More progress on all entites in MW data load

lib/definition.py

@@ -521,8 +521,10 @@ class Definition(QWidget):
                 self.pronounce.emit(url)
             elif url.scheme() == 'word':
                 self.newWord.emit(url.path())
+            elif url.scheme() == 'sense':
+                self.newWord.emit(url.path())
             else:
-                print(f"{clk['fmt'].anchorHref()}: {url.scheme()}")
+                print(f"{clk['fmt'].anchorHref()}")
                 self.alert.emit()
             self._downClickable = None
             return

lib/read.py

@@ -1,12 +1,9 @@
-import json
+from typing import Dict, List, Optional, cast
-from typing import Any, Dict, List, Optional, cast
 
-import requests
 from PyQt6.QtCore import QPoint, QResource, Qt, QTimer, pyqtSignal, pyqtSlot
 from PyQt6.QtGui import (
     QBrush,
     QColor,
-    QCursor,
     QKeyEvent,
     QPainter,
     QPainterPath,
@@ -15,7 +12,7 @@ from PyQt6.QtGui import (
     QTextCursor,
 )
 from PyQt6.QtSql import QSqlQuery
-from PyQt6.QtWidgets import QDialog, QTextEdit, QWidget
+from PyQt6.QtWidgets import QDialog, QWidget
 
 from lib import query_error
 from lib.preferences import Preferences
@@ -89,6 +86,7 @@ class ReadDialog(QDialog, Ui_ReadDialog):
         self.playSound.connect(self.sound.playSound)
         self.playAlert.connect(self.sound.alert)
         self.definition.pronounce.connect(self.sound.playSound)
+        self.definition.newWord.connect(self.newWord)
         return
 
     #
@@ -98,6 +96,15 @@ class ReadDialog(QDialog, Ui_ReadDialog):
     #
     # slots
     #
+    @pyqtSlot(str)
+    def newWord(self, word: str) -> None:
+        w = Word(word)
+        if not w.isValid():
+            self.playAlert.emit()
+            return
+        self.definition.setWord(w)
+        return
+    
     @pyqtSlot()
     def timerAction(self) -> None:
         if self.session.isActive():  # We are stopping
@@ -127,6 +134,9 @@ class ReadDialog(QDialog, Ui_ReadDialog):
             cursor.select(QTextCursor.SelectionType.WordUnderCursor)
             text = cursor.selectedText().strip()
         word = Word(text)
+        if not word.isValid():
+            self.playAlert.emit()
+            return
         word.playPRS()
         return
 
@@ -221,6 +231,9 @@ class ReadDialog(QDialog, Ui_ReadDialog):
             cursor.select(cursor.SelectionType.WordUnderCursor)
             text = cursor.selectedText().strip()
         word = Word(text)
+        if not word.isValid():
+            self.playAlert.emit()
+            return
         self.definition.setWord(word)
         self.showDefinition()
         return

lib/words.py

@@ -40,6 +40,7 @@ class Word:
     """All processing of a dictionary word."""
 
     _words: dict[str, WordType] = {}
+    _valid = False
 
     def __init__(self, word: str) -> None:
         #
@@ -62,14 +63,18 @@ class Word:
                 "definition": json.loads(query.value("definition")),
             }
             self.current = Word._words[word]
+            self._valid = True
             return
         #
         # The code should look at our settings to see if we have an API
         # key for MW to decide on the source to use.
         #
         source = "mw"
-
+        
         self._words[word] = discovered_plugins[source].fetch(word)
+        if self._words[word] is None:
+            self._valid = False
+            return
         self.current = Word._words[word]
         query.prepare(
             "INSERT INTO words "
@@ -81,8 +86,12 @@ class Word:
         query.bindValue(":definition", json.dumps(self.current["definition"]))
         if not query.exec():
             query_error(query)
+        self._valid = True
         return
 
+    def isValid(self) -> bool:
+        return self._valid
+    
     @pyqtSlot()
     def playSound(self) -> None:
         url = discovered_plugins[self.current["source"]].getFirstSound(

plugins/merriam-webster.py

@@ -4,7 +4,7 @@ from typing import Any, Literal, NotRequired, TypedDict, cast
 
 from PyQt6.QtCore import QEventLoop, QUrl
 from PyQt6.QtGui import QFont, QFontDatabase, QTextCharFormat, QTextLayout
-from PyQt6.QtNetwork import QNetworkRequest
+from PyQt6.QtNetwork import QNetworkReply, QNetworkRequest
 from trycast import trycast
 
 from lib.definition import Fragment, Line
@@ -91,7 +91,7 @@ Inflection = TypedDict(
 
 
 class CrossReferenceTarget(TypedDict):
-    cxl: str
+    cxl: NotRequired[str]
     cxr: NotRequired[str]
     cxt: str
     cxn: NotRequired[str]
@@ -277,7 +277,7 @@ class WordType(TypedDict):
     definition: Any
 
 
-def fetch(word: str) -> WordType:
+def fetch(word: str) -> WordType|None:
     request = QNetworkRequest()
     url = QUrl(API.format(word=word, key=key))
     request.setUrl(url)
@@ -287,6 +287,9 @@ def fetch(word: str) -> WordType:
     loop = QEventLoop()
     reply.finished.connect(loop.quit)
     loop.exec()
+    if reply.error() != QNetworkReply.NetworkError.NoError:
+        print(f"Error fetching {word}: {reply.errorString()}")
+        return None
     content = reply.readAll()
     data = json.loads(content.data().decode("utf-8"))
     return {
@@ -295,7 +298,6 @@ def fetch(word: str) -> WordType:
         "definition": data,
     }
 
