X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/6377de3da9998c6305a5b0651ac9922da15fd33e..0ec6a4292ed4ccc4c3c3c24fe9402621066b8c10:/bk_man.c

diff --git a/bk_man.c b/bk_man.c
index 343bf06..dce77d4 100644
--- a/bk_man.c
+++ b/bk_man.c
@@ -215,8 +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) {
@@ -247,6 +248,12 @@ 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) {
@@ -475,7 +482,8 @@ static int man_convert(wchar_t const *s, int maxlen,
 		 */
 		rdaddc(&out, '\\');
 		rdaddc(&out, '&');
-	    } else if (*q == '`' || *q == ' ') {
+	    }
+	    if (*q == '`' || *q == ' ') {
 		/* Quote backticks and nonbreakable spaces always. */
 		rdaddc(&out, '\\');
 	    } else if (*q == '\\') {
@@ -483,8 +491,26 @@ static int man_convert(wchar_t const *s, int maxlen,
 		rdaddsc(&out, "\\e");
 		continue;
 	    } else if (*q == '-') {
-		/* Turn nonbreakable hyphens into \(hy. */
-		rdaddsc(&out, "\\(hy");
+		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)) {
 		/*
@@ -537,6 +563,13 @@ static int man_rdaddwc_reset(rdstringc *rs, int quote_props, manconfig *conf,
     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) {
@@ -569,16 +602,18 @@ 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)) {
-	    quote_props = man_rdaddwc_reset(rs, quote_props, conf, 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)) {
-	    quote_props = man_rdaddwc_reset(rs, quote_props, conf, 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;
@@ -599,9 +634,8 @@ static int man_rdaddwc(rdstringc *rs, word *text, word *end,
 		}
 		if (hyphen) {
 		    quote_props =
-			man_rdaddwc_reset(rs, quote_props, conf, state);
-		    rdaddc(rs, '-');
-		    quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
+			man_rdaddctrl(rs, "-", quote_props, conf, state);
+		    quote_props &= ~QUOTE_INITCTRL;
 		}
 	    } else {
 		quote_props = man_rdaddwc(rs, text->alt, NULL,
@@ -610,9 +644,8 @@ static int man_rdaddwc(rdstringc *rs, word *text, word *end,
 	    if (len != 0)
 		sfree(c);
 	} else if (removeattr(text->type) == word_WhiteSpace) {
-	    quote_props = man_rdaddwc_reset(rs, quote_props, conf, state);
-	    rdaddc(rs, ' ');
-	    quote_props &= ~QUOTE_INITCTRL;   /* not at start any more */
+	    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,
@@ -625,8 +658,7 @@ 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)) {
-	    quote_props = man_rdaddwc_reset(rs, quote_props, conf, state);
-	    rdaddsc(rs, "\\fP");
+	    quote_props = man_rdaddctrl(rs, "\\fP", quote_props, conf, state);
 	}
 	break;
     }
@@ -652,7 +684,7 @@ 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) {