[MOD/FIX] Licensewerk, start counting by codepoint

XEP-0426 came out of the blue, and it *hit* me!

LICENSE

@@ -1,4 +1,4 @@
-For the files src/Parsee/HMAC.c, src/XML/*, tools/*, src/include/XML.h, etc/*, and Makefile, 
+For the files src/include/Unistring.h, src/Unistr.h rc/Parsee/HMAC.c, src/XML/*, tools/*, src/include/XML.h, etc/*, and Makefile, 
 see COPYING.CC0.
 For any other file in src/, see COPYING.AGPL as the primary license.
 

src/Parsee/Utils/String.c

@@ -3,6 +3,8 @@
 #include <Cytoplasm/Memory.h>
 #include <Cytoplasm/Str.h>
 
+#include <Unistring.h>
+
 #include <stdbool.h>
 #include <string.h>
     
@@ -37,6 +39,22 @@ ParseeFindDatastart(char *data)
 
     return (int) (startline - data);
 }
+int 
+ParseeFindDatastartU(char *data)
+{
+    Unistr *str;
+    size_t ret;
+    if (!data)
+    {
+        return 0;
+    }
+
+    str = UnistrCreate(data);
+    ret = UnistrGetOffset(str, (uint32_t) '>');
+    UnistrFree(str);
+
+    return (int) ret;
+}
 
 char *
 ParseeStringifyDate(uint64_t millis)

src/StanzaBuilder.c

@@ -185,7 +185,7 @@ ExportStanza(StanzaBuilder *builder)
         builder->replying_to_sender && 
         builder->body)
     {
-        int off = ParseeFindDatastart(builder->body);
+        int off = ParseeFindDatastartU(builder->body);
         char *ostr = StrInt(off);
         XMLElement *reply = XMLCreateTag("reply");
         XMLElement *fallback = XMLCreateTag("fallback");

src/Unistr.c

@@ -3,8 +3,10 @@
 #include <Cytoplasm/Memory.h>
 #include <Cytoplasm/Str.h>
 #include <Cytoplasm/Log.h>
+#include <Cytoplasm/Log.h>
 
 #include <string.h>
+#include <stdarg.h>
 
 struct Unistr {
     size_t length;
@@ -222,3 +224,46 @@ UnistrFilter(Unistr *str, UnistrFilterFunc filter)
 
     return unistr;
 }
+
+Unistr *
+UnistrConcat(size_t n, ...)
+{
+    va_list list;
+    size_t i;
+    Unistr *ret = UnistrCreate("");
+
+    va_start(list, n);
+    for (i = 0; i < n; i++)
+    {
+        Unistr *to_concat = va_arg(list, Unistr *);
+        size_t j;
+        for (j = 0; j < UnistrSize(to_concat); j++)
+        {
+            UnistrAddch(ret, UnistrGetch(to_concat, j));
+        }
+    }
+
+    va_end(list);
+    return ret;
+}
+size_t 
+UnistrGetOffset(Unistr *str, uint32_t sep)
+{
+    size_t i;
+    uint32_t prev = 0x0A;
+    if (!str || !sep)
+    {
+        return 0;
+    }
+
+    for (i = 0; i < str->length; i++)
+    {
+        uint32_t curr = str->codepoints[i];
+        if (prev == 0x0A && curr != sep)
+        {
+            return i;
+        }
+        prev = curr;
+    }
+    return 0;
+}

src/XMPPThread/Stanzas/Message.c

@@ -9,6 +9,20 @@
 
 #include <string.h>
 
+static void
+LazyRegister(ParseeData *data, char *mxid, char *name)
+{
+    if (!DbExists(data->db, 2, "users", mxid))
+    {
+        ASRegisterUser(data->config, mxid);
+        DbUnlock(data->db, DbCreate(data->db, 2, "users", mxid));
+    }
+    if (name)
+    {
+        ASSetName(data->config, mxid, name);
+    }
+}
+
 static void
 ProcessChatstates(ParseeData *args, XMLElement *stanza)
 {
@@ -211,7 +225,7 @@ end_error:
         HashMap *room_json;
         char *from = HashMapGet(stanza->attrs, "from");
 
-        ASRegisterUser(args->config, from_matrix);
+        LazyRegister(args, from_matrix, NULL);
         room = ASCreateDM(args->config, from_matrix, to);
         mroom_id = StrDuplicate(room);
         Log(LOG_INFO, "Creating a DM to '%s'(%s)...", to, mroom_id);
@@ -273,11 +287,10 @@ end_error:
             XMLElement *oob, *oob_data;
 
             pthread_mutex_unlock(&thr->info->chk_lock);
-            ASRegisterUser(args->config, encoded);
-            if (!chat)
-            {
-                ASSetName(args->config, encoded, res);
-            }
+
+            LazyRegister(args, encoded, !chat ? res : NULL);
+            /* TODO: I don't think we can quite remove that. Maybe 
+             * the user was kicked and the bridge is unaware? */
             ASInvite(args->config, mroom_id, encoded);
             Free(ASJoin(args->config, mroom_id, encoded));
 

src/include/Parsee.h

@@ -286,6 +286,10 @@ extern void ParseeCleanup(void *data);
 /* Finds the offset of the first line without a '>' at its start. */
 extern int ParseeFindDatastart(char *data);
 
+/* Finds the offset of the first line without a '>' at its start(as 
+ * Unicode codepoints). */
+extern int ParseeFindDatastartU(char *data);
+
 
 /* XMPP-ifies a message event to XEP-0393 if possible. */
 extern char * ParseeXMPPify(HashMap *event);

src/include/Unistring.h

@@ -39,6 +39,13 @@ extern uint32_t UnistrGetch(Unistr *unistr, size_t i);
  * Modifies: unistr */
 extern void UnistrAddch(Unistr *unistr, uint32_t u);
 
+/** Concats N unistrings into a new, separate unistring.
+ * ---------------------
+ * Returns: a new unistring[HEAP]
+ * Modifies: NOTHING
+ * Thrasher: UnistrFree */
+extern Unistr * UnistrConcat(size_t n, ...);
+
 /** Encodes a unistring into a C UTF-8 string
  * --------------
  * Returns: a valid NULL-terminated string[HEAP] | NULL
@@ -64,4 +71,9 @@ typedef bool (*UnistrFilterFunc)(uint32_t u);
  * Returns: a new unistring with filtered characters removed */
 extern Unistr * UnistrFilter(Unistr *str, UnistrFilterFunc filter);
 
+/** Finds the offset of the first line not starting with a specific 
+ * characters in terms of Unicode codepoints.
+ * --------
+ * Returns: an offset of the first line to not start by {c} | 0 */
+extern size_t UnistrGetOffset(Unistr *str, uint32_t sep);
 #endif