Supporting unicode <-> slackcode emoji conversions on message text

.gitignore

@@ -3,3 +3,6 @@
 *.so
 /Makefile.dep
 .vscode
+*.gresource
+*~
+test-slack-emoji

Makefile

@@ -12,12 +12,25 @@ C_SRCS = slack.c \
 	 slack-api.c \
 	 slack-object.c \
 	 slack-json.c \
+	 slack-emoji.c \
 	 purple-websocket.c \
 	 json.c
 
+RESOURCES_XMLS = slack-emoji.gresource.xml
+RESOURCES_C_SRCS = $(RESOURCES_XMLS:.gresource.xml=-resource.c)
+RESOURCES_C_HDRS = $(RESOURCES_XMLS:.gresource.xml=-resource.h)
+
+C_SRCS += $(RESOURCES_C_SRCS)
+
 # Object file names using 'Substitution Reference'
 C_OBJS = $(C_SRCS:.c=.o)
 
+%-resource.c: %.gresource.xml %-resource.h
+	glib-compile-resources --internal --generate $< --target $@
+
+%-resource.h: %.gresource.xml
+	glib-compile-resources --internal --generate $< --target $@
+
 PURPLE_MOD=purple
 
 ifeq ($(OS),Windows_NT)
@@ -53,7 +66,7 @@ LIBNAME=libslack.so
 
 PLUGIN_DIR_PURPLE:=$(DESTDIR)$(shell pkg-config --variable=plugindir $(PURPLE_MOD))
 DATA_ROOT_DIR_PURPLE:=$(DESTDIR)$(shell pkg-config --variable=datarootdir $(PURPLE_MOD))
-PKGS=$(PURPLE_MOD) glib-2.0 gobject-2.0
+PKGS=$(PURPLE_MOD) glib-2.0 gobject-2.0 gio-2.0
 
 CFLAGS = \
     -g \
@@ -119,4 +132,14 @@ modversion:
 Makefile.dep: $(C_SRCS)
 	$(CC) -MM $(CFLAGS) $^ > Makefile.dep
 
+
+test-slack-emoji: test-slack-emoji.c slack-emoji.o json.o slack-json.o $(RESOURCES_C_SRCS)
+	$(CC) -o $@ \
+	    $(shell pkg-config --cflags --libs $(PKGS)) \
+	    -lm $^
+
+.PHONY: test
+test: test-slack-emoji
+	for i in $< ; do ./$$i ; done
+
 include Makefile.dep

