X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/2e78832ed88bc518b46e0b248d22da061bea46ed..0ec6a4292ed4ccc4c3c3c24fe9402621066b8c10:/bk_man.c

diff --git a/bk_man.c b/bk_man.c
index 1146150..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)) {
 		/*
@@ -584,6 +610,10 @@ static int man_rdaddwc(rdstringc *rs, word *text, word *end,
 	    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;
@@ -654,7 +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) {