E-book viewer: Highlights: Make URLs in the notes for highlights clickable

src/calibre/gui2/library/annotations.py

@@ -5,7 +5,8 @@
 import codecs
 import json
 import os
-from functools import partial
+import re
+from functools import lru_cache, partial
 from qt.core import (
     QAbstractItemView, QApplication, QCheckBox, QComboBox, QCursor, QDateTime,
     QDialog, QDialogButtonBox, QFont, QFormLayout, QFrame, QHBoxLayout, QIcon,
@@ -16,16 +17,17 @@ from qt.core import (
 from urllib.parse import quote
 
 from calibre import prepare_string_for_xml
-from calibre.ebooks.metadata import authors_to_string, fmt_sidx
 from calibre.db.backend import FTSQueryError
-from calibre.gui2 import Application, choose_save_file, config, error_dialog, gprefs
+from calibre.ebooks.metadata import authors_to_string, fmt_sidx
+from calibre.gui2 import (
+    Application, choose_save_file, config, error_dialog, gprefs, safe_open_url
+)
 from calibre.gui2.dialogs.confirm_delete import confirm
 from calibre.gui2.viewer.widgets import ResultsDelegate, SearchBox
 from calibre.gui2.widgets2 import Dialog
 
 
 # rendering {{{
-
 def render_highlight_as_text(hl, lines, as_markdown=False, link_prefix=None):
     lines.append(hl['highlighted_text'])
     date = QDateTime.fromString(hl['timestamp'], Qt.DateFormat.ISODate).toLocalTime().toString(Qt.DateFormat.SystemLocaleShortDate)
@@ -62,11 +64,57 @@ def render_bookmark_as_text(b, lines, as_markdown=False, link_prefix=None):
     lines.append('')
 
 
+url_prefixes = 'http', 'https'
+url_delimiters = (
+    '\x00-\x09\x0b-\x20\x7f-\xa0\xad\u0600-\u0605\u061c\u06dd\u070f\u08e2\u1680\u180e\u2000-\u200f\u2028-\u202f'
+    '\u205f-\u2064\u2066-\u206f\u3000\ud800-\uf8ff\ufeff\ufff9-\ufffb\U000110bd\U000110cd\U00013430-\U00013438'
+    '\U0001bca0-\U0001bca3\U0001d173-\U0001d17a\U000e0001\U000e0020-\U000e007f\U000f0000-\U000ffffd\U00100000-\U0010fffd'
+)
+url_pattern = r'\b(?:{})://[^{}]{{3,}}'.format('|'.join(url_prefixes), url_delimiters)
+
+
+@lru_cache(maxsize=2)
+def url_pat():
+    return re.compile(url_pattern, flags=re.I)
+
+
+closing_bracket_map = {'(': ')', '[': ']', '{': '}', '<': '>', '*': '*', '"': '"', "'": "'"}
+
+
+def url(text: str, s: int, e: int):
+    while text[e - 1] in '.,?!' and e > 1:  # remove trailing punctuation
+        e -= 1
+    # truncate url at closing bracket/quote
+    if s > 0 and e <= len(text) and text[s-1] in closing_bracket_map:
+        q = closing_bracket_map[text[s-1]]
+        idx = text.find(q, s)
+        if idx > s:
+            e = idx
+    return s, e
+
+
+def render_note_line(line):
+    urls = []
+    for m in url_pat().finditer(line):
+        s, e = url(line, m.start(), m.end())
+        urls.append((s, e))
+    if not urls:
+        yield prepare_string_for_xml(line)
+        return
+    pos = 0
+    for (s, e) in urls:
+        if s > pos:
+            yield prepare_string_for_xml(line[pos:s])
+        yield '<a href="{0}">{0}</a>'.format(prepare_string_for_xml(line[s:e], True))
+    if urls[-1][1] < len(line):
+        yield prepare_string_for_xml(line[urls[-1][1]:])
+
+
 def render_notes(notes, tag='p'):
     current_lines = []
     for line in notes.splitlines():
         if line:
-            current_lines.append(line)
+            current_lines.append(''.join(render_note_line(line)))
         else:
             if current_lines:
                 yield '<{0}>{1}</{0}>'.format(tag, '\n'.join(current_lines))
@@ -697,7 +745,10 @@ class DetailsPanel(QWidget):
         self.show_result(None)
 
     def link_clicked(self, qurl):
-        getattr(self, qurl.host())()
+        if qurl.scheme() == 'calibre':
+            getattr(self, qurl.host())()
+        else:
+            safe_open_url(qurl)
 
     def open_result(self):
         if self.current_result is not None:

