Metadata download: Allow specifying rules to transform publisher names in addition to author and tag names. Fixes #2012304 [[Enhancement] in "Metadata Download," add "rules to filter/transform publisher"](https://bugs.launchpad.net/calibre/+bug/2012304)

src/calibre/ebooks/metadata/sources/identify.py

@@ -501,7 +501,8 @@ def identify(log, abort,  # {{{
     log('We have %d merged results, merging took: %.2f seconds' %
             (len(results), time.time() - start_time))
     tm_rules = msprefs['tag_map_rules']
-    if tm_rules:
+    pm_rules = msprefs['publisher_map_rules']
+    if tm_rules or pm_rules:
         from calibre.ebooks.metadata.tag_mapper import map_tags
     am_rules = msprefs['author_map_rules']
     if am_rules:
@@ -531,6 +532,9 @@ def identify(log, abort,  # {{{
         r.tags = r.tags[:max_tags]
         if getattr(r.pubdate, 'year', 2000) <= UNDEFINED_DATE.year:
             r.pubdate = None
+        if pm_rules and r.publisher:
+            pubs = map_tags([r.publisher], pm_rules)
+            r.publisher = pubs[0] if pubs else ''
 
     if msprefs['swap_author_names']:
         for r in results:

src/calibre/ebooks/metadata/sources/prefs.py

@@ -18,8 +18,9 @@ msprefs.defaults['swap_author_names'] = False
 msprefs.defaults['fewer_tags'] = True
 msprefs.defaults['find_first_edition_date'] = False
 msprefs.defaults['append_comments'] = False
-msprefs.defaults['tag_map_rules'] = []
-msprefs.defaults['author_map_rules'] = []
+msprefs.defaults['tag_map_rules'] = ()
+msprefs.defaults['author_map_rules'] = ()
+msprefs.defaults['publisher_map_rules'] = ()
 msprefs.defaults['id_link_rules'] = {}
 msprefs.defaults['keep_dups'] = False
 

src/calibre/gui2/author_mapper.py

@@ -34,7 +34,7 @@ class RuleEdit(RuleEditBase):
         ('not_matches', _('does not match regex pattern')),
     ))
 
-    MSG = _('Create the rule below, the rule can be used to add or ignore files')
+    MSG = _('Create the rule below, the rule can be used to add or ignore authors')
     SUBJECT = _('the author, if the author name')
     VALUE_ERROR = _('You must provide a value for the author name to match')
     REPLACE_TEXT = _('with the name:')

src/calibre/gui2/preferences/metadata_sources.py

@@ -332,10 +332,11 @@ class ConfigWidget(ConfigWidgetBase, Ui_Form):
         self.select_default_button.clicked.connect(self.fields_model.select_user_defaults)
         self.select_default_button.clicked.connect(self.changed_signal)
         self.set_as_default_button.clicked.connect(self.fields_model.commit_user_defaults)
-        self.tag_map_rules = self.author_map_rules = None
+        self.tag_map_rules = self.author_map_rules = self.publisher_map_rules = None
         m = QMenu(self)
         m.addAction(_('Tags')).triggered.connect(self.change_tag_map_rules)
         m.addAction(_('Authors')).triggered.connect(self.change_author_map_rules)
+        m.addAction(_('Publisher')).triggered.connect(self.change_publisher_map_rules)
         self.map_rules_button.setMenu(m)
         l = self.page.layout()
         l.setStretch(0, 1)
@@ -376,16 +377,25 @@ class ConfigWidget(ConfigWidgetBase, Ui_Form):
         from calibre.gui2.tag_mapper import RulesDialog
         d = RulesDialog(self)
         if msprefs.get('tag_map_rules'):
-            d.rules = msprefs['tag_map_rules']
+            d.rules = list(msprefs['tag_map_rules'])
         if d.exec() == QDialog.DialogCode.Accepted:
             self.tag_map_rules = d.rules
             self.changed_signal.emit()
 
+    def change_publisher_map_rules(self):
+        from calibre.gui2.publisher_mapper import RulesDialog
+        d = RulesDialog(self)
+        if msprefs.get('publisher_map_rules'):
+            d.rules = list(msprefs['publisher_map_rules'])
+        if d.exec() == QDialog.DialogCode.Accepted:
+            self.publisher_map_rules = d.rules
+            self.changed_signal.emit()
+
     def change_author_map_rules(self):
         from calibre.gui2.author_mapper import RulesDialog
         d = RulesDialog(self)
         if msprefs.get('author_map_rules'):
-            d.rules = msprefs['author_map_rules']
+            d.rules = list(msprefs['author_map_rules'])
         if d.exec() == QDialog.DialogCode.Accepted:
             self.author_map_rules = d.rules
             self.changed_signal.emit()
@@ -395,7 +405,7 @@ class ConfigWidget(ConfigWidgetBase, Ui_Form):
         self.sources_model.initialize()
         self.sources_view.resizeColumnsToContents()
         self.fields_model.initialize()
-        self.tag_map_rules = self.author_map_rules = None
+        self.tag_map_rules = self.author_map_rules = self.publisher_map_rules = None
 
     def restore_defaults(self):
         ConfigWidgetBase.restore_defaults(self)
@@ -410,6 +420,8 @@ class ConfigWidget(ConfigWidgetBase, Ui_Form):
             msprefs['tag_map_rules'] = self.tag_map_rules or []
         if self.author_map_rules is not None:
             msprefs['author_map_rules'] = self.author_map_rules or []
+        if self.publisher_map_rules is not None:
+            msprefs['publisher_map_rules'] = self.publisher_map_rules or []
         return ConfigWidgetBase.commit(self)
 
