X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/e555d8e3eeac56c977154eb42228a1a240edd59b..a80c47a8f5806b2163f99fadc374897519499670:/bk_man.c

diff --git a/bk_man.c b/bk_man.c
index b380a1d..bb4e5b5 100644
--- a/bk_man.c
+++ b/bk_man.c
@@ -13,7 +13,7 @@ typedef struct {
     int mindepth;
     char *filename;
     int charset;
-    wchar_t *bullet, *lquote, *rquote;
+    wchar_t *bullet, *rule, *lquote, *rquote;
 } manconfig;
 
 static void man_text(FILE *, word *,
@@ -127,6 +127,7 @@ static manconfig man_configure(paragraph *source) {
     ret.filename = dupstr("output.1");
     ret.charset = CS_ASCII;
     ret.bullet = L"\x2022\0o\0\0";
+    ret.rule = L"\x2500\0-\0\0";
     ret.lquote = L"\x2018\0\x2019\0\"\0\"\0\0";
     ret.rquote = uadv(ret.lquote);
 
@@ -170,6 +171,8 @@ static manconfig man_configure(paragraph *source) {
 		ret.filename = dupstr(adv(p->origkeyword));
 	    } else if (!ustricmp(p->keyword, L"man-bullet")) {
 		ret.bullet = uadv(p->keyword);
+	    } else if (!ustricmp(p->keyword, L"man-rule")) {
+		ret.rule = uadv(p->keyword);
 	    } else if (!ustricmp(p->keyword, L"man-quotes")) {
 		if (*uadv(p->keyword) && *uadv(uadv(p->keyword))) {
 		    ret.lquote = uadv(p->keyword);
@@ -180,7 +183,8 @@ static manconfig man_configure(paragraph *source) {
     }
 
     /*
-     * Now process fallbacks on quote characters and bullets.
+     * Now process fallbacks on quote characters, bullets, and the
+     * rule character.
      */
     while (*uadv(ret.rquote) && *uadv(uadv(ret.rquote)) &&
 	   (!troff_ok(ret.charset, ret.lquote) ||
@@ -193,6 +197,10 @@ static manconfig man_configure(paragraph *source) {
 	   !troff_ok(ret.charset, ret.bullet))
 	ret.bullet = uadv(ret.bullet);
 
+    while (*ret.rule && *uadv(ret.rule) &&
+	   !troff_ok(ret.charset, ret.rule))
+	ret.rule = uadv(ret.rule);
+
     return ret;
 }
 
@@ -207,8 +215,9 @@ paragraph *man_config_filename(char *filename)
     return cmdline_cfg_simple("man-filename", filename, NULL);
 }
 
-#define QUOTE_INITCTRL 1 /* quote initial . and ' on a line */
-#define QUOTE_QUOTES   2 /* quote double quotes by doubling them */
+#define QUOTE_INITCTRL    1 /* quote initial . and ' on a line */
+#define QUOTE_QUOTES      2 /* quote double quotes by doubling them */
+#define QUOTE_LITERAL     4 /* defeat special meaning of `, ', - in troff */
 
 void man_backend(paragraph *sourceform, keywordlist *keywords,
 		 indexdata *idx, void *unused) {
@@ -226,7 +235,10 @@ void man_backend(paragraph *sourceform, keywordlist *keywords,
     /*
      * Open the output file.
      */
-    fp = fopen(conf.filename, "w");
+    if (!strcmp(conf.filename, "-"))
+	fp = stdout;
+    else
+	fp = fopen(conf.filename, "w");
     if (!fp) {
 	error(err_cantopenw, conf.filename);
 	return;
@@ -239,6 +251,12 @@ void man_backend(paragraph *sourceform, keywordlist *keywords,
 	    man_text(fp, p->words, TRUE, 0, &conf);
 	}
 