slack-emoji.c

@@ -0,0 +1,213 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <gio/gio.h>
+#include <glib.h>
+#include <stdio.h>
+
+#include "slack-emoji.h"
+#include "slack-json.h"
+
+static void slack_emoji_load_json(SlackEmojiContext *ctx, json_value *emoji, gboolean skins, GString *unicode_re);
+
+static void slack_emoji_load_json_one(SlackEmojiContext *ctx, json_value *emoji, gboolean skins, GString *unicode_re)
+{
+	gchar *slackcode;
+	gchar *unicode;
+
+	if (emoji->type != json_object) {
+		return;
+	}
+
+	slackcode = json_get_prop_strptr(emoji, "name");
+	unicode = json_get_prop_strptr(emoji, "unicode");
+
+	g_hash_table_insert(ctx->slackcode_to_unicode,
+			    g_strdup(slackcode),
+			    g_strdup(unicode));
+	g_hash_table_insert(ctx->unicode_to_slackcode,
+			    g_strdup(unicode),
+			    g_strdup(slackcode));
+
+	if (unicode_re->len != 1) {
+		g_string_append(unicode_re, "|");
+	}
+	while (*unicode) {
+		gunichar c = g_utf8_get_char(unicode);
+		g_string_append_printf(unicode_re, "\\N{U+%04x}", c);
+		unicode = g_utf8_next_char(unicode);
+	}
+
+	if (!skins) {
+		json_value *skins = json_get_prop(emoji, "skinVariations");
+		if (skins) {
+			slack_emoji_load_json(ctx, skins, TRUE, unicode_re);
+		}
+	}
+}
+
+static void slack_emoji_load_json(SlackEmojiContext *ctx, json_value *emoji, gboolean skins, GString *unicode_re)
+{
+	gsize i;
+	if (emoji->type != json_object) {
+		return;
+	}
+
+	for (i = 0; i < emoji->u.object.length; i++) {
+		slack_emoji_load_json_one(ctx,
+					  emoji->u.object.values[i].value,
+					  skins,
+					  unicode_re);
+	}
+}
+
+SlackEmojiContext *slack_emoji_new(void)
+{
+	GError *err = NULL;
+	GBytes *emojidata;
+	gsize emojilen;
+	gconstpointer emojibytes;
+	json_value *emoji;
+	SlackEmojiContext *ctx;
+	GString *unicode_re_match = NULL;
+
+	ctx = g_new0(SlackEmojiContext, 1);
+	ctx->slackcode_to_unicode = g_hash_table_new_full(g_str_hash,
+							  g_str_equal,
+							  g_free,
+							  g_free);
+	ctx->unicode_to_slackcode = g_hash_table_new_full(g_str_hash,
+							  g_str_equal,
+							  g_free,
+							  g_free);
+
+	emojidata = g_resources_lookup_data("/im/pidgin/plugins/slack-libpurple/weemoji.json",
+					    G_RESOURCE_LOOKUP_FLAGS_NONE,
+					    &err);
+	if (err) {
+		g_warning("Unable to load weemoji.json resource: %s", err->message);
+		g_error_free(err);
+		goto error;
+	}
+
+	emojibytes = g_bytes_get_data(emojidata, &emojilen);
+	emoji = json_parse(emojibytes, emojilen);
+	g_bytes_unref(emojidata);
+
+
+	unicode_re_match = g_string_new("(");
+	slack_emoji_load_json(ctx, emoji, FALSE, unicode_re_match);
+	g_string_append(unicode_re_match, ")");
+
+
+	ctx->slackcode_re = g_regex_new(":([a-z0-9_+-]+):",
+					G_REGEX_DEFAULT,
+					G_REGEX_MATCH_DEFAULT,
+					&err);
+	if (err) {
+		g_warning("Unable to compile emoji unicode regex: %s", err->message);
+		g_error_free(err);
+		goto error;
+	}
+
+	ctx->unicode_re = g_regex_new(unicode_re_match->str,
+				      G_REGEX_DEFAULT,
+				      G_REGEX_MATCH_DEFAULT,
+				      &err);
+	if (err) {
+		fprintf(stderr, "Unable to compile emoji unicode regex: %s", err->message);
+		g_error_free(err);
+		goto error;
+	}
+
+	json_value_free(emoji);
+	g_string_free(unicode_re_match, TRUE);
+
+	return ctx;
+
+ error:
+	g_string_free(unicode_re_match, TRUE);
+	slack_emoji_free(ctx);
+	return NULL;
+}
+
+void slack_emoji_free(SlackEmojiContext *ctx)
+{
+	if (!ctx)
+		return;
+	if (ctx->slackcode_to_unicode)
+		g_hash_table_destroy(ctx->slackcode_to_unicode);
+	if (ctx->unicode_to_slackcode)
+		g_hash_table_destroy(ctx->unicode_to_slackcode);
+	if (ctx->slackcode_re)
+		g_regex_unref(ctx->slackcode_re);
+	if (ctx->unicode_re)
+		g_regex_unref(ctx->unicode_re);
+	g_free(ctx);
+}
+
+static gboolean slack_emoji_replace(const GMatchInfo *info,
+				    GString *res,
+				    GHashTable *mapping,
+				    gboolean slackcode)
+{
+	gchar *match;
+	gchar *r;
+
+	match = g_match_info_fetch(info, 1);
+	r = g_hash_table_lookup(mapping, match);
+
+	if (r != NULL) {
+		if (!slackcode)
+			g_string_append(res, ":");
+		g_string_append(res, r);
+		if (!slackcode)
+			g_string_append(res, ":");
+	} else {
+		if (slackcode)
+			g_string_append(res, ":");
+		g_string_append(res, match);
+		if (slackcode)
+			g_string_append(res, ":");
+	}
+	g_free(match);
+
+	return FALSE;
+}
+
+static gboolean slack_emoji_replace_slackcode(const GMatchInfo *info,
+					      GString *res,
+					      gpointer data)
+{
+	return slack_emoji_replace(info, res, data, TRUE);
+}
+
+static gboolean slack_emoji_replace_unicode(const GMatchInfo *info,
+					    GString *res,
+					    gpointer data)
+{
+	return slack_emoji_replace(info, res, data, FALSE);
+}
+
+char *slack_emoji_slackcode_to_unicode(SlackEmojiContext *ctx, const char *msg)
+{
+	return g_regex_replace_eval(ctx->slackcode_re,
+				    msg,
+				    -1,
+				    0,
+				    0,
+				    slack_emoji_replace_slackcode,
+				    ctx->slackcode_to_unicode,
+				    NULL);
+}
+
+char *slack_emoji_unicode_to_slackcode(SlackEmojiContext *ctx, const char *msg)
+{
+	return g_regex_replace_eval(ctx->unicode_re,
+				    msg,
+				    -1,
+				    0,
+				    0,
+				    slack_emoji_replace_unicode,
+				    ctx->unicode_to_slackcode,
+				    NULL);
+}