 

src/calibre/gui2/preferences/metadata_sources.ui

@@ -234,7 +234,7 @@
                 </sizepolicy>
                </property>
                <property name="text">
-                <string>Create &amp;rules to transform tags/authors</string>
+                <string>Create &amp;rules to transform tags/authors/publishers</string>
                </property>
                <property name="popupMode">
                 <enum>QToolButton::InstantPopup</enum>

src/calibre/gui2/publisher_mapper.py

@@ -0,0 +1,137 @@
+#!/usr/bin/env python
+# License: GPL v3 Copyright: 2018, Kovid Goyal <kovid at kovidgoyal.net>
+
+
+from collections import OrderedDict
+
+from calibre.ebooks.metadata.tag_mapper import map_tags
+from calibre.gui2 import Application, elided_text
+from calibre.gui2.tag_mapper import (
+    RuleEdit as RuleEditBase, RuleEditDialog as RuleEditDialogBase,
+    RuleItem as RuleItemBase, Rules as RulesBase, RulesDialog as RulesDialogBase,
+    Tester as TesterBase,
+)
+from calibre.utils.config import JSONConfig
+
+publisher_maps = JSONConfig('publisher-mapping-rules')
+
+
+class RuleEdit(RuleEditBase):
+
+    ACTION_MAP = OrderedDict((
+        ('replace', _('Change')),
+        ('capitalize', _('Capitalize')),
+        ('titlecase', _('Title-case')),
+        ('lower', _('Lower-case')),
+        ('upper', _('Upper-case')),
+    ))
+
+    MATCH_TYPE_MAP = OrderedDict((
+        ('one_of', _('is one of')),
+        ('not_one_of', _('is not one of')),
+        ('has', _('contains')),
+        ('matches', _('matches regex pattern')),
+        ('not_matches', _('does not match regex pattern')),
+    ))
+
+    MSG = _('Create the rule below, the rule can be used to modify publishers')
+    SUBJECT = _('the publisher, if the publisher name')
+    VALUE_ERROR = _('You must provide a value for the publisher name to match')
+    REPLACE_TEXT = _('with the name:')
+    SINGLE_EDIT_FIELD_NAME = 'publisher'
+
+    @property
+    def can_use_tag_editor(self):
+        return False
+
+    def update_state(self):
+        a = self.action.currentData()
+        replace = a == 'replace'
+        self.la3.setVisible(replace), self.replace.setVisible(replace)
+        m = self.match_type.currentData()
+        is_match = 'matches' in m
+        self.regex_help.setVisible(is_match)
+
+    @property
+    def rule(self):
+        return {
+            'action': self.action.currentData(),
+            'match_type': self.match_type.currentData(),
+            'query': self.query.text().strip(),
+            'replace': self.replace.text().strip(),
+        }
+
+    @rule.setter
+    def rule(self, rule):
+        def sc(name):
+            c = getattr(self, name)
+            idx = c.findData(str(rule.get(name, '')))
+            if idx < 0:
+                idx = 0
+            c.setCurrentIndex(idx)
+        sc('match_type'), sc('action')
+        self.query.setText(str(rule.get('query', '')).strip())
+        self.replace.setText(str(rule.get('replace', '')).strip())
+
+
+class RuleEditDialog(RuleEditDialogBase):
+
+    PREFS_NAME = 'edit-publisher-mapping-rule'
+    RuleEditClass = RuleEdit
+
+
+class RuleItem(RuleItemBase):
+
+    @staticmethod
+    def text_from_rule(rule, parent):
+        query = elided_text(rule['query'], font=parent.font(), width=200, pos='right')
+        text = _(
+            '<b>{action}</b> the publisher name, if it <i>{match_type}</i>: <b>{query}</b>').format(
+                action=RuleEdit.ACTION_MAP[rule['action']], match_type=RuleEdit.MATCH_TYPE_MAP[rule['match_type']], query=query)
+        if rule['action'] == 'replace':
+            text += '<br>' + _('to the name') + ' <b>%s</b>' % rule['replace']
+        return '<div style="white-space: nowrap">' + text + '</div>'
+
+
+class Rules(RulesBase):
+
+    RuleItemClass = RuleItem
+    RuleEditDialogClass = RuleEditDialog
+    MSG = _('You can specify rules to manipulate publisher names here.'
+            ' Click the "Add Rule" button'
+            ' below to get started. The rules will be processed in order for every publisher.')
+
+
+class Tester(TesterBase):
+
+    DIALOG_TITLE = _('Test publisher mapping rules')
+    PREFS_NAME = 'test-publisher-mapping-rules'
+    LABEL = _('Enter an publisher name to test:')
+    PLACEHOLDER = _('Enter publisher and click the "Test" button')
+    EMPTY_RESULT = '<p>&nbsp;</p>'
+
+    def do_test(self):
+        publisher = self.value.strip()
+        ans = map_tags([publisher], self.rules)
+        self.result.setText((ans or ('',))[0])
+
+
+class RulesDialog(RulesDialogBase):
+
+    DIALOG_TITLE = _('Edit publisher mapping rules')
+    PREFS_NAME = 'edit-publisher-mapping-rules'
+    RulesClass = Rules
+    TesterClass = Tester
+    PREFS_OBJECT = publisher_maps
+
+
+if __name__ == '__main__':
+    app = Application([])
+    d = RulesDialog()
+    d.rules = [
+            {'action':'replace', 'query':'alice Bob', 'match_type':'one_of', 'replace':'Alice Bob'},
+    ]
+    d.exec()
+    from pprint import pprint
+    pprint(d.rules)
+    del d, app