+    /* Standard preamble */
+    /* Dodge to try to get literal U+0027 in output when required,
+     * bypassing groff's Unicode transform; pinched from pod2man */
+    fprintf(fp, ".ie \\n(.g .ds Aq \\(aq\n"
+	        ".el       .ds Aq '\n");
+
     /* .TH name-of-program manual-section */
     fprintf(fp, ".TH");
     if (conf.th && *conf.th) {
@@ -379,13 +397,19 @@ void man_backend(paragraph *sourceform, keywordlist *keywords,
 	break;
 
       case para_Rule:
-	/*
-	 * New paragraph containing a horizontal line 1/2em above the
-	 * baseline whose length is the line length minus the current
-	 * indent.
-	 */
-	cleanup_described_thing;
-	fprintf(fp, ".PP\n\\u\\l'\\n(.lu-\\n(.iu'\\d\n");
+	{
+	    char *ruletext;
+	    /*
+	     * New paragraph containing a horizontal line 1/2em above
+	     * the baseline, or a line of rule characters, whose
+	     * length is the line length minus the current indent.
+	     */
+	    cleanup_described_thing;
+	    man_convert(conf.rule, -1, &ruletext, 0, conf.charset, NULL);
+	    fprintf(fp, ".PP\n.ie t \\u\\l'\\n(.lu-\\n(.iu'\\d\n"
+		    ".el \\l'\\n(.lu-\\n(.iu\\&%s'\n", ruletext);
+	    sfree(ruletext);
+	}
 	break;
 
       case para_LcontPush:
@@ -404,7 +428,8 @@ void man_backend(paragraph *sourceform, keywordlist *keywords,
     /*
      * Tidy up.
      */
-    fclose(fp);
+    if (fp != stdout)
+	fclose(fp);
     man_conf_cleanup(conf);
 }
 
@@ -461,11 +486,36 @@ static int man_convert(wchar_t const *s, int maxlen,
 		 */
 		rdaddc(&out, '\\');
 		rdaddc(&out, '&');
-	    } else if (*q == '\\' || *q == '`') {
-		/*
-		 * Quote backslashes and backticks always.
-		 */
+	    }
+	    if (*q == '`' || *q == ' ') {
+		/* Quote backticks and nonbreakable spaces always. */
 		rdaddc(&out, '\\');
+	    } else if (*q == '\\') {
+		/* Turn backslashes into \e. */
+		rdaddsc(&out, "\\e");
+		continue;
+	    } else if (*q == '-') {
+		if (quote_props & QUOTE_LITERAL) {
+		    /*
+		     * Try to preserve literal U+002D.
+		     * This is quite awkward. Debian hacks groff so that
+		     * \- and - both produce it; elsewhere it's not necessarily
+		     * possible to get it.
+		     * Apparently \- is the preferred compromise despite
+		     * having minus-sign semantics, as it is non-breaking.
+		     * (pod2man uses it, anyway.)
+		     */
+		    rdaddc(&out, '\\');
+		} else {
+		    /* Turn nonbreakable hyphens into \(hy. */
+		    rdaddsc(&out, "\\(hy");
+		    continue;
+		}
+	    } else if (*q == '\'' && (quote_props & QUOTE_LITERAL)) {
+		/* Try to preserve literal U+0027 (using string defined
+		 * in preamble) */
+		rdaddsc(&out, "\\*(Aq"); /* "apostrophe quote" */
+		continue;
 	    } else if (*q == '"' && (quote_props & QUOTE_QUOTES)) {
 		/*
 		 * Double quote within double quotes. Quote it by
@@ -504,6 +554,26 @@ static int man_convert(wchar_t const *s, int maxlen,
     return !anyerr;
 }
 
+static int man_rdaddwc_reset(rdstringc *rs, int quote_props, manconfig *conf,
+			     charset_state *state) {
+    char *c;
+
+    man_convert(NULL, 0, &c, quote_props, conf->charset, state);
+    rdaddsc(rs, c);
+    if (*c)
+	quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
+    sfree(c);
+    *state = charset_init_state;
+    return quote_props;
+}
+
+static int man_rdaddctrl(rdstringc *rs, char *c, int quote_props,
+			 manconfig *conf, charset_state *state) {
+    quote_props = man_rdaddwc_reset(rs, quote_props, conf, state);
+    rdaddsc(rs, c);
+    return quote_props;
+}
+
 static int man_rdaddwc(rdstringc *rs, word *text, word *end,
 		       int quote_props, manconfig *conf,
 		       charset_state *state) {
@@ -536,46 +606,50 @@ static int man_rdaddwc(rdstringc *rs, word *text, word *end,
 	if (towordstyle(text->type) == word_Emph &&
 	    (attraux(text->aux) == attr_First ||
 	     attraux(text->aux) == attr_Only)) {
-	    man_convert(NULL, 0, &c, quote_props, conf->charset, state);
-	    rdaddsc(rs, c);
-	    if (*c)
-		quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
-	    sfree(c);
-	    *state = charset_init_state;
-	    rdaddsc(rs, "\\fI");
+	    quote_props = man_rdaddctrl(rs, "\\fI", quote_props, conf, state);
 	} else if ((towordstyle(text->type) == word_Code ||
 		    towordstyle(text->type) == word_WeakCode) &&
 		   (attraux(text->aux) == attr_First ||
 		    attraux(text->aux) == attr_Only)) {
-	    man_convert(NULL, 0, &c, quote_props, conf->charset, state);
-	    rdaddsc(rs, c);
-	    if (*c)
-		quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
-	    sfree(c);
-	    *state = charset_init_state;
-	    rdaddsc(rs, "\\fB");
+	    quote_props = man_rdaddctrl(rs, "\\fB", quote_props, conf, state);
 	}
 