-
 def soundUrl(sound: Sound, fmt="ogg") -> QUrl:
     """Create a URL from a PRS structure."""
     base = f"audio://media.merriam-webster.com/audio/prons/en/us/{fmt}"
@@ -343,16 +345,12 @@ def do_prs(frag: Fragment, prs: list[Pronunciation] | None) -> None:
             fmt.setAnchorHref(soundUrl(pr["sound"]).toString())
             fmt.setForeground(r.linkColor)
             #text = pr["mw"] +' \N{SPEAKER} '
-            text = pr["mw"] +' '
+            text = ' '+pr["mw"] +' '
         else:
             text = pr['mw'] + ' '
-        print(f"text: {text}, length: {len(text)}")
         frag.addText(text, fmt)
         if "l2" in pr:
             frag.addText(pun + pr["l2"], r.subduedLabelFormat)
-    text = frag.layout().text()
-    for fmt in frag.layout().formats():
-        print(f"start: {fmt.start}, length: {fmt.length}, text: \"{text[fmt.start:fmt.start+fmt.length]}\"")
     return
 
 
@@ -741,6 +739,42 @@ def do_uros(uros: list[UndefinedRunOn]|None) -> list[Line]:
                         lines.append(line)
                         lines += newLines
     return lines
+
+def do_cxs(cxs: list[CognateCrossRef]|None) -> list[Line]:
+    assert cxs is not None
+    r = Resources()
+    lines: list[Line] = []
+    for cx in cxs:
+        frag = Fragment()
+        frag.addText(cx['cxl']+' ', r.italicFormat)
+        for cxt in cx['cxtis']:
+            if 'cxl' in cxt:
+                frag.addText(cxt['cxl'], r.italicFormat)
+            text = cxt['cxt']
+            anchor = text
+            if 'cxr' in cxt:
+                anchor = cxt['cxr']
+            if 'cxn' in cxt:
+                anchor += f"/{cxt['cxn']}"
+
+            fmt = QTextCharFormat(r.smallCapsFormat)
+            fmt.setAnchor(True)
+            fmt.setForeground(r.linkColor)
+            fmt.setFontUnderline(True)
+            fmt.setUnderlineColor(r.linkColor)
+            fmt.setFontUnderline(True)
+            fmt.setAnchorHref('sense:///'+anchor)
+            #
+            # XXX - Capitalization does not work
+            #
+            text = text.upper()
+            fmt.setFontPointSize(fmt.fontPointSize() * 0.90)
+            frag.addText(text, fmt)
+        line = Line()
+        line.addFragment(frag)
+        lines.append(line)
+    return lines
+
 def getDef(defines: Any) -> list[Line]:
     Line.setParseText(parseText)
     workList = restructure(defines)
@@ -831,15 +865,17 @@ def getDef(defines: Any) -> list[Line]:
             lines.append(line)
             line = Line()
             frag = Fragment()
-        defines = trycast(list[DefinitionSection], work["def"])
+        if 'def' in work:
-        assert defines is not None
+            defines = trycast(list[DefinitionSection], work["def"])
-        for define in defines:
+            assert defines is not None
-            try:
+            for define in defines:
-                lines += do_def(define)
+                try:
-            except NotImplementedError:
+                    lines += do_def(define)
-                raise
+                except NotImplementedError:
+                    pass
+        if 'cxs' in work:
+            lines += do_cxs(trycast(list[CognateCrossRef], work['cxs']))
         if "uros" in work:
-            print(json.dumps(work['uros'],indent=2))
             uros = trycast(list[UndefinedRunOn], work['uros'])
             lines += do_uros(uros)
         if "dros" in work:
@@ -852,10 +888,6 @@ def getDef(defines: Any) -> list[Line]:
                 phrases.append(line)
             phrases += do_dros(dros)
         if "et" in work:
-            line = Line()
-            frag = Fragment('', r.textFont)
-            frag.addText(f"{work['fl']} ({used[work['fl']]})",r.labelFormat)
-            line.addFragment(frag)
             ets += do_ets(trycast(list[list[Pair]], work["et"]))
         for k in work.keys():
             if k not in [
@@ -872,8 +904,9 @@ def getDef(defines: Any) -> list[Line]:
                     "vrs",
                     "dros",
                     'uros',
+                    'cxs',
             ]:
-                raise NotImplementedError(f"Unknown key {k} in work")
+                print( NotImplementedError(f"Unknown key {k} in work"))
     if len(phrases) > 0:
         lines += phrases
     if len(ets) > 0:
@@ -914,6 +947,11 @@ def replaceCode(code:str) -> tuple[str, QTextCharFormat]:
             fmt.setFontItalic(True)
         elif token == 'sx':
             fmt.setFontCapitalization(QFont.Capitalization.SmallCaps)
+            #
+            # XXX - Capitalization does not work
+            #
+            text = text.upper()
+            fmt.setFontPointSize(fmt.fontPointSize() * 0.90)
         elif token == 'dxt':
             if fields[3] == 'illustration':
                 fmt.setAnchorHref('article:///'+fields[2])
@@ -928,10 +966,14 @@ def replaceCode(code:str) -> tuple[str, QTextCharFormat]:
                 fmt.setAnchorHref('etymology:///'+fields[2])
             else:
                 fmt.setAnchorHref('etymology:///' + fields[1])
+        elif token == 'd_link':
+            if fields[2] != '':
+                fmt.setAnchorHref('direct:///' + fields[2])
+            else:
+                fmt.setAnchorHref('direct:///' + fields[1])
         else:
             raise NotImplementedError(f"Token {code} not implimented")
         fmt.setForeground(r.linkColor)
-        print(f"Format.capitalization(): {fmt.fontCapitalization()}")
     return (text,fmt)
 
 def markup(offset: int, text:str) -> tuple[str, list[QTextLayout.FormatRange]]: