkeys > 0)
-		    error(err_kwillegal, &fp);
+		    err_kwillegal(&fp);
 		if ((needkw & 11) && nkeys == 0)
-		    error(err_kwexpected, &fp);
+		    err_kwexpected(&fp);
 		if ((needkw & 5) && nkeys > 1)
-		    error(err_kwtoomany, &fp);
+		    err_kwtoomany(&fp);
 
 		if (is_macro) {
 		    /*
@@ -954,7 +977,8 @@ 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 */
@@ -971,7 +995,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		    if (t.type != tok_eop && t.type != tok_eof &&
 			(start_cmd == c__invalid ||
 			 t.type != tok_cmd || t.cmd != start_cmd)) {
-			error(err_bodyillegal, &t.pos);
+			err_bodyillegal(&t.pos);
 			/* Error recovery: eat the rest of the paragraph */
 			while (t.type != tok_eop && t.type != tok_eof &&
 			       (start_cmd == c__invalid ||
@@ -1088,7 +1112,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		}
 		break;
 	      case tok_lbrace:
-		error(err_unexbrace, &t.pos);
+		err_unexbrace(&t.pos);
 		/* Error recovery: push nop */
 		sitem = snew(struct stack_item);
 		sitem->type = stack_nop;
@@ -1174,7 +1198,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		     */
 		    dtor(t), t = get_token(in);
 		    if (t.type != tok_lbrace) {
-			error(err_explbr, &t.pos);
+			err_explbr(&t.pos);
 		    } else {
 			int braces = 1;
 			while (braces > 0) {
@@ -1184,7 +1208,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			    else if (t.type == tok_rbrace)
 				braces--;
 			    else if (t.type == tok_eof) {
-				error(err_commenteof, &t.pos);
+				err_commenteof(&t.pos);
 				break;
 			    }
 			}
@@ -1203,7 +1227,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
                     type = t.cmd;
 		    dtor(t), t = get_token(in);
 		    if (t.type != tok_lbrace) {
-			error(err_explbr, &t.pos);
+			err_explbr(&t.pos);
 		    } else {
 			/*
 			 * Enforce that \q may not be used anywhere
@@ -1232,7 +1256,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			    }
 			    stype = stack_quote;
 			} else {
-			    error(err_codequote, &t.pos);
+			    err_codequote(&t.pos);
 			    stype = stack_nop;
 			}
 			sitem = snew(struct stack_item);
@@ -1240,7 +1264,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			sitem->type = stype;
                         if (type == c_cq) {
                             if (style != word_Normal) {
-                                error(err_nestedstyles, &t.pos);
+                                err_nestedstyles(&t.pos);
                             } else {
                                 style = word_WeakCode;
                                 spcstyle = tospacestyle(style);
@@ -1278,7 +1302,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			    wdtext = ustrftime(NULL, broken);
 			    wd.type = style;
 			} else {
-			    error(err_explbr, &t.pos);
+			    err_explbr(&t.pos);
 			    wdtext = NULL;
 			}
 		    } else {
@@ -1300,7 +1324,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			}
 			sfree(rs.text);
 			if (t.type != tok_rbrace) {
-			    error(err_kwexprbr, &t.pos);
+			    err_kwexprbr(&t.pos);
 			}
 		    }
 		    wd.alt = NULL;
@@ -1330,7 +1354,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			if (t.type == tok_cmd &&
 			    (t.cmd == c_i || t.cmd == c_ii)) {
 			    if (indexing) {
-				error(err_nestedindex, &t.pos);
+				err_nestedindex(&t.pos);
 			    } else {
 				/* Add an index-reference word with no
 				 * text as yet */
@@ -1361,7 +1385,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			if (t.type == tok_cmd &&
 			    (t.cmd == c_e || t.cmd == c_c || t.cmd == c_cw)) {
 			    if (style != word_Normal)
-				error(err_nestedstyles, &t.pos);
+				err_nestedstyles(&t.pos);
 			    else {
 				style = (t.cmd == c_c ? word_Code :
 					 t.cmd == c_cw ? word_WeakCode :
@@ -1372,7 +1396,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 			    dtor(t), t = get_token(in);
 			}
 			if (t.type != tok_lbrace) {
-			    error(err_explbr, &t.pos);
+			    err_explbr(&t.pos);
 			    sfree(sitem);
 			} else {
 			    stk_push(parsestk, sitem);
@@ -1384,7 +1408,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		  case c_e:
 		    type = t.cmd;
 		    if (style != word_Normal) {
-			error(err_nestedstyles, &t.pos);
+			err_nestedstyles(&t.pos);
 			/* Error recovery: eat lbrace, push nop. */
 			dtor(t), t = get_token(in);
 			sitem = snew(struct stack_item);
@@ -1394,7 +1418,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		    }
 		    dtor(t), t = get_token(in);
 		    if (t.type != tok_lbrace) {
-			error(err_explbr, &t.pos);
+			err_explbr(&t.pos);
 		    } else {
 			style = (type == c_c ? word_Code :
 				 type == c_cw ? word_WeakCode :
@@ -1411,7 +1435,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		  case c_I:
 		    type = t.cmd;
 		    if (indexing) {
-			error(err_nestedindex, &t.pos);
+			err_nestedindex(&t.pos);
 			/* Error recovery: eat lbrace, push nop. */
 			dtor(t), t = get_token(in);
 			sitem = snew(struct stack_item);
@@ -1430,7 +1454,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		    if (t.type == tok_cmd &&
 			(t.cmd == c_e || t.cmd == c_c || t.cmd == c_cw)) {
 			if (style != word_Normal)
-			    error(err_nestedstyles, &t.pos);
+			    err_nestedstyles(&t.pos);
 			else {
 			    style = (t.cmd == c_c ? word_Code :
 				     t.cmd == c_cw ? word_WeakCode :
@@ -1442,7 +1466,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		    }
 		    if (t.type != tok_lbrace) {
 			sfree(sitem);
-			error(err_explbr, &t.pos);
+			err_explbr(&t.pos);
 		    } else {
 			/* Add an index-reference word with no text as yet */
 			wd.type = word_IndexRef;
@@ -1505,7 +1529,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 		    break;
 		  default:
 		    if (!macrolookup(macros, in, t.text, &t.pos))
-			error(err_badmidcmd, t.text, &t.pos);
+			err_badmidcmd(t.text, &t.pos);
 		    break;
 		}
 	    }
@@ -1518,11 +1542,21 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
 	if (stk_top(parsestk)) {
 	    while ((sitem = stk_pop(parsestk)))
 		sfree(sitem);
-	    error(err_missingrbrace, &t.pos);
+	    err_missingrbrace(&t.pos);
 	}
 	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;
     }
@@ -1530,7 +1564,7 @@ static void read_file(paragraph ***ret, input *in, indexdata *idx) {
     if (stk_top(crossparastk)) {
 	void *p;
 
-	error(err_missingrbrace2, &t.pos);
+	err_missingrbrace2(&t.pos);
 	while ((p = stk_pop(crossparastk)))
 	    sfree(p);
     }
@@ -1540,27 +1574,89 @@ 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;
 }