From: chris mikkelson <chris@mikk.net>
Date: Thu, 11 Mar 2010 20:53:25 +0000 (-0600)
Subject: Merged text module initialization, updated regexps.
X-Git-Url: https://git.mikk.net/?a=commitdiff_plain;h=7a0e1227daf6d62954ad8fe17c0a3a843d87bb59;p=liburl

Merged text module initialization, updated regexps.
---

diff --git a/text.c b/text.c
index 19f39d6..8c77eb2 100644
--- a/text.c
+++ b/text.c
@@ -4,30 +4,31 @@
  */
 
 #include <pcre.h>
+#include <err.h>
 #include "re_stream.h"
 
 #include "msgproc.h"
 
-static const char *text_url_pattern = "https?://\\S+";
+static const char *text_url_pattern = "https?://[^\\s]+(?=\\s|$|\\))";
 static pcre *text_url_re;
-static const char *html_url_pattern = "<(a href|img src)=\"https?://[^\"]+";
+static const char *html_url_pattern = "(?<=<a href=\")https?://[^\"]+(?=\")"; 
 static pcre *html_url_re;
 
 static void
-text_plain_init(void) {
+text_init(void) {
 	const char *etxt; int epos;
+	static int initialized = 0;
+
+	if (initialized) return;
+	initialized = 1;
+
 	text_url_re = pcre_compile(text_url_pattern, 0, &etxt, &epos, 0);
 	if (!text_url_re) {
-		/* die */
+		errx(1, "text_url_pattern compile error\n");
 	}
-}
-
-static void
-text_html_init(void) {
-	const char *etxt; int epos;
 	html_url_re = pcre_compile(html_url_pattern, 0, &etxt, &epos, 0);
 	if (!html_url_re) {
-		/* die */
+		errx(1, "html_url_pattern compile error\n");
 	}
 }
 
@@ -80,7 +81,7 @@ text_finish(msgproc *m)
 
 msgproc_module msgproc_text = {
 	MSGPROC_TEXT,			/* type */
-	text_plain_init,		/* module init */
+	text_init,			/* module init */
 	NULL,				/* set module parameter */
 	text_plain_start,		/* start module instance */
 	NULL,				/* set module instance parameter */
@@ -91,7 +92,7 @@ msgproc_module msgproc_text = {
 
 msgproc_module msgproc_html = {
 	MSGPROC_HTML,			/* type */
-	text_html_init,			/* module init */
+	text_init,			/* module init */
 	NULL,				/* set module parameter */
 	text_html_start,		/* start module instance */
 	NULL,				/* set module instance parameter */