+	if (towordstyle(text->type) == word_Code ||
+	    towordstyle(text->type) == word_WeakCode)
+	    quote_props |= QUOTE_LITERAL;
+
 	if (removeattr(text->type) == word_Normal) {
 	    charset_state s2 = *state;
+	    int len = ustrlen(text->text), hyphen = FALSE;
 
-	    if (man_convert(text->text, 0, &c, quote_props, conf->charset, &s2) ||
+	    if (text->breaks && text->text[len - 1] == '-') {
+		len--;
+		hyphen = TRUE;
+	    }
+	    if (len == 0 ||
+		man_convert(text->text, len, &c, quote_props, conf->charset,
+			    &s2) ||
 		!text->alt) {
-		rdaddsc(rs, c);
-		if (*c)
-		    quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
-		*state = s2;
+		if (len != 0) {
+		    rdaddsc(rs, c);
+		    if (*c)
+			quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
+		    *state = s2;
+		}
+		if (hyphen) {
+		    quote_props =
+			man_rdaddctrl(rs, "-", quote_props, conf, state);
+		    quote_props &= ~QUOTE_INITCTRL;
+		}
 	    } else {
 		quote_props = man_rdaddwc(rs, text->alt, NULL,
 					  quote_props, conf, state);
 	    }
-	    sfree(c);
+	    if (len != 0)
+		sfree(c);
 	} else if (removeattr(text->type) == word_WhiteSpace) {
-	    man_convert(L" ", 1, &c, quote_props, conf->charset, state);
-	    rdaddsc(rs, c);
-	    if (*c)
-		quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
-	    sfree(c);
+	    quote_props = man_rdaddctrl(rs, " ", quote_props, conf, state);
+	    quote_props &= ~QUOTE_INITCTRL;
 	} else if (removeattr(text->type) == word_Quote) {
 	    man_convert(quoteaux(text->aux) == quote_Open ?
 			conf->lquote : conf->rquote, 0,
@@ -588,21 +662,11 @@ static int man_rdaddwc(rdstringc *rs, word *text, word *end,
 	if (towordstyle(text->type) != word_Normal &&
 	    (attraux(text->aux) == attr_Last ||
 	     attraux(text->aux) == attr_Only)) {
-	    man_convert(NULL, 0, &c, quote_props, conf->charset, state);
-	    rdaddsc(rs, c);
-	    if (*c)
-		quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
-	    sfree(c);
-	    *state = charset_init_state;
-	    rdaddsc(rs, "\\fP");
+	    quote_props = man_rdaddctrl(rs, "\\fP", quote_props, conf, state);
 	}
 	break;
     }
-    man_convert(NULL, 0, &c, quote_props, conf->charset, state);
-    rdaddsc(rs, c);
-    if (*c)
-	quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
-    sfree(c);
+    quote_props = man_rdaddwc_reset(rs, quote_props, conf, state);
 
     return quote_props;
 }
@@ -624,7 +688,7 @@ static void man_codepara(FILE *fp, word *text, int charset) {
     for (; text; text = text->next) if (text->type == word_WeakCode) {
 	char *c;
 	wchar_t *t, *e;
-	int quote_props = QUOTE_INITCTRL;
+	int quote_props = QUOTE_INITCTRL | QUOTE_LITERAL;
 
 	t = text->text;
 	if (text->next && text->next->type == word_Emph) {