slack-emoji.gresource.xml

@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<gresources>
+  <gresource prefix="/im/pidgin/plugins/slack-libpurple">
+    <file>weemoji.json</file>
+  </gresource>
+</gresources>

slack-emoji.h

@@ -0,0 +1,18 @@
+#ifndef _PURPLE_SLACK_EMOJI_H
+#define _PURPLE_SLACK_EMOJI_H
+
+#include <glib.h>
+
+typedef struct SlackEmojiContext {
+	GHashTable *slackcode_to_unicode; /* char *slackcode -> char *unicode */
+	GHashTable *unicode_to_slackcode; /* char *unicode -> char *slackcode */
+	GRegex *slackcode_re;
+	GRegex *unicode_re;
+} SlackEmojiContext;
+
+SlackEmojiContext *slack_emoji_new();
+void slack_emoji_free(SlackEmojiContext *ctx);
+char *slack_emoji_slackcode_to_unicode(SlackEmojiContext *ctx, const char *msg);
+char *slack_emoji_unicode_to_slackcode(SlackEmojiContext *ctx, const char *msg);
+
+#endif

slack-message.c

@@ -10,6 +10,7 @@
 #include "slack-conversation.h"
 #include "slack-message.h"
 #include "slack-thread.h"
+#include "slack-emoji.h"
 
 gchar *slack_html_to_message(SlackAccount *sa, const char *s, PurpleMessageFlags flags) {
 
@@ -71,7 +72,9 @@ gchar *slack_html_to_message(SlackAccount *sa, const char *s, PurpleMessageFlags
 		g_string_append_c(msg, *s++);
 	}
 
-	return g_string_free(msg, FALSE);
+	gchar *converted = slack_emoji_unicode_to_slackcode(sa->emoji_ctx, msg->str);
+	g_string_free(msg, TRUE);
+	return converted;
 }
 
 void slack_message_to_html(GString *html, SlackAccount *sa, gchar *s, PurpleMessageFlags *flags, gchar *prepend_newline_str) {
@@ -81,6 +84,8 @@ void slack_message_to_html(GString *html, SlackAccount *sa, gchar *s, PurpleMess
 	if (flags)
 		*flags |= PURPLE_MESSAGE_NO_LINKIFY;
 
+	gchar *converted = slack_emoji_slackcode_to_unicode(sa->emoji_ctx, s);
+	s = converted;
 	size_t l = strlen(s);
 	char *end = &s[l];
 
@@ -176,6 +181,8 @@ void slack_message_to_html(GString *html, SlackAccount *sa, gchar *s, PurpleMess
 		g_free(replace);
 		g_free(newHtml);
 	}
+
+	g_free(converted);
 }
 
 /*

slack.c

@@ -274,6 +274,8 @@ static void slack_login(PurpleAccount *account) {
 
 	sa->buddies = g_hash_table_new_full(/* slack_object_id_hash, slack_object_id_equal, */ g_str_hash, g_str_equal, NULL, NULL);
 
+	sa->emoji_ctx = slack_emoji_new();
+
 	sa->mark_list = MARK_LIST_END;
 
 	purple_connection_set_display_name(gc, account->alias ?: account->username);
@@ -397,6 +399,8 @@ static void slack_close(PurpleConnection *gc) {
 	g_hash_table_destroy(sa->user_names);
 	g_hash_table_destroy(sa->users);
 
+	slack_emoji_free(sa->emoji_ctx);
+
 #if GLIB_CHECK_VERSION(2,60,0)
 	g_queue_clear_full(&sa->avatar_queue, g_object_unref);
 #else

slack.h

@@ -8,6 +8,7 @@
 #include "glibcompat.h"
 #include "purple-websocket.h"
 #include "slack-object.h"
+#include "slack-emoji.h"
 
 #define SLACK_PLUGIN_ID "prpl-slack"
 
@@ -48,6 +49,8 @@ typedef struct _SlackAccount {
 	GHashTable *buddies; /* char *slack_id -> PurpleBListNode */
 	gboolean roomlist_stop;
 
+	SlackEmojiContext *emoji_ctx;
+
 	guint mark_timer;
 	SlackObject *mark_list;