From 7284c9339449a5f022c4ba3adf40335447cb5007 Mon Sep 17 00:00:00 2001
From: Jaakko Keränen <jaakko.keranen@iki.fi>
Date: Wed, 24 Feb 2021 04:41:15 +0200
Subject: Bookmarks: Looking up site icons

Bookmark icons apply to site roots instead of domains. A root includes the user directory if one is found in the URL.
---
 src/bookmarks.c | 35 +++++++++++++++++------------------
 src/gmutil.c    | 26 +++++++++++++++++++++-----
 src/gmutil.h    |  1 +
 3 files changed, 39 insertions(+), 23 deletions(-)

diff --git a/src/bookmarks.c b/src/bookmarks.c
index 60048d36..91280f3c 100644
--- a/src/bookmarks.c
+++ b/src/bookmarks.c
@@ -254,27 +254,26 @@ iChar siteIcon_Bookmarks(const iBookmarks *d, const iString *url) {
     if (isEmpty_String(url)) {
         return 0;
     }
-    iChar        icon         = 0;
-    iRangecc     hostName     = urlHost_String(url);
-    const size_t hostSize     = size_Range(&hostName);
-    size_t       matchingSize = iInvalidSize; /* we'll pick the shortest matching */
+    static iRegExp *tagPattern_;
+    if (!tagPattern_) {
+        tagPattern_ = new_RegExp("\\busericon\\b", caseSensitive_RegExpOption);
+    }
+    const iRangecc urlRoot      = urlRoot_String(url);
+    size_t         matchingSize = iInvalidSize; /* we'll pick the shortest matching */
+    iChar          icon         = 0;
     lock_Mutex(d->mtx);
     iConstForEach(Hash, i, &d->bookmarks) {
         const iBookmark *bm = (const iBookmark *) i.value;
-        iUrl parts;
-        init_Url(&parts, &bm->url);
-        if (!hasTag_Bookmark(bm, "usericon")) { /* TODO: Inefficient! RegExp rebuilt every time. */
-            continue;
-        }
-        if (size_Range(&hostName) == size_Range(&parts.host) &&
-            iCmpStrNCase(hostName.start, parts.host.start, hostSize) == 0) {
-            const size_t n = size_String(&bm->url);
-            if (n > size_String(url)) {
-                continue; /* Bookmark must be higher up in the path tree. */
-            }
-            if (n < matchingSize && bm->icon) {
-                matchingSize = n;
-                icon = bm->icon;
+        iRegExpMatch m;
+        init_RegExpMatch(&m);
+        if (bm->icon && matchString_RegExp(tagPattern_, &bm->tags, &m)) {
+            const iRangecc bmRoot = urlRoot_String(&bm->url);
+            if (equalRangeCase_Rangecc(urlRoot, bmRoot)) {
+                const size_t n = size_String(&bm->url);
+                if (n < matchingSize) {
+                    matchingSize = n;
+                    icon = bm->icon;
+                }
             }
         }
     }
diff --git a/src/gmutil.c b/src/gmutil.c
index 1edc41ae..2b40367d 100644
--- a/src/gmutil.c
+++ b/src/gmutil.c
@@ -153,20 +153,36 @@ iRangecc urlHost_String(const iString *d) {
 }
 
 iRangecc urlUser_String(const iString *d) {
-    iRegExp *userPats[2] = { new_RegExp("~([^/?]+)", 0),
-                             new_RegExp("/users/([^/?]+)", caseInsensitive_RegExpOption) };
+    static iRegExp *userPats_[2];
+    if (!userPats_[0]) {
+        userPats_[0] = new_RegExp("~([^/?]+)", 0);
+        userPats_[1] = new_RegExp("/users/([^/?]+)", caseInsensitive_RegExpOption);
+    }
     iRegExpMatch m;
     init_RegExpMatch(&m);
     iRangecc found = iNullRange;
-    iForIndices(i, userPats) {
-        if (matchString_RegExp(userPats[i], d, &m)) {
+    iForIndices(i, userPats_) {
+        if (matchString_RegExp(userPats_[i], d, &m)) {
             found = capturedRange_RegExpMatch(&m, 1);
         }
-        iRelease(userPats[i]);
     }
     return found;
 }
 
+iRangecc urlRoot_String(const iString *d) {
+    const char *rootEnd;
+    const iRangecc user = urlUser_String(d);
+    if (!isEmpty_Range(&user)) {
+        rootEnd = user.end;
+    }
+    else {
+        iUrl parts;
+        init_Url(&parts, d);
+        rootEnd = parts.path.start;
+    }
+    return (iRangecc){ constBegin_String(d), rootEnd };
+}
+
 static iBool isAbsolutePath_(iRangecc path) {
     return isAbsolute_Path(collect_String(urlDecode_String(collect_String(newRange_String(path)))));
 }
diff --git a/src/gmutil.h b/src/gmutil.h
index de8c1e65..b2cee61a 100644
--- a/src/gmutil.h
+++ b/src/gmutil.h
@@ -104,6 +104,7 @@ void            init_Url                (iUrl *, const iString *text);
 iRangecc        urlScheme_String        (const iString *);
 iRangecc        urlHost_String          (const iString *);
 iRangecc        urlUser_String          (const iString *);
+iRangecc        urlRoot_String          (const iString *);
 const iString * absoluteUrl_String      (const iString *, const iString *urlMaybeRelative);
 iBool           isLikelyUrl_String      (const iString *);
 void            punyEncodeUrlHost_String(iString *);
-- 
cgit v1.2.3