Merge branch 'release-4.16.1'

[i3/i3] / src / regex.c

diff --git a/src/regex.c b/src/regex.c
index df01dae4ed8f72a639cbf821122b6d867df25f81..8f039157f6f7547d57e66733aa4bffdb23f72345 100644 (file)
--- a/src/regex.c
+++ b/src/regex.c
@@ -2,11 +2,11 @@
  * vim:ts=4:sw=4:expandtab
  *
  * i3 - an improved dynamic tiling window manager
- * © 2009-2011 Michael Stapelberg and contributors (see also: LICENSE)
+ * © 2009 Michael Stapelberg and contributors (see also: LICENSE)
  *
+ * regex.c: Interface to libPCRE (perl compatible regular expressions).
  *
  */
-
 #include "all.h"
 
 /*
@@ -21,30 +21,60 @@
  */
 struct regex *regex_new(const char *pattern) {
     const char *error;
-    int offset;
+    int errorcode, offset;
 
-    struct regex *re = scalloc(sizeof(struct regex));
+    struct regex *re = scalloc(1, sizeof(struct regex));
     re->pattern = sstrdup(pattern);
-    if (!(re->regex = pcre_compile(pattern, 0, &error, &offset, NULL))) {
-        ELOG("PCRE regular expression compilation failed at %d: %s",
+    int options = PCRE_UTF8;
+    /* We use PCRE_UCP so that \B, \b, \D, \d, \S, \s, \W, \w and some POSIX
+     * character classes play nicely with Unicode */
+    options |= PCRE_UCP;
+    while (!(re->regex = pcre_compile2(pattern, options, &errorcode, &error, &offset, NULL))) {
+        /* If the error is that PCRE was not compiled with UTF-8 support we
+         * disable it and try again */
+        if (errorcode == 32) {
+            options &= ~PCRE_UTF8;
+            continue;
+        }
+        ELOG("PCRE regular expression compilation failed at %d: %s\n",
              offset, error);
+        regex_free(re);
         return NULL;
     }
     re->extra = pcre_study(re->regex, 0, &error);
+    /* If an error happened, we print the error message, but continue.
+     * Studying the regular expression leads to faster matching, but it’s not
+     * absolutely necessary. */
+    if (error) {
+        ELOG("PCRE regular expression studying failed: %s\n", error);
+    }
     return re;
 }
 
+/*
+ * Frees the given regular expression. It must not be used afterwards!
+ *
+ */
+void regex_free(struct regex *regex) {
+    if (!regex)
+        return;
+    FREE(regex->pattern);
+    FREE(regex->regex);
+    FREE(regex->extra);
+    FREE(regex);
+}
+
 /*
  * Checks if the given regular expression matches the given input and returns
  * true if it does. In either case, it logs the outcome using LOG(), so it will
- * be visible without any debug loglevel.
+ * be visible without debug logging.
  *
  */
 bool regex_matches(struct regex *regex, const char *input) {
     int rc;
 
-    /* TODO: is strlen(input) correct for UTF-8 matching? */
-    /* TODO: enable UTF-8 */
+    /* We use strlen() because pcre_exec() expects the length of the input
+     * string in bytes */
     if ((rc = pcre_exec(regex->regex, regex->extra, input, strlen(input), 0, 0, NULL, 0)) == 0) {
         LOG("Regular expression \"%s\" matches \"%s\"\n",
             regex->pattern, input);
@@ -57,7 +87,7 @@ bool regex_matches(struct regex *regex, const char *input) {
         return false;
     }
 
-    /* TODO: handle the other error codes */
-    LOG("PCRE error\n");
+    ELOG("PCRE error %d while trying to use regular expression \"%s\" on input \"%s\", see pcreapi(3)\n",
+         rc, regex->pattern, input);
     return false;
 }