add option to strip html in export

anki/exporting.py

@@ -6,11 +6,13 @@ import re, os, zipfile, shutil, unicodedata
 import json
 
 from anki.lang import _
-from anki.utils import ids2str, splitFields, namedtmp
+from anki.utils import ids2str, splitFields, namedtmp, stripHTML
 from anki.hooks import runHook
 from anki import Collection
 
 class Exporter:
+    includeHTML = None
+
     def __init__(self, col, did=None):
         self.col = col
         self.did = did
@@ -24,6 +26,14 @@ class Exporter:
         self.doExport(file)
         file.close()
 
+    def processText(self, text):
+        if self.includeHTML is False:
+            text = self.stripHTML(text)
+
+        text = self.escapeText(text)
+
+        return text
+
     def escapeText(self, text):
         "Escape newlines, tabs, CSS and quotechar."
         # fixme: we should probably quote fields with newlines
@@ -36,6 +46,16 @@ class Exporter:
             text = "\"" + text.replace("\"", "\"\"") + "\""
         return text
 
+    def stripHTML(self, text):
+        # very basic conversion to text
+        s = text
+        s = re.sub(r"(?i)<(br ?/?|div|p)>", " ", s)
+        s = re.sub(r"\[sound:[^]]+\]", "", s)
+        s = stripHTML(s)
+        s = re.sub(r"[ \n\t]+", " ", s)
+        s = s.strip()
+        return s
+
     def cardIds(self):
         if not self.did:
             cids = self.col.db.list("select id from cards")
@@ -51,6 +71,7 @@ class TextCardExporter(Exporter):
 
     key = _("Cards in Plain Text")
     ext = ".txt"
+    includeHTML = True
 
     def __init__(self, col):
         Exporter.__init__(self, col)
@@ -61,7 +82,7 @@ class TextCardExporter(Exporter):
         def esc(s):
             # strip off the repeated question in answer if exists
             s = re.sub("(?si)^.*<hr id=answer>\n*", "", s)
-            return self.escapeText(s)
+            return self.processText(s)
         out = ""
         for cid in ids:
             c = self.col.getCard(cid)
@@ -77,6 +98,7 @@ class TextNoteExporter(Exporter):
     key = _("Notes in Plain Text")
     ext = ".txt"
     includeTags = True
+    includeHTML = True
 
     def __init__(self, col):
         Exporter.__init__(self, col)
@@ -95,7 +117,7 @@ where cards.id in %s)""" % ids2str(cardIds)):
             if self.includeID:
                 row.append(str(id))
             # fields
-            row.extend([self.escapeText(f) for f in splitFields(flds)])
+            row.extend([self.processText(f) for f in splitFields(flds)])
             # tags
             if self.includeTags:
                 row.append(tags.strip())

aqt/exporting.py

@@ -61,6 +61,12 @@ class ExportDialog(QDialog):
             getattr(self.exporter, "includeMedia", None) is not None)
         self.frm.includeTags.setVisible(
             getattr(self.exporter, "includeTags", None) is not None)
+        html = getattr(self.exporter, "includeHTML", None)
+        if html is not None:
+            self.frm.includeHTML.setVisible(True)
+            self.frm.includeHTML.setChecked(html)
+        else:
+            self.frm.includeHTML.setVisible(False)
         # show deck list?
         self.frm.deck.setVisible(not self.isVerbatim)
 
@@ -71,6 +77,8 @@ class ExportDialog(QDialog):
             self.frm.includeMedia.isChecked())
         self.exporter.includeTags = (
             self.frm.includeTags.isChecked())
+        self.exporter.includeHTML = (
+            self.frm.includeHTML.isChecked())
         if not self.frm.deck.currentIndex():
             self.exporter.did = None
         else:

designer/exporting.ui

@@ -7,7 +7,7 @@
     <x>0</x>
     <y>0</y>
     <width>295</width>
-    <height>202</height>
+    <height>223</height>
    </rect>
   </property>
   <property name="windowTitle">
@@ -76,6 +76,13 @@
        </property>
       </widget>
      </item>
+     <item>
+      <widget class="QCheckBox" name="includeHTML">
+       <property name="text">
+        <string>Include HTML and media references</string>
+       </property>
+      </widget>
+     </item>
     </layout>
    </item>
    <item>

tests/test_exporting.py

@@ -14,7 +14,7 @@ def setup1():
     global deck
     deck = getEmptyCol()
     f = deck.newNote()
-    f['Front'] = "foo"; f['Back'] = "bar"; f.tags = ["tag", "tag2"]
+    f['Front'] = "foo"; f['Back'] = "bar<br>"; f.tags = ["tag", "tag2"]
     deck.addNote(f)
     # with a different deck
     f = deck.newNote()
@@ -129,8 +129,11 @@ def test_export_textnote():
     os.close(fd)
     os.unlink(f)
     e.exportInto(f)
-    e.includeTags = True
+    assert open(f).readline() == "foo\tbar<br>\ttag tag2\n"
+    e.includeTags = False
+    e.includeHTML = False
     e.exportInto(f)
+    assert open(f).readline() == "foo\tbar\n"
 
 def test_exporters():
     assert "*.apkg" in str(exporters())