src/calibre/gui2/viewer/highlights.py

@@ -18,7 +18,7 @@ from calibre.constants import (
     builtin_colors_dark, builtin_colors_light, builtin_decorations
 )
 from calibre.ebooks.epub.cfi.parse import cfi_sort_key
-from calibre.gui2 import error_dialog, is_dark_theme
+from calibre.gui2 import error_dialog, is_dark_theme, safe_open_url
 from calibre.gui2.dialogs.confirm_delete import confirm
 from calibre.gui2.library.annotations import (
     Details, Export as ExportBase, render_highlight_as_text, render_notes
@@ -397,7 +397,7 @@ class NotesDisplay(Details):
     def __init__(self, parent=None):
         Details.__init__(self, parent)
         self.setSizePolicy(QSizePolicy.Policy.Expanding, QSizePolicy.Policy.Maximum)
-        self.anchorClicked.connect(self.edit_notes)
+        self.anchorClicked.connect(self.anchor_clicked)
         self.current_notes = ''
 
     def show_notes(self, text=''):
@@ -405,10 +405,17 @@ class NotesDisplay(Details):
         self.setVisible(bool(text))
         self.current_notes = text
         html = '\n'.join(render_notes(text))
-        self.setHtml('<div><a href="edit://moo" style="text-decoration: none">{}</a></div>{}'.format(_('Edit notes'), html))
+        self.setHtml('<div><a href="edit://moo">{}</a></div>{}'.format(_('Edit notes'), html))
+        self.document().setDefaultStyleSheet('a[href] { text-decoration: none }')
         h = self.document().size().height() + 2
         self.setMaximumHeight(h)
 
+    def anchor_clicked(self, qurl):
+        if qurl.scheme() == 'edit':
+            self.edit_notes()
+        else:
+            safe_open_url(qurl)
+
     def edit_notes(self):
         current_text = self.current_notes
         d = NotesEditDialog(current_text, self)

src/pyj/read_book/highlights.pyj

@@ -14,7 +14,7 @@ from modals import (
     create_custom_dialog, error_dialog, get_text_dialog, question_dialog,
     warning_dialog
 )
-from read_book.globals import current_book, is_dark_theme, runtime
+from read_book.globals import current_book, is_dark_theme, runtime, ui_operations
 from widgets import create_button
 
 ICON_SIZE_VAL = 3
@@ -712,23 +712,60 @@ add_extra_css(def():
     return ans
 )
 
+url_pat = /\bhttps?:\/\/\S{3,}/ig
+closing_bracket_map = {'(': ')', '[': ']', '{': '}', '<': '>', '*': '*', '"': '"', "'": "'"}
+opening_brackets = Object.keys(closing_bracket_map).join('')
+
+
+def url(text: str, s: int, e: int):
+    while '.,?!'.indexOf(text[e-1]) > -1 and e > 1:  # remove trailing punctuation
+        e -= 1
+    # truncate url at closing bracket/quote
+    if s > 0 and e <= text.length and closing_bracket_map[text[s-1]]:
+        q = closing_bracket_map[text[s-1]]
+        idx = text.indexOf(q, s)
+        if idx > s:
+            e = idx
+    return s, e
+
+
+def render_notes(notes, container, make_urls_clickable):
+    current_para = E.p()
 
-def render_notes(notes, container):
-    current_block = ''
     def add_para():
-        nonlocal current_block
+        nonlocal current_para
-        container.appendChild(E.p(current_block))
+        container.appendChild(current_para)
         if container.childNodes.length > 1:
             container.lastChild.style.marginTop = '2ex'
-        current_block = ''
+        current_para = E.p()
+
+    def add_line(line):
+        url_pat.lastIndex = 0
+        urls = v'[]'
+
+        while make_urls_clickable:
+            m = url_pat.exec(line)
+            if not m:
+                break
+            urls.push(url(line, m.index, url_pat.lastIndex))
+        if not urls.length:
+            current_para.appendChild(document.createTextNode(line))
+            return
+        pos = 0
+        for (s, e) in urls:
+            if s > pos:
+                current_para.appendChild(document.createTextNode(line[pos:s]))
+            current_para.appendChild(E.a(line[s:e], href='javascript: void(0)', class_='blue-link', onclick=ui_operations.open_url.bind(None, line[s:e])))
+        if urls[-1][1] < line.length:
+            current_para.appendChild(document.createTextNode(line[urls[-1][1]:]))
 
     for line in notes.splitlines():
         if not line or not line.strip():
-            if current_block:
+            if current_para.childNodes.length:
                 add_para()
                 continue
-        current_block += line + '\n'
+        add_line(line)
-    if current_block:
+    if current_para.childNodes.length:
         add_para()
     return container
 
@@ -847,7 +884,7 @@ def highlight_entry(h, onclick, view, hide_panel):
         E.div(class_='notes')
     )
     if h.notes:
-        render_notes(h.notes, ans.querySelector('.notes'))
+        render_notes(h.notes, ans.querySelector('.notes'), True)
     ans.dataset.notes = h.notes or ''
     ans.dataset.title = h.highlighted_text or ''
     return ans

src/pyj/read_book/selection_bar.pyj

@@ -366,7 +366,7 @@ class SelectionBar:
             c.style.maxHeight = 'min(20ex, 40vh)'
         else:
             c.style.maxHeight = '20ex'
-        render_notes(notes, c)
+        render_notes(notes, c, True)
     # }}}
 
     # accessors {{{